exiftool

#!/usr/bin/perl -w
#------------------------------------------------------------------------------
# File:         exiftool
#
# Description:  Read/write meta information
#
# Revisions:    Nov. 12/03 - P. Harvey Created
#               (See html/history.html for revision history)
#------------------------------------------------------------------------------
use strict;
use warnings;
require 5.004;

my $version = '12.86';

# add our 'lib' directory to the include list BEFORE 'use Image::ExifTool'
my $exePath;
BEGIN {
    # (undocumented -xpath option added in 11.91, must come before other options)
    $exePath = @ARGV && lc($ARGV[0]) eq '-xpath' && shift() ? $^X : $0;
    # get exe directory
    my $exeDir = ($exePath =~ /(.*)[\\\/]/) ? $1 : '.';
    my $incDir = ($0 =~ /(.*)[\\\/]/) ? "$1/lib" : './lib';
    if (-l $0) {
        my $lnk = eval { readlink $0 };
        if (defined $lnk) {
            my $lnkDir = ($lnk =~ /(.*)[\\\/]/) ? $1 : '.';
            $exeDir = (($lnk =~ m(^/)) ? '' : $exeDir . '/') . $lnkDir;
            $incDir = "$exeDir/lib";
        }
    }
    $Image::ExifTool::exeDir = $exeDir; # use our exeDir for loading config file
    # add lib directory at start of include path
    unshift @INC, $incDir;
    # load or disable config file if specified
    if (@ARGV and lc($ARGV[0]) eq '-config') {
        shift;
        $Image::ExifTool::configFile = shift;
    }
}
use Image::ExifTool qw{:Public};

# function prototypes
sub SigInt();
sub SigCont();
sub Cleanup();
sub GetImageInfo($$);
sub SetImageInfo($$$);
sub DoHardLink($$$$$);
sub CleanXML($);
sub EncodeXML($);
sub FormatXML($$$);
sub EscapeJSON($;$);
sub FormatJSON($$$);
sub PrintCSV();
sub AddGroups($$$$);
sub ConvertBinary($);
sub IsEqual($$);
sub Infile($;$);
sub AddSetTagsFile($;$);
sub Warning($$);
sub DoSetFromFile($$$);
sub CleanFilename($);
sub SetWindowTitle($);
sub ProcessFiles($;$);
sub ScanDir($$;$);
sub FindFileWindows($$);
sub FileNotFound($);
sub PreserveTime();
sub AbsPath($);
sub MyConvertFileName($$);
sub SuggestedExtension($$$);
sub LoadPrintFormat($;$);
sub FilenameSPrintf($;$@);
sub NextUnusedFilename($;$);
sub CreateDirectory($);
sub OpenOutputFile($;@);
sub AcceptFile($);
sub SlurpFile($$);
sub FilterArgfileLine($);
sub ReadStayOpen($);
sub Progress($$);
sub PrintTagList($@);
sub PrintErrors($$$);

$SIG{INT}  = 'SigInt';  # do cleanup on Ctrl-C
$SIG{CONT} = 'SigCont'; # (allows break-out of delays)
END {
    Cleanup();
}

# declare all static file-scope variables
my @commonArgs;     # arguments common to all commands
my @condition;      # conditional processing of files
my @csvFiles;       # list of files when reading with CSV option (in ExifTool Charset)
my @csvTags;        # order of tags for first file with CSV option (lower case)
my @delFiles;       # list of files to delete
my @dynamicFiles;   # list of -tagsFromFile files with dynamic names and -TAG<=FMT pairs
my @efile;          # files for writing list of error/fail/same file names
my @exclude;        # list of excluded tags
my (@echo3, @echo4);# stdout and stderr echo after processing is complete
my @files;          # list of files and directories to scan
my @moreArgs;       # more arguments to process after -stay_open -@
my @newValues;      # list of new tag values to set
my @requestTags;    # tags to request (for -p or -if option arguments)
my @srcFmt;         # source file name format strings
my @tags;           # list of tags to extract
my %altFile;        # alternate files to extract information (keyed by lower-case family 8 group)
my %appended;       # list of files appended to
my %countLink;      # count hard and symbolic links made
my %created;        # list of files we created
my %csvTags;        # lookup for all found tags with CSV option (lower case keys)
my %database;       # lookup for database information based on file name (in ExifTool Charset)
my %filterExt;      # lookup for filtered extensions
my %ignore;         # directory names to ignore
my $ignoreHidden;   # flag to ignore hidden files
my %outComma;       # flag that output text file needs a comma
my %outTrailer;     # trailer for output text file
my %preserveTime;   # preserved timestamps for files
my %printFmt;       # the contents of the print format file
my %seqFileDir;     # file sequence number in each directory
my %setTags;        # hash of list references for tags to set from files
my %setTagsList;    # list of other tag lists for multiple -tagsFromFile from the same file
my %usedFileName;   # lookup for file names we already used in TestName feature
my %utf8FileName;   # lookup for file names that are UTF-8 encoded
my %warnedOnce;     # lookup for once-only warnings
my %wext;           # -W extensions to write
my $allGroup;       # show group name for all tags
my $altEnc;         # alternate character encoding if not UTF-8
my $argFormat;      # use exiftool argument-format output
my $binaryOutput;   # flag for binary output (undef or 1, or 0 for binary XML/PHP)
my $binaryStdout;   # flag set if we output binary to stdout
my $binSep;         # separator used for list items in binary output
my $binTerm;        # terminator used for binary output
my $comma;          # flag set if we need a comma in JSON output
my $count;          # count of files scanned when reading or deleting originals
my $countBad;       # count of files with errors
my $countBadCr;     # count files not created due to errors
my $countBadWr;     # count write errors
my $countCopyWr;    # count of files copied without being changed
my $countDir;       # count of directories scanned
my $countFailed;    # count files that failed condition
my $countGoodCr;    # count files created OK
my $countGoodWr;    # count files written OK
my $countNewDir;    # count of directories created
my $countSameWr;    # count files written OK but not changed
my $critical;       # flag for critical operations (disable CTRL-C)
my $csv;            # flag for CSV option (set to "CSV", or maybe "JSON" when writing)
my $csvAdd;         # flag to add CSV information to existing lists
my $csvDelim;       # delimiter for CSV files
my $csvSaveCount;   # save counter for last CSV file loaded
my $deleteOrig;     # 0=restore original files, 1=delete originals, 2=delete w/o asking
my $disableOutput;  # flag to disable normal output
my $doSetFileName;  # flag set if FileName may be written
my $doUnzip;        # flag to extract info from .gz and .bz2 files
my ($end,$endDir,%endDir);  # flags to end processing
my $escapeC;        # C-style escape
my $escapeHTML;     # flag to escape printed values for html
my $evalWarning;    # warning from eval
my $executeID;      # -execute ID number
my $failCondition;  # flag to fail -if condition
my $fastCondition;  # flag for fast -if condition
my $fileHeader;     # header to print to output file (or console, once)
my $fileTrailer;    # trailer for output file
my $filtered;       # flag indicating file was filtered by name
my $filterFlag;     # file filter flag (0x01=deny extensions, 0x02=allow extensions, 0x04=add ext)
my $fixLen;         # flag to fix description lengths when writing alternate languages
my $forcePrint;     # string to use for missing tag values (undef to not print them)
my $geoOnly;        # flag to extract Geolocation tags only
my $helped;         # flag to avoid printing help if no tags specified
my $html;           # flag for html-formatted output (2=html dump)
my $interrupted;    # flag set if CTRL-C is pressed during a critical process
my $isBinary;       # true if value is a SCALAR ref
my $isWriting;      # flag set if we are writing tags
my $joinLists;      # flag set to join list values into a single string
my $json;           # flag for JSON/PHP output format (1=JSON, 2=PHP)
my $langOpt;        # language option
my $listDir;        # treat a directory as a regular file
my $listItem;       # item number for extracting single item from a list
my $listSep;        # list item separator (', ' by default)
my $mt;             # main ExifTool object
my $multiFile;      # non-zero if we are scanning multiple files
my $noBinary;       # flag set to ignore binary tags
my $outFormat;      # -1=Canon format, 0=same-line, 1=tag names, 2=values only
my $outOpt;         # output file or directory name
my $overwriteOrig;  # flag to overwrite original file (1=overwrite, 2=in place)
my $pause;          # pause before returning
my $preserveTime;   # flag to preserve times of updated files (2=preserve FileCreateDate only)
my $progress;       # flag to calculate total files to process (0=calculate but don't display)
my $progressCount;  # count of files processed
my $progressIncr;   # increment for progress counter
my $progressMax;    # total number of files to process
my $progressNext;   # next progress count to output
my $progStr;        # progress message string
my $quiet;          # flag to disable printing of informational messages / warnings
my $rafStdin;       # File::RandomAccess for stdin (if necessary to rewind)
my $recurse;        # recurse into subdirectories (2=also hidden directories)
my $rtnVal;         # command return value (0=success)
my $rtnValPrev;     # previous command return value (0=success)
my $saveCount;      # count the number of times we will/did call SaveNewValues()
my $scanWritable;   # flag to process only writable file types
my $sectHeader;     # current section header for -p option
my $sectTrailer;    # section trailer for -p option
my $seqFileDir;     # sequential file number used for %-C
my $seqFileNum;     # sequential file number used for %C
my $setCharset;     # character set setting ('default' if not set and -csv -b used)
my $showGroup;      # number of group to show (may be zero or '')
my $showTagID;      # non-zero to show tag ID's
my $stayOpenBuff='';# buffer for -stay_open file
my $stayOpenFile;   # name of the current -stay_open argfile
my $structOpt;      # output structured XMP information (JSON and XML output only)
my $tabFormat;      # non-zero for tab output format
my $tagOut;         # flag for separate text output file for each tag
my $textOut;        # extension for text output file (or undef for no output)
my $textOverwrite;  # flag to overwrite existing text output file (2=append, 3=over+append)
my $tmpFile;        # temporary file to delete on exit
my $tmpText;        # temporary text file
my $validFile;      # flag indicating we processed a valid file
my $verbose;        # verbose setting
my $vout;           # verbose output file reference (\*STDOUT or \*STDERR by default)
my $windowTitle;    # title for console window
my %wroteHEAD;      # list of output txt files to which we wrote HEAD
my $xml;            # flag for XML-formatted output

# flag to keep the input -@ argfile open:
# 0 = normal behaviour
# 1 = received "-stay_open true" and waiting for argfile to keep open
# 2 = currently reading from STAYOPEN argfile
# 3 = waiting for -@ to switch to a new STAYOPEN argfile
my $stayOpen = 0;

my $rtnValApp = 0;  # app return value (0=success)
my $curTitle = '';  # current window title

# lookup for O/S names which may use a backslash as a directory separator
# (ref File::Spec of PathTools-3.2701)
my %hasBackslash = ( MSWin32 => 1, os2 => 1, dos => 1, NetWare => 1, symbian => 1, cygwin => 1 );

# lookup for O/S names which use CR/LF newlines
my $isCRLF = { MSWin32 => 1, os2 => 1, dos => 1 }->{$^O};

# lookup for JSON characters that we escape specially
my %jsonChar = ( '"'=>'"', '\\'=>'\\', "\t"=>'t', "\n"=>'n', "\r"=>'r' );

# lookup for C-style escape sequences
my %escC = ( "\n" => '\n', "\r" => '\r', "\t" => '\t', '\\' => '\\\\');
my %unescC = ( a => "\a", b => "\b", f => "\f", n => "\n", r => "\r",
               t => "\t", 0 => "\0", '\\' => '\\' );

# options requiring additional arguments
# (used only to skip over these arguments when reading -stay_open ARGFILE)
# (arg is converted to lower case then tested again unless an entry was found with the same case)
my %optArgs = (
    '-tagsfromfile' => 1, '-addtagsfromfile' => 1, '-alltagsfromfile' => 1,
    '-@' => 1,
    '-api' => 1,
    '-c' => 1, '-coordformat' => 1,
    '-charset' => 0, # (optional arg; OK because arg cannot begin with "-")
    '-config' => 1,
    '-csvdelim' => 1,
    '-d' => 1, '-dateformat' => 1,
    '-D' => 0, # necessary to avoid matching lower-case equivalent
    '-echo' => 1, '-echo#' => 1,
    '-efile' => 1, '-efile#' => 1, '-efile!' => 1, '-efile#!' => 1,
    '-ext' => 1, '--ext' => 1, '-ext+' => 1, '--ext+' => 1,
        '-extension' => 1, '--extension' => 1, '-extension+' => 1, '--extension+' => 1,
    '-fileorder' => 1, '-fileorder#' => 1,
    '-geotag' => 1,
    '-globaltimeshift' => 1,
    '-i' => 1, '-ignore' => 1,
    '-if' => 1, '-if#' => 1,
    '-lang' => 0, # (optional arg; cannot begin with "-")
    '-listitem' => 1,
    '-o' => 1, '-out' => 1,
    '-p' => 1, '-printformat' => 1, '-p-' => 1, '-printformat-' => 1,
    '-P' => 0,
    '-password' => 1,
    '-require' => 1,
    '-sep' => 1, '-separator' => 1,
    '-srcfile' => 1,
    '-stay_open' => 1,
    '-use' => 1,
    '-userparam' => 1,
    '-w' => 1, '-w!' => 1, '-w+' => 1, '-w+!' => 1, '-w!+' => 1,
        '-textout' => 1, '-textout!' => 1, '-textout+' => 1, '-textout+!' => 1, '-textout!+' => 1,
         '-tagout' => 1,  '-tagout!' => 1,  '-tagout+' => 1,  '-tagout+!' => 1,  '-tagout!+' => 1,
    '-wext' => 1,
    '-wm' => 1, '-writemode' => 1,
    '-x' => 1, '-exclude' => 1,
    '-X' => 0,
);

# recommended packages and alternatives
my @recommends = qw(
    Archive::Zip
    Compress::Zlib
    Digest::MD5
    Digest::SHA
    IO::Compress::Bzip2
    POSIX::strptime
    Time::Local
    Unicode::LineBreak
    Compress::Raw::Lzma
    IO::Compress::RawDeflate
    IO::Uncompress::RawInflate
    IO::Compress::Brotli
    IO::Uncompress::Brotli
    Win32::API
    Win32::FindFile
    Win32API::File
);
my %altRecommends = (
   'POSIX::strptime' => 'Time::Piece', # (can use Time::Piece instead of POSIX::strptime)
);

my %unescapeChar = ( 't'=>"\t", 'n'=>"\n", 'r'=>"\r" );

# special subroutines used in -if condition
sub Image::ExifTool::EndDir() { return $endDir = 1 }
sub Image::ExifTool::End()    { return $end = 1 }

# exit routine
sub Exit {
    if ($pause) {
        if (eval { require Term::ReadKey }) {
            print STDERR "-- press any key --";
            Term::ReadKey::ReadMode('cbreak');
            Term::ReadKey::ReadKey(0);
            Term::ReadKey::ReadMode(0);
            print STDERR "\b \b" x 20;
        } else {
            print STDERR "-- press RETURN --\n";
            <STDIN>;
        }
    }
    exit shift;
}
# my warning and error routines (NEVER say "die"!)
sub Warn  {
    if ($quiet < 2 or $_[0] =~ /^Error/) {
        my $oldWarn = $SIG{'__WARN__'};
        delete $SIG{'__WARN__'};
        warn(@_);
        $SIG{'__WARN__'} = $oldWarn if defined $oldWarn;
    }
}
sub Error { Warn @_; $rtnVal = 1; }
sub WarnOnce($) {
    Warn(@_) and $warnedOnce{$_[0]} = 1 unless $warnedOnce{$_[0]};
}

# define signal handlers and cleanup routine
sub SigInt()  {
    $critical and $interrupted = 1, return;
    Cleanup();
    exit 1;
}
sub SigCont() { }
sub Cleanup() {
    $mt->Unlink($tmpFile) if defined $tmpFile;
    $mt->Unlink($tmpText) if defined $tmpText;
    undef $tmpFile;
    undef $tmpText;
    PreserveTime() if %preserveTime;
    SetWindowTitle('');
}

#------------------------------------------------------------------------------
# main script
#

# isolate arguments common to all commands
if (grep /^-common_args$/i, @ARGV) {
    my (@newArgs, $common, $end);
    foreach (@ARGV) {
        if (/^-common_args$/i and not $end) {
            $common = 1;
        } elsif ($common) {
            push @commonArgs, $_;
        } else {
            $end = 1 if $_ eq '--';
            push @newArgs, $_;
        }
    }
    @ARGV = @newArgs if $common;
}

#..............................................................................
# loop over sets of command-line arguments separated by "-execute"
Command: for (;;) {

if (@echo3) {
    my $str = join("\n", @echo3) . "\n";
    $str =~ s/\$\{status\}/$rtnVal/ig;
    print STDOUT $str;
}
if (@echo4) {
    my $str = join("\n", @echo4) . "\n";
    $str =~ s/\$\{status\}/$rtnVal/ig;
    print STDERR $str;
}

$rafStdin->Close() if $rafStdin;
undef $rafStdin;

# save our previous return codes
$rtnValPrev = $rtnVal;
$rtnValApp = $rtnVal if $rtnVal;

# exit Command loop now if we are all done processing commands
last unless @ARGV or not defined $rtnVal or $stayOpen >= 2 or @commonArgs;

# attempt to restore text mode for STDOUT if necessary
if ($binaryStdout) {
    binmode(STDOUT,':crlf') if $] >= 5.006 and $isCRLF;
    $binaryStdout = 0;
}

# flush console and print "{ready}" message if -stay_open is in effect
if ($stayOpen >= 2) {
    if ($quiet and not defined $executeID) {
        # flush output if possible
        eval { require IO::Handle } and STDERR->flush(), STDOUT->flush();
    } else {
        eval { require IO::Handle } and STDERR->flush();
        my $id = defined $executeID ? $executeID : '';
        my $save = $|;
        $| = 1;     # turn on output autoflush for stdout
        print "{ready$id}\n";
        $| = $save; # restore original autoflush setting
    }
}

# initialize necessary static file-scope variables
# (not done: @commonArgs, @moreArgs, $critical, $binaryStdout, $helped,
#  $interrupted, $mt, $pause, $rtnValApp, $rtnValPrev, $stayOpen, $stayOpenBuff, $stayOpenFile)
undef @condition;
undef @csvFiles;
undef @csvTags;
undef @delFiles;
undef @dynamicFiles;
undef @echo3;
undef @echo4;
undef @efile;
undef @exclude;
undef @files;
undef @newValues;
undef @srcFmt;
undef @tags;
undef %appended;
undef %countLink;
undef %created;
undef %csvTags;
undef %database;
undef %endDir;
undef %filterExt;
undef %ignore;
undef %outComma;
undef %outTrailer;
undef %printFmt;
undef %preserveTime;
undef %seqFileDir;
undef %setTags;
undef %setTagsList;
undef %usedFileName;
undef %utf8FileName;
undef %warnedOnce;
undef %wext;
undef $allGroup;
undef $altEnc;
undef $argFormat;
undef $binaryOutput;
undef $binSep;
undef $binTerm;
undef $comma;
undef $csv;
undef $csvAdd;
undef $deleteOrig;
undef $disableOutput;
undef $doSetFileName;
undef $doUnzip;
undef $end;
undef $endDir;
undef $escapeHTML;
undef $escapeC;
undef $evalWarning;
undef $executeID;
undef $failCondition;
undef $fastCondition;
undef $fileHeader;
undef $filtered;
undef $fixLen;
undef $forcePrint;
undef $geoOnly;
undef $ignoreHidden;
undef $joinLists;
undef $langOpt;
undef $listItem;
undef $multiFile;
undef $noBinary;
undef $outOpt;
undef $preserveTime;
undef $progress;
undef $progressCount;
undef $progressIncr;
undef $progressMax;
undef $progressNext;
undef $recurse;
undef $scanWritable;
undef $sectHeader;
undef $setCharset;
undef $showGroup;
undef $showTagID;
undef $structOpt;
undef $tagOut;
undef $textOut;
undef $textOverwrite;
undef $tmpFile;
undef $tmpText;
undef $validFile;
undef $verbose;
undef $windowTitle;

$count = 0;
$countBad = 0;
$countBadCr = 0;
$countBadWr = 0;
$countCopyWr = 0;
$countDir = 0;
$countFailed = 0;
$countGoodCr = 0;
$countGoodWr = 0;
$countNewDir = 0;
$countSameWr = 0;
$csvDelim = ',';
$csvSaveCount = 0;
$fileTrailer = '';
$filterFlag = 0;
$html = 0;
$isWriting = 0;
$json = 0;
$listSep = ', ';
$outFormat = 0;
$overwriteOrig = 0;
$progStr = '';
$quiet = 0;
$rtnVal = 0;
$saveCount = 0;
$sectTrailer = '';
$seqFileDir = 0;
$seqFileNum = 0;
$tabFormat = 0;
$vout = \*STDOUT;
$xml = 0;

# define local variables used only in this command loop
my @fileOrder;      # tags to use for ordering of input files
my $fileOrderFast;  # -fast level for -fileOrder option
my $addGeotime;     # automatically added geotime argument
my $doGlob;         # flag set to do filename wildcard expansion
my $endOfOpts;      # flag set if "--" option encountered
my $escapeXML;      # flag to escape printed values for xml
my $setTagsFile;    # filename for last TagsFromFile option
my $sortOpt;        # sort option is used
my $srcStdin;       # one of the source files is STDIN
my $useMWG;         # flag set if we are using any MWG tag

my ($argsLeft, @nextPass, $badCmd);
my $pass = 0;

# for Windows, use globbing for wildcard expansion if available - MK/20061010
if ($^O eq 'MSWin32' and eval { require File::Glob }) {
    # override the core glob forcing case insensitivity
    import File::Glob qw(:globally :nocase);
    $doGlob = 1;
}

$mt = Image::ExifTool->new;      # create ExifTool object

# don't extract duplicates by default unless set by UserDefined::Options
$mt->Options(Duplicates => 0) unless %Image::ExifTool::UserDefined::Options
    and defined $Image::ExifTool::UserDefined::Options{Duplicates};

# default is to join lists if the List option was set to zero in the config file
$joinLists = 1 if defined $mt->Options('List') and not $mt->Options('List');

# preserve FileCreateDate if possible
if (not $preserveTime and $^O eq 'MSWin32') {
    $preserveTime = 2 if  eval { require Win32::API } and eval { require Win32API::File };
}

# add user-defined command-line arguments
if (@Image::ExifTool::UserDefined::Arguments) {
    unshift @ARGV, @Image::ExifTool::UserDefined::Arguments;
}

# parse command-line options in 2 passes...
# pass 1: set all of our ExifTool options
# pass 2: print all of our help and informational output (-list, -ver, etc)
for (;;) {

  # execute the command now if no more arguments or -execute is used
  if (not @ARGV or ($ARGV[0] =~ /^(-|\xe2\x88\x92)execute(\d+)?$/i and not $endOfOpts)) {
    if (@ARGV) {
        $executeID = $2;        # save -execute number for "{ready}" response
        $helped = 1;            # don't show help if we used -execute
        $badCmd and shift, $rtnVal=1, next Command;
    } elsif ($stayOpen >= 2) {
        ReadStayOpen(\@ARGV);   # read more arguments from -stay_open file
        next;
    } elsif ($badCmd) {
        undef @commonArgs;      # all done.  Flush common arguments
        $rtnVal = 1;
        next Command;
    }
    if ($pass == 0) {
        # insert common arguments now if not done already
        if (@commonArgs and not defined $argsLeft) {
            # count the number of arguments remaining for subsequent commands
            $argsLeft = scalar(@ARGV) + scalar(@moreArgs);
            unshift @ARGV, @commonArgs;
            # all done with commonArgs if this is the end of the command
            undef @commonArgs unless $argsLeft;
            next;
        }
        # check if we have more arguments now than we did before we processed
        # the common arguments.  If so, then we have an infinite processing loop
        if (defined $argsLeft and $argsLeft < scalar(@ARGV) + scalar(@moreArgs)) {
            Warn "Ignoring -common_args from $ARGV[0] onwards to avoid infinite recursion\n";
            while ($argsLeft < scalar(@ARGV) + scalar(@moreArgs)) {
                @ARGV and shift(@ARGV), next;
                shift @moreArgs;
            }
        }
        # require MWG module if used in any argument
        # (note: doesn't cover the -p option because these tags will be parsed on the 2nd pass)
        $useMWG = 1 if not $useMWG and grep /^mwg:/i, @tags, @requestTags;
        if ($useMWG) {
            require Image::ExifTool::MWG;
            Image::ExifTool::MWG::Load();
        }
        # update necessary variables for 2nd pass
        if (defined $forcePrint) {
            unless (defined $mt->Options('MissingTagValue')) {
                $mt->Options(MissingTagValue => '-');
            }
            $forcePrint = $mt->Options('MissingTagValue');
        }
    }
    if (@nextPass) {
        # process arguments which were deferred to the next pass
        unshift @ARGV, @nextPass;
        undef @nextPass;
        undef $endOfOpts;
        ++$pass;
        next;
    }
    @ARGV and shift;    # remove -execute from argument list
    last;               # process the command now
  }
  $_ = shift;
  next if $badCmd;      # flush remaining arguments if aborting this command

  # allow funny dashes (nroff dash bug for cut-n-paste from pod)
  if (not $endOfOpts and s/^(-|\xe2\x88\x92)//) {
    s/^\xe2\x88\x92/-/;         # translate double-dash too
    if ($_ eq '-') {
        $pass or push @nextPass, '--';
        $endOfOpts = 1;
        next;
    }
    my $a = lc $_;
    if (/^list([wfrdx]|wf|g(\d*)|geo)?$/i) {
        $pass or push @nextPass, "-$_";
        my $type = lc($1 || '');
        if (not $type or $type eq 'w' or $type eq 'x') {
            my $group;
            if ($ARGV[0] and $ARGV[0] =~ /^(-|\xe2\x88\x92)(.+):(all|\*)$/i) {
                if ($pass == 0) {
                    $useMWG = 1 if lc($2) eq 'mwg';
                    push @nextPass, shift;
                    next;
                }
                $group = $2;
                shift;
                $group =~ /IFD/i and Warn("Can't list tags for specific IFD\n"), $helped=1, next;
                $group =~ /^(all|\*)$/ and undef $group;
            } else {
                $pass or next;
            }
            $helped = 1;
            if ($type eq 'x') {
                require Image::ExifTool::TagInfoXML;
                my %opts;
                $opts{Flags} = 1 if defined $forcePrint;
                $opts{NoDesc} = 1 if $outFormat > 0;
                $opts{Lang} = $langOpt;
                Image::ExifTool::TagInfoXML::Write(undef, $group, %opts);
                next;
            }
            my $wr = ($type eq 'w');
            my $msg = ($wr ? 'Writable' : 'Available') . ($group ? " $group" : '') . ' tags';
            PrintTagList($msg, $wr ? GetWritableTags($group) : GetAllTags($group));
            # also print shortcuts if listing all tags
            next if $group or $wr;
            my @tagList = GetShortcuts();
            PrintTagList('Command-line shortcuts', @tagList) if @tagList;
            next;
        }
        $pass or next;
        $helped = 1;
        if ($type eq 'wf') {
            my @wf;
            CanWrite($_) and push @wf, $_ foreach GetFileType();
            PrintTagList('Writable file extensions', @wf);
        } elsif ($type eq 'f') {
            PrintTagList('Supported file extensions', GetFileType());
        } elsif ($type eq 'r') {
            PrintTagList('Recognized file extensions', GetFileType(undef, 0));
        } elsif ($type eq 'd') {
            PrintTagList('Deletable groups', GetDeleteGroups());
        } elsif ($type eq 'geo') {
            require Image::ExifTool::Geolocation;
            my ($i, $entry);
            print "Geolocation database:\n" unless $quiet;
            my $isAlt = $mt->Options('GeolocAltNames') ? ',AltNames' : '';
            $isAlt = '' if $isAlt and not Image::ExifTool::Geolocation::ReadAltNames();
            print "City,Region,Subregion,CountryCode,Country,TimeZone,FeatureCode,Population,Latitude,Longitude$isAlt\n";
            Image::ExifTool::Geolocation::SortDatabase('City') if $sortOpt;
            my $minPop = $mt->Options('GeolocMinPop');
            my $feature = $mt->Options('GeolocFeature') || '';
            my $neg = $feature =~ s/^-//;
            my %fcodes = map { lc($_) => 1 } split /\s*,\s*/, $feature;
            my @isUTF8 = (0,1,2,4);   # items that need converting from UTF8
            push @isUTF8, 10 if $isAlt;
            for ($i=0; ; ++$i) {
                my @entry = Image::ExifTool::Geolocation::GetEntry($i,$langOpt,1) or last;
                $#entry = 9;    # remove everything after latitude (eg. feature type)
                next if $minPop and $entry[7] < $minPop;
                next if %fcodes and $neg ? $fcodes{lc $entry[6]} : not $fcodes{lc $entry[6]};
                push @entry, Image::ExifTool::Geolocation::GetAltNames($i,1) if $isAlt;
                $_ = defined $_ ? $mt->Decode($_, 'UTF8') : '' foreach @entry[@isUTF8];
                pop @entry if $isAlt and not $entry[10];
                print join(',', @entry), "\n";
            }
        } else { # 'g(\d*)'
            # list all groups in specified family
            my $family = $2 || 0;
            PrintTagList("Groups in family $family", $mt->GetAllGroups($family));
        }
        next;
    }
    if ($a eq 'ver') {
        $pass or push(@nextPass,'-ver'), next;
        my $libVer = $Image::ExifTool::VERSION;
        my $str = $libVer eq $version ? '' : " [Warning: Library version is $libVer]";
        if ($verbose) {
            print "ExifTool version $version$str$Image::ExifTool::RELEASE\n";
            printf "Perl version %s%s\n", $], (defined ${^UNICODE} ? " (-C${^UNICODE})" : '');
            print "Platform: $^O\n";
            if ($verbose > 8) {
                print "Current Dir: " . Cwd::getcwd() . "\n" if (eval { require Cwd });
                print "Script Name: $0\n";
                print "Exe Name:    $^X\n";
                print "Exe Dir:     $Image::ExifTool::exeDir\n";
                print "Exe Path:    $exePath\n";
            }
            print "Optional libraries:\n";
            foreach (@recommends) {
                next if /^Win32/ and $^O ne 'MSWin32';
                my $ver = eval "require $_ and \$${_}::VERSION";
                my $alt = $altRecommends{$_};
                # check for alternative if primary not available
                $ver = eval "require $alt and \$${alt}::VERSION" and $_ = $alt if not $ver and $alt;
                printf "  %-28s %s\n", $_, $ver || '(not installed)';
            }
            if ($verbose > 1) {
                print "Include directories:\n";
                ref $_ or print "  $_\n" foreach @INC;
            }
        } else {
            print "$version$str$Image::ExifTool::RELEASE\n";
        }
        $helped = 1;
        next;
    }
    if (/^(all|add)?tagsfromfile(=.*)?$/i) {
        $setTagsFile = $2 ? substr($2,1) : (@ARGV ? shift : '');
        if ($setTagsFile eq '') {
            Error("File must be specified for -tagsFromFile option\n");
            $badCmd = 1;
            next;
        }
        # create necessary lists, etc for this new -tagsFromFile file
        AddSetTagsFile($setTagsFile, { Replace => ($1 and lc($1) eq 'add') ? 0 : 1 } );
        next;
    }
    if ($a eq '@') {
        my $argFile = shift or Error("Expecting filename for -\@ option\n"), $badCmd=1, next;
        # switch to new ARGFILE if using chained -stay_open options
        if ($stayOpen == 1) {
            # defer remaining arguments until we close this argfile
            @moreArgs = @ARGV;
            undef @ARGV;
        } elsif ($stayOpen == 3) {
            if ($stayOpenFile and $stayOpenFile ne '-' and $argFile eq $stayOpenFile) {
                # don't allow user to switch to the same -stay_open argfile
                # because it will result in endless recursion
                $stayOpen = 2;
                Warn "Ignoring request to switch to the same -stay_open ARGFILE ($argFile)\n";
                next;
            }
            close STAYOPEN;
            $stayOpen = 1;  # switch to this -stay_open file
        }
        my $fp = ($stayOpen == 1 ? \*STAYOPEN : \*ARGFILE);
        unless ($mt->Open($fp, $argFile)) {
            unless ($argFile !~ /^\// and $mt->Open($fp, "$Image::ExifTool::exeDir/$argFile")) {
                Error "Error opening arg file $argFile\n";
                $badCmd = 1;
                next
            }
        }
        if ($stayOpen == 1) {
            $stayOpenFile = $argFile;   # remember the name of the file we have open
            $stayOpenBuff = '';         # initialize buffer for reading this file
            $stayOpen = 2;
            $helped = 1;
            ReadStayOpen(\@ARGV);
            next;
        }
        my (@newArgs, $didBOM);
        foreach (<ARGFILE>) {
            # filter Byte Order Mark if it exists from start of UTF-8 text file
            unless ($didBOM) {
                s/^\xef\xbb\xbf//;
                $didBOM = 1;
            }
            $_ = FilterArgfileLine($_);
            push @newArgs, $_ if defined $_;
        }
        close ARGFILE;
        unshift @ARGV, @newArgs;
        next;
    }
    /^(-?)(a|duplicates)$/i and $mt->Options(Duplicates => ($1 ? 0 : 1)), next;
    if ($a eq 'api') {
        my $opt = shift;
        if (defined $opt and length $opt) {
            my $val = ($opt =~ s/=(.*)//s) ? $1 : 1;
            # empty string means an undefined value unless ^= is used
            $val = undef unless $opt =~ s/\^$// or length $val;
            $mt->Options($opt => $val);
        } else {
            print "Available API Options:\n";
            my $availableOptions = Image::ExifTool::AvailableOptions();
            printf("  %-17s - %s\n", $$_[0], $$_[2]) foreach @$availableOptions;
            $helped = 1;
        }
        next;
    }
    /^arg(s|format)$/i and $argFormat = 1, next;
    if (/^(-?)b(inary)?$/i) {
        ($binaryOutput, $noBinary) = $1 ? (undef, 1) : (1, undef);
        $mt->Options(Binary => $binaryOutput, NoPDFList => $binaryOutput);
        next;
    }
    if (/^c(oordFormat)?$/i) {
        my $fmt = shift;
        $fmt or Error("Expecting coordinate format for -c option\n"), $badCmd=1, next;
        $mt->Options('CoordFormat', $fmt);
        next;
    }
    if ($a eq 'charset') {
        my $charset = (@ARGV and $ARGV[0] !~ /^(-|\xe2\x88\x92)/) ? shift : undef;
        if (not $charset) {
            $pass or push(@nextPass, '-charset'), next;
            my %charsets;
            $charsets{$_} = 1 foreach values %Image::ExifTool::charsetName;
            PrintTagList('Available character sets', sort keys %charsets);
            $helped = 1;
        } elsif ($charset !~ s/^(\w+)=// or lc($1) eq 'exiftool') {
            {
                local $SIG{'__WARN__'} = sub { $evalWarning = $_[0] };
                undef $evalWarning;
                $mt->Options(Charset => $charset);
            }
            if ($evalWarning) {
                Warn $evalWarning;
            } else {
                $setCharset = $mt->Options('Charset');
            }
        } else {
            # set internal encoding of specified metadata type
            my $type = { id3 => 'ID3', iptc => 'IPTC', exif => 'EXIF', filename => 'FileName',
                         photoshop => 'Photoshop', quicktime => 'QuickTime', riff=>'RIFF' }->{lc $1};
            $type or Warn("Unknown type for -charset option: $1\n"), next;
            $mt->Options("Charset$type" => $charset);
        }
        next;
    }
    /^config$/i and Warn("Ignored -config option (not first on command line)\n"), shift, next;
    if (/^csv(\+?=.*)?$/i) {
        my $csvFile = $1;
        # must process on 2nd pass so -f and -charset options are available
        unless ($pass) {
            push @nextPass, "-$_";
            if ($csvFile) {
                push @newValues, { SaveCount => ++$saveCount }; # marker to save new values now
                $csvSaveCount = $saveCount;
            }
            next;
        }
        if ($csvFile) {
            $csvFile =~ s/^(\+?=)//;
            $csvAdd = 2 if $1 eq '+=';
            $vout = \*STDERR if $srcStdin;
            $verbose and print $vout "Reading CSV file $csvFile\n";
            my $msg;
            if ($mt->Open(\*CSVFILE, $csvFile)) {
                binmode CSVFILE;
                require Image::ExifTool::Import;
                $msg = Image::ExifTool::Import::ReadCSV(\*CSVFILE, \%database, $forcePrint, $csvDelim);
                close(CSVFILE);
            } else {
                $msg = "Error opening CSV file '${csvFile}'";
            }
            $msg and Warn("$msg\n");
            $isWriting = 1;
        }
        $csv = 'CSV';
        next;
    }
    if (/^csvdelim$/i) {
        $csvDelim = shift;
        defined $csvDelim or Error("Expecting argument for -csvDelim option\n"), $badCmd=1, next;
        $csvDelim =~ /"/ and Error("CSV delimiter can not contain a double quote\n"), $badCmd=1, next;
        my %unescape = ( 't'=>"\t", 'n'=>"\n", 'r'=>"\r", '\\' => '\\' );
        $csvDelim =~ s/\\(.)/$unescape{$1}||"\\$1"/sge;
        $mt->Options(CSVDelim => $csvDelim);
        next;
    }
    if (/^d$/ or $a eq 'dateformat') {
        my $fmt = shift;
        $fmt or Error("Expecting date format for -d option\n"), $badCmd=1, next;
        $mt->Options('DateFormat', $fmt);
        next;
    }
    (/^D$/ or $a eq 'decimal') and $showTagID = 'D', next;
    /^delete_original(!?)$/i and $deleteOrig = ($1 ? 2 : 1), next;
    /^list_dir$/i and $listDir = 1, next;
    (/^e$/ or $a eq '-composite') and $mt->Options(Composite => 0), next;
    (/^-e$/ or $a eq 'composite') and $mt->Options(Composite => 1), next;
    (/^E$/ or $a eq 'escapehtml') and require Image::ExifTool::HTML and $escapeHTML = 1, next;
    ($a eq 'ec' or $a eq 'escapec') and $escapeC = 1, next;
    ($a eq 'ex' or $a eq 'escapexml') and $escapeXML = 1, next;
    if (/^echo(\d)?$/i) {
        my $n = $1 || 1;
        my $arg = shift;
        next unless defined $arg;
        $n > 4 and Warn("Invalid -echo number\n"), next;
        if ($n > 2) {
            $n == 3 ? push(@echo3, $arg) : push(@echo4, $arg);
        } else {
            print {$n==2 ? \*STDERR : \*STDOUT} $arg, "\n";
        }
        $helped = 1;
        next;
    }
    if (/^(ee|extractembedded)(\d*)$/i) {
        $mt->Options(ExtractEmbedded => $2 || 1);
        $mt->Options(Duplicates => 1);
        next;
    }
    if (/^efile(\d+)?(!)?$/i) {
        my $arg = shift;
        defined $arg or Error("Expecting file name for -$_ option\n"), $badCmd=1, next;
        $efile[0] = $arg if not $1 or $1 & 0x01;# error
        $efile[1] = $arg if $1 and $1 & 0x02;   # unchanged
        $efile[2] = $arg if $1 and $1 & 0x04;   # failed -if condition
        $efile[3] = $arg if $1 and $1 & 0x08;   # updated
        $efile[4] = $arg if $1 and $1 & 0x016;  # created
        unlink $arg if $2;
        next;
    }
    # (-execute handled at top of loop)
    if (/^-?ext(ension)?(\+)?$/i) {
        my $ext = shift;
        defined $ext or Error("Expecting extension for -ext option\n"), $badCmd=1, next;
        my $flag = /^-/ ? 0 : ($2 ? 2 : 1);
        $filterFlag |= (0x01 << $flag);
        $ext =~ s/^\.//;    # remove leading '.' if it exists
        $filterExt{uc($ext)} = $flag ? 1 : 0;
        next;
    }
    if (/^f$/ or $a eq 'forceprint') {
        $forcePrint = 1;
        next;
    }
    if (/^F([-+]?\d*)$/ or /^fixbase([-+]?\d*)$/i) {
        $mt->Options(FixBase => $1);
        next;
    }
    if (/^fast(\d*)$/i) {
        $mt->Options(FastScan => (length $1 ? $1 : 1));
        next;
    }
    if (/^(file\d+)$/i) {
        $altFile{lc $1} = shift or Error("Expecting file name for -file option\n"), $badCmd=1, next;
        next;
    }
    if (/^fileorder(\d*)$/i) {
        push @fileOrder, shift if @ARGV;
        my $num = $1 || 0;
        $fileOrderFast = $num if not defined $fileOrderFast or $fileOrderFast > $num;
        next;
    }
    $a eq 'globaltimeshift' and $mt->Options(GlobalTimeShift => shift), next;
    if (/^(g)(roupHeadings|roupNames)?([\d:]*)$/i) {
        $showGroup = $3 || 0;
        $allGroup = ($2 ? lc($2) eq 'roupnames' : $1 eq 'G');
        $mt->Options(SavePath => 1) if $showGroup =~ /\b5\b/;
        $mt->Options(SaveFormat => 1) if $showGroup =~ /\b6\b/;
        next;
    }
    if ($a eq 'geotag') {
        my $trkfile = shift;
        unless ($pass) {
            # defer to next pass so the filename charset is available
            push @nextPass, '-geotag', $trkfile;
            next;
        }
        $trkfile or Error("Expecting file name for -geotag option\n"), $badCmd=1, next;
        # allow wildcards in filename
        if ($trkfile =~ /[*?]/) {
            # CORE::glob() splits on white space, so use File::Glob if possible
            my @trks;
            if ($^O eq 'MSWin32' and eval { require Win32::FindFile }) {
                # ("-charset filename=UTF8" must be set for this to work with Unicode file names)
                @trks = FindFileWindows($mt, $trkfile);
            } elsif (eval { require File::Glob }) {
                @trks = File::Glob::bsd_glob($trkfile);
            } else {
                @trks = glob($trkfile);
            }
            @trks or Error("No matching file found for -geotag option\n"), $badCmd=1, next;
            push @newValues, 'geotag='.shift(@trks) while @trks > 1;
            $trkfile = pop(@trks);
        }
        $_ = "geotag=$trkfile";
        # (fall through!)
    }
    if (/^h$/ or $a eq 'htmlformat') {
        require Image::ExifTool::HTML;
        $html = $escapeHTML = 1;
        $json = $xml = 0;
        next;
    }
    (/^H$/ or $a eq 'hex') and $showTagID = 'H', next;
    if (/^htmldump([-+]?\d+)?$/i) {
        $verbose = ($verbose || 0) + 1;
        $html = 2;
        $mt->Options(HtmlDumpBase => $1) if defined $1;
        next;
    }
    if (/^i(gnore)?$/i) {
        my $dir = shift;
        defined $dir or Error("Expecting directory name for -i option\n"), $badCmd=1, next;
        $ignore{$dir} = 1;
        $dir eq 'HIDDEN' and $ignoreHidden = 1;
        next;
    }
    if (/^if(\d*)$/i) {
        my $cond = shift;
        my $fast = length($1) ? $1 : undef;
        defined $cond or Error("Expecting expression for -if option\n"), $badCmd=1, next;
        # use lowest -fast setting if multiple conditions
        if (not @condition or not defined $fast or (defined $fastCondition and $fastCondition > $fast)) {
            $fastCondition = $fast;
        }
        # prevent processing file unnecessarily for simple case of failed '$ok' or 'not $ok'
        $cond =~ /^\s*(not\s*)\$ok\s*$/i and ($1 xor $rtnValPrev) and $failCondition=1;
        # add to list of requested tags
        push @requestTags, $cond =~ /\$\{?((?:[-\w]+:)*[-\w?*]+)/g;
        push @condition, $cond;
        next;
    }
    if (/^j(son)?(\+?=.*)?$/i) {
        if ($2) {
            # must process on 2nd pass because we need -f and -charset options
            unless ($pass) {
                push @nextPass, "-$_";
                push @newValues, { SaveCount => ++$saveCount }; # marker to save new values now
                $csvSaveCount = $saveCount;
                next;
            }
            my $jsonFile = $2;
            $jsonFile =~ s/^(\+?=)//;
            $csvAdd = 2 if $1 eq '+=';
            $vout = \*STDERR if $srcStdin;
            $verbose and print $vout "Reading JSON file $jsonFile\n";
            my $chset = $mt->Options('Charset');
            my $msg;
            if ($mt->Open(\*JSONFILE, $jsonFile)) {
                binmode JSONFILE;
                require Image::ExifTool::Import;
                $msg = Image::ExifTool::Import::ReadJSON(\*JSONFILE, \%database, $forcePrint, $chset);
                close(JSONFILE);
            } else {
                $msg = "Error opening JSON file '${jsonFile}'";
            }
            $msg and Warn("$msg\n");
            $isWriting = 1;
            $csv = 'JSON';
        } else {
            $json = 1;
            $html = $xml = 0;
            $mt->Options(Duplicates => 1);
            require Image::ExifTool::XMP;   # for FixUTF8()
        }
        next;
    }
    /^(k|pause)$/i and $pause = 1, next;
    (/^l$/ or $a eq 'long') and --$outFormat, next;
    (/^L$/ or $a eq 'latin') and $mt->Options(Charset => 'Latin'), next;
    if ($a eq 'lang') {
        $langOpt = (@ARGV and $ARGV[0] !~ /^(-|\xe2\x88\x92)/) ? shift : undef;
        if ($langOpt) {
            # make lower case and use underline as a separator (eg. 'en_ca')
            $langOpt =~ tr/-A-Z/_a-z/;
            $mt->Options(Lang => $langOpt);
            next if $langOpt eq $mt->Options('Lang');
        } else {
            $pass or push(@nextPass, '-lang'), next;
        }
        my $langs = $quiet ? '' : "Available languages:\n";
        $langs .= "  $_ - $Image::ExifTool::langName{$_}\n" foreach @Image::ExifTool::langs;
        $langs =~ tr/_/-/;  # display dashes instead of underlines in language codes
        $langs = Image::ExifTool::HTML::EscapeHTML($langs) if $escapeHTML;
        $langs = $mt->Decode($langs, 'UTF8');
        $langOpt and Error("Invalid or unsupported language '${langOpt}'.\n$langs"), $badCmd=1, next;
        print $langs;
        $helped = 1;
        next;
    }
    if ($a eq 'listitem') {
        my $li = shift;
        defined $li and Image::ExifTool::IsInt($li) or Warn("Expecting integer for -listItem option\n"), next;
        $mt->Options(ListItem => $li);
        $listItem = $li;
        next;
    }
    /^(m|ignoreminorerrors)$/i and $mt->Options(IgnoreMinorErrors => 1), next;
    /^(n|-printconv)$/i and $mt->Options(PrintConv => 0), next;
    /^(-n|printconv)$/i and $mt->Options(PrintConv => 1), next;
    $a eq 'nop' and $helped=1, next; # (undocumented) no operation, added in 11.25
    if (/^o(ut)?$/i) {
        $outOpt = shift;
        defined $outOpt or Error("Expected output file or directory name for -o option\n"), $badCmd=1, next;
        CleanFilename($outOpt);
        # verbose messages go to STDERR of output is to console
        $vout = \*STDERR if $vout =~ /^-(\.\w+)?$/;
        next;
    }
    /^overwrite_original$/i and $overwriteOrig = 1, next;
    /^overwrite_original_in_place$/i and $overwriteOrig = 2, next;
    if (/^p(-?)$/ or /^printformat(-?)$/i) {
        my $fmt = shift;
        if ($pass) {
            LoadPrintFormat($fmt, $1 || $binaryOutput);
            # load MWG module now if necessary
            if (not $useMWG and grep /^mwg:/i, @requestTags) {
                $useMWG = 1;
                require Image::ExifTool::MWG;
                Image::ExifTool::MWG::Load();
            }
        } else {
            # defer to next pass so the filename charset is available
            push @nextPass, "-$_", $fmt;
        }
        next;
    }
    (/^P$/ or $a eq 'preserve') and $preserveTime = 1, next;
    /^password$/i and $mt->Options(Password => shift), next;
    if (/^progress(\d*)(:.*)?$/i) {
        $progressIncr = $1 || 1;
        $progressNext = 0; # start showing progress at the first file
        if ($2) {
            $windowTitle = substr $2, 1;
            $windowTitle = 'ExifTool %p%%' unless length $windowTitle;
            $windowTitle =~ /%\d*[bpr]/ and $progress = 0 unless defined $progress;
        } else {
            $progress = 1;
            $verbose = 0 unless defined $verbose;
        }
        $progressCount = 0;
        next;
    }
    /^q(uiet)?$/i and ++$quiet, next;
    /^r(ecurse)?(\.?)$/i and $recurse = ($2 ? 2 : 1), next;
    if ($a eq 'require') { # (undocumented) added in version 8.65
        my $ver = shift;
        unless (defined $ver and Image::ExifTool::IsFloat($ver)) {
            Error("Expecting version number for -require option\n");
            $badCmd = 1;
            next;
        }
        unless ($Image::ExifTool::VERSION >= $ver) {
            Error("Requires ExifTool version $ver or later\n");
            $badCmd = 1;
        }
        next;
    }
    /^restore_original$/i and $deleteOrig = 0, next;
    (/^S$/ or $a eq 'veryshort') and $outFormat+=2, next;
    /^s(hort)?(\d*)$/i and $outFormat = $2 eq '' ? $outFormat + 1 : $2, next;
    /^scanforxmp$/i and $mt->Options(ScanForXMP => 1), next;
    if (/^sep(arator)?$/i) {
        my $sep = $listSep = shift;
        defined $listSep or Error("Expecting list item separator for -sep option\n"), $badCmd=1, next;
        $sep =~ s/\\(.)/$unescapeChar{$1}||$1/sge;   # translate escape sequences
        (defined $binSep ? $binTerm : $binSep) = $sep;
        $mt->Options(ListSep => $listSep);
        $joinLists = 1;
        # also split when writing values
        my $listSplit = quotemeta $listSep;
        # a space in the string matches zero or more whitespace characters
        $listSplit =~ s/(\\ )+/\\s\*/g;
        # but a single space alone matches one or more whitespace characters
        $listSplit = '\\s+' if $listSplit eq '\\s*';
        $mt->Options(ListSplit => $listSplit);
        next;
    }
    /^(-)?sort$/i and $sortOpt = $1 ? 0 : 1, next;
    if ($a eq 'srcfile') {
        @ARGV or Warn("Expecting FMT for -srcfile option\n"), next;
        push @srcFmt, shift;
        next;
    }
    if ($a eq 'stay_open') {
        my $arg = shift;
        defined $arg or Warn("Expecting argument for -stay_open option\n"), next;
        if ($arg =~ /^(1|true)$/i) {
            if (not $stayOpen) {
                $stayOpen = 1;
            } elsif ($stayOpen == 2) {
                $stayOpen = 3;  # chained -stay_open options
            } else {
                Warn "-stay_open already active\n";
            }
        } elsif ($arg =~ /^(0|false)$/i) {
            if ($stayOpen >= 2) {
                # close -stay_open argfile and process arguments up to this point
                close STAYOPEN;
                push @ARGV, @moreArgs;
                undef @moreArgs;
            } elsif (not $stayOpen) {
                Warn("-stay_open wasn't active\n");
            }
            $stayOpen = 0;
        } else {
            Warn "Invalid argument for -stay_open\n";
        }
        next;
    }
    if (/^(-)?struct$/i) {
        $mt->Options(Struct => $1 ? 0 : 1);
        next;
    }
    /^t(ab)?$/  and $tabFormat = 1, next;
    if (/^T$/ or $a eq 'table') {
        $tabFormat = $forcePrint = 1; $outFormat+=2; ++$quiet;
        next;
    }
    if (/^(u)(nknown(2)?)?$/i) {
        my $inc = ($3 or (not $2 and $1 eq 'U')) ? 2 : 1;
        $mt->Options(Unknown => $mt->Options('Unknown') + $inc);
        next;
    }
    if ($a eq 'use') {
        my $module = shift;
        $module or Error("Expecting module name for -use option\n"), $badCmd=1, next;
        lc $module eq 'mwg' and $useMWG = 1, next;
        $module =~ /[^\w:]/ and Error("Invalid module name: $module\n"), $badCmd=1, next;
        local $SIG{'__WARN__'} = sub { $evalWarning = $_[0] };
        unless (eval "require Image::ExifTool::$module" or
                eval "require $module" or
                eval "require '${module}'")
        {
            Error("Error using module $module\n");
            $badCmd = 1;
        }
        next;
    }
    if ($a eq 'userparam') {
        my $opt = shift;
        defined $opt or Error("Expected parameter for -userParam option\n"), $badCmd=1, next;
        $opt =~ /=/ or $opt .= '=1';
        $mt->Options(UserParam => $opt);
        next;
    }
    if (/^v(erbose)?(\d*)$/i) {
        $verbose = ($2 eq '') ? ($verbose || 0) + 1 : $2;
        next;
    }
    if (/^(w|textout|tagout)([!+]*)$/i) {
        # (note: all logic ignores $textOut of 0 or '')
        $textOut = shift || Warn("Expecting argument for -$_ option\n");
        my ($t1, $t2) = ($1, $2);
        $textOverwrite = 0;
        $textOverwrite += 1 if $t2 =~ /!/;  # overwrite
        $textOverwrite += 2 if $t2 =~ /\+/; # append
        if ($t1 ne 'W' and lc($t1) ne 'tagout') {
            undef $tagOut;
        } elsif ($textOverwrite >= 2 and $textOut !~ /%[-+]?\d*[.:]?\d*[lu]?[tgso]/) {
            $tagOut = 0; # append tags to one file
        } else {
            $tagOut = 1; # separate file for each tag
        }
        next;
    }
    if (/^(-?)(wext|tagoutext)$/i) {
        my $ext = shift;
        defined $ext or Error("Expecting extension for -wext option\n"), $badCmd=1, next;
        my $flag = 1;
        $1 and $wext{'*'} = 1, $flag = -1;
        $ext =~ s/^\.//;
        $wext{lc $ext} = $flag;
        next;
    }
    if ($a eq 'wm' or $a eq 'writemode') {
        my $wm = shift;
        defined $wm or Error("Expecting argument for -$_ option\n"), $badCmd=1, next;
        $wm =~ /^[wcg]*$/i or Error("Invalid argument for -$_ option\n"), $badCmd=1, next;
        $mt->Options(WriteMode => $wm);
        next;
    }
    if (/^x$/ or $a eq 'exclude') {
        my $tag = shift;
        defined $tag or Error("Expecting tag name for -x option\n"), $badCmd=1, next;
        $tag =~ s/\ball\b/\*/ig;    # replace 'all' with '*' in tag names
        if ($setTagsFile) {
            push @{$setTags{$setTagsFile}}, "-$tag";
        } else {
            push @exclude, $tag;
        }
        next;
    }
    (/^X$/ or $a eq 'xmlformat') and $xml = 1, $html = $json = 0, $mt->Options(Duplicates => 1), next;
    if (/^php$/i) {
        $json = 2;
        $html = $xml = 0;
        $mt->Options(Duplicates => 1);
        next;
    }
    if (/^z(ip)?$/i) {
        $doUnzip = 1;
        $mt->Options(Compress => 1, XMPShorthand => 1);
        $mt->Options(Compact => 1) unless $mt->Options('Compact');
        next;
    }
    $_ eq '' and push(@files, '-'), $srcStdin = 1, next;   # read STDIN
    length $_ eq 1 and $_ ne '*' and Error("Unknown option -$_\n"), $badCmd=1, next;
    if (/^[^<]+(<?)=(.*)/s) {
        my $val = $2;
        if ($1 and length($val) and ($val eq '@' or not defined FilenameSPrintf($val))) {
            # save count of new values before a dynamic value
            push @newValues, { SaveCount => ++$saveCount };
        }
        push @newValues, $_;
        if (/^mwg:/i) {
            $useMWG = 1;
        } elsif (/^([-\w]+:)*(filename|directory|testname)\b/i) {
            $doSetFileName = 1;
        } elsif (/^([-\w]+:)*(geotag|geotime|geosync|geolocate)\b/i) {
            if (lc $2 eq 'geotime') {
                $addGeotime = '';
            } else {
                # add geotag/geosync/geolocate commands first
                unshift @newValues, pop @newValues;
                if (lc $2 eq 'geotag' and (not defined $addGeotime or $addGeotime) and length $val) {
                    $addGeotime = ($1 || '') . 'Geotime<DateTimeOriginal#';
                }
            }
        }
    } else {
        # assume '-tagsFromFile @' if tags are being redirected
        # and -tagsFromFile hasn't already been specified
        AddSetTagsFile($setTagsFile = '@') if not $setTagsFile and /(<|>)/;
        if ($setTagsFile) {
            push @{$setTags{$setTagsFile}}, $_;
            if ($1 eq '>') {
                $useMWG = 1 if /^(.*>\s*)?mwg:/si;
                if (/\b(filename|directory|testname)#?$/i) {
                    $doSetFileName = 1;
                } elsif (/\bgeotime#?$/i) {
                    $addGeotime = '';
                }
            } else {
                $useMWG = 1 if /^([^<]+<\s*(.*\$\{?)?)?mwg:/si;
                if (/^([-\w]+:)*(filename|directory|testname)\b/i) {
                    $doSetFileName = 1;
                } elsif (/^([-\w]+:)*geotime\b/i) {
                    $addGeotime = '';
                }
            }
        } else {
            my $lst = s/^-// ? \@exclude : \@tags;
            unless (/^([-\w*]+:)*([-\w*?]+)#?$/) {
                Warn(qq(Invalid TAG name: "$_"\n));
            }
            push @$lst, $_; # (push everything for backward compatibility)
        }
    }
  } else {
    unless ($pass) {
        # defer to next pass so the filename charset is available
        push @nextPass, $_;
        next;
    }
    if ($doGlob and /[*?]/) {
        if ($^O eq 'MSWin32' and eval { require Win32::FindFile }) {
            push @files, FindFileWindows($mt, $_);
        } else {
            # glob each filespec if necessary - MK/20061010
            push @files, File::Glob::bsd_glob($_);
        }
        $doGlob = 2;
    } else {
        push @files, $_;
        $srcStdin = 1 if $_ eq '-';
    }
  }
}

# set "OK" UserParam based on result of last command
$mt->Options(UserParam => 'OK=' . (not $rtnValPrev));

# set verbose output to STDERR if output could be to console
$vout = \*STDERR if $srcStdin and ($isWriting or @newValues);
$mt->Options(TextOut => $vout) if $vout eq \*STDERR;

# change default EXIF string encoding if MWG used
if ($useMWG and not defined $mt->Options('CharsetEXIF')) {
    $mt->Options(CharsetEXIF => 'UTF8');
}

# allow geolocation without input file if set to a position
if (not @files and not $outOpt and not @newValues) {
    my $loc = $mt->Options('Geolocation');
    # use undocumented feature to input JSON file directly from command line
    $loc and $loc ne '1' and push(@files, qq(\@JSON:{})), $geoOnly = 1;
}

# print help
unless ((@tags and not $outOpt) or @files or @newValues or $geoOnly) {
    if ($doGlob and $doGlob == 2) {
        Warn "No matching files\n";
        $rtnVal = 1;
        next;
    }
    if ($outOpt) {
        Warn "Nothing to write\n";
        $rtnVal = 1;
        next;
    }
    unless ($helped) {
        # catch warnings if we have problems running perldoc
        local $SIG{'__WARN__'} = sub { $evalWarning = $_[0] };
        my $dummy = \*SAVEERR;  # avoid "used only once" warning
        unless ($^O eq 'os2') {
            open SAVEERR, ">&STDERR";
            open STDERR, '>/dev/null';
        }
        if (system('perldoc',$0)) {
            print "Syntax:  exiftool [OPTIONS] FILE\n\n";
            print "Consult the exiftool documentation for a full list of options.\n";
        }
        unless ($^O eq 'os2') {
            close STDERR;
            open STDERR, '>&SAVEERR';
        }
    }
    next;
}

# do sanity check on -delete_original and -restore_original
if (defined $deleteOrig and (@newValues or @tags)) {
    if (not @newValues) {
        my $verb = $deleteOrig ? 'deleting' : 'restoring from';
        Warn "Can't specify tags when $verb originals\n";
    } elsif ($deleteOrig) {
        Warn "Can't use -delete_original when writing.\n";
        Warn "Maybe you meant -overwrite_original ?\n";
    } else {
        Warn "It makes no sense to use -restore_original when writing\n";
    }
    $rtnVal = 1;
    next;
}

if ($overwriteOrig > 1 and $outOpt) {
    Warn "Can't overwrite in place when -o option is used\n";
    $rtnVal = 1;
    next;
}

if ($tagOut and ($csv or %printFmt or $tabFormat or $xml or ($verbose and $html))) {
    Warn "Sorry, -W may not be combined with -csv, -htmlDump, -j, -p, -t or -X\n";
    $rtnVal = 1;
    next;
}

if ($csv and $csv eq 'CSV' and not $isWriting) {
    if ($textOut) {
        Warn "Sorry, -w may not be combined with -csv\n";
        $rtnVal = 1;
        next;
    }
    if ($binaryOutput) {
        $binaryOutput = 0;
        $setCharset = 'default' unless defined $setCharset;
    }
    if (%printFmt) {
        Warn "The -csv option has no effect when -p is used\n";
        undef $csv;
    }
    require Image::ExifTool::XMP if $setCharset;
}

if ($escapeHTML or $json) {
    # must be UTF8 for HTML conversion and JSON output
    $mt->Options(Charset => 'UTF8') if $json;
    # use Escape option to do our HTML escaping unless XML output
    $mt->Options(Escape => 'HTML') if $escapeHTML and not $xml;
} elsif ($escapeXML and not $xml) {
    $mt->Options(Escape => 'XML');
}

# set sort option
if ($sortOpt) {
    # (note that -csv sorts alphabetically by default anyway if more than 1 file)
    my $sort = ($outFormat > 0 or $xml or $json or $csv) ? 'Tag' : 'Descr';
    $mt->Options(Sort => $sort, Sort2 => $sort);
}

# set $structOpt in case set by API option
if ($mt->Options('Struct') and not $structOpt) {
    $structOpt = $mt->Options('Struct');
    require 'Image/ExifTool/XMPStruct.pl';
}

# set up for RDF/XML, JSON and PHP output formats
if ($xml) {
    require Image::ExifTool::XMP;   # for EscapeXML()
    my $charset = $mt->Options('Charset');
    # standard XML encoding names for supported Charset settings
    # (ref http://www.iana.org/assignments/character-sets)
    my %encoding = (
        UTF8     => 'UTF-8',
        Latin    => 'windows-1252',
        Latin2   => 'windows-1250',
        Cyrillic => 'windows-1251',
        Greek    => 'windows-1253',
        Turkish  => 'windows-1254',
        Hebrew   => 'windows-1255',
        Arabic   => 'windows-1256',
        Baltic   => 'windows-1257',
        Vietnam  => 'windows-1258',
        MacRoman => 'macintosh',
    );
    # switch to UTF-8 if we don't have a standard encoding name
    unless ($encoding{$charset}) {
        $charset = 'UTF8';
        $mt->Options(Charset => $charset);
    }
    # set file header/trailer for XML output
    $fileHeader = "<?xml version='1.0' encoding='$encoding{$charset}'?>\n" .
                  "<rdf:RDF xmlns:rdf='http://www.w3.org/1999/02/22-rdf-syntax-ns#'>\n";
    $fileTrailer = "</rdf:RDF>\n";
    # extract as a list unless short output format
    $joinLists = 1 if $outFormat > 0;
    $mt->Options(List => 1) unless $joinLists;
    $showGroup = $allGroup = 1;         # always show group 1
    # set binaryOutput flag to 0 or undef (0 = output encoded binary in XML)
    $binaryOutput = ($outFormat > 0 ? undef : 0) if $binaryOutput;
    $showTagID = 'D' if $tabFormat and not $showTagID;
} elsif ($json) {
    if ($json == 1) { # JSON
        $fileHeader = '[';
        $fileTrailer = "]\n";
    } else { # PHP
        $fileHeader = 'Array(';
        $fileTrailer = ");\n";
    }
    # allow binary output in a text-mode file when -php/-json and -b used together
    # (this works because PHP strings are simple arrays of bytes, and CR/LF
    #  won't be messed up in the text mode output because they are converted
    #  to escape sequences in the strings)
    if ($binaryOutput) {
        $binaryOutput = 0;
        require Image::ExifTool::XMP if $json == 1;  # (for EncodeBase64)
    }
    $mt->Options(List => 1) unless $joinLists;
    $showTagID = 'D' if $tabFormat and not $showTagID;
} elsif ($structOpt) {
    $mt->Options(List => 1);
} else {
    $joinLists = 1;     # join lists for all other unstructured output formats
}

if ($argFormat) {
    $outFormat = 3;
    $allGroup = 1 if defined $showGroup;
}

# change to forward slashes if necessary in all filenames (like CleanFilename)
if ($hasBackslash{$^O}) {
    tr/\\/\// foreach @files;
}

# can't do anything if no file specified
unless (@files) {
    unless ($outOpt) {
        if ($doGlob and $doGlob == 2) {
            Warn "No matching files\n";
        } else {
            Warn "No file specified\n";
        }
        $rtnVal = 1;
        next;
    }
    push @files, '';    # create file from nothing
}

# set Verbose and HtmlDump options
if ($verbose) {
    $disableOutput = 1 unless @tags or @exclude or $tagOut;
    undef $binaryOutput unless $tagOut;    # disable conflicting option
    if ($html) {
        $html = 2;    # flag for html dump
        $mt->Options(HtmlDump => $verbose);
    } else {
        $mt->Options(Verbose => $verbose) unless $tagOut;
    }
} elsif (defined $verbose) {
    # auto-flush output when -v0 is used
    require FileHandle;
    STDOUT->autoflush(1);
    STDERR->autoflush(1);
}

# validate all tags we're writing
my $needSave = 1;
if (@newValues) {
    # assume -geotime value if -geotag specified without -geotime
    if ($addGeotime) {
        AddSetTagsFile($setTagsFile = '@') unless $setTagsFile and $setTagsFile eq '@';
        push @{$setTags{$setTagsFile}}, $addGeotime;
        $verbose and print $vout qq{Argument "-$addGeotime" is assumed\n};
    }
    my %setTagsIndex;
    # add/delete option lookup
    my %addDelOpt = ( '+' => 'AddValue', '-' => 'DelValue', "\xe2\x88\x92" => 'DelValue' );
    $saveCount = 0;
    foreach (@newValues) {
        if (ref $_ eq 'HASH') {
            # save new values now if we stored a "SaveCount" marker
            if ($$_{SaveCount}) {
                $saveCount = $mt->SaveNewValues();
                $needSave = 0;
                # insert marker to load values from CSV file now if this was the CSV file
                push @dynamicFiles, \$csv if $$_{SaveCount} == $csvSaveCount;
            }
            next;
        }
        /(.*?)=(.*)/s or next;
        my ($tag, $newVal) = ($1, $2);
        $tag =~ s/\ball\b/\*/ig;    # replace 'all' with '*' in tag names
        $newVal eq '' and undef $newVal unless $tag =~ s/\^([-+]*)$/$1/;  # undefined to delete tag
        if ($tag =~ /^(All)?TagsFromFile$/i) {
            defined $newVal or Error("Need file name for -tagsFromFile\n"), next Command;
            ++$isWriting;
            if ($newVal eq '@' or not defined FilenameSPrintf($newVal) or
                # can't set tags yet if we are using tags from other files with the -fileNUM option
                grep /\bfile\d+:/i, @{$setTags{$newVal}})
            {
                push @dynamicFiles, $newVal;
                next;   # set tags from dynamic file later
            }
            unless ($mt->Exists($newVal) or $newVal eq '-') {
                Warn "File '${newVal}' does not exist for -tagsFromFile option\n";
                $rtnVal = 1;
                next Command;
            }
            my $setTags = $setTags{$newVal};
            # do we have multiple -tagsFromFile options with this file?
            if ($setTagsList{$newVal}) {
                # use the tags set in the i-th occurrence
                my $i = $setTagsIndex{$newVal} || 0;
                $setTagsIndex{$newVal} = $i + 1;
                $setTags = $setTagsList{$newVal}[$i] if $setTagsList{$newVal}[$i];
            }
            # set specified tags from this file
            unless (DoSetFromFile($mt, $newVal, $setTags)) {
                $rtnVal = 1;
                next Command;
            }
            $needSave = 1;
            next;
        }
        my %opts = ( Shift => 0 );  # shift values if possible instead of adding/deleting
        # allow writing of 'Unsafe' tags unless specified by wildcard
        $opts{Protected} = 1 unless $tag =~ /[?*]/;

        if ($tag =~ s/<// and defined $newVal) {
            if (defined FilenameSPrintf($newVal)) {
                SlurpFile($newVal, \$newVal) or next;   # read file data into $newVal
            } else {
                $tag =~ s/([-+]|\xe2\x88\x92)$// and $opts{$addDelOpt{$1}} = 1;
                # verify that this tag can be written
                my $result = Image::ExifTool::IsWritable($tag);
                if ($result) {
                    $opts{ProtectSaved} = $saveCount;   # protect new values set after this
                    # add to list of dynamic tag values
                    push @dynamicFiles, [ $tag, $newVal, \%opts ];
                    ++$isWriting;
                } elsif (defined $result) {
                    Warn "Tag '${tag}' is not writable\n";
                } else {
                    Warn "Tag '${tag}' does not exist\n";
                }
                next;
            }
        }
        if ($tag =~ s/([-+]|\xe2\x88\x92)$//) {
            $opts{$addDelOpt{$1}} = 1;  # set AddValue or DelValue option
            # set $newVal to '' if deleting nothing
            $newVal = '' if $1 eq '-' and not defined $newVal;
        }
        if ($escapeC and defined $newVal) {
            $newVal =~ s/\\(x([0-9a-fA-F]{2})|.)/$2 ? chr(hex($2)) : $unescC{$1} || $1/seg;
        }
        my ($rtn, $wrn) = $mt->SetNewValue($tag, $newVal, %opts);
        $needSave = 1;
        ++$isWriting if $rtn;
        $wrn and Warning($mt, $wrn);
    }
    # exclude specified tags
    foreach (@exclude) {
        $mt->SetNewValue($_, undef, Replace => 2);
        $needSave = 1;
    }
    unless ($isWriting or $outOpt or @tags) {
        Warn "Nothing to do.\n";
        $rtnVal = 1;
        next;
    }
} elsif (grep /^(\*:)?\*$/, @exclude) {
    Warn "All tags excluded -- nothing to do.\n";
    $rtnVal = 1;
    next;
}
if ($isWriting and @tags and not $outOpt) {
    my ($tg, $s) = @tags > 1 ? ("$tags[0] ...", 's') : ($tags[0], '');
    Warn "Ignored superfluous tag name$s or invalid option$s: -$tg\n";
}
# save current state of new values if setting values from target file
# or if we may be translating to a different format
$mt->SaveNewValues() if $outOpt or (@dynamicFiles and $needSave);

$multiFile = 1 if @files > 1;
@exclude and $mt->Options(Exclude => \@exclude);

undef $binaryOutput if $html;

if ($binaryOutput) {
    $outFormat = 99;    # shortest possible output format
    $mt->Options(PrintConv => 0);
    unless ($textOut or $binaryStdout) {
        binmode(STDOUT);
        $binaryStdout = 1;
        $mt->Options(TextOut => ($vout = \*STDERR));
    }
    # disable conflicting options
    undef $showGroup;
}

# sort by groups to look nicer depending on options
if (defined $showGroup and not (@tags and $allGroup) and ($sortOpt or not defined $sortOpt)) {
    $mt->Options(Sort => "Group$showGroup");
}

if ($textOut) {
    CleanFilename($textOut);  # make all forward slashes
    # add '.' before output extension if necessary
    $textOut = ".$textOut" unless $textOut =~ /[.%]/ or defined $tagOut;
}

# determine if we should scan for only writable files
if ($outOpt) {
    my $type = GetFileType($outOpt);
    if ($type) {
        # (must test original file name because we can write .webp but not other RIFF types)
        unless (CanWrite($outOpt)) {
            Warn "Can't write $type files\n";
            $rtnVal = 1;
            next;
        }
        $scanWritable = $type unless CanCreate($type);
    } else {
        $scanWritable = 1;
    }
    $isWriting = 1;     # set writing flag
} elsif ($isWriting or defined $deleteOrig) {
    $scanWritable = 1;
}

# initialize alternate encoding flag
$altEnc = $mt->Options('Charset');
undef $altEnc if $altEnc eq 'UTF8';

# set flag to fix description lengths if necessary
if (not $altEnc and $mt->Options('Lang') ne 'en' and eval { require Encode }) {
    # (note that Unicode::GCString is part of the Unicode::LineBreak package)
    $fixLen = eval { require Unicode::GCString } ? 2 : 1;
}

# sort input files if specified
if (@fileOrder) {
    my @allFiles;
    ProcessFiles($mt, \@allFiles);
    my $sortTool = Image::ExifTool->new;
    $sortTool->Options(FastScan => $fileOrderFast) if $fileOrderFast;
    $sortTool->Options(PrintConv => $mt->Options('PrintConv'));
    $sortTool->Options(Duplicates => 0);
    my (%sortBy, %isFloat, @rev, $file);
    # save reverse sort flags
    push @rev, (s/^-// ? 1 : 0) foreach @fileOrder;
    foreach $file (@allFiles) {
        my @tags;
        my $info = $sortTool->ImageInfo(Infile($file,1), @fileOrder, \@tags);
        # get values of all tags (or '~' to sort last if not defined)
        foreach (@tags) {
            $_ = $$info{$_};    # put tag value into @tag list
            defined $_ or $_ = '~', next;
            $isFloat{$_} = Image::ExifTool::IsFloat($_);
            # pad numbers to 12 digits to keep them sequential
            s/(\d+)/(length($1) < 12 ? '0'x(12-length($1)) : '') . $1/eg unless $isFloat{$_};
        }
        $sortBy{$file} = \@tags;    # save tag values for each file
    }
    # sort in specified order
    @files = sort {
        my ($i, $cmp);
        for ($i=0; $i<@rev; ++$i) {
            my $u = $sortBy{$a}[$i];
            my $v = $sortBy{$b}[$i];
            if (not $isFloat{$u} and not $isFloat{$v}) {
                $cmp = $u cmp $v;               # alphabetically
            } elsif ($isFloat{$u} and $isFloat{$v}) {
                $cmp = $u <=> $v;               # numerically
            } else {
                $cmp = $isFloat{$u} ? -1 : 1;   # numbers first
            }
            return $rev[$i] ? -$cmp : $cmp if $cmp;
        }
        return $a cmp $b;   # default to sort by name
    } @allFiles;
} elsif (defined $progress) {
    # expand FILE argument to count the number of files to process
    my @allFiles;
    ProcessFiles($mt, \@allFiles);
    @files = @allFiles;
}
# set file count for progress message
$progressMax = scalar @files if defined $progress;

# store duplicate database information under absolute path
my @dbKeys = keys %database;
if (@dbKeys) {
    if (eval { require Cwd }) {
        undef $evalWarning;
        local $SIG{'__WARN__'} = sub { $evalWarning = $_[0] };
        foreach (@dbKeys) {
            my $db = $database{$_};
            tr/\\/\// and $database{$_} = $db;  # allow for backslashes in SourceFile
            # (punt on using ConvertFileName here, so $absPath may be a mix of encodings)
            my $absPath = AbsPath($_);
            if (defined $absPath) {
                $database{$absPath} = $db unless $database{$absPath};
                if ($verbose and $verbose > 1) {
                    print $vout "Imported entry for '${_}' (full path: '${absPath}')\n";
                }
            } elsif ($verbose and $verbose > 1) {
                print $vout "Imported entry for '${_}' (non-existent file)\n";
            }
        }
    }
}

# process all specified files
ProcessFiles($mt);

if ($filtered and not $validFile) {
    Warn "No file with specified extension\n";
    $rtnVal = 1;
}

# print CSV information if necessary
PrintCSV() if $csv and not $isWriting;

# print folder/file trailer if necessary
if ($textOut) {
    foreach (keys %outTrailer) {
        next unless $outTrailer{$_};
        if ($mt->Open(\*OUTTRAIL, $_, '>>')) {
            my $fp = \*OUTTRAIL;
            print $fp $outTrailer{$_};
            close $fp;
        } else {
            Error("Error appending to $_\n");
        }
    }
} else {
    print $sectTrailer if $sectTrailer;
    print $fileTrailer if $fileTrailer and not $fileHeader;
}

my $totWr = $countGoodWr + $countBadWr + $countSameWr + $countCopyWr +
            $countGoodCr + $countBadCr;

if (defined $deleteOrig) {

    # print summary and delete requested files
    unless ($quiet) {
        printf "%5d directories scanned\n", $countDir if $countDir;
        printf "%5d directories created\n", $countNewDir if $countNewDir;
        printf "%5d files failed condition\n", $countFailed if $countFailed;
        printf "%5d image files found\n", $count;
    }
    if (@delFiles) {
        # verify deletion unless "-delete_original!" was specified
        if ($deleteOrig == 1) {
            printf '%5d originals will be deleted!  Are you sure [y/n]? ', scalar(@delFiles);
            my $response = <STDIN>;
            unless ($response =~ /^(y|yes)\s*$/i) {
                Warn "Originals not deleted.\n";
                next;
            }
        }
        $countGoodWr = $mt->Unlink(@delFiles);
        $countBad = scalar(@delFiles) - $countGoodWr;
    }
    if ($quiet) {
        # no more messages
    } elsif ($count and not $countGoodWr and not $countBad) {
        printf "%5d original files found\n", $countGoodWr; # (this will be 0)
    } elsif ($deleteOrig) {
        printf "%5d original files deleted\n", $countGoodWr if $count;
        printf "%5d originals not deleted due to errors\n", $countBad if $countBad;
    } else {
        printf "%5d image files restored from original\n", $countGoodWr if $count;
        printf "%5d files not restored due to errors\n", $countBad if $countBad;
    }

} elsif ((not $binaryStdout or $verbose) and not $quiet) {

    # print summary
    my $tot = $count + $countBad;
    if ($countDir or $totWr or $countFailed or $tot > 1 or $textOut or %countLink) {
        my $o = (($html or $json or $xml or %printFmt or $csv) and not $textOut) ? \*STDERR : $vout;
        printf($o "%5d directories scanned\n", $countDir) if $countDir;
        printf($o "%5d directories created\n", $countNewDir) if $countNewDir;
        printf($o "%5d files failed condition\n", $countFailed) if $countFailed;
        printf($o "%5d image files created\n", $countGoodCr) if $countGoodCr;
        printf($o "%5d image files updated\n", $countGoodWr) if $totWr - $countGoodCr - $countBadCr - $countCopyWr;
        printf($o "%5d image files unchanged\n", $countSameWr) if $countSameWr;
        printf($o "%5d image files %s\n", $countCopyWr, $overwriteOrig ? 'moved' : 'copied') if $countCopyWr;
        printf($o "%5d files weren't updated due to errors\n", $countBadWr) if $countBadWr;
        printf($o "%5d files weren't created due to errors\n", $countBadCr) if $countBadCr;
        printf($o "%5d image files read\n", $count) if ($tot+$countFailed)>1 or ($countDir and not $totWr);
        printf($o "%5d files could not be read\n", $countBad) if $countBad;
        printf($o "%5d output files created\n", scalar(keys %created)) if $textOut;
        printf($o "%5d output files appended\n", scalar(keys %appended)) if %appended;
        printf($o "%5d hard links created\n", $countLink{Hard} || 0) if $countLink{Hard} or $countLink{BadHard};
        printf($o "%5d hard links could not be created\n", $countLink{BadHard}) if $countLink{BadHard};
        printf($o "%5d symbolic links created\n", $countLink{Sym} || 0) if $countLink{Sym} or $countLink{BadSym};
        printf($o "%5d symbolic links could not be created\n", $countLink{BadSym}) if $countLink{BadSym};
    }
}

# set error status if we had any errors or if all files failed the "-if" condition
if ($countBadWr or $countBadCr or $countBad) {
    $rtnVal = 1;
} elsif ($countFailed and not ($count or $totWr) and not $rtnVal) {
    $rtnVal = 2;
}

# clean up after each command
Cleanup();

} # end "Command" loop ........................................................

close STAYOPEN if $stayOpen >= 2;

Exit $rtnValApp;    # all done


#------------------------------------------------------------------------------
# Get image information from EXIF data in file (or write file if writing)
# Inputs: 0) ExifTool object reference, 1) file name
sub GetImageInfo($$)
{
    my ($et, $orig) = @_;
    my (@foundTags, $info, $file, $ind, $g8);

    # set window title for this file if necessary
    if (defined $windowTitle) {
        if ($progressCount >= $progressNext) {
            my $prog = $progressMax ? "$progressCount/$progressMax" : '0/0';
            my $title = $windowTitle;
            my ($num, $denom) = split '/', $prog;
            my $frac = $num / ($denom || 1);
            my $n = $title =~ s/%(\d+)b/%b/ ? $1 : 20;  # length of bar
            my $bar = int($frac * $n + 0.5);
            my %lkup = (
                b => ('I' x $bar) . ('.' x ($n - $bar)),
                f => $orig,
                p => int(100 * $frac + 0.5),
                r => $prog,
               '%'=> '%',
            );
            $title =~ s/%([%bfpr])/$lkup{$1}/eg;
            SetWindowTitle($title);
            if (defined $progressMax) {
                undef $progressNext;
            } else {
                $progressNext += $progressIncr;
            }
        }
        # ($progressMax is not defined for "-progress:%f")
        ++$progressCount unless defined $progressMax;
    }
    unless (length $orig or $outOpt) {
        Warn qq(Error: Zero-length file name - ""\n);
        ++$countBad;
        return;
    }
    # determine the name of the source file based on the original input file name
    if (@srcFmt) {
        my ($fmt, $first);
        foreach $fmt (@srcFmt) {
            $file = $fmt eq '@' ? $orig : FilenameSPrintf($fmt, $orig);
            # use this file if it exists
            $et->Exists($file) and undef($first), last;
            $verbose and print $vout "Source file $file does not exist\n";
            $first = $file unless defined $first;
        }
        $file = $first if defined $first;
        my ($d, $f) = Image::ExifTool::SplitFileName($orig);
        $et->Options(UserParam => "OriginalDirectory#=$d");
        $et->Options(UserParam => "OriginalFileName#=$f");
    } else {
        $file = $orig;
    }
    # set alternate file names
    foreach $g8 (sort keys %altFile) {
        my $altName = $orig;
        # must double any '$' symbols in the original file name because
        # they are used for tag names in a -fileNUM argument
        $altName =~ s/\$/\$\$/g;
        $altName = FilenameSPrintf($altFile{$g8}, $altName);
        $et->SetAlternateFile($g8, $altName);
    }

    my $pipe = $file;
    if ($doUnzip) {
        # pipe through gzip or bzip2 if necessary
        if ($file =~ /\.(gz|bz2)$/i) {
            my $type = lc $1;
            if ($file =~ /[^-_.'A-Za-z0-9\/\\]/) {
                Warn "Error: Insecure zip file name. Skipped\n";
                EFile($file);
                ++$countBad;
                return;
            }
            if ($type eq 'gz') {
                $pipe = qq{gzip -dc "$file" |};
            } else {
                $pipe = qq{bzip2 -dc "$file" |};
            }
            $$et{TRUST_PIPE} = 1;
        }
    }
    # evaluate -if expression for conditional processing
    if (@condition) {
        unless ($file eq '-' or $et->Exists($file)) {
            Warn "Error: File not found - $file\n";
            EFile($file);
            FileNotFound($file);
            ++$countBad;
            return;
        }
        my $result;

        unless ($failCondition) {
            # catch run time errors as well as compile errors
            undef $evalWarning;
            local $SIG{'__WARN__'} = sub { $evalWarning = $_[0] };

            my (%info, $condition);
            # extract information and build expression for evaluation
            my $opts = { Duplicates => 1, RequestTags => \@requestTags, Verbose => 0, HtmlDump => 0 };
            $$opts{FastScan} = $fastCondition if defined $fastCondition;
            # return all tags but explicitly mention tags on command line so
            # requested images will generate the appropriate warnings
            @foundTags = ('*', @tags) if @tags;
            $info = $et->ImageInfo(Infile($pipe,$isWriting), \@foundTags, $opts);
            foreach $condition (@condition) {
                my $cond = $et->InsertTagValues($condition, \@foundTags, \%info);
                {
                    # set package so eval'd functions are in Image::ExifTool namespace
                    package Image::ExifTool;

                    my $self = $et;
                    #### eval "-if" condition (%info, $self)
                    $result = eval $cond;

                    $@ and $evalWarning = $@;
                }
                if ($evalWarning) {
                    # fail condition if warning is issued
                    undef $result;
                    if ($verbose) {
                        chomp $evalWarning;
                        $evalWarning =~ s/ at \(eval .*//s;
                        Warn "Condition: $evalWarning - $file\n";
                    }
                }
                last unless $result;
            }
            undef @foundTags if $fastCondition; # ignore if we didn't get all tags
        }
        unless ($result) {
            Progress($vout, "-------- $file (failed condition)") if $verbose;
            EFile($file, 2);
            ++$countFailed;
            return;
        }
        # can't make use of $info if verbose because we must reprocess
        # the file anyway to generate the verbose output
        undef $info if $verbose or defined $fastCondition;
    } elsif ($file =~ s/^(\@JSON:)(.*)/$1/) {
        # read JSON file from command line
        my $dat = $2;
        $info = $et->ImageInfo(\$dat, \@foundTags);
        if ($geoOnly) { /^Geolocation/ or delete $$info{$_} foreach keys %$info; $file = ' ' }
    }
    if (defined $deleteOrig) {
        Progress($vout, "======== $file") if defined $verbose;
        ++$count;
        my $original = "${file}_original";
        $et->Exists($original) or return;
        if ($deleteOrig) {
            $verbose and print $vout "Scheduled for deletion: $original\n";
            push @delFiles, $original;
        } elsif ($et->Rename($original, $file)) {
            $verbose and print $vout "Restored from $original\n";
            EFile($file, 3);
            ++$countGoodWr;
        } else {
            Warn "Error renaming $original\n";
            EFile($file);
            ++$countBad;
        }
        return;
    }
    ++$seqFileNum;  # increment our file counter
    my ($dir) = Image::ExifTool::SplitFileName($orig);
    $seqFileDir = $seqFileDir{$dir} = ($seqFileDir{$dir} || 0) + 1;

    my $lineCount = 0;
    my ($fp, $outfile, $append);
    if ($textOut and ($verbose or $et->Options('PrintCSV')) and not $tagOut) {
        ($fp, $outfile, $append) = OpenOutputFile($orig);
        $fp or EFile($file), ++$countBad, return;
        # delete file if we exit prematurely (unless appending)
        $tmpText = $outfile unless $append;
        $et->Options(TextOut => $fp);
    }

    if ($isWriting) {
        Progress($vout, "======== $file") if defined $verbose;
        SetImageInfo($et, $file, $orig);
        $info = $et->GetInfo('Warning', 'Error');
        PrintErrors($et, $info, $file);
        # close output text file if necessary
        if (defined $outfile) {
            undef $tmpText;
            close($fp);
            $et->Options(TextOut => $vout);
            if ($info->{Error}) {
                $et->Unlink($outfile);  # erase bad file
            } elsif ($append) {
                $appended{$outfile} = 1 unless $created{$outfile};
            } else {
                $created{$outfile} = 1;
            }
        }
        return;
    }

    # extract information from this file
    unless ($file eq '-' or $et->Exists($file) or $info) {
        Warn "Error: File not found - $file\n";
        FileNotFound($file);
        defined $outfile and close($fp), undef($tmpText), $et->Unlink($outfile);
        EFile($file);
        ++$countBad;
        return;
    }
    # print file/progress message
    my $o;
    unless ($binaryOutput or $textOut or %printFmt or $html > 1 or $csv) {
        if ($html) {
            require Image::ExifTool::HTML;
            my $f = Image::ExifTool::HTML::EscapeHTML($file);
            print "<!-- $f -->\n";
        } elsif (not ($json or $xml)) {
            $o = \*STDOUT if ($multiFile and not $quiet) or $progress;
        }
    }
    $o = \*STDERR if $progress and not $o;
    Progress($o, "======== $file") if $o;
    if ($info) {
        # get the information we wanted
        if (@tags and not %printFmt) {
            @foundTags = @tags;
            $info = $et->GetInfo(\@foundTags);
        }
    } else {
        # request specified tags unless using print format option
        my $oldDups = $et->Options('Duplicates');
        if (%printFmt) {
            $et->Options(Duplicates => 1);
            $et->Options(RequestTags => \@requestTags);
            if ($printFmt{SetTags}) {
                # initialize options so we can set any tags we want
                $$et{TAGS_FROM_FILE} = 1;
                $et->Options(MakerNotes => 1);
                $et->Options(Struct => 2);
                $et->Options(List => 1);
                $et->Options(CoordFormat => '%d %d %.8f') unless $et->Options('CoordFormat');
            }
        } else {
            @foundTags = @tags;
        }
        # extract the information
        $info = $et->ImageInfo(Infile($pipe), \@foundTags);
        $et->Options(Duplicates => $oldDups);
    }
    # all done now if we already wrote output text file (eg. verbose option)
    if ($fp) {
        if (defined $outfile) {
            $et->Options(TextOut => \*STDOUT);
            undef $tmpText;
            if ($info->{Error}) {
                close($fp);
                $et->Unlink($outfile);  # erase bad file
            } else {
                ++$lineCount;       # output text file (likely) is not empty
            }
        }
        if ($info->{Error}) {
            Warn "Error: $info->{Error} - $file\n";
            EFile($file);
            ++$countBad;
            return;
        }
    }

    # print warnings to stderr if using binary output
    # (because we are likely ignoring them and piping stdout to file)
    # or if there is none of the requested information available
    if ($binaryOutput or not %$info) {
        my $errs = $et->GetInfo('Warning', 'Error');
        PrintErrors($et, $errs, $file) and EFile($file), $rtnVal = 1;
    } elsif ($et->GetValue('Error') or ($$et{Validate} and $et->GetValue('Warning'))) {
        $rtnVal = 1;
    }

    # open output file (or stdout if no output file) if not done already
    unless (defined $outfile or $tagOut) {
        ($fp, $outfile, $append) = OpenOutputFile($orig);
        $fp or EFile($file), ++$countBad, return;
        $tmpText = $outfile unless $append;
    }

    # restore state of comma flag for this file if appending
    $comma = $outComma{$outfile} if $append and ($textOverwrite & 0x02);

    # print the results for this file
    if (%printFmt) {
        # output using print format file (-p) option
        my ($type, $doc, $grp, $lastDoc, $cache);
        $fileTrailer = '';
        # repeat for each sub-document if necessary
        if ($$et{DOC_COUNT}) {
            # (cache tag keys if there are sub-documents)
            $lastDoc = $$et{DOC_COUNT} and $cache = { };
        } else {
            $lastDoc = 0;
        }
        for ($doc=0; $doc<=$lastDoc; ++$doc) {
            my ($skipBody, $opt);
            foreach $type (qw(HEAD SECT IF BODY ENDS TAIL)) {
                my $prf = $printFmt{$type} or next;
                if ($type eq 'HEAD' and defined $outfile) {
                    next if $wroteHEAD{$outfile};
                    $wroteHEAD{$outfile} = 1;
                }
                next if $type eq 'BODY' and $skipBody;
                # silence "IF" warnings and warnings for subdocuments > 1
                if ($type eq 'IF' or ($doc > 1 and not $$et{OPTIONS}{IgnoreMinorErrors})) {
                    $opt = 'Silent';
                } else {
                    $opt = 'Warn';
                }
                if ($lastDoc) {
                    if ($doc) {
                        next if $type eq 'HEAD' or $type eq 'TAIL'; # only repeat SECT/IF/BODY/ENDS
                        $grp = "Doc$doc";
                    } else {
                        $grp = 'Main';
                    }
                }
                my @lines;
                foreach (@$prf) {
                    my $line = $et->InsertTagValues($_, \@foundTags, $opt, $grp, $cache);
                    if ($type eq 'IF') {
                        $skipBody = 1 unless defined $line;
                    } elsif (defined $line) {
                        push @lines, $line;
                    }
                }
                $lineCount += scalar @lines;
                if ($type eq 'SECT') {
                    my $thisHeader = join '', @lines;
                    if ($sectHeader and $sectHeader ne $thisHeader) {
                        print $fp $sectTrailer if $sectTrailer;
                        undef $sectHeader;
                    }
                    $sectTrailer = '';
                    print $fp $sectHeader = $thisHeader unless $sectHeader;
                } elsif ($type eq 'ENDS') {
                    $sectTrailer .= join '', @lines if defined $sectHeader;
                } elsif ($type eq 'TAIL') {
                    $fileTrailer .= join '', @lines;
                } elsif (@lines) {
                    print $fp @lines;
                }
            }
        }
        delete $printFmt{HEAD} unless defined $outfile; # print header only once per output file
        my $errs = $et->GetInfo('Warning', 'Error');
        PrintErrors($et, $errs, $file) and EFile($file);
    } elsif (not $disableOutput) {
        my ($tag, $line, %noDups, %csvInfo, $bra, $ket, $sep);
        if ($fp) {
            # print file header (only once)
            if ($fileHeader) {
                print $fp $fileHeader unless defined $outfile and ($created{$outfile} or $appended{$outfile});
                undef $fileHeader unless $textOut;
            }
            if ($html) {
                print $fp "<table>\n";
            } elsif ($xml) {
                my $f = $file;
                CleanXML(\$f);
                print $fp "\n<rdf:Description rdf:about='${f}'";
                print $fp "\n  xmlns:et='http://ns.exiftool.org/1.0/'";
                print $fp " et:toolkit='Image::ExifTool $Image::ExifTool::VERSION'";
                # define namespaces for all tag groups
                my (%groups, @groups, $grp0, $grp1);
                foreach $tag (@foundTags) {
                    ($grp0, $grp1) = $et->GetGroup($tag);
                    unless ($grp1) {
                        next unless defined $forcePrint;
                        $grp0 = $grp1 = 'Unknown';
                    }
                    # add groups from structure fields
                    AddGroups($$info{$tag}, $grp0, \%groups, \@groups) if ref $$info{$tag};
                    next if $groups{$grp1};
                    # include family 0 and 1 groups in URI except for internal tags
                    # (this will put internal tags in the "XML" group on readback)
                    $groups{$grp1} = $grp0;
                    push @groups, $grp1;
                }
                foreach $grp1 (@groups) {
                    my $grp = $groups{$grp1};
                    unless ($grp eq $grp1 and $grp =~ /^(ExifTool|File|Composite|Unknown)$/) {
                        $grp .= "/$grp1";
                    }
                    print $fp "\n  xmlns:$grp1='http://ns.exiftool.org/$grp/1.0/'";
                }
                print $fp '>' if $outFormat < 1; # finish rdf:Description token unless short format
                $ind = $outFormat >= 0 ? ' ' : '   ';
            } elsif ($json) {
                # set delimiters for JSON or PHP output
                ($bra, $ket, $sep) = $json == 1 ? ('{','}',':') : ('Array(',')',' =>');
                print $fp ",\n" if $comma;
                print $fp qq($bra\n  "SourceFile"$sep ), EscapeJSON(MyConvertFileName($et,$file),1);
                $comma = 1;
                $ind = (defined $showGroup and not $allGroup) ? '    ' : '  ';
            } elsif ($csv) {
                my $file2 = MyConvertFileName($et, $file);
                $database{$file2} = \%csvInfo;
                push @csvFiles, $file2;
            }
        }
        # suppress duplicates manually in JSON and short XML output
        my $noDups = ($json or ($xml and $outFormat > 0));
        my $printConv = $et->Options('PrintConv');
        my $lastGroup = '';
        my $i = -1;
TAG:    foreach $tag (@foundTags) {
            ++$i;   # keep track on index in @foundTags
            my $tagName = GetTagName($tag);
            my ($group, $valList);
            # get the value for this tag
            my $val = $$info{$tag};
            # set flag if this is binary data
            $isBinary = (ref $val eq 'SCALAR' and defined $binaryOutput);
            if (ref $val) {
                # happens with -X, -j or -php when combined with -b:
                if (defined $binaryOutput and not $binaryOutput and $$et{TAG_INFO}{$tag}{Protected}) {
                    # avoid extracting Unsafe binary tags (eg. data blocks) [insider information]
                    my $lcTag = lc $tag;
                    $lcTag =~ s/ .*//;
                    next unless $$et{REQ_TAG_LOOKUP}{$lcTag} or ($$et{OPTIONS}{RequestAll} || 0) > 2;
                }
                $val = ConvertBinary($val); # convert SCALAR references
                next unless defined $val;
                if ($structOpt and ref $val) {
                    # serialize structure if necessary
                    $val = Image::ExifTool::XMP::SerializeStruct($et, $val) unless $xml or $json;
                } elsif (ref $val eq 'ARRAY') {
                    if (defined $listItem) {
                        # take only the specified item
                        $val = $$val[$listItem];
                    # join arrays of simple values (with newlines for binary output)
                    } elsif ($binaryOutput) {
                        if ($tagOut) {
                            $valList = $val;
                            $val = shift @$valList;
                        } else {
                            $val = join defined $binSep ? $binSep : "\n", @$val;
                        }
                    } elsif ($joinLists) {
                        $val = join $listSep, @$val;
                    }
                }
            }
            if (not defined $val) {
                # ignore tags that weren't found unless necessary
                next if $binaryOutput;
                if (defined $forcePrint) {
                    $val = $forcePrint; # forced to print all tag values
                } elsif (not $csv) {
                    next;
                }
            }
            if (defined $showGroup) {
                $group = $et->GetGroup($tag, $showGroup);
                # look ahead to see if this tag may suppress a priority tag in
                # the same group, and if so suppress this tag instead
                # (note that the tag key may look like "TAG #(1)" when the "#" feature is used)
                next if $noDups and $tag =~ /^(.*?) ?\(/ and defined $$info{$1} and
                        $group eq $et->GetGroup($1, $showGroup);
                $group = 'Unknown' if not $group and ($xml or $json or $csv);
                if ($fp and not ($allGroup or $csv)) {
                    if ($lastGroup ne $group) {
                        if ($html) {
                            my $cols = 1;
                            ++$cols if $outFormat==0 or $outFormat==1;
                            ++$cols if $showTagID;
                            print $fp "<tr><td colspan=$cols bgcolor='#dddddd'>$group</td></tr>\n";
                        } elsif ($json) {
                            print $fp "\n  $ket" if $lastGroup;
                            print $fp ',' if $lastGroup or $comma;
                            print $fp qq(\n  "$group"$sep $bra);
                            undef $comma;
                            undef %noDups;  # allow duplicate names in different groups
                        } else {
                            print $fp "---- $group ----\n";
                        }
                        $lastGroup = $group;
                    }
                    undef $group;   # undefine so we don't print it below
                }
            } elsif ($noDups) {
                # don't allow duplicates, but avoid suppressing the priority tag
                next if $tag =~ /^(.*?) ?\(/ and defined $$info{$1};
            }

            ++$lineCount;           # we are printing something meaningful

            # loop through list values when -b -W used
            for (;;) {
                if ($tagOut) {
                    # determine suggested extension for output file
                    my $ext = SuggestedExtension($et, \$val, $tagName);
                    if (%wext and ($wext{$ext} || $wext{'*'} || -1) < 0) {
                        if ($verbose and $verbose > 1) {
                            print $vout "Not writing $ext output file for $tagName\n";
                        }
                        next TAG;
                    }
                    my @groups = $et->GetGroup($tag);
                    defined $outfile and close($fp), undef($tmpText); # (shouldn't happen)
                    my $org = $et->GetValue('OriginalRawFileName') || $et->GetValue('OriginalFileName');
                    ($fp, $outfile, $append) = OpenOutputFile($orig, $tagName, \@groups, $ext, $org);
                    $fp or ++$countBad, next TAG;
                    $tmpText = $outfile unless $append;
                }
                # write binary output
                if ($binaryOutput) {
                    print $fp $val;
                    print $fp $binTerm if defined $binTerm;
                    if ($tagOut) {
                        if ($append) {
                            $appended{$outfile} = 1 unless $created{$outfile};
                        } else {
                            $created{$outfile} = 1;
                        }
                        close($fp);
                        undef $tmpText;
                        $verbose and print $vout "Wrote $tagName to $outfile\n";
                        undef $outfile;
                        undef $fp;
                        next TAG unless $valList and @$valList;
                        $val = shift @$valList;
                        next; # loop over values of List tag
                    }
                    next TAG;
                }
                last;
            }
            # save information for CSV output
            if ($csv) {
                my $tn = $tagName;
                $tn .= '#' if $tag =~ /#/;  # add ValueConv "#" suffix if used
                my $gt = $group ? "$group:$tn" : $tn;
                # (tag-name case may be different if some tags don't exist
                # in a file, so all logic must use lower-case tag names)
                my $lcTag = lc $gt;
                # override existing entry only if top priority
                next if defined $csvInfo{$lcTag} and $tag =~ /\(/;
                $csvInfo{$lcTag} = $val;
                if (defined $csvTags{$lcTag}) {
                    # overwrite with actual extracted tag name
                    # (note: can't check "if defined $val" here because -f may be used)
                    $csvTags{$lcTag} = $gt if defined $$info{$tag};
                    next;
                }
                # must check for "Unknown" group (for tags that don't exist)
                if ($group and defined $csvTags[$i] and $csvTags[$i] =~ /^(.*):$tn$/i) {
                     next if $group eq 'Unknown';   # nothing more to do if we don't know tag group
                     if ($1 eq 'unknown') {
                        # replace unknown entry in CSV tag lookup and list
                        delete $csvTags{$csvTags[$i]};
                        $csvTags{$lcTag} = defined($val) ? $gt : '';
                        $csvTags[$i] = $lcTag;
                        next;
                     }
                }
                # (don't save unextracted tag name unless -f was used)
                $csvTags{$lcTag} = defined($val) ? $gt : '';
                if (@csvFiles == 1) {
                    push @csvTags, $lcTag; # save order of tags for first file
                } elsif (@csvTags) {
                    undef @csvTags;
                }
                next;
            }

            # get description if we need it (use tag name if $outFormat > 0)
            my $desc = $outFormat > 0 ? $tagName : $et->GetDescription($tag);

            if ($xml) {
                # RDF/XML output format
                my $tok = "$group:$tagName";
                if ($outFormat > 0) {
                    if ($structOpt and ref $val) {
                        $val = Image::ExifTool::XMP::SerializeStruct($et, $val);
                    }
                    if ($escapeHTML) {
                        $val =~ tr/\0-\x08\x0b\x0c\x0e-\x1f/./;
                        Image::ExifTool::XMP::FixUTF8(\$val) unless $altEnc;
                        $val = Image::ExifTool::HTML::EscapeHTML($val, $altEnc);
                    } else {
                        CleanXML(\$val);
                    }
                    unless ($noDups{$tok}) {
                        # manually un-do CR/LF conversion in Windows because output
                        # is in text mode, which will re-convert newlines to CR/LF
                        $isCRLF and $val =~ s/\x0d\x0a/\x0a/g;
                        print $fp "\n $tok='${val}'";
                        # XML does not allow duplicate attributes
                        $noDups{$tok} = 1;
                    }
                    next;
                }
                my ($xtra, $valNum, $descClose);
                if ($showTagID) {
                    my ($id, $lang) = $et->GetTagID($tag);
                    if ($id =~ /^\d+$/) {
                        $id = sprintf("0x%.4x", $id) if $showTagID eq 'H';
                    } else {
                        $id = Image::ExifTool::XMP::FullEscapeXML($id);
                    }
                    $xtra = " et:id='${id}'";
                    $xtra .= " xml:lang='${lang}'" if $lang;
                } else {
                    $xtra = '';
                }
                if ($tabFormat) {
                    my $table = $et->GetTableName($tag);
                    my $index = $et->GetTagIndex($tag);
                    $xtra .= " et:table='${table}'";
                    $xtra .= " et:index='${index}'" if defined $index;
                }
                # Note: New $xtra attributes must be added to %ignoreEtProp in XMP.pm!
                my $lastVal = $val;
                for ($valNum=0; $valNum<2; ++$valNum) {
                    $val = FormatXML($val, $ind, $group);
                    # manually un-do CR/LF conversion in Windows because output
                    # is in text mode, which will re-convert newlines to CR/LF
                    $isCRLF and $val =~ s/\x0d\x0a/\x0a/g;
                    if ($outFormat >= 0) {
                        # normal output format (note: this will give
                        # non-standard RDF/XML if there are any attributes)
                        print $fp "\n <$tok$xtra$val</$tok>";
                        last;
                    } elsif ($valNum == 0) {
                        CleanXML(\$desc);
                        if ($xtra) {
                            print $fp "\n <$tok>";
                            print $fp "\n  <rdf:Description$xtra>";
                            $descClose = "\n  </rdf:Description>";
                        } else {
                            print $fp "\n <$tok rdf:parseType='Resource'>";
                            $descClose = '';
                        }
                        # print tag Description
                        print $fp "\n   <et:desc>$desc</et:desc>";
                        if ($printConv) {
                            # print PrintConv value
                            print $fp "\n   <et:prt$val</et:prt>";
                            $val = $et->GetValue($tag, 'ValueConv');
                            $val = '' unless defined $val;
                            # go back to print ValueConv value only if different
                            next unless IsEqual($val, $lastVal);
                            print $fp "$descClose\n </$tok>";
                            last;
                        }
                    }
                    # print ValueConv value
                    print $fp "\n   <et:val$val</et:val>";
                    print $fp "$descClose\n </$tok>";
                    last;
                }
                next;
            } elsif ($json) {
                # JSON or PHP output format
                my $tok = $allGroup ? "$group:$tagName" : $tagName;
                # (removed due to backward incompatibility)
                # $tok .= '#' if $tag =~ /#/; # add back '#' suffix if used
                next if $noDups{$tok};
                $noDups{$tok} = 1;
                print $fp ',' if $comma;
                print $fp qq(\n$ind"$tok"$sep );
                if ($showTagID or $outFormat < 0) {
                    $val = { val => $val };
                    if ($showTagID) {
                        my ($id, $lang) = $et->GetTagID($tag);
                        $id = sprintf('0x%.4x', $id) if $showTagID eq 'H' and $id =~ /^\d+$/;
                        $$val{lang} = $lang if $lang;
                        $$val{id} = $id;
                    }
                    if ($tabFormat) {
                        $$val{table} = $et->GetTableName($tag);
                        my $index = $et->GetTagIndex($tag);
                        $$val{index} = $index if defined $index;
                    }
                    if ($outFormat < 0) {
                        $$val{desc} = $desc;
                        if ($printConv) {
                            my $num = $et->GetValue($tag, 'ValueConv');
                            $$val{num} = $num if defined $num and not IsEqual($num, $$val{val});
                        }
                    }
                }
                FormatJSON($fp, $val, $ind);
                $comma = 1;
                next;
            }
            my $id;
            if ($showTagID) {
                $id = $et->GetTagID($tag);
                if ($id =~ /^(\d+)(\.\d+)?$/) { # only print numeric ID's
                    $id = sprintf("0x%.4x", $1) if $showTagID eq 'H';
                } else {
                    $id = '-';
                }
            }

            if ($escapeC) {
                $val =~ s/([\0-\x1f\\\x7f])/$escC{$1} || sprintf('\x%.2x', ord $1)/eg;
            } else {
                # translate unprintable chars in value and remove trailing spaces
                $val =~ tr/\x01-\x1f\x7f/./;
                $val =~ s/\x00//g;
                $val =~ s/\s+$//;
            }

            if ($html) {
                print $fp "<tr>";
                print $fp "<td>$group</td>" if defined $group;
                print $fp "<td>$id</td>" if $showTagID;
                print $fp "<td>$desc</td>" if $outFormat <= 1;
                print $fp "<td>$val</td></tr>\n";
            } else {
                my $buff = '';
                if ($tabFormat) {
                    $buff = "$group\t" if defined $group;
                    $buff .= "$id\t" if $showTagID;
                    if ($outFormat <= 1) {
                        $buff .= "$desc\t$val\n";
                    } elsif (defined $line) {
                        $line .= "\t$val";
                    } else {
                        $line = $val;
                    }
                } elsif ($outFormat < 0) {    # long format
                    $buff = "[$group] " if defined $group;
                    $buff .= "$id " if $showTagID;
                    $buff .= "$desc\n      $val\n";
                } elsif ($outFormat == 0 or $outFormat == 1) {
                    my $wid;
                    my $len = 0;
                    if (defined $group) {
                        $buff = sprintf("%-15s ", "[$group]");
                        $len = 16;
                    }
                    if ($showTagID) {
                        $wid = ($showTagID eq 'D') ? 5 : 6;
                        $len += $wid + 1;
                        ($wid = $len - length($buff) - 1) < 1 and $wid = 1;
                        $buff .= sprintf "%${wid}s ", $id;
                    }
                    $wid = 32 - (length($buff) - $len);
                    # pad description to a constant length
                    # (get actual character length when using alternate languages
                    # because these descriptions may contain UTF8-encoded characters)
                    my $padLen = $wid;
                    if (not $fixLen) {
                        $padLen -= length $desc;
                    } elsif ($fixLen == 1) {
                        $padLen -= length Encode::decode_utf8($desc);
                    } else {
                        my $gcstr = eval { Unicode::GCString->new(Encode::decode_utf8($desc)) };
                        if ($gcstr) {
                            $padLen -= $gcstr->columns;
                        } else {
                            $padLen -= length Encode::decode_utf8($desc);
                            Warning($et, 'Unicode::GCString problem.  Columns may be misaligned');
                            $fixLen = 1;
                        }
                    }
                    $padLen = 0 if $padLen < 0;
                    $buff .= $desc . (' ' x $padLen) . ": $val\n";
                } elsif ($outFormat == 2) {
                    $buff = "[$group] " if defined $group;
                    $buff .= "$id " if $showTagID;
                    $buff .= "$tagName: $val\n";
                } elsif ($argFormat) {
                    $buff = '-';
                    $buff .= "$group:" if defined $group;
                    $tagName .= '#' if $tag =~ /#/; # add '#' suffix if used
                    $buff .= "$tagName=$val\n";
                } else {
                    $buff = "$group " if defined $group;
                    $buff .= "$id " if $showTagID;
                    $buff .= "$val\n";
                }
                print $fp $buff;
            }
            if ($tagOut) {
                if ($append) {
                    $appended{$outfile} = 1 unless $created{$outfile};
                } else {
                    $created{$outfile} = 1;
                }
                close($fp);
                undef $tmpText;
                $verbose and print $vout "Wrote $tagName to $outfile\n";
                undef $outfile;
                undef $fp;
            }
        }
        if ($fp) {
            if ($html) {
                print $fp "</table>\n";
            } elsif ($xml) {
                # close rdf:Description element
                print $fp $outFormat < 1 ? "\n</rdf:Description>\n" : "/>\n";
            } elsif ($json) {
                print $fp "\n  $ket" if $lastGroup;
                print $fp "\n$ket";
                $comma = 1;
            } elsif ($tabFormat and $outFormat > 1) {
                print $fp "$line\n" if defined $line;
            }
        }
    }
    if (defined $outfile) {
        if ($textOverwrite & 0x02) {
            # save state of this file if we may be appending
            $outComma{$outfile} = $comma;
            $outTrailer{$outfile} = '';
            $outTrailer{$outfile} .= $sectTrailer and $sectTrailer = '' if $sectTrailer;
            $outTrailer{$outfile} .= $fileTrailer if $fileTrailer;
        } else {
            # write section and file trailers before closing the file
            print $fp $sectTrailer and $sectTrailer = '' if $sectTrailer;
            print $fp $fileTrailer if $fileTrailer;
        }
        close($fp);
        undef $tmpText;
        if ($lineCount) {
            if ($append) {
                $appended{$outfile} = 1 unless $created{$outfile};
            } else {
                $created{$outfile} = 1;
            }
        } else {
            $et->Unlink($outfile) unless $append; # don't keep empty output files
        }
        undef $comma;
    }
    ++$count;
}

#------------------------------------------------------------------------------
# Set information in file
# Inputs: 0) ExifTool object reference, 1) source file name
#         2) original source file name ('' to create from scratch)
# Returns: true on success
sub SetImageInfo($$$)
{
    my ($et, $file, $orig) = @_;
    my ($outfile, $restored, $isTemporary, $isStdout, $outType, $tagsFromSrc);
    my ($hardLink, $symLink, $testName, $sameFile);
    my $infile = $file;    # save infile in case we change it again

    # clean up old temporary file if necessary
    if (defined $tmpFile) {
        $et->Unlink($tmpFile);
        undef $tmpFile;
    }
    # clear any existing errors or warnings since we check these on return
    delete $$et{VALUE}{Error};
    delete $$et{VALUE}{Warning};

    # first, try to determine our output file name so we can return quickly
    # if it already exists (note: this test must be delayed until after we
    # set tags from dynamic files if writing FileName or Directory)
    if (defined $outOpt) {
        if ($outOpt =~ /^-(\.\w+)?$/) {
            # allow output file type to be specified with "-o -.EXT"
            $outType = GetFileType($outOpt) if $1;
            $outfile = '-';
            $isStdout = 1;
        } else {
            $outfile = FilenameSPrintf($outOpt, $orig);
            if ($outfile eq '') {
                Warn "Error: Can't create file with zero-length name from $orig\n";
                EFile($infile);
                ++$countBadCr;
                return 0;
            }
        }
        if (not $isStdout and (($et->IsDirectory($outfile) and not $listDir) or $outfile =~ /\/$/)) {
            $outfile .= '/' unless $outfile =~ /\/$/;
            my $name = $file;
            $name =~ s/^.*\///s;    # remove directory name
            $outfile .= $name;
        } else {
            my $srcType = GetFileType($file) || '';
            $outType or $outType = GetFileType($outfile);
            if ($outType and ($srcType ne $outType or $outType eq 'ICC') and $file ne '-') {
                unless (CanCreate($outType)) {
                    my $what = $srcType ? 'other types' : 'scratch';
                    WarnOnce "Error: Can't create $outType files from $what\n";
                    EFile($infile);
                    ++$countBadCr;
                    return 0;
                }
                if ($file ne '') {
                    # restore previous new values unless done already
                    $et->RestoreNewValues() unless $restored;
                    $restored = 1;
                    # translate to this type by setting specified tags from file
                    my @setTags = @tags;
                    foreach (@exclude) {
                        push @setTags, "-$_";
                    }
                    # force some tags to be copied for certain file types
                    my %forceCopy = (
                        ICC => 'ICC_Profile',
                        VRD => 'CanonVRD',
                        DR4 => 'CanonDR4',
                    );
                    push @setTags, $forceCopy{$outType} if $forceCopy{$outType};
                    # assume "-tagsFromFile @" unless -tagsFromFile already specified
                    # (%setTags won't be empty if -tagsFromFile used)
                    if (not %setTags or (@setTags and not $setTags{'@'})) {
                        return 0 unless DoSetFromFile($et, $file, \@setTags);
                    } elsif (@setTags) {
                        # add orphaned tags to existing "-tagsFromFile @" for this file only
                        push @setTags, @{$setTags{'@'}};
                        $tagsFromSrc = \@setTags;
                    }
                    # all done with source file -- create from meta information alone
                    $file = '';
                }
            }
        }
        unless ($isStdout) {
            $outfile = NextUnusedFilename($outfile);
            if ($et->Exists($outfile, 1) and not $doSetFileName) {
                Warn "Error: '${outfile}' already exists - $infile\n";
                EFile($infile);
                ++$countBadWr;
                return 0;
            }
        }
    } elsif ($file eq '-') {
        $isStdout = 1;
    }
    # set tags from destination file if required
    if (@dynamicFiles) {
        # restore previous values if necessary
        $et->RestoreNewValues() unless $restored;
        my ($dyFile, %setTagsIndex);
        foreach $dyFile (@dynamicFiles) {
            if (not ref $dyFile) {
                my ($fromFile, $setTags);
                if ($dyFile eq '@') {
                    $fromFile = $orig;
                    $setTags = $tagsFromSrc || $setTags{$dyFile};
                } else {
                    $fromFile = FilenameSPrintf($dyFile, $orig);
                    defined $fromFile or EFile($infile), ++$countBadWr, return 0;
                    $setTags = $setTags{$dyFile};
                }
                # do we have multiple -tagsFromFile options with this file?
                if ($setTagsList{$dyFile}) {
                    # use the tags set in the i-th occurrence
                    my $i = $setTagsIndex{$dyFile} || 0;
                    $setTagsIndex{$dyFile} = $i + 1;
                    $setTags = $setTagsList{$dyFile}[$i] if $setTagsList{$dyFile}[$i];
                }
                # set new values values from file
                return 0 unless DoSetFromFile($et, $fromFile, $setTags);
            } elsif (ref $dyFile eq 'ARRAY') {
                # a dynamic file containing a simple tag value
                my $fname = FilenameSPrintf($$dyFile[1], $orig);
                my ($buff, $rtn, $wrn);
                my $opts = $$dyFile[2];
                if (defined $fname and SlurpFile($fname, \$buff)) {
                    $verbose and print $vout "Reading $$dyFile[0] from $fname\n";
                    ($rtn, $wrn) = $et->SetNewValue($$dyFile[0], $buff, %$opts);
                    $wrn and Warn "$wrn\n";
                }
                # remove this tag if we couldn't set it properly
                $rtn or $et->SetNewValue($$dyFile[0], undef, Replace => 2,
                                         ProtectSaved => $$opts{ProtectSaved});
                next;
            } elsif (ref $dyFile eq 'SCALAR') {
                # set new values from CSV or JSON database
                my ($f, $found, $tag);
                undef $evalWarning;
                local $SIG{'__WARN__'} = sub { $evalWarning = $_[0] };
                # force UTF-8 if the database was JSON
                my $old = $et->Options('Charset');
                $et->Options(Charset => 'UTF8') if $csv eq 'JSON';
                # read tags for SourceFile '*' plus the specific file
                foreach $f ('*', MyConvertFileName($et, $file)) {
                    my $csvInfo = $database{$f};
                    unless ($csvInfo) {
                        next if $f eq '*';
                        # check absolute path
                        # (punt on using ConvertFileName here, so $absPath may be a mix of encodings)
                        my $absPath = AbsPath($f);
                        next unless defined $absPath and $csvInfo = $database{$absPath};
                    }
                    $found = 1;
                    $verbose and print $vout "Setting new values from $csv database\n";
                    foreach $tag (sort keys %$csvInfo) {
                        next if $tag =~ /\b(SourceFile|Directory|FileName)$/i; # don't write these
                        my ($rtn, $wrn) = $et->SetNewValue($tag, $$csvInfo{$tag},
                                          Protected => 1, AddValue => $csvAdd,
                                          ProtectSaved => $csvSaveCount);
                        $wrn and Warn "$wrn\n" if $verbose;
                    }
                }
                $et->Options(Charset => $old) if $csv eq 'JSON';
                unless ($found) {
                    Warn("No SourceFile '${file}' in imported $csv database\n");
                    my $absPath = AbsPath($file);
                    Warn("(full path: '${absPath}')\n") if defined $absPath and $absPath ne $file;
                    return 0;
                }
            }
        }
    }
    if ($isStdout) {
        # write to STDOUT
        $outfile = \*STDOUT;
        unless ($binaryStdout) {
            binmode(STDOUT);
            $binaryStdout = 1;
        }
    } else {
        # get name of hard link if we are creating one
        $hardLink = $et->GetNewValues('HardLink');
        $symLink = $et->GetNewValues('SymLink');
        $testName = $et->GetNewValues('TestName');
        $hardLink = FilenameSPrintf($hardLink, $orig) if defined $hardLink;
        $symLink = FilenameSPrintf($symLink, $orig) if defined $symLink;
        # determine what our output file name should be
        my $newFileName = $et->GetNewValues('FileName');
        my $newDir = $et->GetNewValues('Directory');
        if (defined $newFileName and not length $newFileName) {
            Warning($et,"New file name is empty - $infile");
            undef $newFileName;
        }
        if (defined $testName) {
            my $err;
            $err = "You shouldn't write FileName or Directory with TestFile" if defined $newFileName or defined $newDir;
            $err = "The -o option shouldn't be used with TestFile" if defined $outfile;
            $err and Warn("Error: $err - $infile\n"), EFile($infile), ++$countBadWr, return 0;
            $testName = FilenameSPrintf($testName, $orig);
            $testName = Image::ExifTool::GetNewFileName($file, $testName) if $file ne '';
        }
        if (defined $newFileName or defined $newDir or ($doSetFileName and defined $outfile)) {
            if ($newFileName) {
                $newFileName = FilenameSPrintf($newFileName, $orig);
                if (defined $outfile) {
                    $outfile = Image::ExifTool::GetNewFileName($file, $outfile) if $file ne '';
                    $outfile = Image::ExifTool::GetNewFileName($outfile, $newFileName);
                } elsif ($file ne '') {
                    $outfile = Image::ExifTool::GetNewFileName($file, $newFileName);
                }
            }
            if ($newDir) {
                $newDir = FilenameSPrintf($newDir, $orig);
                $outfile = Image::ExifTool::GetNewFileName(defined $outfile ? $outfile : $file, $newDir);
            }
            $outfile = NextUnusedFilename($outfile, $infile);
            if ($et->Exists($outfile, 1)) {
                if ($infile eq $outfile) {
                    undef $outfile;     # not changing the file name after all
                # (allow for case-insensitive filesystems)
                } elsif ($et->IsSameFile($infile, $outfile)) {
                    $sameFile = $outfile;   # same file, but the name has a different case
                } else {
                    Warn "Error: '${outfile}' already exists - $infile\n";
                    EFile($infile);
                    ++$countBadWr;
                    return 0;
                }
            }
        }
        if (defined $outfile) {
            defined $verbose and print $vout "'${infile}' --> '${outfile}'\n";
            # create output directory if necessary
            CreateDirectory($outfile);
            # set temporary file (automatically erased on abnormal exit)
            $tmpFile = $outfile if defined $outOpt;
        }
        unless (defined $tmpFile) {
            # count the number of tags and pseudo-tags we are writing
            my ($numSet, $numPseudo) = $et->CountNewValues();
            if ($numSet != $numPseudo and $et->IsDirectory($file)) {
                print $vout "Can't write real tags to a directory - $infile\n" if defined $verbose;
                $numSet = $numPseudo;
            }
            if ($et->Exists($file)) {
                unless ($numSet) {
                    # no need to write if no tags set
                    print $vout "Nothing changed in $file\n" if defined $verbose;
                    EFile($infile, 1);
                    ++$countSameWr;
                    return 1;
                }
            } elsif (CanCreate($file)) {
                if ($numSet == $numPseudo) {
                    # no need to write if no real tags
                    Warn("Error: Nothing to write - $file\n");
                    EFile($infile, 1);
                    ++$countBadWr;
                    return 0;
                }
                unless (defined $outfile) {
                    # create file from scratch
                    $outfile = $file;
                    $file = '';
                }
            } else {
                # file doesn't exist, and we can't create it
                Warn "Error: File not found - $file\n";
                EFile($infile);
                FileNotFound($file);
                ++$countBadWr;
                return 0;
            }
            # quickly rename file and/or set file date if this is all we are doing
            if ($numSet == $numPseudo) {
                my ($r0, $r1, $r2, $r3) = (0, 0, 0, 0);
                if (defined $outfile) {
                    $r0 = $et->SetFileName($file, $outfile);
                    $file = $$et{NewName} if $r0 > 0;   # continue with new name if changed
                }
                unless ($r0 < 0) {
                    $r1 = $et->SetFileModifyDate($file,undef,'FileCreateDate');
                    $r2 = $et->SetFileModifyDate($file);
                    $r3 = $et->SetSystemTags($file);
                }
                if ($r0 > 0 or $r1 > 0 or $r2 > 0 or $r3 > 0) {
                    EFile($infile, 3);
                    ++$countGoodWr;
                } elsif ($r0 < 0 or $r1 < 0 or $r2 < 0 or $r3 < 0) {
                    EFile($infile);
                    ++$countBadWr;
                    return 0;
                } else {
                    EFile($infile, 1);
                    ++$countSameWr;
                }
                if (defined $hardLink or defined $symLink or defined $testName) {
                    DoHardLink($et, $file, $hardLink, $symLink, $testName);
                }
                return 1;
            }
            if (not defined $outfile or defined $sameFile) {
                # write to a truly temporary file
                $outfile = "${file}_exiftool_tmp";
                if ($et->Exists($outfile)) {
                    Warn("Error: Temporary file already exists: $outfile\n");
                    EFile($infile);
                    ++$countBadWr;
                    return 0;
                }
                $isTemporary = 1;
            }
            # new output file is temporary until we know it has been written properly
            $tmpFile = $outfile;
        }
    }
    # rewrite the file
    my $success = $et->WriteInfo(Infile($file), $outfile, $outType);

    # create hard link if specified
    if ($success and (defined $hardLink or defined $symLink or defined $testName)) {
        my $src = defined $outfile ? $outfile : $file;
        DoHardLink($et, $src, $hardLink, $symLink, $testName);
    }

    # get file time if preserving it
    my ($aTime, $mTime, $cTime, $doPreserve);
    $doPreserve = $preserveTime unless $file eq '';
    if ($doPreserve and $success) {
        ($aTime, $mTime, $cTime) = $et->GetFileTime($file);
        # don't override date/time values written by the user
        undef $cTime if $$et{WRITTEN}{FileCreateDate};
        if ($$et{WRITTEN}{FileModifyDate} or $doPreserve == 2) {
            if (defined $cTime) {
                undef $aTime;       # only preserve FileCreateDate
                undef $mTime;
            } else {
                undef $doPreserve;  # (nothing to preserve)
            }
        }
    }

    if ($success == 1) {
        # preserve the original file times
        if (defined $tmpFile) {
            if ($et->Exists($file)) {
                $et->SetFileTime($tmpFile, $aTime, $mTime, $cTime) if $doPreserve;
                if ($isTemporary) {
                    # preserve original file attributes if possible
                    $et->CopyFileAttrs($file, $outfile);
                    # move original out of the way
                    my $original = "${file}_original";
                    if (not $overwriteOrig and not $et->Exists($original)) {
                        # rename the file and check again to be sure the file doesn't exist
                        # (in case, say, the filesystem truncated the file extension)
                        if (not $et->Rename($file, $original) or $et->Exists($file)) {
                            Error "Error renaming $file\n";
                            return 0;
                        }
                    }
                    my $dstFile = defined $sameFile ? $sameFile : $file;
                    if ($overwriteOrig > 1) {
                        # copy temporary file over top of original to preserve attributes
                        my ($err, $buff);
                        my $newFile = $tmpFile;
                        $et->Open(\*NEW_FILE, $newFile) or Error("Error opening $newFile\n"), return 0;
                        binmode(NEW_FILE);

                        #..........................................................
                        # temporarily disable CTRL-C during this critical operation
                        $critical = 1;
                        undef $tmpFile;     # handle deletion of temporary file ourself
                        if ($et->Open(\*ORIG_FILE, $file, '+<')) {
                            binmode(ORIG_FILE);
                            while (read(NEW_FILE, $buff, 65536)) {
                                print ORIG_FILE $buff or $err = 1;
                            }
                            close(NEW_FILE);
                            # Handle files being shorter than the original
                            eval { truncate(ORIG_FILE, tell(ORIG_FILE)) } or $err = 1;
                            close(ORIG_FILE) or $err = 1;
                            if ($err) {
                                Warn "Couldn't overwrite in place - $file\n";
                                unless ($et->Rename($newFile, $file) or
                                    ($et->Unlink($file) and $et->Rename($newFile, $file)))
                                {
                                    Error("Error renaming $newFile to $file\n");
                                    undef $critical;
                                    SigInt() if $interrupted;
                                    return 0;
                                }
                            } else {
                                $et->SetFileModifyDate($file, $cTime, 'FileCreateDate', 1);
                                $et->SetFileModifyDate($file, $mTime, 'FileModifyDate', 1);
                                $et->Unlink($newFile);
                                if ($doPreserve) {
                                    $et->SetFileTime($file, $aTime, $mTime, $cTime);
                                    # save time to set it later again to patch OS X 10.6 bug
                                    $preserveTime{$file} = [ $aTime, $mTime, $cTime ];
                                }
                            }
                            EFile($infile, 3);
                            ++$countGoodWr;
                        } else {
                            close(NEW_FILE);
                            Warn "Error opening $file for writing\n";
                            EFile($infile);
                            $et->Unlink($newFile);
                            ++$countBadWr;
                        }
                        undef $critical;            # end critical section
                        SigInt() if $interrupted;   # issue delayed SIGINT if necessary
                        #..........................................................

                    # simply rename temporary file to replace original
                    # (if we didn't already rename it to add "_original")
                    } elsif ($et->Rename($tmpFile, $dstFile)) {
                        EFile($infile, 3);
                        ++$countGoodWr;
                    } else {
                        my $newFile = $tmpFile;
                        undef $tmpFile; # (avoid deleting file if we get interrupted)
                        # unlink may fail if already renamed or no permission
                        if (not $et->Unlink($file)) {
                            Warn "Error renaming temporary file to $dstFile\n";
                            EFile($infile);
                            $et->Unlink($newFile);
                            ++$countBadWr;
                        # try renaming again now that the target has been deleted
                        } elsif (not $et->Rename($newFile, $dstFile)) {
                            Warn "Error renaming temporary file to $dstFile\n";
                            EFile($infile);
                            # (don't delete tmp file now because it is all we have left)
                            ++$countBadWr;
                        } else {
                            EFile($infile, 3);
                            ++$countGoodWr;
                        }
                    }
                } elsif ($overwriteOrig) {
                    # erase original file
                    EFile($infile, 3);
                    $et->Unlink($file) or Warn "Error erasing original $file\n";
                    ++$countGoodWr;
                } else {
                    EFile($infile, 4);
                    ++$countGoodCr;
                }
            } else {
                # this file was created from scratch, not edited
                EFile($infile, 4);
                ++$countGoodCr;
            }
        } else {
            EFile($infile, 3);
            ++$countGoodWr;
        }
    } elsif ($success) {
        EFile($infile, 1);
        if ($isTemporary) {
            # just erase the temporary file since no changes were made
            $et->Unlink($tmpFile);
            ++$countSameWr;
        } else {
            $et->SetFileTime($outfile, $aTime, $mTime, $cTime) if $doPreserve;
            if ($overwriteOrig) {
                $et->Unlink($file) or Warn "Error erasing original $file\n";
            }
            ++$countCopyWr;
        }
        print $vout "Nothing changed in $file\n" if defined $verbose;
    } else {
        EFile($infile);
        $et->Unlink($tmpFile) if defined $tmpFile;
        ++$countBadWr;
    }
    undef $tmpFile;
    return $success;
}

#------------------------------------------------------------------------------
# Make hard link and handle TestName if specified
# Inputs: 0) ExifTool ref, 1) source file name, 2) HardLink name,
#         3) SymLink name, 4) TestFile name
sub DoHardLink($$$$$)
{
    my ($et, $src, $hardLink, $symLink, $testName) = @_;
    if (defined $hardLink) {
        $hardLink = NextUnusedFilename($hardLink);
        if ($et->SetFileName($src, $hardLink, 'Link') > 0) {
            $countLink{Hard} = ($countLink{Hard} || 0) + 1;
        } else {
            $countLink{BadHard} = ($countLink{BadHard} || 0) + 1;
        }
    }
    if (defined $symLink) {
        $symLink = NextUnusedFilename($symLink);
        if ($et->SetFileName($src, $symLink, 'SymLink') > 0) {
            $countLink{Sym} = ($countLink{Sym} || 0) + 1;
        } else {
            $countLink{BadSym} = ($countLink{BadSym} || 0) + 1;
        }
    }
    if (defined $testName) {
        $testName = NextUnusedFilename($testName, $src);
        if ($usedFileName{$testName}) {
            $et->Warn("File '${testName}' would exist");
        } elsif ($et->SetFileName($src, $testName, 'Test', $usedFileName{$testName}) == 1) {
            $usedFileName{$testName} = 1;
            $usedFileName{$src} = 0;
        }
    }
}

#------------------------------------------------------------------------------
# Clean string for XML (also removes invalid control chars and malformed UTF-8)
# Inputs: 0) string ref
# Returns: nothing, but input string is escaped
sub CleanXML($)
{
    my $strPt = shift;
    # translate control characters that are invalid in XML
    $$strPt =~ tr/\0-\x08\x0b\x0c\x0e-\x1f/./;
    # fix malformed UTF-8 characters
    Image::ExifTool::XMP::FixUTF8($strPt) unless $altEnc;
    # escape necessary characters for XML
    $$strPt = Image::ExifTool::XMP::EscapeXML($$strPt);
}

#------------------------------------------------------------------------------
# Encode string for XML
# Inputs: 0) string ref
# Returns: encoding used (and input string is translated)
sub EncodeXML($)
{
    my $strPt = shift;
    if ($$strPt =~ /[\0-\x08\x0b\x0c\x0e-\x1f]/ or
        (not $altEnc and Image::ExifTool::IsUTF8($strPt) < 0))
    {
        # encode binary data and non-UTF8 with special characters as base64
        $$strPt = Image::ExifTool::XMP::EncodeBase64($$strPt);
        # #ATV = Alexander Vonk, private communication
        return 'http://www.w3.org/2001/XMLSchema#base64Binary'; #ATV
    } elsif ($escapeHTML) {
        $$strPt = Image::ExifTool::HTML::EscapeHTML($$strPt, $altEnc);
    } else {
        $$strPt = Image::ExifTool::XMP::EscapeXML($$strPt);
    }
    return '';  # not encoded
}

#------------------------------------------------------------------------------
# Format value for XML output
# Inputs: 0) value, 1) indentation, 2) group
# Returns: formatted value
sub FormatXML($$$)
{
    local $_;
    my ($val, $ind, $grp) = @_;
    my $gt = '>';
    if (ref $val eq 'ARRAY') {
        # convert ARRAY into an rdf:Bag
        my $val2 = "\n$ind <rdf:Bag>";
        foreach (@$val) {
            $val2 .= "\n$ind  <rdf:li" . FormatXML($_, "$ind  ", $grp) . "</rdf:li>";
        }
        $val = "$val2\n$ind </rdf:Bag>\n$ind";
    } elsif (ref $val eq 'HASH') {
        $gt = " rdf:parseType='Resource'>";
        my $val2 = '';
        my @keys = $$val{_ordered_keys_} ? @{$$val{_ordered_keys_}} : sort keys %$val;
        foreach (@keys) {
            # (some variable-namespace XML structure fields may have a different group)
            my ($ns, $tg) = ($grp, $_);
            if (/^(.*?):(.*)/) {
                if ($grp eq 'JSON') {
                    $tg =~ tr/:/_/; # colons in JSON structure elements are not namespaces
                } else {
                    ($ns, $tg) = ($1, $2);
                }
            }
            # validate XML attribute name
            my $name;
            foreach $name ($ns, $tg) {
                # make sure name is valid for XML
                $name =~ tr/-_A-Za-z0-9.//dc;
                $name = '_' . $name if $name !~ /^[_A-Za-z]/;
            }
            my $tok = $ns . ':' . $tg;
            $val2 .= "\n$ind <$tok" . FormatXML($$val{$_}, "$ind ", $grp) . "</$tok>";
        }
        $val = "$val2\n$ind";
    } else {
        # (note: SCALAR reference should have already been converted)
        my $enc = EncodeXML(\$val);
        $gt = " rdf:datatype='${enc}'>\n" if $enc; #ATV
    }
    return $gt . $val;
}

#------------------------------------------------------------------------------
# Escape string for JSON or PHP
# Inputs: 0) string, 1) flag to force numbers to be quoted too
# Returns: Escaped string (quoted if necessary)
sub EscapeJSON($;$)
{
    my ($str, $quote) = @_;
    unless ($quote) {
        # JSON boolean (true or false)
        return lc($str) if $str =~ /^(true|false)$/i and $json < 2;
        # JSON/PHP number (see json.org for numerical format)
        # return $str if $str =~ /^-?(\d|[1-9]\d+)(\.\d+)?(e[-+]?\d+)?$/i;
        # (these big numbers caused problems for some JSON parsers, so be more conservative)
        return $str if $str =~ /^-?(\d|[1-9]\d{1,14})(\.\d{1,16})?(e[-+]?\d{1,3})?$/i;
    }
    # encode JSON string in base64 if necessary
    if ($json < 2 and defined $binaryOutput and Image::ExifTool::IsUTF8(\$str) < 0) {
        return '"base64:' . Image::ExifTool::XMP::EncodeBase64($str, 1) . '"';
    }
    # escape special characters
    $str =~ s/(["\t\n\r\\])/\\$jsonChar{$1}/sg;
    if ($json < 2) { # JSON
        $str =~ tr/\0//d;   # remove all nulls
        # escape other control characters with \u
        $str =~ s/([\0-\x1f])/sprintf("\\u%.4X",ord $1)/sge;
        # JSON strings must be valid UTF8
        Image::ExifTool::XMP::FixUTF8(\$str) unless $altEnc;
    } else { # PHP
        $str =~ s/\0+$// unless $isBinary;  # remove trailing nulls unless binary
        # must escape "$" too for PHP
        $str =~ s/\$/\\\$/sg;
        # escape other control characters with \x
        $str =~ s/([\0-\x1f])/sprintf("\\x%.2X",ord $1)/sge;
    }
    return '"' . $str . '"';    # return the quoted string
}

#------------------------------------------------------------------------------
# Print JSON or PHP value
# Inputs: 0) file reference, 1) value, 2) indentation
sub FormatJSON($$$)
{
    local $_;
    my ($fp, $val, $ind) = @_;
    my $comma;
    if (not ref $val) {
        print $fp EscapeJSON($val);
    } elsif (ref $val eq 'ARRAY') {
        if ($joinLists and not ref $$val[0]) {
            print $fp EscapeJSON(join $listSep, @$val);
        } else {
            my ($bra, $ket) = $json == 1 ? ('[',']') : ('Array(',')');
            print $fp $bra;
            foreach (@$val) {
                print $fp ',' if $comma;
                FormatJSON($fp, $_, $ind);
                $comma = 1,
            }
            print $fp $ket,
        }
    } elsif (ref $val eq 'HASH') {
        my ($bra, $ket, $sep) = $json == 1 ? ('{','}',':') : ('Array(',')',' =>');
        print $fp $bra;
        my @keys = $$val{_ordered_keys_} ? @{$$val{_ordered_keys_}} : sort keys %$val;
        foreach (@keys) {
            print $fp ',' if $comma;
            my $key = EscapeJSON($_, 1);
            print $fp qq(\n$ind  $key$sep );
            # hack to force decimal id's to be printed as strings with -H
            if ($showTagID and $_ eq 'id' and $showTagID eq 'H' and $$val{$_} =~ /^\d+\.\d+$/) {
                print $fp qq{"$$val{$_}"};
            } else {
                FormatJSON($fp, $$val{$_}, "$ind  ");
            }
            $comma = 1,
        }
        print $fp "\n$ind$ket",
    } else {
        # (note: SCALAR reference should have already been converted)
        print $fp '"<err>"';
    }
}

#------------------------------------------------------------------------------
# Format value for CSV file
# Inputs: value
# Returns: value quoted if necessary
sub FormatCSV($)
{
    my $val = shift;
    # check for valid encoding if the Charset option was used
    if ($setCharset and ($val =~ /[^\x09\x0a\x0d\x20-\x7e\x80-\xff]/ or
        ($setCharset eq 'UTF8' and Image::ExifTool::IsUTF8(\$val) < 0)))
    {
        $val = 'base64:' . Image::ExifTool::XMP::EncodeBase64($val, 1);
    }
    # currently, there is a chance that the value may contain NULL characters unless
    # the -b option is used to encode as Base64.  It is unclear whether or not this
    # is valid CSV, but some readers may not like it.  (If this becomes a problem,
    # in the future values may need to be truncated at the first NULL character.)
    $val = qq{"$val"} if $val =~ s/"/""/g or $val =~ /(^\s+|\s+$)/ or $val =~ /[\n\r]|\Q$csvDelim/;
    return $val;
}

#------------------------------------------------------------------------------
# Print accumulated CSV information
sub PrintCSV()
{
    my ($file, $lcTag, @tags);

    @csvTags or @csvTags = sort keys %csvTags;
    # make a list of tags actually found
    foreach $lcTag (@csvTags) {
        push @tags, FormatCSV($csvTags{$lcTag}) if $csvTags{$lcTag};
    }
    print join($csvDelim, 'SourceFile', @tags), "\n";
    my $empty = defined($forcePrint) ? $forcePrint : '';
    foreach $file (@csvFiles) {
        my @vals = (FormatCSV($file)); # start with full file name
        my $csvInfo = $database{$file};
        foreach $lcTag (@csvTags) {
            next unless $csvTags{$lcTag};
            my $val = $$csvInfo{$lcTag};
            defined $val or push(@vals,$empty), next;
            push @vals, FormatCSV($val);
        }
        print join($csvDelim, @vals), "\n";
    }
}

#------------------------------------------------------------------------------
# Add tag groups from structure fields to a list for xmlns
# Inputs: 0) tag value, 1) parent group, 2) group hash ref, 3) group list ref
sub AddGroups($$$$)
{
    my ($val, $grp, $groupHash, $groupList) = @_;
    my ($key, $val2);
    if (ref $val eq 'HASH') {
        foreach $key (sort keys %$val) {
            if ($key =~ /^(.*?):/ and not $$groupHash{$1} and $grp ne 'JSON') {
                $$groupHash{$1} = $grp;
                push @$groupList, $1;
            }
            AddGroups($$val{$key}, $grp, $groupHash, $groupList) if ref $$val{$key};
        }
    } elsif (ref $val eq 'ARRAY') {
        foreach $val2 (@$val) {
            AddGroups($val2, $grp, $groupHash, $groupList) if ref $val2;
        }
    }
}

#------------------------------------------------------------------------------
# Convert binary data (SCALAR references) for printing
# Inputs: 0) object reference
# Returns: converted object, or undef if we don't want binary objects
sub ConvertBinary($)
{
    my $obj = shift;
    my ($key, $val);
    if (ref $obj eq 'HASH') {
        foreach $key (keys %$obj) {
            next unless ref $$obj{$key};
            $$obj{$key} = ConvertBinary($$obj{$key});
            return undef unless defined $$obj{$key};
        }
    } elsif (ref $obj eq 'ARRAY') {
        foreach $val (@$obj) {
            next unless ref $val;
            $val = ConvertBinary($val);
            return undef unless defined $val;
        }
    } elsif (ref $obj eq 'SCALAR') {
        return undef if $noBinary;
        # (binaryOutput flag is set to 0 for binary mode of XML/PHP/JSON output formats)
        if (defined $binaryOutput) {
            $obj = $$obj;
            # encode in base64 if necessary (0xf7 allows for up to 21-bit UTF-8 code space)
            if ($json == 1 and ($obj =~ /[^\x09\x0a\x0d\x20-\x7e\x80-\xf7]/ or
                                Image::ExifTool::IsUTF8(\$obj) < 0))
            {
                $obj = 'base64:' . Image::ExifTool::XMP::EncodeBase64($obj, 1);
            }
        } else {
            # (-b is not valid for HTML output)
            my $bOpt = $html ? '' : ', use -b option to extract';
            if ($$obj =~ /^Binary data \d+ bytes$/) {
                $obj = "($$obj$bOpt)";
            } else {
                $obj = '(Binary data ' . length($$obj) . " bytes$bOpt)";
            }
        }
    }
    return $obj;
}

#------------------------------------------------------------------------------
# Compare ValueConv and PrintConv values of a tag to see if they are equal
# Inputs: 0) value1, 1) value2
# Returns: true if they are equal
sub IsEqual($$)
{
    my ($a, $b) = @_;
    # (scalar values are not print-converted)
    return 1 if $a eq $b or ref $a eq 'SCALAR';
    if (ref $a eq 'HASH' and ref $b eq 'HASH') {
        return 0 if scalar(keys %$a) != scalar(keys %$b);
        my $key;
        foreach $key (keys %$a) {
            return 0 unless IsEqual($$a{$key}, $$b{$key});
        }
    } else {
        return 0 if ref $a ne 'ARRAY' or ref $b ne 'ARRAY' or @$a != @$b;
        my $i;
        for ($i=0; $i<scalar(@$a); ++$i) {
            return 0 unless IsEqual($$a[$i], $$b[$i]);
        }
    }
    return 1;
}

#------------------------------------------------------------------------------
# Add tag list for copying tags from specified file
# Inputs: 0) set tags file name (or FMT), 1) options for SetNewValuesFromFile()
# Returns: nothing
# Notes: Uses global variables: %setTags, %setTagsList, @newValues, $saveCount
sub AddSetTagsFile($;$)
{
    my ($setFile, $opts) = @_;
    if ($setTags{$setFile}) {
        # move these tags aside and make a new list for the next invocation of this file
        $setTagsList{$setFile} or $setTagsList{$setFile} = [ ];
        push @{$setTagsList{$setFile}}, $setTags{$setFile};
    }
    $setTags{$setFile} = [];    # create list for tags to copy from this file
    # insert marker to save new values now (necessary even if this is not a dynamic
    # file in case the same file is source'd multiple times in a single command)
    push @newValues, { SaveCount => ++$saveCount }, "TagsFromFile=$setFile";
    # add option to protect the tags which are assigned after this
    # (this is the mechanism by which the command-line order-of-operations is preserved)
    $opts or $opts = { };
    $$opts{ProtectSaved} = $saveCount;
    push @{$setTags{$setFile}}, $opts;
}

#------------------------------------------------------------------------------
# Get input file name or reference for calls to the ExifTool API
# Inputs: 0) file name ('-' for STDIN), 1) flag to buffer STDIN
# Returns: file name, or RAF reference for buffering STDIN
sub Infile($;$)
{
    my ($file, $bufferStdin) = @_;
    if ($file eq '-' and ($bufferStdin or $rafStdin)) {
        if ($rafStdin) {
            $rafStdin->Seek(0); # rewind
        } elsif (open RAF_STDIN, '-') {
            $rafStdin = File::RandomAccess->new(\*RAF_STDIN);
            $rafStdin->BinMode();
        }
        return $rafStdin if $rafStdin;
    }
    return $file;
}

#------------------------------------------------------------------------------
# Issue warning to stderr, adding leading "Warning: " and trailing newline
# if the warning isn't suppressed by the API NoWarning option
# Inputs: 0) ExifTool ref, 1) warning string
sub Warning($$)
{
    my ($et, $str) = @_;
    my $noWarn = $et->Options('NoWarning');
    if (not defined $noWarn or not eval { $str =~ /$noWarn/ }) {
        Warn "Warning: $str\n";
    }
}

#------------------------------------------------------------------------------
# Set new values from file
# Inputs: 0) ExifTool ref, 1) filename, 2) reference to list of values to set
# Returns: 0 on error (and increments $countBadWr)
sub DoSetFromFile($$$)
{
    local $_;
    my ($et, $file, $setTags) = @_;
    $verbose and print $vout "Setting new values from $file\n";
    my $info = $et->SetNewValuesFromFile(Infile($file,1), @$setTags);
    my $numSet = scalar(keys %$info);
    if ($$info{Error}) {
        # delete all error and warning tags
        my @warns = grep /^(Error|Warning)\b/, keys %$info;
        $numSet -= scalar(@warns);
        # issue a warning only for the main error
        my $err = $$info{Error};
        delete $$info{$_} foreach @warns;
        my $noWarn = $et->Options('NoWarning');
        $$info{Warning} = $err unless defined $noWarn and eval { $err =~ /$noWarn/ };
    } elsif ($$info{Warning}) {
        my $warns = 1;
        ++$warns while $$info{"Warning ($warns)"};
        $numSet -= $warns;
    }
    PrintErrors($et, $info, $file) and EFile($file), ++$countBadWr, return 0;
    Warning($et,"No writable tags set from $file") unless $numSet;
    return 1;
}

#------------------------------------------------------------------------------
# Translate backslashes to forward slashes in filename if necessary
# Inputs: 0) Filename
# Returns: nothing, but changes filename if necessary
sub CleanFilename($)
{
    $_[0] =~ tr/\\/\// if $hasBackslash{$^O};
}

#------------------------------------------------------------------------------
# Check for valid UTF-8 of a file name
# Inputs: 0) string, 1) original encoding
# Returns: 0=plain ASCII, 1=valid UTF-8, -1=invalid UTF-8 (and print warning)
sub CheckUTF8($$)
{
    my ($file, $enc) = @_;
    my $isUTF8 = 0;
    if ($file =~ /[\x80-\xff]/) {
        $isUTF8 = Image::ExifTool::IsUTF8(\$file);
        if ($isUTF8 < 0) {
            if ($enc) {
                Warn("Invalid filename encoding for $file\n");
            } elsif (not defined $enc) {
                WarnOnce(qq{FileName encoding not specified.  Use "-charset FileName=CHARSET"\n});
            }
        }
    }
    return $isUTF8;
}

#------------------------------------------------------------------------------
# Set window title
# Inputs: title string or '' to reset title
sub SetWindowTitle($)
{
    my $title = shift;
    if ($curTitle ne $title) {
        $curTitle = $title;
        if ($^O eq 'MSWin32') {
            $title =~ s/([&\/\?:|"<>])/^$1/g;   # escape special chars
            eval { system qq{title $title} };
        } else {
            # (this only works for XTerm terminals, and STDERR must go to the console)
            printf STDERR "\033]0;%s\007", $title;
        }
    }
}

#------------------------------------------------------------------------------
# Process files in our @files list
# Inputs: 0) ExifTool ref, 1) list ref to just return full file names
sub ProcessFiles($;$)
{
    my ($et, $list) = @_;
    my $enc = $et->Options('CharsetFileName');
    my $file;
    foreach $file (@files) {
        $et->Options(CharsetFileName => 'UTF8') if $utf8FileName{$file};
        if (defined $progressMax) {
            unless (defined $progressNext) {
                $progressNext = $progressCount + $progressIncr;
                $progressNext -= $progressNext % $progressIncr; # (show even multiples)
                $progressNext = $progressMax if $progressNext > $progressMax;
            }
            ++$progressCount;
            if ($progress) {
                if ($progressCount >= $progressNext) {
                    $progStr = " [$progressCount/$progressMax]";
                } else {
                    undef $progStr; # don't update progress yet
                }
            }
        }
        if ($et->IsDirectory($file) and not $listDir) {
            $multiFile = $validFile = 1;
            ScanDir($et, $file, $list);
        } elsif ($filterFlag and not AcceptFile($file)) {
            if ($et->Exists($file)) {
                $filtered = 1;
                Progress($vout, "-------- $file (wrong extension)") if $verbose;
            } else {
                Warn "Error: File not found - $file\n";
                FileNotFound($file);
                $rtnVal = 1;
            }
        } else {
            $validFile = 1;
            if ($list) {
                push(@$list, $file);
            } else {
                if (%endDir) {
                    my ($d, $f) = Image::ExifTool::SplitFileName($file);
                    next if $endDir{$d};
                }
                GetImageInfo($et, $file);
                $end and Warn("End called - $file\n");
                if ($endDir) {
                    Warn("EndDir called - $file\n");
                    my ($d, $f) = Image::ExifTool::SplitFileName($file);
                    $endDir{$d} = 1;
                    undef $endDir;
                }
            }
        }
        $et->Options(CharsetFileName => $enc) if $utf8FileName{$file};
        last if $end;
    }
}

#------------------------------------------------------------------------------
# Scan directory for image files
# Inputs: 0) ExifTool ref, 1) directory name, 2) list ref to return file names
sub ScanDir($$;$)
{
    local $_;
    my ($et, $dir, $list) = @_;
    my (@fileList, $done, $file, $utf8Name, $winSurrogate, $endThisDir);
    my $enc = $et->Options('CharsetFileName');
    # recode as UTF-8 if necessary
    if ($enc) {
        unless ($enc eq 'UTF8') {
            $dir = $et->Decode($dir, $enc, undef, 'UTF8');
            $et->Options(CharsetFileName => 'UTF8');    # now using UTF8
        }
        $utf8Name = 1;
    }
    return if $ignore{$dir};
    # use Win32::FindFile on Windows if available
    # (ReadDir will croak if there is a wildcard, so check for this)
    if ($^O eq 'MSWin32' and $dir !~ /[*?]/) {
        undef $evalWarning;
        local $SIG{'__WARN__'} = sub { $evalWarning = $_[0] };;
        if (CheckUTF8($dir, $enc) >= 0) {
            if (eval { require Win32::FindFile }) {
                eval {
                    @fileList = Win32::FindFile::ReadDir($dir);
                    $_ = $_->cFileName foreach @fileList;
                };
                $@ and $evalWarning = $@;
                if ($evalWarning) {
                    chomp $evalWarning;
                    $evalWarning =~ s/ at .*//s;
                    Warning($et,"[Win32::FindFile] $evalWarning - $dir");
                    $winSurrogate = 1 if $evalWarning =~ /surrogate/;
                } else {
                    $et->Options(CharsetFileName => 'UTF8');    # now using UTF8
                    $utf8Name = 1;  # ReadDir returns UTF-8 file names
                    $done = 1;
                }
            } else {
                $done = 0;
            }
        }
    }
    unless ($done) {
        # use standard perl library routines to read directory
        unless (opendir(DIR_HANDLE, $dir)) {
            Warn("Error opening directory $dir\n");
            return;
        }
        @fileList = readdir(DIR_HANDLE);
        closedir(DIR_HANDLE);
        if (defined $done) {
            # issue warning if some names would have required Win32::FindFile
            foreach $file ($dir, @fileList) {
                next unless $file =~ /[\?\x80-\xff]/;
                WarnOnce("Install Win32::FindFile to support Windows Unicode file names in directories\n");
                last;
            }
        }
    }
    $dir =~ /\/$/ or $dir .= '/';   # make sure directory name ends with '/'
    foreach $file (@fileList) {
        my $path = "$dir$file";
        if ($et->IsDirectory($path)) {
            next unless $recurse;
            # ignore directories starting with "." by default
            next if $file =~ /^\./ and ($recurse == 1 or $file eq '.' or $file eq '..');
            next if $ignore{$file} or ($ignore{SYMLINKS} and -l $path);
            ScanDir($et, $path, $list);
            last if $end;
            next;
        }
        next if $endThisDir;
        next if $ignoreHidden and $file =~ /^\./;   # ignore hidden files if specified
        # apply rules from -ext options
        my $accepted;
        if ($filterFlag) {
            $accepted = AcceptFile($file) or next;
            # must be specifically accepted to bypass selection logic
            $accepted &= 0x01;
        }
        unless ($accepted) {
            # read/write this file if it is a supported type
            if ($scanWritable) {
                if ($scanWritable eq '1') {
                    next unless CanWrite($file);
                } else {
                    my $type = GetFileType($file);
                    next unless defined $type and $type eq $scanWritable;
                }
            } elsif (not GetFileType($file)) {
                next unless $doUnzip;
                next unless $file =~ /\.(gz|bz2)$/i;
            }
        }
        # Windows patch to avoid replacing filename containing Unicode surrogate with 8.3 name
        if ($winSurrogate and $isWriting and
            (not $overwriteOrig or $overwriteOrig != 2) and
            not $doSetFileName and $file =~ /~/)   # (8.3 name will contain a tilde)
        {
            Warn("Not writing $path\n");
            WarnOnce("Use -overwrite_original_in_place to write files with Unicode surrogate characters\n");
            EFile($file);
            ++$countBad;
            next;
        }
        $utf8FileName{$path} = 1 if $utf8Name;
        if ($list) {
            push(@$list, $path);
        } else {
            GetImageInfo($et, $path);
            if ($end) {
                Warn("End called - $file\n");
                last;
            }
            if ($endDir) {
                $path =~ s(/$)();
                Warn("EndDir called - $path\n");
                $endDir{$path} = 1;
                $endThisDir = 1;
                undef $endDir;
            }
        }
    }
    ++$countDir;
    $et->Options(CharsetFileName => $enc);  # restore original setting
}

#------------------------------------------------------------------------------
# Find files with wildcard expression on Windows
# Inputs: 0) ExifTool ref, 1) file name with wildcards
# Returns: list of matching file names
# Notes:
# 1) Win32::FindFile must already be loaded
# 2) Sets flag in %utf8FileName for each file found
sub FindFileWindows($$)
{
    my ($et, $wildfile) = @_;

    # recode file name as UTF-8 if necessary
    my $enc = $et->Options('CharsetFileName');
    $wildfile = $et->Decode($wildfile, $enc, undef, 'UTF8') if $enc and $enc ne 'UTF8';
    $wildfile =~ tr/\\/\//; # use forward slashes
    my ($dir, $wildname) = ($wildfile =~ m{(.*[:/])(.*)}) ? ($1, $2) : ('', $wildfile);
    if ($dir =~ /[*?]/) {
        Warn "Wildcards don't work in the directory specification\n";
        return ();
    }
    CheckUTF8($wildfile, $enc) >= 0 or return ();
    undef $evalWarning;
    local $SIG{'__WARN__'} = sub { $evalWarning = $_[0] };
    my @files;
    eval {
        my @names = Win32::FindFile::FindFile($wildfile) or return;
        # (apparently this isn't always sorted, so do a case-insensitive sort here)
        @names = sort { uc($a) cmp uc($b) } @names;
        my ($rname, $nm);
        # replace "\?" with ".", and "\*" with ".*" for regular expression
        ($rname = quotemeta $wildname) =~ s/\\\?/./g;
        $rname =~ s/\\\*/.*/g;
        foreach $nm (@names) {
            $nm = $nm->cFileName;
            # make sure that FindFile behaves
            # (otherwise "*.jpg" matches things like "a.jpg_original"!)
            next unless $nm =~ /^$rname$/i;
            next if $nm eq '.' or $nm eq '..';  # don't match "." and ".."
            my $file = "$dir$nm";       # add back directory name
            push @files, $file;
            $utf8FileName{$file} = 1;   # flag this file name as UTF-8 encoded
        }
    };
    $@ and $evalWarning = $@;
    if ($evalWarning) {
        chomp $evalWarning;
        $evalWarning =~ s/ at .*//s;
        Warn "Error: [Win32::FindFile] $evalWarning - $wildfile\n";
        undef @files;
        EFile($wildfile);
        ++$countBad;
    }
    return @files;
}

#------------------------------------------------------------------------------
# Handle missing file on the command line
# Inputs: 0) file name
sub FileNotFound($)
{
    my $file = shift;
    if ($file =~ /^(DIR|FILE)$/) {
        my $type = { DIR => 'directory', FILE => 'file' }->{$file};
        Warn qq{You were meant to enter any valid $type name, not "$file" literally.\n};
    }
}

#------------------------------------------------------------------------------
# Patch for OS X 10.6 to preserve file modify date
# (this probably isn't a 100% fix, but it may solve a majority of the cases)
sub PreserveTime()
{
    local $_;
    $mt->SetFileTime($_, @{$preserveTime{$_}}) foreach keys %preserveTime;
    undef %preserveTime;
}

#------------------------------------------------------------------------------
# Return absolute path for a file
# Inputs: 0) file name
# Returns: absolute path string, or undef if path could not be determined
# Note: Warnings should be suppressed when calling this routine
sub AbsPath($)
{
    my $file = shift;
    my $path;
    if (defined $file and eval { require Cwd }) {
        $path = eval { Cwd::abs_path($file) };
        # make the delimiters and case consistent
        # (abs_path is very inconsistent about what it returns in Windows)
        if (defined $path and $hasBackslash{$^O}) {
            $path =~ tr/\\/\//;
            $path = lc $path;
        }
    }
    return $path;
}

#------------------------------------------------------------------------------
# Convert file name to ExifTool Charset
# Inputs: 0) ExifTool ref, 1) file name in CharsetFileName
# Returns: file name in ExifTool Charset
sub MyConvertFileName($$)
{
    my ($et, $file) = @_;
    my $enc = $et->Options('CharsetFileName');
    $et->Options(CharsetFileName => 'UTF8') if $utf8FileName{$file};
    my $convFile = $et->ConvertFileName($file);
    $et->Options(CharsetFileName => $enc) if $utf8FileName{$file};
    return $convFile;
}

#------------------------------------------------------------------------------
# Add print format entry
# Inputs: 0) expression string
sub AddPrintFormat($)
{
    my $expr = shift;
    my $type;
    if ($expr =~ /^#/) {
        $expr =~ s/^#\[(HEAD|SECT|IF|BODY|ENDS|TAIL)\]// or return; # ignore comments
        $type = $1;
    } else {
        $type = 'BODY';
    }
    $printFmt{$type} or $printFmt{$type} = [ ];
    push @{$printFmt{$type}}, $expr;
    # add to list of requested tags
    push @requestTags, $expr =~ /\$\{?((?:[-\w]+:)*[-\w?*]+)/g;
    $printFmt{SetTags} = 1 if $expr =~ /\bSetTags\b/;
}

#------------------------------------------------------------------------------
# Get suggested file extension based on tag value for binary output
# Inputs: 0) ExifTool ref, 1) data ref, 2) tag name
# Returns: file extension (lower case), or 'dat' if unknown
sub SuggestedExtension($$$)
{
    my ($et, $valPt, $tag) = @_;
    my $ext;
    if (not $binaryOutput) {
        $ext = 'txt';
    } elsif ($$valPt =~ /^\xff\xd8\xff/) {
        $ext = 'jpg';
    } elsif ($$valPt =~ /^(\0\0\0\x0cjP(  |\x1a\x1a)\x0d\x0a\x87\x0a|\xff\x4f\xff\x51\0)/) {
        $ext = 'jp2';
    } elsif ($$valPt =~ /^(\x89P|\x8aM|\x8bJ)NG\r\n\x1a\n/) {
        $ext = 'png';
    } elsif ($$valPt =~ /^GIF8[79]a/) {
        $ext = 'gif';
    } elsif ($$valPt =~ /^<\?xpacket/ or $tag eq 'XMP') {
        $ext = 'xmp';
    } elsif ($$valPt =~ /^<\?xml/ or $tag eq 'XML') {
        $ext = 'xml';
    } elsif ($$valPt =~ /^RIFF....WAVE/s) {
        $ext = 'wav';
    } elsif ($tag eq 'OriginalRawImage' and defined($ext = $et->GetValue('OriginalRawFileName'))) {
        $ext =~ s/^.*\.//s;
        $ext = $ext ? lc($ext) : 'raw';
    } elsif ($tag eq 'EXIF') {
        $ext = 'exif';
    } elsif ($tag eq 'ICC_Profile') {
        $ext = 'icc';
    } elsif ($$valPt =~ /^(MM\0\x2a|II\x2a\0)/) {
        $ext = 'tiff';
    } elsif ($$valPt =~ /^.{4}ftyp(3gp|mp4|f4v|qt  )/s) {
        my %movType = ( 'qt  ' => 'mov' );
        $ext = $movType{$1} || $1;
    } elsif ($$valPt !~ /^.{0,4096}\0/s) {
        $ext = 'txt';
    } elsif ($$valPt =~ /^BM.{15}\0/s) {
        $ext = 'bmp';
    } elsif ($$valPt =~ /^CANON OPTIONAL DATA\0/) {
        $ext = 'vrd';
    } elsif ($$valPt =~ /^IIII\x04\0\x04\0/) {
        $ext = 'dr4';
    } elsif ($$valPt =~ /^(.{10}|.{522})(\x11\x01|\x00\x11)/s) {
        $ext = 'pict';
    } elsif ($$valPt =~ /^\xff\x0a|\0\0\0\x0cJXL \x0d\x0a......ftypjxl/s) {
        $ext = 'jxl';
    } elsif ($$valPt =~ /^.{4}jumb\0.{3}jumdc2pa/s) {
        $ext = 'c2pa';
    } elsif ($tag eq 'JUMBF') {
        $ext = 'jumbf';
    } else {
        $ext = 'dat';
    }
    return $ext;
}

#------------------------------------------------------------------------------
# Load print format file
# Inputs: 0) file name, 1) flag to avoid adding newline to input argument
# - saves lines of file to %printFmt list
# - adds tag names to @tags list
sub LoadPrintFormat($;$)
{
    my ($arg, $noNL) = @_;
    if (not defined $arg) {
        Error "Must specify file or expression for -p option\n";
    } elsif ($arg !~ /\n/ and -f $arg and $mt->Open(\*FMT_FILE, $arg)) {
        foreach (<FMT_FILE>) {
            AddPrintFormat($_);
        }
        close(FMT_FILE);
    } else {
        $arg .= "\n" unless $noNL;
        AddPrintFormat($arg);
    }
}

#------------------------------------------------------------------------------
# A sort of sprintf for filenames
# Inputs: 0) format string (%d=dir, %f=file name, %e=ext),
#         1) source filename or undef to test format string
#         2-4) [%t %g %s %o only] tag name, ref to array of group names,
#              suggested extension, original raw file name
# Returns: new filename or undef on error (or if no file and fmt contains token)
sub FilenameSPrintf($;$@)
{
    my ($fmt, $file, @extra) = @_;
    local $_;
    # return format string straight away if no tokens
    return $fmt unless $fmt =~ /%[-+]?\d*[.:]?\d*[lu]?[dDfFeEtgso]/;
    return undef unless defined $file;
    CleanFilename($file);   # make sure we are using forward slashes
    # split filename into directory, file, extension
    my %part;
    @part{qw(d f E)} = ($file =~ /^(.*?)([^\/]*?)(\.[^.\/]*)?$/);
    defined $part{f} or Warn("Error: Bad pattern match for file $file\n"), return undef;
    if ($part{E}) {
        $part{e} = substr($part{E}, 1);
    } else {
        @part{qw(e E)} = ('', '');
    }
    $part{F} = $part{f} . $part{E};
    ($part{D} = $part{d}) =~ s{/+$}{};
    @part{qw(t g s o)} = @extra;
    my ($filename, $pos) = ('', 0);
    while ($fmt =~ /(%([-+]?)(\d*)([.:]?)(\d*)([lu]?)([dDfFeEtgso]))/g) {
        $filename .= substr($fmt, $pos, pos($fmt) - $pos - length($1));
        $pos = pos($fmt);
        my ($sign, $wid, $dot, $skip, $mod, $code) = ($2, $3, $4, $5 || 0, $6, $7);
        my (@path, $part, $len, $groups);
        if (lc $code eq 'd' and $dot and $dot eq ':') {
            # field width applies to directory levels instead of characters
            @path = split '/', $part{$code};
            $len = scalar @path;
        } else {
            if ($code eq 'g') {
                $groups = $part{g} || [ ] unless defined $groups;
                $fmt =~ /\G(\d?)/g; # look for %g1, %g2, etc
                $part{g} = $$groups[$1 || 0];
                $pos = pos($fmt);
            }
            $part{$code} = '' unless defined $part{$code};
            $len = length $part{$code};
        }
        next unless $skip < $len;
        $wid = $len - $skip if $wid eq '' or $wid + $skip > $len;
        $skip = $len - $wid - $skip if $sign eq '-';
        if (@path) {
            $part = join('/', @path[$skip..($skip+$wid-1)]);
            $part .= '/' unless $code eq 'D';
        } else {
            $part = substr($part{$code}, $skip, $wid);
        }
        $part = ($mod eq 'u') ? uc($part) : lc($part) if $mod;
        $filename .= $part;
    }
    $filename .= substr($fmt, $pos); # add rest of file name
    # remove double slashes (except at beginning to allow Windows UNC paths)
    $filename =~ s{(?!^)//}{/}g;
    return $filename;
}

#------------------------------------------------------------------------------
# Convert number to alphabetical index: a, b, c, ... z, aa, ab ...
# Inputs: 0) number
# Returns: alphabetical index string
sub Num2Alpha($)
{
    my $num = shift;
    my $alpha = chr(97 + ($num % 26));
    while ($num >= 26) {
        $num = int($num / 26) - 1;
        $alpha = chr(97 + ($num % 26)) . $alpha;
    }
    return $alpha;
}

#------------------------------------------------------------------------------
# Expand '%c' and '%C' codes if filename to get next unused file name
# Inputs: 0) file name format string, 1) filename ok to use even if it exists
# Returns: new file name
sub NextUnusedFilename($;$)
{
    my ($fmt, $okfile) = @_;
    return $fmt unless $fmt =~ /%[-+]?\d*\.?\d*[lun]?[cC]/;
    my %sep = ( '-' => '-', '+' => '_' );
    my ($copy, $alpha) = (0, 'a');
    for (;;) {
        my ($filename, $pos) = ('', 0);
        while ($fmt =~ /(%([-+]?)(\d*)(\.?)(\d*)([lun]?)([cC]))/g) {
            $filename .= substr($fmt, $pos, pos($fmt) - $pos - length($1));
            $pos = pos($fmt);
            my ($sign, $wid, $dec, $wid2, $mod, $tok) = ($2, $3 || 0, $4, $5 || 0, $6, $7);
            my $seq;
            if ($tok eq 'C') {
                $seq = $wid + ($sign eq '-' ? $seqFileDir : $seqFileNum) - 1;
                $wid = $wid2;
            } else {
                next unless $dec or $copy;
                $wid = $wid2 if $wid < $wid2;
                # add dash or underline separator if '-' or '+' specified
                $filename .= $sep{$sign} if $sign;
            }
            if ($mod and $mod ne 'n') {
                my $a = $tok eq 'C' ? Num2Alpha($seq) : $alpha;
                my $str = ($wid and $wid > length $a) ? 'a' x ($wid - length($a)) : '';
                $str .= $a;
                $str = uc $str if $mod eq 'u';
                $filename .= $str;
            } else {
                my $c = $tok eq 'C' ? $seq : $copy;
                my $num = $c + ($mod ? 1 : 0);
                $filename .= $wid ? sprintf("%.${wid}d",$num) : $num;
            }
        }
        $filename .= substr($fmt, $pos); # add rest of file name
        # return now with filename unless file exists
        return $filename unless ($mt->Exists($filename, 1) and not defined $usedFileName{$filename}) or $usedFileName{$filename};
        if (defined $okfile) {
            return $filename if $filename eq $okfile;
            my ($fn, $ok) = (AbsPath($filename), AbsPath($okfile));
            return $okfile if defined $fn and defined $ok and $fn eq $ok;
        }
        ++$copy;
        ++$alpha;
    }
}

#------------------------------------------------------------------------------
# Create directory for specified file
# Inputs: 0) complete file name including path
# Returns: true if a directory was created
my $k32CreateDir;
sub CreateDirectory($)
{
    my $file = shift;
    my ($dir, $created);
    ($dir = $file) =~ s/[^\/]*$//;  # remove filename from path specification
    if ($dir and not $mt->IsDirectory($dir)) {
        my @parts = split /\//, $dir;
        $dir = '';
        foreach (@parts) {
            $dir .= $_;
            if (length $dir and not $mt->IsDirectory($dir) and
                # don't try to create a network drive root directory
                not ($hasBackslash{$^O} and $dir =~ m{^//[^/]*$}))
            {
                my $success;
                # create directory since it doesn't exist
                my $d2 = $dir; # (must make a copy in case EncodeFileName recodes it)
                if ($mt->EncodeFileName($d2)) {
                    # handle Windows Unicode directory names
                    unless (eval { require Win32::API }) {
                        Error('Install Win32::API to create directories with Unicode names');
                        return 0;
                    }
                    unless ($k32CreateDir) {
                        $k32CreateDir = Win32::API->new('KERNEL32', 'CreateDirectoryW', 'PP', 'I');
                    }
                    $success = $k32CreateDir->Call($d2, 0) if $k32CreateDir;
                } else {
                    $success = mkdir($d2, 0777);
                }
                $success or Error("Error creating directory $dir\n"), return 0;
                $verbose and print $vout "Created directory $dir\n";
                $created = 1;
            }
            $dir .= '/';
        }
        ++$countNewDir if $created;
    }
    return $created;
}

#------------------------------------------------------------------------------
# Open output text file
# Inputs: 0) file name format string, 1-N) extra arguments for FilenameSPrintf
# Returns: 0) file reference (or undef on error), 1) file name if opened, 2) append flag
# Notes: returns reference to STDOUT and no file name if no textOut file needed
sub OpenOutputFile($;@)
{
    my ($file, @args) = @_;
    my ($fp, $outfile, $append);
    if ($textOut) {
        $outfile = $file;
        CleanFilename($outfile);
        if ($textOut =~ /%[-+]?\d*[.:]?\d*[lun]?[dDfFeEtgsocC]/ or defined $tagOut) {
            # make filename from printf-like $textOut
            $outfile = FilenameSPrintf($textOut, $file, @args);
            return () unless defined $outfile;
            $outfile = NextUnusedFilename($outfile);
            CreateDirectory($outfile);  # create directory if necessary
        } else {
            $outfile =~ s/\.[^.\/]*$//; # remove extension if it exists
            $outfile .= $textOut;
        }
        my $mode = '>';
        if ($mt->Exists($outfile, 1)) {
            unless ($textOverwrite) {
                Warn "Output file $outfile already exists for $file\n";
                return ();
            }
            if ($textOverwrite == 2 or ($textOverwrite == 3 and $created{$outfile})) {
                $mode = '>>';
                $append = 1;
            }
        }
        unless ($mt->Open(\*OUTFILE, $outfile, $mode)) {
            my $what = $mode eq '>' ? 'creating' : 'appending to';
            Error("Error $what $outfile\n");
            return ();
        }
        binmode(OUTFILE) if $binaryOutput;
        $fp = \*OUTFILE;
    } else {
        $fp = \*STDOUT;
    }
    return($fp, $outfile, $append);
}

#------------------------------------------------------------------------------
# Filter files based on extension
# Inputs: 0) file name
# Returns: 0 = rejected, 1 = specifically accepted, 2 = accepted by default
# Notes: This routine should only be called if $filterFlag is set
sub AcceptFile($)
{
    my $file = shift;
    my $ext = ($file =~ /^.*\.(.+)$/s) ? uc($1) : '';
    return $filterExt{$ext} if defined $filterExt{$ext};
    return $filterExt{'*'} if defined $filterExt{'*'};
    return 0 if $filterFlag & 0x02; # reject if accepting specific extensions
    return 2;   # accept by default
}

#------------------------------------------------------------------------------
# Slurp file into buffer
# Inputs: 0) file name, 1) buffer reference
# Returns: 1 on success
sub SlurpFile($$)
{
    my ($file, $buffPt) = @_;
    $mt->Open(\*INFILE, $file) or Warn("Error opening file $file\n"), return 0;
    binmode(INFILE);
    # (CAREFUL!:  must clear buffer first to reset possible utf8 flag because the data
    #  would be corrupted if it was read into a buffer which had the utf8 flag set!)
    undef $$buffPt;
    my $bsize = 1024 * 1024;
    my $num = read(INFILE, $$buffPt, $bsize);
    unless (defined $num) {
        close(INFILE);
        Warn("Error reading $file\n");
        return 0;
    }
    my $bmax = 64 * $bsize;
    while ($num == $bsize) {
        $bsize *= 2 if $bsize < $bmax;
        my $buff;
        $num = read(INFILE, $buff, $bsize);
        last unless $num;
        $$buffPt .= $buff;
    }
    close(INFILE);
    return 1;
}


#------------------------------------------------------------------------------
# Filter argfile line
# Inputs: 0) line of argfile
# Returns: filtered line or undef to ignore
sub FilterArgfileLine($)
{
    my $arg = shift;
    if ($arg =~ /^#/) {             # comment lines begin with '#'
        return undef unless $arg =~ s/^#\[CSTR\]//;
        $arg =~ s/[\x0d\x0a]+$//s;  # remove trailing newline
        # escape double quotes, dollar signs and ampersands if they aren't already
        # escaped by an odd number of backslashes, and escape a single backslash
        # if it occurs at the end of the string
        $arg =~ s{\\(.)|(["\$\@]|\\$)}{'\\'.($2 || $1)}sge;
        # un-escape characters in C string
        my %esc = ( a => "\a", b => "\b", f => "\f", n => "\n",
                    r => "\r", t => "\t", '"' => '"', '\\' => '\\' );
        $arg =~ s/\\(.)/$esc{$1}||'\\'.$1/egs;
    } else {
        $arg =~ s/^\s+//;           # remove leading white space
        $arg =~ s/[\x0d\x0a]+$//s;  # remove trailing newline
        # remove white space before, and single space after '=', '+=', '-=' or '<='
        $arg =~ s/^(-[-:\w]+#?)\s*([-+<]?=) ?/$1$2/;
        return undef if $arg eq '';
    }
    return $arg;
}

#------------------------------------------------------------------------------
# Read arguments from -stay_open argfile
# Inputs: 0) argument list ref
# Notes: blocks until -execute, -stay_open or -@ option is available
#        (or until there was an error reading from the file)
sub ReadStayOpen($)
{
    my $args = shift;
    my (@newArgs, $processArgs, $result, $optArgs);
    my $lastOpt = '';
    my $unparsed = length $stayOpenBuff;
    for (;;) {
        if ($unparsed) {
            # parse data already read from argfile
            $result = $unparsed;
            undef $unparsed;
        } else {
            # read more data from argfile
            # - this read may block (which is good) if reading from a pipe
            $result = sysread(STAYOPEN, $stayOpenBuff, 65536, length($stayOpenBuff));
        }
        if ($result) {
            my $pos = 0;
            while ($stayOpenBuff =~ /\n/g) {
                my $len = pos($stayOpenBuff) - $pos;
                my $arg = substr($stayOpenBuff, $pos, $len);
                $pos += $len;
                $arg = FilterArgfileLine($arg);
                next unless defined $arg;
                push @newArgs, $arg;
                if ($optArgs) {
                    # this is an argument for the last option
                    undef $optArgs;
                    next unless $lastOpt eq '-stay_open' or $lastOpt eq '-@';
                } else {
                    $lastOpt = lc $arg;
                    $optArgs = $optArgs{$arg};
                    unless (defined $optArgs) {
                        $optArgs = $optArgs{$lastOpt};
                        # handle options with trailing numbers
                        $optArgs = $optArgs{"$1#$2"} if not defined $optArgs and $lastOpt =~ /^(.*?)\d+(!?)$/;
                    }
                    next unless $lastOpt =~ /^-execute\d*$/;
                }
                $processArgs = 1;
                last;   # process arguments up to this point
            }
            next unless $pos;   # nothing to do if we didn't read any arguments
            # keep unprocessed data in buffer
            $stayOpenBuff = substr($stayOpenBuff, $pos);
            if ($processArgs) {
                # process new arguments after -execute or -stay_open option
                unshift @$args, @newArgs;
                last;
            }
        } elsif ($result == 0) {
            # sysread() didn't block (eg. when reading from a file),
            # so wait for a short time (1/100 sec) then try again
            # Note: may break out of this early if SIGCONT is received
            select(undef,undef,undef,0.01);
        } else {
            Warn "Error reading from ARGFILE\n";
            close STAYOPEN;
            $stayOpen = 0;
            last;
        }
    }
}

#------------------------------------------------------------------------------
# Add new entry to -efile output file
# Inputs: 0) file name, 1) -efile option number (0=error, 1=same, 2=failed, 3=updated, 4=created)
sub EFile($$)
{
    my $entry = shift;
    my $efile = $efile[shift || 0];
    if (defined $efile and length $entry and $entry ne '-') {
        my $err;
        CreateDirectory($efile);
        if ($mt->Open(\*EFILE_FILE, $efile, '>>')) {
            print EFILE_FILE $entry, "\n" or Warn("Error writing to $efile\n"), $err = 1;
            close EFILE_FILE;
        } else {
            Warn("Error opening '${efile}' for append\n");
            $err = 1;
        }
        if ($err) {
            defined $_ and $_ eq $efile and undef $_ foreach @efile;
        }
    }
}

#------------------------------------------------------------------------------
# Print progress message if it is time for it
# Inputs: 0) file ref, 1) message
sub Progress($$)
{
    my ($file, $msg) = @_;
    if (defined $progStr) {
        print $file $msg, $progStr, "\n";
        undef $progressNext if defined $progressMax;
    }
}

#------------------------------------------------------------------------------
# Print list of tags
# Inputs: 0) message, 1-N) list of tag names
sub PrintTagList($@)
{
    my $msg = shift;
    print $msg, ":\n" unless $quiet;
    my $tag;
    if ($outFormat < 0 and $msg =~ /file extensions$/ and @_) {
        foreach $tag (@_) {
            printf("  %-11s %s\n", $tag, GetFileType($tag, 1));
        }
        return;
    }
    my ($len, $pad) = (0, $quiet ? '' : '  ');
    foreach $tag (@_) {
        my $taglen = length($tag);
        if ($len + $taglen > 77) {
            print "\n";
            ($len, $pad) = (0, $quiet ? '' : '  ');
        }
        print $pad, $tag;
        $len += $taglen + 1;
        $pad = ' ';
    }
    @_ or print $pad, '[empty list]';
    print "\n";
}

#------------------------------------------------------------------------------
# Print warnings and errors from info hash
# Inputs: 0) ExifTool object ref, 1) info hash, 2) file name
# Returns: true if there was an Error
sub PrintErrors($$$)
{
    my ($et, $info, $file) = @_;
    my ($tag, $key);
    foreach $tag (qw(Warning Error)) {
        next unless $$info{$tag};
        my @keys = ( $tag );
        push @keys, sort(grep /^$tag /, keys %$info) if $et->Options('Duplicates');
        foreach $key (@keys) {
            Warn "$tag: $info->{$key} - $file\n";
        }
    }
    return $$info{Error};
}

__END__

=head1 NAME

exiftool - Read and write meta information in files

=head1 SYNOPSIS

=head2 Reading

B<exiftool> [I<OPTIONS>] [-I<TAG>...] [--I<TAG>...] I<FILE>...

=head2 Writing

B<exiftool> [I<OPTIONS>] -I<TAG>[+-E<lt>]=[I<VALUE>]... I<FILE>...

=head2 Copying

B<exiftool> [I<OPTIONS>] B<-tagsFromFile> I<SRCFILE>
[-[I<DSTTAG>E<lt>]I<SRCTAG>...] I<FILE>...

=head2 Other

B<exiftool> [ B<-ver> |
B<-list>[B<w>|B<f>|B<r>|B<wf>|B<g>[I<NUM>]|B<d>|B<x>|B<geo>] ]

For specific examples, see the L<EXAMPLES|/READING EXAMPLES> sections below.

This documentation is displayed if exiftool is run without an input I<FILE>
when one is expected.

=head1 DESCRIPTION

A command-line interface to L<Image::ExifTool|Image::ExifTool>, used for
reading and writing meta information in a variety of file types.  I<FILE> is
one or more source file names, directory names, or C<-> for the standard
input.  Metadata is read from source files and printed in readable form to
the console (or written to output text files with B<-w>).

To write or delete metadata, tag values are assigned using
-I<TAG>=[I<VALUE>], and/or the B<-geotag>, B<-csv=> or B<-json=> options. 
To copy or move metadata, the B<-tagsFromFile> feature is used.  By default
the original files are preserved with C<_original> appended to their names
-- be sure to verify that the new files are OK before erasing the originals.
Once in write mode, exiftool will ignore any read-specific options.

Note:  If I<FILE> is a directory name then only supported file types in the
directory are processed (in write mode only writable types are processed).
However, files may be specified by name, or the B<-ext> option may be used
to force processing of files with any extension.  Hidden files in the
directory are also processed.  Adding the B<-r> option causes subdirectories
to be processed recursively, but subdirectories with names beginning with
"." are skipped unless B<-r.> is used.

Below is a list of file types and meta information formats currently
supported by ExifTool (r = read, w = write, c = create):

  File Types
  ------------+-------------+-------------+-------------+------------
  360   r/w   | DOCX  r     | ITC   r     | NUMBERS r   | RAW   r/w
  3FR   r     | DPX   r     | J2C   r     | NXD   r     | RIFF  r
  3G2   r/w   | DR4   r/w/c | JNG   r/w   | O     r     | RSRC  r
  3GP   r/w   | DSS   r     | JP2   r/w   | ODP   r     | RTF   r
  7Z    r     | DV    r     | JPEG  r/w   | ODS   r     | RW2   r/w
  A     r     | DVB   r/w   | JSON  r     | ODT   r     | RWL   r/w
  AA    r     | DVR-MS r    | JXL   r/w   | OFR   r     | RWZ   r
  AAC   r     | DYLIB r     | K25   r     | OGG   r     | RM    r
  AAE   r     | EIP   r     | KDC   r     | OGV   r     | SEQ   r
  AAX   r/w   | EPS   r/w   | KEY   r     | ONP   r     | SKETCH r
  ACR   r     | EPUB  r     | LA    r     | OPUS  r     | SO    r
  AFM   r     | ERF   r/w   | LFP   r     | ORF   r/w   | SR2   r/w
  AI    r/w   | EXE   r     | LIF   r     | ORI   r/w   | SRF   r
  AIFF  r     | EXIF  r/w/c | LNK   r     | OTF   r     | SRW   r/w
  APE   r     | EXR   r     | LRV   r/w   | PAC   r     | SVG   r
  ARQ   r/w   | EXV   r/w/c | M2TS  r     | PAGES r     | SWF   r
  ARW   r/w   | F4A/V r/w   | M4A/V r/w   | PBM   r/w   | THM   r/w
  ASF   r     | FFF   r/w   | MACOS r     | PCD   r     | TIFF  r/w
  AVI   r     | FITS  r     | MAX   r     | PCX   r     | TORRENT r
  AVIF  r/w   | FLA   r     | MEF   r/w   | PDB   r     | TTC   r
  AZW   r     | FLAC  r     | MIE   r/w/c | PDF   r/w   | TTF   r
  BMP   r     | FLIF  r/w   | MIFF  r     | PEF   r/w   | TXT   r
  BPG   r     | FLV   r     | MKA   r     | PFA   r     | VCF   r
  BTF   r     | FPF   r     | MKS   r     | PFB   r     | VNT   r
  C2PA  r     | FPX   r     | MKV   r     | PFM   r     | VRD   r/w/c
  CHM   r     | GIF   r/w   | MNG   r/w   | PGF   r     | VSD   r
  COS   r     | GLV   r/w   | MOBI  r     | PGM   r/w   | WAV   r
  CR2   r/w   | GPR   r/w   | MODD  r     | PLIST r     | WDP   r/w
  CR3   r/w   | GZ    r     | MOI   r     | PICT  r     | WEBP  r/w
  CRM   r/w   | HDP   r/w   | MOS   r/w   | PMP   r     | WEBM  r
  CRW   r/w   | HDR   r     | MOV   r/w   | PNG   r/w   | WMA   r
  CS1   r/w   | HEIC  r/w   | MP3   r     | PPM   r/w   | WMV   r
  CSV   r     | HEIF  r/w   | MP4   r/w   | PPT   r     | WPG   r
  CUR   r     | HTML  r     | MPC   r     | PPTX  r     | WTV   r
  CZI   r     | ICC   r/w/c | MPG   r     | PS    r/w   | WV    r
  DCM   r     | ICO   r     | MPO   r/w   | PSB   r/w   | X3F   r/w
  DCP   r/w   | ICS   r     | MQV   r/w   | PSD   r/w   | XCF   r
  DCR   r     | IDML  r     | MRC   r     | PSP   r     | XISF  r
  DFONT r     | IIQ   r/w   | MRW   r/w   | QTIF  r/w   | XLS   r
  DIVX  r     | IND   r/w   | MXF   r     | R3D   r     | XLSX  r
  DJVU  r     | INSP  r/w   | NEF   r/w   | RA    r     | XMP   r/w/c
  DLL   r     | INSV  r     | NKA   r     | RAF   r/w   | ZIP   r
  DNG   r/w   | INX   r     | NKSC  r/w   | RAM   r     |
  DOC   r     | ISO   r     | NRW   r/w   | RAR   r     |

  Meta Information
  ----------------------+----------------------+---------------------
  EXIF           r/w/c  |  CIFF           r/w  |  Ricoh RMETA    r
  GPS            r/w/c  |  AFCP           r/w  |  Picture Info   r
  IPTC           r/w/c  |  Kodak Meta     r/w  |  Adobe APP14    r
  XMP            r/w/c  |  FotoStation    r/w  |  MPF            r
  MakerNotes     r/w/c  |  PhotoMechanic  r/w  |  Stim           r
  Photoshop IRB  r/w/c  |  JPEG 2000      r    |  DPX            r
  ICC Profile    r/w/c  |  DICOM          r    |  APE            r
  MIE            r/w/c  |  Flash          r    |  Vorbis         r
  JFIF           r/w/c  |  FlashPix       r    |  SPIFF          r
  Ducky APP12    r/w/c  |  QuickTime      r    |  DjVu           r
  PDF            r/w/c  |  Matroska       r    |  M2TS           r
  PNG            r/w/c  |  MXF            r    |  PE/COFF        r
  Canon VRD      r/w/c  |  PrintIM        r    |  AVCHD          r
  Nikon Capture  r/w/c  |  FLAC           r    |  ZIP            r
  GeoTIFF        r/w/c  |  ID3            r    |  (and more)

=head1 OPTIONS

Case is not significant for any command-line option (including tag and group
names), except for single-character options when the corresponding
upper-case option exists.  Many single-character options have equivalent
long-name versions (shown in brackets), and some options have inverses which
are invoked with a leading double-dash.  Unrecognized options are
interpreted as tag names (for this reason, multiple single-character options
may NOT be combined into one argument).  Contrary to standard practice,
options may appear after source file names on the exiftool command line.

=head2 Option Overview

L<Tag operations|/Tag operations>

  -TAG or --TAG                    Extract or exclude specified tag
  -TAG[+-^]=[VALUE]                Write new value for tag
  -TAG[+-]<=DATFILE                Write tag value from contents of file
  -[+]TAG[+-]<SRCTAG               Copy tag value (see -tagsFromFile)

  -tagsFromFile SRCFILE            Copy tag values from file
  -x TAG      (-exclude)           Exclude specified tag

L<Input-output text formatting|/Input-output text formatting>

  -args       (-argFormat)         Format metadata as exiftool arguments
  -b          (-binary)            Output metadata in binary format
  -c FMT      (-coordFormat)       Set format for GPS coordinates
  -charset [[TYPE=]CHARSET]        Specify encoding for special characters
  -csv[[+]=CSVFILE]                Export/import tags in CSV format
  -csvDelim STR                    Set delimiter for CSV file
  -d FMT      (-dateFormat)        Set format for date/time values
  -D          (-decimal)           Show tag ID numbers in decimal
  -E,-ex,-ec  (-escape(HTML|XML|C))Escape tag values for HTML, XML or C
  -f          (-forcePrint)        Force printing of all specified tags
  -g[NUM...]  (-groupHeadings)     Organize output by tag group
  -G[NUM...]  (-groupNames)        Print group name for each tag
  -h          (-htmlFormat)        Use HTML formatting for output
  -H          (-hex)               Show tag ID numbers in hexadecimal
  -htmlDump[OFFSET]                Generate HTML-format binary dump
  -j[[+]=JSONFILE] (-json)         Export/import tags in JSON format
  -l          (-long)              Use long 2-line output format
  -L          (-latin)             Use Windows Latin1 encoding
  -lang [LANG]                     Set current language
  -listItem INDEX                  Extract specific item from a list
  -n          (--printConv)        No print conversion
  -p[-] STR   (-printFormat)       Print output in specified format
  -php                             Export tags as a PHP Array
  -s[NUM]     (-short)             Short output format (-s for tag names)
  -S          (-veryShort)         Very short output format
  -sep STR    (-separator)         Set separator string for list items
  -sort                            Sort output alphabetically
  -struct                          Enable output of structured information
  -t          (-tab)               Output in tab-delimited list format
  -T          (-table)             Output in tabular format
  -v[NUM]     (-verbose)           Print verbose messages
  -w[+|!] EXT (-textOut)           Write (or overwrite!) output text files
  -W[+|!] FMT (-tagOut)            Write output text file for each tag
  -Wext EXT   (-tagOutExt)         Write only specified file types with -W
  -X          (-xmlFormat)         Use RDF/XML output format

L<Processing control|/Processing control>

  -a          (-duplicates)        Allow duplicate tags to be extracted
  -e          (--composite)        Do not generate composite tags
  -ee[NUM]    (-extractEmbedded)   Extract information from embedded files
  -ext[+] EXT (-extension)         Process files with specified extension
  -F[OFFSET]  (-fixBase)           Fix the base for maker notes offsets
  -fast[NUM]                       Increase speed when extracting metadata
  -fileOrder[NUM] [-]TAG           Set file processing order
  -i DIR      (-ignore)            Ignore specified directory name
  -if[NUM] EXPR                    Conditionally process files
  -m          (-ignoreMinorErrors) Ignore minor errors and warnings
  -o OUTFILE  (-out)               Set output file or directory name
  -overwrite_original              Overwrite original by renaming tmp file
  -overwrite_original_in_place     Overwrite original by copying tmp file
  -P          (-preserve)          Preserve file modification date/time
  -password PASSWD                 Password for processing protected files
  -progress[NUM][:[TITLE]]         Show file progress count
  -q          (-quiet)             Quiet processing
  -r[.]       (-recurse)           Recursively process subdirectories
  -scanForXMP                      Brute force XMP scan
  -u          (-unknown)           Extract unknown tags
  -U          (-unknown2)          Extract unknown binary tags too
  -wm MODE    (-writeMode)         Set mode for writing/creating tags
  -z          (-zip)               Read/write compressed information

L<Other options|/Other options>

  -@ ARGFILE                       Read command-line arguments from file
  -k          (-pause)             Pause before terminating
  -list[w|f|wf|g[NUM]|d|x]         List various exiftool capabilities
  -ver                             Print exiftool version number
  --                               End of options

L<Special features|/Special features>

  -geotag TRKFILE                  Geotag images from specified GPS log
  -globalTimeShift SHIFT           Shift all formatted date/time values
  -use MODULE                      Add features from plug-in module

L<Utilities|/Utilities>

  -delete_original[!]              Delete "_original" backups
  -restore_original                Restore from "_original" backups

L<Advanced options|/Advanced options>

  -api OPT[[^]=[VAL]]              Set ExifTool API option
  -common_args                     Define common arguments
  -config CFGFILE                  Specify configuration file name
  -echo[NUM] TEXT                  Echo text to stdout or stderr
  -efile[NUM][!] TXTFILE           Save names of files with errors
  -execute[NUM]                    Execute multiple commands on one line
  -fileNUM ALTFILE                 Load tags from alternate file
  -list_dir                        List directories, not their contents
  -srcfile FMT                     Process a different source file
  -stay_open FLAG                  Keep reading -@ argfile even after EOF
  -userParam PARAM[[^]=[VAL]]      Set user parameter (API UserParam opt)

=head2 Option Details

=head3 Tag operations

=over 5

=item B<->I<TAG>

Extract information for the specified tag (eg. C<-CreateDate>).  Multiple
tags may be specified in a single command.  A tag name is the handle by
which a piece of information is referenced.  See
L<Image::ExifTool::TagNames|Image::ExifTool::TagNames> for documentation on
available tag names.  A tag name may include leading group names separated
by colons (eg. C<-EXIF:CreateDate>, or C<-Doc1:XMP:Creator>), and each group
name may be prefixed by a digit to specify family number (eg.
C<-1IPTC:City>).  (Note that the API SavePath and SaveFormat options must be
used for the family 5 and 6 groups respectively to be available.)  Use the
B<-listg> option to list available group names by family.

A special tag name of C<All> may be used to indicate all meta information
(ie. B<-All>).  This is particularly useful when a group name is specified
to extract all information in a group (but beware that unless the B<-a>
option is also used, some tags in the group may be suppressed by same-named
tags in other groups).  The wildcard characters C<?> and C<*> may be used in
a tag name to match any single character and zero or more characters
respectively.  These may not be used in a group name, with the exception
that a group name of C<*> (or C<All>) may be used to extract all instances
of a tag (as if B<-a> was used).  Note that arguments containing wildcards
must be quoted on the command line of most systems to prevent shell
globbing.

A C<#> may be appended to the tag name to disable the print conversion on a
per-tag basis (see the B<-n> option).  This may also be used when writing or
copying tags.

If no tags are specified, all available information is extracted (as if
C<-All> had been specified).

Note:  Descriptions, not tag names, are shown by default when extracting
information.  Use the B<-s> option to see the tag names instead.

=item B<-->I<TAG>

Exclude specified tag from extracted information.  Same as the B<-x> option.
Group names and wildcards are permitted as described above for B<-TAG>. 
Once excluded from the output, a tag may not be re-included by a subsequent
option.  May also be used following a B<-tagsFromFile> option to exclude
tags from being copied (when redirecting to another tag, it is the source
tag that should be excluded), or to exclude groups from being deleted when
deleting all information (eg. C<-all= --exif:all> deletes all but EXIF
information).  But note that this will not exclude individual tags from a
group delete (unless a family 2 group is specified, see note 4 below).
Instead, individual tags may be recovered using the B<-tagsFromFile> option
(eg. C<-all= -tagsfromfile @ -artist>).

To speed processing when reading XMP, exclusions in XMP groups also bypass
processing of the corresponding XMP property and any contained properties.
For example, C<--xmp-crs:all> may speed processing significantly in cases
where a large number of XMP-crs tags exist.  To use this feature to bypass
processing of a specific XMP property, the property name must be used
instead of the ExifTool tag name (eg. C<--xmp-crs:dabs>).  Also, C<XMP-all>
may be used to to indicate any XMP namespace (eg. C<--xmp-all:dabs>).

=item B<->I<TAG>[+-^]B<=>[I<VALUE>]

Write a new value for the specified tag (eg. C<-comment=wow>), or delete the
tag if no I<VALUE> is given (eg. C<-comment=>).  C<+=> and C<-=> are used to
add or remove existing entries from a list, or to shift date/time values
(see L<Image::ExifTool::Shift.pl|Image::ExifTool::Shift.pl> and note 6 below
for more details).  C<+=> may also be used to increment numerical values (or
decrement if I<VALUE> is negative), and C<-=> may be used to conditionally
delete or replace a tag (see L</WRITING EXAMPLES> for examples).  C<^=> is
used to write an empty string instead of deleting the tag when no I<VALUE>
is given, but otherwise it is equivalent to C<=>.  (Note that the caret must
be quoted on the Windows command line.)

I<TAG> may contain one or more leading family 0, 1, 2 or 7 group names,
prefixed by optional family numbers, and separated colons.  If no group name
is specified, the tag is created in the preferred group, and updated in any
other location where a same-named tag already exists.  The preferred group
in JPEG and TIFF-format images is the first group in the following list
where I<TAG> is valid: 1) EXIF, 2) IPTC, 3) XMP.

The wildcards C<*> and C<?> may be used in tag names to assign the same
value to multiple tags.  When specified with wildcards, "Unsafe" tags are
not written.  A tag name of C<All> is equivalent to C<*> (except that it
doesn't require quoting, while arguments with wildcards do on systems with
shell globbing), and is often used when deleting all metadata (ie. C<-All=>)
or an entire group (eg. C<-XMP-dc:All=>, see note 4 below).  Note that not
all groups are deletable, and that the JPEG APP14 "Adobe" group is not
removed by default with C<-All=> because it may affect the appearance of the
image.  However, color space information is removed, so the colors may be
affected (but this may be avoided by copying back the tags defined by the
ColorSpaceTags shortcut).  Use the B<-listd> option for a complete list of
deletable groups, and see note 5 below regarding the "APP" groups.  Also,
within an image some groups may be contained within others, and these groups
are removed if the containing group is deleted:

  JPEG Image:
  - Deleting EXIF or IFD0 also deletes ExifIFD, GlobParamIFD,
    GPS, IFD1, InteropIFD, MakerNotes, PrintIM and SubIFD.
  - Deleting ExifIFD also deletes InteropIFD and MakerNotes.
  - Deleting Photoshop also deletes IPTC.

  TIFF Image:
  - Deleting EXIF only removes ExifIFD which also deletes
    InteropIFD and MakerNotes.

  MOV/MP4 Video:
  - Deleting ItemList also deletes Keys tags.

Notes:

1) B<Many tag values may be assigned in a single command>.  If two
assignments affect the same tag, the latter takes precedence (except for
list-type tags, for which both values are written).

2) In general, MakerNotes tags are considered "Permanent", and may be edited
but not created or deleted individually.  This avoids many potential
problems, including the inevitable compatibility problems with OEM software
which may be very inflexible about the information it expects to find in the
maker notes.

3) Changes to PDF files by ExifTool are reversible (by deleting the update
with C<-PDF-update:all=>) because the original information is never actually
deleted from the file.  So ExifTool alone may not be used to securely edit
metadata in PDF files.

4) Specifying C<-GROUP:all=> deletes the entire group as a block only if a
single family 0 or 1 group is specified.  Otherwise all deletable tags in
the specified group(s) are removed individually, and in this case is it
possible to exclude individual tags from a mass delete.  For example,
C<-time:all --Exif:Time:All> removes all deletable Time tags except those in
the EXIF.  This difference also applies if family 2 is specified when
deleting all groups.  For example, C<-2all:all=> deletes tags individually,
while C<-all:all=> deletes entire blocks.

5) The "APP" group names ("APP0" through "APP15") are used to delete JPEG
application segments which are not associated with another deletable group. 
For example, specifying C<-APP14:All=> will NOT delete the APP14 "Adobe"
segment because this is accomplished with C<-Adobe:All>.  But note that
these unnamed APP segments may not be excluded with C<--APPxx:all> when
deleting all information.

6) When shifting a value, the shift is applied to the original value of the
tag, overriding any other values previously assigned to the tag on the same
command line.  To shift a date/time value and copy it to another tag in the
same operation, use the B<-globalTimeShift> option.

Special feature:  Integer values may be specified in hexadecimal with a
leading C<0x>, and simple rational values may be specified as fractions.

=item B<->I<TAG>E<lt>=I<DATFILE> or B<->I<TAG>E<lt>=I<FMT>

Set the value of a tag from the contents of file I<DATFILE>.  The file name
may also be given by a I<FMT> string where %d, %f and %e represent the
directory, file name and extension of the original I<FILE> (see the B<-w>
option for more details).  Note that quotes are required around this
argument to prevent shell redirection since it contains a C<E<lt>> symbol.
If I<DATFILE>/I<FMT> is not provided, the effect is the same as C<-TAG=>,
and the tag is simply deleted.  C<+E<lt>=> or C<-E<lt>=> may also be used to
add or delete specific list entries, or to shift date/time values.

=item B<-tagsFromFile> I<SRCFILE> or I<FMT>

Copy tag values from I<SRCFILE> to I<FILE>.  Tag names on the command line
after this option specify the tags to be copied, or excluded from the copy.
Wildcards are permitted in these tag names.  If no tags are specified, then
all possible tags (see note 1 below) from the source file are copied to
same-named tags in the preferred location of the output file (the same as
specifying C<-all>).  More than one B<-tagsFromFile> option may be used to
copy tags from multiple files.

By default, this option will update any existing and writable same-named
tags in the output I<FILE>, but will create new tags only in their preferred
groups.  This allows some information to be automatically transferred to the
appropriate group when copying between images of different formats. However,
if a group name is specified for a tag then the information is written only
to this group (unless redirected to another group, see below).  If C<All> is
used as a group name, then the specified tag(s) are written to the same
family 1 group they had in the source file (ie. the same specific location,
like ExifIFD or XMP-dc).  For example, the common operation of copying all
writable tags to the same specific locations in the output I<FILE> is
achieved by adding C<-all:all>.  A different family may be specified by
adding a leading family number to the group name (eg. C<-0all:all> preserves
the same general location, like EXIF or XMP).

I<SRCFILE> may be the same as I<FILE> to move information around within a
single file.  In this case, C<@> may be used to represent the source file
(ie. C<-tagsFromFile @>), permitting this feature to be used for batch
processing multiple files.  Specified tags are then copied from each file in
turn as it is rewritten.  For advanced batch use, the source file name may
also be specified using a I<FMT> string in which %d, %f and %e represent the
directory, file name and extension of I<FILE>.  (eg. the current I<FILE>
would be represented by C<%d%f.%e>, with the same effect as C<@>).  See the
B<-w> option for I<FMT> string examples.

A powerful redirection feature allows a destination tag to be specified for
each copied tag.  With this feature, information may be written to a tag
with a different name or group.  This is done using
E<quot>'-I<DSTTAG>E<lt>I<SRCTAG>'E<quot> or
E<quot>'-I<SRCTAG>E<gt>I<DSTTAG>'E<quot> on the command line after
B<-tagsFromFile>, and causes the value of I<SRCTAG> to be copied from
I<SRCFILE> and written to I<DSTTAG> in I<FILE>.  Has no effect unless
I<SRCTAG> exists in I<SRCFILE>.  Note that this argument must be quoted to
prevent shell redirection, and there is no C<=> sign as when assigning new
values.  Source and/or destination tags may be prefixed by a group name
and/or suffixed by C<#>.  Wildcards are allowed in both the source and
destination tag names.  A destination group and/or tag name of C<All> or
C<*> writes to the same family 1 group and/or tag name as the source (but
the family may be specified by adding a leading number to the group name,
eg. C<0All> writes to the same family 0 group as the source).  If no
destination group is specified, the information is written to the preferred
group.  Whitespace around the C<E<gt>> or C<E<lt>> is ignored. As a
convenience, C<-tagsFromFile @> is assumed for any redirected tags which are
specified without a prior B<-tagsFromFile> option.  Copied tags may also be
added or deleted from a list with arguments of the form
E<quot>'-I<SRCTAG>+E<lt>I<DSTTAG>'E<quot> or
E<quot>'-I<SRCTAG>-E<lt>I<DSTTAG>'E<quot> (but see Note 5 below).

An extension of the redirection feature allows strings involving tag names
to be used on the right hand side of the C<E<lt>> symbol with the syntax
E<quot>'-I<DSTTAG>E<lt>I<STR>'E<quot>, where tag names in I<STR> are
prefixed with a C<$> symbol.  See the B<-p> option and the
L</Advanced formatting feature> section for more details about this syntax. 
Strings starting with a C<=> sign must insert a single space after the
C<E<lt>> to avoid confusion with the C<E<lt>=> operator which sets the tag
value from the contents of a file.  A single space at the start of the
string is removed if it exists, but all other whitespace in the string is
preserved.  See note 8 below about using the redirection feature with
list-type stags, shortcuts or when using wildcards in tag names.

See L</COPYING EXAMPLES> for examples using B<-tagsFromFile>.

Notes:

1) Some tags (generally tags which may affect the appearance of the image)
are considered "Unsafe" to write, and are only copied if specified
explicitly (ie. no wildcards).  See the
L<tag name documentation|Image::ExifTool::TagNames> for more details about
"Unsafe" tags.

2) Be aware of the difference between excluding a tag from being copied
(--I<TAG>), and deleting a tag (-I<TAG>=).  Excluding a tag prevents it from
being copied to the destination image, but deleting will remove a
pre-existing tag from the image.

3) The maker note information is copied as a block, so it isn't affected
like other information by subsequent tag assignments on the command line,
and individual makernote tags may not be excluded from a block copy.  Also,
since the PreviewImage referenced from the maker notes may be rather large,
it is not copied, and must be transferred separately if desired.

4) The order of operations is to copy all specified tags at the point of the
B<-tagsFromFile> option in the command line.  Any tag assignment to the
right of the B<-tagsFromFile> option is made after all tags are copied.  For
example, new tag values are set in the order One, Two, Three then Four with
this command:

    exiftool -One=1 -tagsFromFile s.jpg -Two -Four=4 -Three d.jpg

This is significant in the case where an overlap exists between the copied
and assigned tags because later operations may override earlier ones.

5) The normal behaviour of copied tags differs from that of assigned tags
for list-type tags and conditional replacements because each copy operation
on a tag overrides any previous operations.  While this avoids duplicate
list items when copying groups of tags from a file containing redundant
information, it also prevents values of different tags from being copied
into the same list when this is the intent.  To accumulate values
from different operations into the same list, add a C<+> after the initial
C<-> of the argument.  For example:

    exiftool -tagsfromfile @ '-subject<make' '-+subject<model' ...

Similarly, C<-+DSTTAG> must be used when conditionally replacing a tag to
prevent overriding earlier conditions.

6) The B<-a> option (allow duplicate tags) is always in effect when copying
tags from I<SRCFILE>, but the highest priority tag is always copied last so
it takes precedence.

7) Structured tags are copied by default when copying tags.  See the
B<-struct> option for details.

8) With the redirection feature, copying a tag directly (ie.
E<quot>'-I<DSTTAG>E<lt>I<SRCTAG>'E<quot>) is not the same as interpolating
its value inside a string (ie. E<quot>'-I<DSTTAG>E<lt>$I<SRCTAG>'E<quot>)
for source tags which are list-type tags,
L<shortcut tags|Image::ExifTool::Shortcuts>, tag names containing wildcards,
or UserParam variables.  When copying directly, the values of each matching
source tag are copied individually to the destination tag (as if they were
separate assignments).  However, when interpolated inside a string, list
items and the values of shortcut tags are concatenated (with a separator set
by the B<-sep> option), and wildcards are not allowed.  Also, UserParam
variables are available only when interpolated in a string.  Another
difference is that a minor warning is generated if a tag doesn't exist when
interpolating its value in a string (with C<$>), but isn't when copying the
tag directly.

Finally, the behaviour is different when a destination tag or group of
C<All> is used.  When copying directly, a destination group and/or tag name
of C<All> writes to the same family 1 group and/or tag name as the source. 
But when interpolated in a string, the identity of the source tags are lost
and the value is written to all possible groups/tags.  For example, the
string form must be used in the following command since the intent is to set
the value of all existing date/time tags from C<CreateDate>:

    exiftool '-time:all<$createdate' -wm w FILE

=item B<-x> I<TAG> (B<-exclude>)

Exclude the specified tag.  There may be multiple B<-x> options.  This has
the same effect as --I<TAG> on the command line.  See the --I<TAG>
documentation above for a complete description.

=back

=head3 Input-output text formatting

Note that trailing spaces are removed from extracted values for most output
text formats.  The exceptions are B<-b>, B<-csv>, B<-j> and B<-X>.

=over 5

=item B<-args> (B<-argFormat>)

Output information in the form of exiftool arguments, suitable for use with
the B<-@> option when writing.  May be combined with the B<-G> option to
include group names.  This feature may be used to effectively copy tags
between images, but allows the metadata to be altered by editing the
intermediate file (C<out.args> in this example):

    exiftool -args -G1 --filename --directory src.jpg > out.args
    exiftool -@ out.args -sep ', ' dst.jpg

Note:  Be careful when copying information with this technique since it is
easy to write tags which are normally considered "Unsafe".  For instance,
the FileName and Directory tags are excluded in the example above to avoid
renaming and moving the destination file.  Also note that the second command
above will produce warning messages for any tags which are not writable.

As well, the B<-sep> option should be used as in the second command above to
maintain separate list items when writing metadata back to image files, and
the B<-struct> option may be used when extracting to preserve structured XMP
information.

=item B<-b>, B<--b> (B<-binary>, B<--binary>)

Output requested metadata in binary format without tag names or descriptions
(B<-b> or B<-binary>).  This option is mainly used for extracting embedded
images or other binary data, but it may also be useful for some text strings
since control characters (such as newlines) are not replaced by '.' as they
are in the default output.  By default, list items are separated by a
newline when extracted with the B<-b> option, but this may be changed (see
the B<-sep> option for details).  May be combined with B<-j>, B<-php> or
B<-X> to extract binary data in JSON, PHP or XML format, but note that
"Unsafe" tags are not extracted as binary unless they are specified
explicitly or the API RequestAll option is set to 3 or higher.

With a leading double dash (B<--b> or B<--binary>), tags which contain
binary data are suppressed in the output when reading.

=item B<-c> I<FMT> (B<-coordFormat>)

Set the print format for GPS coordinates.  I<FMT> uses the same syntax as
a C<printf> format string.  The specifiers correspond to degrees, minutes
and seconds in that order, but minutes and seconds are optional.  For
example, the following table gives the output for the same coordinate using
various formats:

            FMT                  Output
    -------------------    ------------------
    "%d deg %d' %.2f"\"    54 deg 59' 22.80"  (default for reading)
    "%d %d %.8f"           54 59 22.80000000  (default for copying)
    "%d deg %.4f min"      54 deg 59.3800 min
    "%.6f degrees"         54.989667 degrees

Notes:

1) To avoid loss of precision, the default coordinate format is different
when copying tags using the B<-tagsFromFile> option.

2) If the hemisphere is known, a reference direction (N, S, E or W) is
appended to each printed coordinate, but adding a C<+> or C<-> to the format
specifier (eg. C<%+.6f> or C<%-.6f>) prints a signed coordinate instead. 
(C<+> adds a leading "+" for positive coordinates, but C<-> does not.)

3) This print formatting may be disabled with the B<-n> option to extract
coordinates as signed decimal degrees.

=item B<-charset> [[I<TYPE>=]I<CHARSET>]

If I<TYPE> is C<ExifTool> or not specified, this option sets the ExifTool
character encoding for output tag values when reading and input values when
writing, with a default of C<UTF8>.  If no I<CHARSET> is given, a list of
available character sets is returned.  Valid I<CHARSET> values are:

    CHARSET     Alias(es)        Description
    ----------  ---------------  ----------------------------------
    UTF8        cp65001, UTF-8   UTF-8 characters (default)
    Latin       cp1252, Latin1   Windows Latin1 (West European)
    Latin2      cp1250           Windows Latin2 (Central European)
    Cyrillic    cp1251, Russian  Windows Cyrillic
    Greek       cp1253           Windows Greek
    Turkish     cp1254           Windows Turkish
    Hebrew      cp1255           Windows Hebrew
    Arabic      cp1256           Windows Arabic
    Baltic      cp1257           Windows Baltic
    Vietnam     cp1258           Windows Vietnamese
    Thai        cp874            Windows Thai
    DOSLatinUS  cp437            DOS Latin US
    DOSLatin1   cp850            DOS Latin1
    DOSCyrillic cp866            DOS Cyrillic
    MacRoman    cp10000, Roman   Macintosh Roman
    MacLatin2   cp10029          Macintosh Latin2 (Central Europe)
    MacCyrillic cp10007          Macintosh Cyrillic
    MacGreek    cp10006          Macintosh Greek
    MacTurkish  cp10081          Macintosh Turkish
    MacRomanian cp10010          Macintosh Romanian
    MacIceland  cp10079          Macintosh Icelandic
    MacCroatian cp10082          Macintosh Croatian

I<TYPE> may be C<FileName> to specify the encoding of file names on the
command line (ie. I<FILE> arguments).  In Windows, this triggers use of
wide-character i/o routines, thus providing support for Unicode file names.
See the L</WINDOWS UNICODE FILE NAMES> section below for details.

Other values of I<TYPE> listed below are used to specify the internal
encoding of various meta information formats.

    TYPE       Description                                  Default
    ---------  -------------------------------------------  -------
    EXIF       Internal encoding of EXIF "ASCII" strings    (none)
    ID3        Internal encoding of ID3v1 information       Latin
    IPTC       Internal IPTC encoding to assume when        Latin
                IPTC:CodedCharacterSet is not defined
    Photoshop  Internal encoding of Photoshop IRB strings   Latin
    QuickTime  Internal encoding of QuickTime strings       MacRoman
    RIFF       Internal encoding of RIFF strings            0

See L<https://exiftool.org/faq.html#Q10> for more information about coded
character sets, and the L<Image::ExifTool Options|Image::ExifTool/Options>
for more details about the B<-charset> settings.

=item B<-csv>[[+]=I<CSVFILE>]

Export information in CSV format, or import information if I<CSVFILE> is
specified.  When importing, the CSV file must be in exactly the same format
as the exported file.  The first row of the I<CSVFILE> must be the ExifTool
tag names (with optional group names) for each column of the file, and
values must be separated by commas.  A special "SourceFile" column specifies
the files associated with each row of information (and a SourceFile of "*"
may be used to define default tags to be imported for all files which are
combined with any tags specified for the specific SourceFile processed). The
B<-csvDelim> option may be used to change the input/output field delimiter
if something other than a comma is required.

The following examples demonstrate basic use of the B<-csv> option:

    # generate CSV file with common tags from all images in a directory
    exiftool -common -csv dir > out.csv

    # update metadata for all images in a directory from CSV file
    exiftool -csv=a.csv dir

When importing, empty values are ignored unless the B<-f> option is used and
the API MissingTagValue is set to an empty string (in which case the tag is
deleted).  Also, FileName and Directory columns are ignored if they exist
(ie. ExifTool will not attempt to write these tags with a CSV import), but
all other columns are imported.  To force a tag to be deleted, use the B<-f>
option and set the value to "-" in the CSV file (or to the MissingTagValue
if this API option was used).  Multiple databases may be imported in a
single command.

When exporting a CSV file, the B<-g> or B<-G> option adds group names to the
tag headings.  If the B<-a> option is used to allow duplicate tag names, the
duplicate tags are only included in the CSV output if the column headings
are unique.  Adding the B<-G4> option ensures a unique column heading for
each tag.  The B<-b> option may be added to output binary data, encoded in
base64 if necessary (indicated by ASCII "base64:" as the first 7 bytes of
the value).  Values may also be encoded in base64 if the B<-charset> option
is used and the value contains invalid characters.

When exporting specific tags, the CSV columns are arranged in the same order
as the specified tags provided the column headings exactly match the
specified tag names, otherwise the columns are sorted in alphabetical order.

When importing from a CSV file, only files specified on the command line are
processed.  Any extra entries in the CSV file are ignored.

List-type tags are stored as simple strings in a CSV file, but the B<-sep>
option may be used to split them back into separate items when importing.

Special feature:  B<-csv>+=I<CSVFILE> may be used to add items to existing
lists.  This affects only list-type tags.  Also applies to the B<-j> option.

Note that this option is fundamentally different than all other output
format options because it requires information from all input files to be
buffered in memory before the output is written.  This may result in
excessive memory usage when processing a very large number of files with a
single command.  Also, it makes this option incompatible with the B<-w> and
B<-W> options.  When processing a large number of files, it is recommended
to either use the JSON (B<-j>) or XML (B<-X>) output format, or use B<-p> to
generate a fixed-column CSV file instead of using the B<-csv> option.

=item B<-csvDelim> I<STR>

Set the delimiter for separating CSV entries for CSV file input/output via
the B<-csv> option.  I<STR> may contain "\t", "\n", "\r" and "\\" to
represent TAB, LF, CR and '\' respectively.  A double quote is not allowed
in the delimiter.  Default is ','.

=item B<-d> I<FMT> (B<-dateFormat>)

Set the format for date/time tag values.  The I<FMT> string may contain
formatting codes beginning with a percent character (C<%>) to represent the
various components of a date/time value.  The specifics of the I<FMT> syntax
are system dependent -- consult the C<strftime> man page on your system for
details.  The default format is equivalent to "%Y:%m:%d %H:%M:%S".  This
option has no effect on date-only or time-only tags and ignores timezone
information if present.  ExifTool adds a C<%f> format code to represent
fractional seconds, and supports an optional width to specify the number of
digits after the decimal point (eg. C<%3f> would give something like
C<.437>), and a minus sign to drop the decimal point (eg. C<%-3f> would give
C<437>).  Only one B<-d> option may be used per command.  Requires
POSIX::strptime or Time::Piece for the inversion conversion when writing.

=item B<-D> (B<-decimal>)

Show tag ID number in decimal when extracting information.

=item B<-E>, B<-ex>, B<-ec> (B<-escapeHTML>, B<-escapeXML>, B<-escapeC>)

Escape characters in output tag values for HTML (B<-E>), XML (B<-ex>) or C
(B<-ec>).  For HTML, all characters with Unicode code points above U+007F
are escaped as well as the following 5 characters: & (&amp;) E<39> (&#39;)
E<quot> (&quot;) E<gt> (&gt;) and E<lt> (&lt;).  For XML, only these 5
characters are escaped.  The B<-E> option is implied with B<-h>, and B<-ex>
is implied with B<-X>.  For C, all control characters and the backslash are
escaped.  The inverse conversion is applied when writing tags.

=item B<-f> (B<-forcePrint>)

Force printing of tags even if they don't exist.  This option applies to
tags specified on the command line, or with the B<-p>, B<-if> or
B<-tagsFromFile> options.  When B<-f> is used, the value of any missing tag
is set to a dash (C<->) by default, but this may be configured via the API
MissingTagValue option.  B<-f> is also used to add a 'flags' attribute to
the B<-listx> output, or to allow tags to be deleted when writing with the
B<-csv>=I<CSVFILE> feature.

=item B<-g>[I<NUM>][:I<NUM>...] (B<-groupHeadings>)

Organize output by tag group.  I<NUM> specifies a group family number, and
may be 0 (general location), 1 (specific location), 2 (category), 3
(document number), 4 (instance number), 5 (metadata path), 6 (EXIF/TIFF
format), 7 (tag ID) or 8 (file number).  B<-g0> is assumed if a family
number is not specified.  May be combined with other options to add group
names to the output.  Multiple families may be specified by separating them
with colons.  By default the resulting group name is simplified by removing
any leading C<Main:> and collapsing adjacent identical group names, but this
can be avoided by placing a colon before the first family number (eg.
B<-g:3:1>).  Use the B<-listg> option to list group names for a specified
family.  The API SavePath and SaveFormat options are automatically enabled
if the respective family 5 or 6 group names are requested.  See the
L<API GetGroup documentation|Image::ExifTool/GetGroup> for more information.

=item B<-G>[I<NUM>][:I<NUM>...] (B<-groupNames>)

Same as B<-g> but print group name for each tag.  B<-G0> is assumed if
I<NUM> is not specified.  May be combined with a number of other options to
add group names to the output.  Note that I<NUM> may be added wherever B<-G>
is mentioned in the documentation.  See the B<-g> option above for details.

=item B<-h> (B<-htmlFormat>)

Use HTML table formatting for output.  Implies the B<-E> option.  The
formatting options B<-D>, B<-H>, B<-g>, B<-G>, B<-l> and B<-s> may be used
in combination with B<-h> to influence the HTML format.

=item B<-H> (B<-hex>)

Show tag ID number in hexadecimal when extracting information.

=item B<-htmlDump>[I<OFFSET>]

Generate a dynamic web page containing a hex dump of the EXIF information.
This can be a very powerful tool for low-level analysis of EXIF information.
The B<-htmlDump> option is also invoked if the B<-v> and B<-h> options are
used together.  The verbose level controls the maximum length of the blocks
dumped.  An I<OFFSET> may be given to specify the base for displayed
offsets.  If not provided, the EXIF/TIFF base offset is used.  Use
B<-htmlDump0> for absolute offsets.  Currently only EXIF/TIFF and JPEG
information is dumped, but the -u option can be used to give a raw hex dump
of other file formats.

=item B<-j>[[+]=I<JSONFILE>] (B<-json>)

Use JSON (JavaScript Object Notation) formatting for console output, or
import JSON file if I<JSONFILE> is specified.  This option may be combined
with B<-g> to organize the output into objects by group, or B<-G> to add
group names to each tag.  List-type tags with multiple items are output as
JSON arrays unless B<-sep> is used.  By default XMP structures are flattened
into individual tags in the JSON output, but the original structure may be
preserved with the B<-struct> option (this also causes all list-type XMP
tags to be output as JSON arrays, otherwise single-item lists would be
output as simple strings).  The B<-a> option is implied when B<-json> is
used, but entries with identical JSON names are suppressed in the output. 
(B<-G4> may be used to ensure that all tags have unique JSON names.)  Adding
the B<-D> or B<-H> option changes tag values to JSON objects with "val" and
"id" fields, and adding B<-l> adds a "desc" field, and a "num" field if the
numerical value is different from the converted "val".  The B<-b> option may
be added to output binary data, encoded in base64 if necessary (indicated by
ASCII "base64:" as the first 7 bytes of the value), and B<-t> may be added
to include tag table information (see B<-t> for details).  The JSON output
is UTF-8 regardless of any B<-L> or B<-charset> option setting, but the
UTF-8 validation is disabled if a character set other than UTF-8 is
specified.  Note that ExifTool quotes JSON values only if they don't look
like numbers (regardless of the original storage format or the relevant
metadata specification).

If I<JSONFILE> is specified, the file is imported and the tag definitions
from the file are used to set tag values on a per-file basis.  The special
"SourceFile" entry in each JSON object associates the information with a
specific target file.  An object with a missing SourceFile or a SourceFile
of "*" defines default tags for all target files which are combined with any
tags specified for the specific SourceFile processed.  The imported JSON
file must have the same format as the exported JSON files with the exception
that options exporting JSON objects instead of simple values are not
compatible with the import file format (ie. export with B<-D>, B<-H>, B<-l>,
or B<-T> is not compatible, and use B<-G> instead of B<-g>).  Additionally,
tag names in the input JSON file may be suffixed with a C<#> to disable
print conversion.

Unlike CSV import, empty values are not ignored, and will cause an empty
value to be written if supported by the specific metadata type.  Tags are
deleted by using the B<-f> option and setting the tag value to "-" (or to
the MissingTagValue setting if this API option was used).  Importing with
B<-j>+=I<JSONFILE> causes new values to be added to existing lists.

=item B<-l> (B<-long>)

Use long 2-line Canon-style output format.  Adds a description and
unconverted value (if it is different from the converted value) to the XML,
JSON or PHP output when B<-X>, B<-j> or B<-php> is used.  May also be
combined with B<-listf>, B<-listr> or B<-listwf> to add descriptions of the
file types.

=item B<-L> (B<-latin>)

Use Windows Latin1 encoding (cp1252) for output tag values instead of the
default UTF-8.  When writing, B<-L> specifies that input text values are
Latin1 instead of UTF-8.  Equivalent to C<-charset latin>.

=item B<-lang> [I<LANG>]

Set current language for tag descriptions and converted values.  I<LANG> is
C<de>, C<fr>, C<ja>, etc.  Use B<-lang> with no other arguments to get a
list of available languages.  The default language is C<en> if B<-lang> is
not specified.  Note that tag/group names are always English, independent of
the B<-lang> setting, and translation of warning/error messages has not yet
been implemented.  May also be combined with B<-listx> to output
descriptions in one language only.

By default, ExifTool uses UTF-8 encoding for special characters, but the
B<-L> or B<-charset> option may be used to invoke other encodings.  Note
that ExifTool uses Unicode::LineBreak if available to help preserve the
column alignment of the plain text output for languages with a
variable-width character set.

Currently, the language support is not complete, but users are welcome to
help improve this by submitting their own translations.  To submit a
translation, follow these steps (you must have Perl installed for this):

1. Download and unpack the latest Image-ExifTool full distribution.

2. 'cd' into the Image-ExifTool directory.

3. Run this command to make an XML file of the desired tags (eg. EXIF):

   ./exiftool -listx -exif:all > out.xml

4. Copy this text into a file called 'import.pl' in the exiftool directory:

    push @INC, 'lib';
    require Image::ExifTool::TagInfoXML;
    my $file = shift or die "Expected XML file name\n";
    $Image::ExifTool::TagInfoXML::makeMissing = shift;
    Image::ExifTool::TagInfoXML::BuildLangModules($file,8);

5. Run the 'import.pl' script to Import the XML file, generating the
'MISSING' entries for your language (eg. Russian):

   perl import.pl out.xml ru

6. Edit the generated language module lib/Image/ExifTool/Lang/ru.pm, and
search and replace all 'MISSING' strings in the file with your translations.

7. Email the module ('ru.pm' in this example) to philharvey66 at gmail.com

8. Thank you!!

=item B<-listItem> I<INDEX>

For list-type tags, this causes only the item with the specified index to be
extracted.  I<INDEX> is 0 for the first item in the list.  Negative indices
may also be used to reference items from the end of the list.  Has no effect
on single-valued tags.  Also applies to tag values when copying from a tag,
and in B<-if> conditions.

=item B<-n> (B<--printConv>)

Disable print conversion for all tags.  By default, extracted values are
converted to a more human-readable format, but the B<-n> option disables
this conversion, revealing the machine-readable values.  For example:

    > exiftool -Orientation -S a.jpg
    Orientation: Rotate 90 CW
    > exiftool -Orientation -S -n a.jpg
    Orientation: 6

The print conversion may also be disabled on a per-tag basis by suffixing
the tag name with a C<#> character:

    > exiftool -Orientation# -Orientation -S a.jpg
    Orientation: 6
    Orientation: Rotate 90 CW

These techniques may also be used to disable the inverse print conversion
when writing.  For example, the following commands all have the same effect:

    > exiftool -Orientation='Rotate 90 CW' a.jpg
    > exiftool -Orientation=6 -n a.jpg
    > exiftool -Orientation#=6 a.jpg

=item B<-p>[-] I<STR> or I<FMTFILE> (B<-printFormat>)

Print output in the format specified by the given string or file.  The
argument is interpreted as a string unless a file of that name exists, in
which case the string is loaded from the contents of the file.  Tag names in
the format string or file begin with a C<$> symbol and may contain leading
group names and/or a trailing C<#> (to disable print conversion).  Case is
not significant.  Braces C<{}> may be used around the tag name to separate
it from subsequent text (and must be used if subsequent text begins with an
alphanumeric character, hyphen, underline, colon or number sign).  Use C<$$>
to represent a C<$> symbol, and C<$/> for a newline.  When the string
argument is used (ie. I<STR>), a newline is added to the end of the string
unless B<-p-> is specified or the B<-b> option is used.

Multiple B<-p> options may be used.  Lines beginning with C<#[HEAD]> and
C<#[TAIL]> are output before the first processed file and after the last
processed file respectively.  Lines beginning with C<#[SECT]> and C<#[ENDS]>
are output before and after each section of files.  A section is defined as
a group of consecutive files with the same section header (eg. files are
grouped by directory if C<#[SECT]> contains C<$directory>).  Lines beginning
with C<#[BODY]> and lines not beginning with C<#> are output for each
processed file.  Lines beginning with C<#[IF]> are not output, but all BODY
lines are skipped if any tag on an IF line doesn't exist.  Other lines
beginning with C<#> are ignored.  (To output a line beginning with C<#>, use
C<#[BODY]#>.) For example, this format file:

    # this is a comment line
    #[HEAD]-- Generated by ExifTool $exifToolVersion --
    File: $FileName - $DateTimeOriginal
    (f/$Aperture, ${ShutterSpeed}s, ISO $EXIF:ISO)
    #[TAIL]-- end --

with this command:

    exiftool -p test.fmt a.jpg b.jpg

produces output like this:

    -- Generated by ExifTool 12.86 --
    File: a.jpg - 2003:10:31 15:44:19
    (f/5.6, 1/60s, ISO 100)
    File: b.jpg - 2006:05:23 11:57:38
    (f/8.0, 1/13s, ISO 100)
    -- end --

The values of List-type tags with multiple items, Shortcut tags representing
multiple tags, and matching tags when the C<All> group is specified are
joined according the B<-sep> option setting when interpolated in the string.
(Note that when C<All> is used as a group name, dupicate tags are included
regardless of the Duplicates option setting.)  When C<All> is used as a tag
name, a value of 1 is returned if any tag exists in the specified group, or
0 otherwise (unless the C<All> group is also specified, in which case the
values of all matching tags are joined).

When B<-ee> (B<-extractEmbedded>) is combined with B<-p>, embedded documents
are effectively processed as separate input files.

If a specified tag does not exist, a minor warning is issued and the line
with the missing tag is not printed.  However, the B<-f> option may be used
to set the value of missing tags to '-' (but this may be configured via the
API MissingTagValue option), or the B<-m> option may be used to ignore minor
warnings and leave the missing values empty.  Alternatively, B<-q -q> may be
used to simply suppress the warning messages.

The L</Advanced formatting feature> may be used to modify the values of
individual tags within the B<-p> option string.

Note that the API RequestTags option is automatically set for all tags used
in the I<FMTFILE> or I<STR>.  This allows all other tags to be ignored using
B<-API IgnoreTags=all>, resulting in reduced memory usage and increased
speed.

=item B<-php>

Format output as a PHP Array.  The B<-g>, B<-G>, B<-D>, B<-H>, B<-l>,
B<-sep> and B<-struct> options combine with B<-php>, and duplicate tags are
handled in the same way as with the B<-json> option.  As well, the B<-b>
option may be added to output binary data, and B<-t> may be added to include
tag table information (see B<-t> for details).  Here is a simple example
showing how this could be used in a PHP script:

    <?php
    eval('$array=' . `exiftool -php -q image.jpg`);
    print_r($array);
    ?>

=item B<-s>[I<NUM>] (B<-short>)

Short output format.  Prints tag names instead of descriptions.  Add I<NUM>
or up to 3 B<-s> options for even shorter formats:

    -s1 or -s        - print tag names instead of descriptions
    -s2 or -s -s     - no extra spaces to column-align values
    -s3 or -s -s -s  - print values only (no tag names)

Also effective when combined with B<-t>, B<-h>, B<-X> or B<-listx> options.

=item B<-S> (B<-veryShort>)

Very short format.  The same as B<-s2> or two B<-s> options.  Tag names are
printed instead of descriptions, and no extra spaces are added to
column-align values.

=item B<-sep> I<STR> (B<-separator>)

Specify separator string for items in list-type tags.  When reading, the
default is to join list items with ", ".  When writing, this option causes
values assigned to list-type tags to be split into individual items at each
substring matching I<STR> (otherwise they are not split by default).  Space
characters in I<STR> match zero or more whitespace characters in the value.

Note that an empty separator ("") is allowed, and will join items with no
separator when reading, or split the value into individual characters when
writing.

For pure binary output (B<-b> used without B<-j>, B<-php> or B<-X>), the
first B<-sep> option specifies a list-item separator, and a second B<-sep>
option specifies a terminator for the end of the list (or after each value
if not a list).  In these strings, C<\n>, C<\r> and C<\t> may be used to
represent a newline, carriage return and tab respectively.  By default,
binary list items are separated by a newline, and no terminator is added.

=item B<-sort>, B<--sort>

Sort output by tag description, or by tag name if the B<-s> option is used.
When sorting by description, the sort order will depend on the B<-lang>
option setting.  Without the B<-sort> option, tags appear in the order they
were specified on the command line, or if not specified, the order they were
extracted from the file.  By default, tags are organized by groups when
combined with the B<-g> or B<-G> option, but this grouping may be disabled
with B<--sort>.

=item B<-struct>, B<--struct>

Output structured XMP information instead of flattening to individual tags.
This option works well when combined with the XML (B<-X>) and JSON (B<-j>)
output formats.  For other output formats, XMP structures and lists are
serialized into the same format as when writing structured information (see
L<https://exiftool.org/struct.html> for details).  When copying, structured
tags are copied by default unless B<--struct> is used to disable this
feature (although flattened tags may still be copied by specifying them
individually unless B<-struct> is used).  These options have no effect when
assigning new values since both flattened and structured tags may always be
used when writing.

=item B<-t> (B<-tab>)

Output a tab-delimited list of description/values (useful for database
import).  May be combined with B<-s> to print tag names instead of
descriptions, or B<-S> to print tag values only, tab-delimited on a single
line.  The B<-t> option may be combined with B<-j>, B<-php> or B<-X> to add
tag table information (C<table>, tag C<id>, and C<index> for cases where
multiple conditional tags exist with the same ID).

=item B<-T> (B<-table>)

Output tag values in table form.  Equivalent to B<-t -S -q -f>.

=item B<-v>[I<NUM>] (B<-verbose>)

Print verbose messages.  I<NUM> specifies the level of verbosity in the
range 0-5, with higher numbers being more verbose.  If I<NUM> is not given,
then each B<-v> option increases the level of verbosity by 1.  With any
level greater than 0, most other options are ignored and normal console
output is suppressed unless specific tags are extracted.  Using B<-v0>
causes the console output buffer to be flushed after each line (which may be
useful to avoid delays when piping exiftool output), and prints the name of
each processed file when writing and the new file name when renaming,
moving or copying.  Verbose levels above B<-v0> do not flush after each
line.  Also see the B<-progress> option.

=item B<-w>[+|!] I<EXT> or I<FMT> (B<-textOut>)

Write console output to files with names ending in I<EXT>, one for each
source file.  The output file name is obtained by replacing the source file
extension (including the '.') with the specified extension (and a '.' is
added to the start of I<EXT> if it doesn't already contain one).
Alternatively, a I<FMT> string may be used to give more control over the
output file name and directory.  In the format string, %d, %f and %e
represent the directory, filename and extension of the source file, and %c
represents a copy number which is automatically incremented if the file
already exists.  %d includes the trailing '/' if necessary, but %e does not
include the leading '.'.  For example:

    -w %d%f.txt       # same effect as "-w txt"
    -w dir/%f_%e.out  # write files to "dir" as "FILE_EXT.out"
    -w dir2/%d%f.txt  # write to "dir2", keeping dir structure
    -w a%c.txt        # write to "a.txt" or "a1.txt" or "a2.txt"...

Existing files will not be changed unless an exclamation point is added to
the option name (ie. B<-w!> or B<-textOut!>) to overwrite the file, or a
plus sign (ie. B<-w+> or B<-textOut+>) to append to the existing file.  Both
may be used (ie. B<-w+!> or B<-textOut+!>) to overwrite output files that
didn't exist before the command was run, and append the output from multiple
source files.  For example, to write one output file for all source files in
each directory:

    exiftool -filename -createdate -T -w+! %d/out.txt -r DIR

Capitalized format codes %D, %F, %E and %C provide slightly different
alternatives to the lower case versions.  %D does not include the trailing
'/', %F is the full filename including extension, %E includes the leading
'.', and %C increments the count for each processed file (see below).

Notes:

1) In a Windows BAT file the C<%> character is represented by C<%%>, so an
argument like C<%d%f.txt> is written as C<%%d%%f.txt>.

2) If the argument for B<-w> does not contain a valid format code (eg. %f),
then it is interpreted as a file extension, but there are three different
ways to create a single output file from multiple source files:

    # 1. Shell redirection
    exiftool FILE1 FILE2 ... > out.txt

    # 2. With the -w option and a zero-width format code
    exiftool -w+! %0fout.txt FILE1 FILE2 ...

    # 3. With the -W option (see the -W option below)
    exiftool -W+! out.txt FILE1 FILE2 ...

Advanced features:

A substring of the original file name, directory or extension may be taken
by specifying a field width immediately following the '%' character.  If the
width is negative, the substring is taken from the end.  The substring
position (characters to ignore at the start or end of the string) may be
given by a second optional value after a decimal point.  For example:

    Input File Name     Format Specifier    Output File Name
    ----------------    ----------------    ----------------
    Picture-123.jpg     %7f.txt             Picture.txt
    Picture-123.jpg     %-.4f.out           Picture.out
    Picture-123.jpg     %7f.%-3f            Picture.123
    Picture-123a.jpg    Meta%-3.1f.txt      Meta123.txt

(Note that special characters may have a width of greater than one.)

For %d and %D, the field width/position specifiers may be applied to the
directory levels instead of substring position by using a colon instead of a
decimal point in the format specifier.  For example:

    Source Dir     Format   Result       Notes
    ------------   ------   ----------   ------------------
    pics/2012/02   %2:d     pics/2012/   take top 2 levels
    pics/2012/02   %-:1d    pics/2012/   up one directory level
    pics/2012/02   %:1d     2012/02/     ignore top level
    pics/2012/02   %1:1d    2012/        take 1 level after top
    pics/2012/02   %-1:D    02           bottom level folder name
    /Users/phil    %:2d     phil/        ignore top 2 levels

(Note that the root directory counts as one level when an absolute path is
used as in the last example above.)

For %c, these modifiers have a different effects.  If a field width is
given, the copy number is padded with zeros to the specified width.  A
leading '-' adds a dash before the copy number, and a '+' adds an underline.
By default, the copy number is omitted from the first file of a given name,
but this can be changed by adding a decimal point to the modifier.  For
example:

    -w A%-cZ.txt      # AZ.txt, A-1Z.txt, A-2Z.txt ...
    -w B%5c.txt       # B.txt, B00001.txt, B00002.txt ...
    -w C%.c.txt       # C0.txt, C1.txt, C2.txt ...
    -w D%-.c.txt      # D-0.txt, D-1.txt, D-2.txt ...
    -w E%-.4c.txt     # E-0000.txt, E-0001.txt, E-0002.txt ...
    -w F%-.4nc.txt    # F-0001.txt, F-0002.txt, F-0003.txt ...
    -w G%+c.txt       # G.txt, G_1.txt G_2.txt ...
    -w H%-lc.txt      # H.txt, H-b.txt, H-c.txt ...
    -w I.%.3uc.txt    # I.AAA.txt, I.AAB.txt, I.AAC.txt ...

A special feature allows the copy number to be incremented for each
processed file by using %C (upper case) instead of %c.  This allows a
sequential number to be added to output file names, even if the names are
different.  For %C, a copy number of zero is not omitted as it is with %c. 
A leading '-' causes the number to be reset at the start of each new
directory (in the original directory structure if the files are being
moved), and '+' has no effect.  The number before the decimal place gives
the starting index, the number after the decimal place gives the field
width.  The following examples show the output filenames when used with the
command C<exiftool rose.jpg star.jpg jet.jpg ...>:

    -w %C%f.txt       # 0rose.txt, 1star.txt, 2jet.txt
    -w %f-%10C.txt    # rose-10.txt, star-11.txt, jet-12.txt
    -w %.3C-%f.txt    # 000-rose.txt, 001-star.txt, 002-jet.txt
    -w %57.4C%f.txt   # 0057rose.txt, 0058star.txt, 0059jet.txt

All format codes may be modified by 'l' or 'u' to specify lower or upper
case respectively (ie. C<%le> for a lower case file extension).  When used
to modify %c or %C, the numbers are changed to an alphabetical base (see
example H above).  Also, %c and %C may be modified by 'n' to count using
natural numbers starting from 1, instead of 0 (see example F above).

This same I<FMT> syntax is used with the B<-o> and B<-tagsFromFile> options,
although %c and %C are only valid for output file names.

=item B<-W>[+|!] I<FMT> (B<-tagOut>)

This enhanced version of the B<-w> option allows a separate output file to
be created for each extracted tag.  See the B<-w> option documentation above
for details of the basic functionality.  Listed here are the differences
between B<-W> and B<-w>:

1) With B<-W>, a new output file is created for each extracted tag.

2) B<-W> supports four additional format codes:  %t, %g and %s represent the
tag name, group name, and suggested extension for the output file (based on
the format of the data), and %o represents the value of the
OriginalRawFileName or OriginalFileName tag from the input file (including
extension).  The %g code may be followed by a single digit to specify the
group family number (eg. %g1), otherwise family 0 is assumed.  The substring
width/position/case specifiers may be used with these format codes in
exactly the same way as with %f and %e.

3) The argument for B<-W> is interpreted as a file name if it contains no
format codes.  (For B<-w>, this would be a file extension.)  This change
allows a simple file name to be specified, which, when combined with the
append feature, provides a method to write metadata from multiple source
files to a single output file without the need for shell redirection.  For
example, the following pairs of commands give the same result:

    # overwriting existing text file
    exiftool test.jpg > out.txt     # shell redirection
    exiftool test.jpg -W+! out.txt  # equivalent -W option

    # append to existing text file
    exiftool test.jpg >> out.txt    # shell redirection
    exiftool test.jpg -W+ out.txt   # equivalent -W option

4) Adding the B<-v> option to B<-W> sends a list of the tags and output file
names to the console instead of giving a verbose dump of the entire file. 
(Unless appending all output to one file for each source file by using
B<-W+> with an output file I<FMT> that does not contain %t, %g, %s or %o.)

5) Individual list items are stored in separate files when B<-W> is combined
with B<-b>, but note that for separate files to be created %c or %C must be
used in I<FMT> to give the files unique names.

=item B<-Wext> I<EXT>, B<--Wext> I<EXT> (B<-tagOutExt>)

This option is used to specify the type of output file(s) written by the
B<-W> option.  An output file is written only if the suggested extension
matches I<EXT>.  Multiple B<-Wext> options may be used to write more than
one type of file.  Use B<--Wext> to write all but the specified type(s).

=item B<-X> (B<-xmlFormat>)

Use ExifTool-specific RDF/XML formatting for console output.  Implies the
B<-a> option, so duplicate tags are extracted.  The formatting options
B<-b>, B<-D>, B<-H>, B<-l>, B<-s>, B<-sep>, B<-struct> and B<-t> may be used
in combination with B<-X> to affect the output, but note that the tag ID
(B<-D>, B<-H> and B<-t>), binary data (B<-b>) and structured output
(B<-struct>) options are not effective for the short output (B<-s>). Another
restriction of B<-s> is that only one tag with a given group and name may
appear in the output.  Note that the tag ID options (B<-D>, B<-H> and B<-t>)
will produce non-standard RDF/XML unless the B<-l> option is also used.

By default, B<-X> outputs flattened tags, so B<-struct> should be added if
required to preserve XMP structures.  List-type tags with multiple values
are formatted as an RDF Bag, but they are combined into a single string when
B<-s> or B<-sep> is used.  Using B<-L> changes the XML encoding from "UTF-8"
to "windows-1252".  Other B<-charset> settings change the encoding only if
there is a corresponding standard XML character set.  The B<-b> option
causes binary data values to be written, encoded in base64 if necessary. 
The B<-t> option adds tag table information to the output (see B<-t> for
details).

Note:  This output is NOT the same as XMP because it uses
dynamically-generated property names corresponding to the ExifTool tag names
with ExifTool family 1 group names as namespaces, and not the standard XMP
properties and namespaces.  To write XMP instead, use the B<-o> option with
an XMP extension for the output file.

=back

=head3 Processing control

=over 5

=item B<-a>, B<--a> (B<-duplicates>, B<--duplicates>)

Allow (B<-a>) or suppress (B<--a>) duplicate tag names to be extracted.  By
default, duplicate tags are suppressed when reading unless the B<-ee> or B<-X>
options are used or the Duplicates option is enabled in the configuration file.
When writing, this option allows multiple Warning messages to be shown.
Duplicate tags are always extracted when copying.

=item B<-e> (B<--composite>)

Extract existing tags only -- don't generate composite tags.

=item B<-ee>[I<NUM>] (B<-extractEmbedded>)

Extract information from embedded documents in EPS files, embedded EPS
information and JPEG and Jpeg2000 images in PDF files, embedded MPF images
in JPEG and MPO files, streaming metadata in AVCHD videos, and the resource
fork of Mac OS files.  Implies the B<-a> option.  Use B<-g3> or B<-G3> to
identify the originating document for extracted information.  Embedded
documents containing sub-documents are indicated with dashes in the family 3
group name.  (eg. C<Doc2-3> is the 3rd sub-document of the 2nd embedded
document.) Note that this option may increase processing time substantially,
especially for PDF files with many embedded images or videos with streaming
metadata.

When used with B<-ee>, the B<-p> option is evaluated for each embedded
document as if it were a separate input file.  This allows, for example,
generation of GPS track logs from timed metadata in videos.  See
L<https://exiftool.org/geotag.html#Inverse> for examples.

Setting I<NUM> to 2 causes the H264 video stream in MP4 videos to be parsed
until the first Supplemental Enhancement Information (SEI) message is
decoded, or 3 to parse the entire H624 stream and decode all SEI
information.  For M2TS videos, a setting of 3 causes the entire file to be
parsed in search of unlisted programs which may contain timed GPS.

=item B<-ext>[+] I<EXT>, B<--ext> I<EXT> (B<-extension>)

Process only files with (B<-ext>) or without (B<--ext>) a specified
extension.  There may be multiple B<-ext> and B<--ext> options.  A plus sign
may be added (ie. B<-ext+>) to add the specified extension to the normally
processed files.  EXT may begin with a leading '.', which is ignored.  Case
is not significant.  C<"*"> may be used to process files with any extension
(or none at all), as in the last three examples:

    exiftool -ext JPG DIR             # process only JPG files
    exiftool --ext cr2 --ext dng DIR  # supported files but CR2/DNG
    exiftool -ext+ txt DIR            # supported files plus TXT
    exiftool -ext "*" DIR             # process all files
    exiftool -ext "*" --ext xml DIR   # process all but XML files
    exiftool -ext "*" --ext . DIR     # all but those with no ext

Using this option has two main advantages over specifying C<*.I<EXT>> on the
command line:  1) It applies to files in subdirectories when combined with
the B<-r> option.  2) The B<-ext> option is case-insensitive, which is
useful when processing files on case-sensitive filesystems.

Note that all files specified on the command line will be processed
regardless of extension unless the B<-ext> option is used.

=item B<-F>[I<OFFSET>] (B<-fixBase>)

Fix the base for maker notes offsets.  A common problem with some image
editors is that offsets in the maker notes are not adjusted properly when
the file is modified.  This may cause the wrong values to be extracted for
some maker note entries when reading the edited file.  This option allows an
integer I<OFFSET> to be specified for adjusting the maker notes base offset.
If no I<OFFSET> is given, ExifTool takes its best guess at the correct base.
Note that exiftool will automatically fix the offsets for images which store
original offset information (eg. newer Canon models).  Offsets are fixed
permanently if B<-F> is used when writing EXIF to an image. eg)

    exiftool -F -exif:resolutionunit=inches image.jpg

=item B<-fast>[I<NUM>]

Increase speed of extracting information.  With B<-fast> (or B<-fast1>),
ExifTool will not scan to the end of a JPEG image to check for an AFCP or
PreviewImage trailer, or past the first comment in GIF images or the
audio/video data in WAV/AVI files to search for additional metadata.  These
speed benefits are small when reading images directly from disk, but can be
substantial if piping images through a network connection.  For more
substantial speed benefits, B<-fast2> also causes exiftool to avoid
extracting any EXIF MakerNote information, and to stop processing at the
IDAT chunk of PNG images and the mdat atom of QuickTime-format files (but
note that some files may store metadata after this).  B<-fast3> avoids
extracting metadata from the file, and returns only pseudo System tags, but
still reads the file header to obtain an educated guess at FileType.
B<-fast4> doesn't even read the file header, and returns only System tags
and a FileType based on the file extension.  B<-fast5> also disables
generation of the Composite tags (like B<-e>).  Has no effect when writing.

Note that a separate B<-fast> setting may be used for evaluation of a B<-if>
condition, or when ordering files with the B<-fileOrder> option.  See the
B<-if> and B<-fileOrder> options for details.

=item B<-fileOrder>[I<NUM>] [-]I<TAG>

Set file processing order according to the sorted value of the specified
I<TAG>.  For example, to process files in order of date:

    exiftool -fileOrder DateTimeOriginal DIR

Additional B<-fileOrder> options may be added for secondary sort keys.
Numbers are sorted numerically, and all other values are sorted
alphabetically.  Files missing the specified tag are sorted last.  The sort
order may be reversed by prefixing the tag name with a C<-> (eg.
C<-fileOrder -createdate>).  Print conversion of the sorted values is
disabled with the B<-n> option, or a C<#> appended to the tag name.  Other
formatting options (eg. B<-d>) have no effect on the sorted values.  Note
that the B<-fileOrder> option can incur large performance penalty since it
involves an additional initial processing pass of all files, but this impact
may be reduced by specifying a I<NUM> to effectively set the B<-fast> level
for the initial pass.  For example, B<-fileOrder4> may be used if I<TAG> is
a pseudo System tag.  If multiple B<-fileOrder> options are used, the
extraction is done at the lowest B<-fast> level.  Note that files are sorted
across directory boundaries if multiple input directories are specified.

=item B<-i> I<DIR> (B<-ignore>)

Ignore specified directory name.  I<DIR> may be either an individual folder
name, or a full path.  If a full path is specified, it must match the
Directory tag exactly to be ignored.  Use multiple B<-i> options to ignore
more than one directory name.  A special I<DIR> value of C<SYMLINKS> (case
sensitive) may be specified to avoid recursing into directories which are
symbolic links when the B<-r> option is used.  As well, a value of C<HIDDEN>
(case sensitive) may be used to ignore files with names that start with a
"." (ie. hidden files on Unix systems) when scanning a directory.

=item B<-if>[I<NUM>] I<EXPR>

Specify a condition to be evaluated before processing each I<FILE>.  I<EXPR>
is a Perl-like logic expression containing tag names prefixed by C<$>
symbols.  It is evaluated with the tags from each I<FILE> in turn, and the
file is processed only if the expression returns true.  Unlike Perl variable
names, tag names are not case sensitive and may contain a hyphen.  As well,
tag names may have a leading group names separated by colons, and/or a
trailing C<#> character to disable print conversion.  The expression
C<$GROUP:all> evaluates to 1 if any tag exists in the specified C<GROUP>, or
0 otherwise (see note 2 below).  When multiple B<-if> options are used, all
conditions must be satisfied to process the file.  Returns an exit status of
2 if all files fail the condition.  Below are a few examples:

    # extract shutterspeed from all Canon images in a directory
    exiftool -shutterspeed -if '$make eq "Canon"' dir

    # add one hour to all images created on or after Apr. 2, 2006
    exiftool -alldates+=1 -if '$CreateDate ge "2006:04:02"' dir

    # set EXIF ISO value if possible, unless it is set already
    exiftool '-exif:iso<iso' -if 'not $exif:iso' dir

    # find images containing a specific keyword (case insensitive)
    exiftool -if '$keywords =~ /harvey/i' -filename dir

Adding I<NUM> to the B<-if> option causes a separate processing pass to be
executed for evaluating I<EXPR> at a B<-fast> level given by I<NUM> (see the
B<-fast> option documentation for details).  Without I<NUM>, only one
processing pass is done at the level specified by the B<-fast> option.  For
example, using B<-if5> is possible if I<EXPR> uses only pseudo System tags,
and may significantly speed processing if enough files fail the condition.

The expression has access to the current ExifTool object through C<$self>,
and the following special functions are available to allow short-circuiting
of the file processing.  Both functions have a return value of 1.  Case is
significant for function names.

    End()    - end processing after this file
    EndDir() - end processing of files in the current directory
               after this file (not compatible with -fileOrder)

Notes:

1) The B<-n> and B<-b> options also apply to tags used in I<EXPR>.

2) Some binary data blocks are not extracted unless specified explicitly.
These tags are not available for use in the B<-if> condition unless they are
also specified on the command line.  The alternative is to use the
C<$GROUP:all> syntax. (eg. Use C<$exif:all> instead of C<$exif> in I<EXPR>
to test for the existence of EXIF tags.)

3) Tags in the string are interpolated in a similar way to B<-p> before the
expression is evaluated.  In this interpolation, C<$/> is converted to a
newline and C<$$> represents a single C<$> symbol.  So Perl variables, if
used, require a double C<$>, and regular expressions ending in C<$/> must
use C<$$/> instead.

4) The condition accesses only tags from the file being processed unless the
B<-fileNUM> option is used to read an alternate file and the corresponding
family 8 group name is specified for the tag.  See the B<-fileNUM> option
details for more information.

5) The B<-a> option has no effect on the evaluation of the expression, and
the values of duplicate tags are accessible only by specifying a group name
(such as a family 4 instance number, eg. C<$Copy1:TAG>, C<$Copy2:TAG>, etc).

6) A special "OK" UserParam is available to test the success of the previous
command when B<-execute> was used, and may be used like any other tag in the
condition (ie. "$OK").

7) The API RequestTags option is automatically set for all tags used in the
B<-if> condition.

=item B<-m> (B<-ignoreMinorErrors>)

Ignore minor errors and warnings.  This enables writing to files with minor
errors and disables some validation checks which could result in minor
warnings.  Generally, minor errors/warnings indicate a problem which usually
won't result in loss of metadata if ignored.  However, there are exceptions,
so ExifTool leaves it up to you to make the final decision.  Minor errors
and warnings are indicated by "[minor]" at the start of the message.
Warnings which affect processing when ignored are indicated by "[Minor]"
(with a capital "M").  Note that this causes missing values in
B<-tagsFromFile>, B<-p> and B<-if> strings to be set to an empty string
rather than an undefined value.

=item B<-o> I<OUTFILE> or I<FMT> (B<-out>)

Set the output file or directory name when writing information.  Without
this option, when any "real" tags are written the original file is renamed
to C<FILE_original> and output is written to I<FILE>.  When writing only
FileName and/or Directory "pseudo" tags, B<-o> causes the file to be copied
instead of moved, but directories specified for either of these tags take
precedence over that specified by the B<-o> option.

I<OUTFILE> may be C<-> to write to stdout.  The output file name may also be
specified using a I<FMT> string in which %d, %f and %e represent the
directory, file name and extension of I<FILE>.  Also, %c may be used to add
a copy number. See the B<-w> option for I<FMT> string examples.

The output file is taken to be a directory name if it already exists as a
directory or if the name ends with '/'.  Output directories are created if
necessary.  Existing files will not be overwritten.  Combining the
B<-overwrite_original> option with B<-o> causes the original source file to
be erased after the output file is successfully written.

A special feature of this option allows the creation of certain types of
files from scratch, or with the metadata from another type of file.  The
following file types may be created using this technique:

    XMP, EXIF, EXV, MIE, ICC/ICM, VRD, DR4

The output file type is determined by the extension of I<OUTFILE> (specified
as C<-.EXT> when writing to stdout).  The output file is then created from a
combination of information in I<FILE> (as if the B<-tagsFromFile> option was
used), and tag values assigned on the command line.  If no I<FILE> is
specified, the output file may be created from scratch using only tags
assigned on the command line.

=item B<-overwrite_original>

Overwrite the original I<FILE> (instead of preserving it by adding
C<_original> to the file name) when writing information to an image.
Caution: This option should only be used if you already have separate backup
copies of your image files.  The overwrite is implemented by renaming a
temporary file to replace the original.  This deletes the original file and
replaces it with the edited version in a single operation.  When combined
with B<-o>, this option causes the original file to be deleted if the output
file was successfully written (ie. the file is moved instead of copied).

=item B<-overwrite_original_in_place>

Similar to B<-overwrite_original> except that an extra step is added to
allow the original file attributes to be preserved.  For example, on a Mac
this causes the original file creation date, type, creator, label color,
icon, Finder tags, other extended attributes and hard links to the file to
be preserved (but note that the Mac OS resource fork is always preserved
unless specifically deleted with C<-rsrc:all=>).  This is implemented by
opening the original file in update mode and replacing its data with a copy
of a temporary file before deleting the temporary.  The extra step results
in slower performance, so the B<-overwrite_original> option should be used
instead unless necessary.

Note that this option reverts to the behaviour of the B<-overwrite_original>
option when also writing the FileName and/or Directory tags.

=item B<-P> (B<-preserve>)

Preserve the filesystem modification date/time (C<FileModifyDate>) of the
original file when writing.  Note that some filesystems store a creation
date (ie. C<FileCreateDate> on Windows and Mac systems) which is not
affected by this option.  This creation date is preserved on Windows systems
where Win32API::File and Win32::API are available regardless of this
setting.  For other systems, the B<-overwrite_original_in_place> option may
be used if necessary to preserve the creation date.  The B<-P> option is
superseded by any value written to the FileModifyDate tag.

=item B<-password> I<PASSWD>

Specify password to allow processing of password-protected PDF documents. 
If a password is required but not given, a warning is issued and the
document is not processed.  This option is ignored if a password is not
required.

=item B<-progress>[NUM][:[I<TITLE>]]

Show the progress when processing files.  Without a colon, the B<-progress>
option adds a progress count in brackets after the name of each processed
file, giving the current file number and the total number of files to be
processed.  Implies the B<-v0> option, causing the names of processed files
to also be printed when writing.  When combined with the B<-if> option, the
total count includes all files before the condition is applied, but files
that fail the condition will not have their names printed.  If NUM is
specified, the progress is shown every NUM input files.

If followed by a colon (ie. B<-progress:>), the console window title is set
according to the specified I<TITLE> string.  If no I<TITLE> is given, a
default I<TITLE> string of "ExifTool %p%%" is assumed.  In the string, %f
represents the file name, %p is the progress as a percent, %r is the
progress as a ratio, %##b is a progress bar of width "##" (where "##" is an
integer specifying the bar width in characters, or 20 characters by default
if "##" is omitted), and %% is a % character.  May be combined with the
normal B<-progress> option to also show the progress count in console
messages. (Note: For this feature to function correctly on Mac/Linux, stderr
must go to the console.)

=item B<-q> (B<-quiet>)

Quiet processing.  One B<-q> suppresses normal informational messages, and a
second B<-q> suppresses warnings as well.  Error messages can not be
suppressed, although minor errors may be downgraded to warnings with the
B<-m> option, which may then be suppressed with C<-q -q>.

=item B<-r>[.] (B<-recurse>)

Recursively process files in subdirectories.  Only meaningful if I<FILE> is
a directory name.  Subdirectories with names beginning with "." are not
processed unless "." is added to the option name (ie. B<-r.> or
B<-recurse.>).  By default, exiftool will also follow symbolic links to
directories if supported by the system, but this may be disabled with
C<-i SYMLINKS> (see the B<-i> option for details).  Combine this with
B<-ext> options to control the types of files processed.

=item B<-scanForXMP>

Scan all files (even unsupported formats) for XMP information unless found
already.  When combined with the B<-fast> option, only unsupported file
types are scanned.  Warning: It can be time consuming to scan large files.

=item B<-u> (B<-unknown>)

Extract values of unknown tags.  Add another B<-u> to also extract unknown
information from binary data blocks.  This option applies to tags with
numerical tag ID's, and causes tag names like "Exif_0xc5d9" to be generated
for unknown information.  It has no effect on information types which have
human-readable tag ID's (such as XMP), since unknown tags are extracted
automatically from these formats.

=item B<-U> (B<-unknown2>)

Extract values of unknown tags as well as unknown information from some
binary data blocks.  This is the same as two B<-u> options.

=item B<-wm> I<MODE> (B<-writeMode>)

Set mode for writing/creating tags.  I<MODE> is a string of one or more
characters from the list below.  The default write mode is C<wcg>.

    w - Write existing tags
    c - Create new tags
    g - create new Groups as necessary

For example, use C<-wm cg> to only create new tags (and avoid editing
existing ones).

The level of the group is the SubDirectory level in the metadata structure.
For XMP or IPTC this is the full XMP/IPTC block (the family 0 group), but
for EXIF this is the individual IFD (the family 1 group).

=item B<-z> (B<-zip>)

When reading, causes information to be extracted from .gz and .bz2
compressed images (only one image per archive; requires gzip and bzip2 to be
available).  When writing, causes compressed information to be written if
supported by the metadata format (eg. PNG supports compressed textual
metadata, JXL supports compressed EXIF and XML, and MIE supports any
compressed metadata), disables the recommended padding in embedded XMP
(saving 2424 bytes when writing XMP in a file), and writes XMP in shorthand
format -- the equivalent of setting the API Compress=1 and
Compact="NoPadding,Shorthand".

=back

=head3 Other options

=over 5

=item B<-@> I<ARGFILE>

Read command-line arguments from the specified file.  The file contains one
argument per line (NOT one option per line -- some options require
additional arguments, and all arguments must be placed on separate lines).
Blank lines and lines beginning with C<#> are ignored (unless they start
with C<#[CSTR]>, in which case the rest of the line is treated as a C
string, allowing standard C escape sequences such as "\n" for a newline). 
White space at the start of a line is removed.  Normal shell processing of
arguments is not performed, which among other things means that arguments
should not be quoted and spaces are treated as any other character.
I<ARGFILE> may exist relative to either the current directory or the
exiftool directory unless an absolute pathname is given.

For example, the following I<ARGFILE> will set the value of Copyright to
"Copyright YYYY, Phil Harvey", where "YYYY" is the year of CreateDate:

    -d
    %Y
    -copyright<Copyright $createdate, Phil Harvey

Arguments in I<ARGFILE> behave exactly the same as if they were entered at
the location of the B<-@> option on the command line, with the exception
that the B<-config> and B<-common_args> options may not be used in an
I<ARGFILE>.

=item B<-k> (B<-pause>)

Pause with the message C<-- press any key --> or C<-- press RETURN -->
(depending on your system) before terminating.  This option is used to
prevent the command window from closing when run as a Windows drag and drop
application.

=item B<-list>, B<-listw>, B<-listf>, B<-listr>, B<-listwf>,
B<-listg>[I<NUM>], B<-listd>, B<-listx>, B<-listgeo>

Print a list of all valid tag names (B<-list>), all writable tag names
(B<-listw>), all supported file extensions (B<-listf>), all recognized file
extensions (B<-listr>), all writable file extensions (B<-listwf>), all tag
groups [in a specified family] (B<-listg>[I<NUM>]), all deletable tag groups
(B<-listd>), an XML database of tag details including language translations
(B<-listx>), or the Geolocation database (B<-listgeo>).  The B<-list>,
B<-listw> and B<-listx> options may be followed by an additional argument of
the form C<-GROUP:All> to list only tags in a specific group, where C<GROUP>
is one or more family 0-2 group names (excepting EXIF IFD groups) separated
by colons.  With B<-listg>, I<NUM> may be given to specify the group family,
otherwise family 0 is assumed.  The B<-l> option may be combined with
B<-listf>, B<-listr> or B<-listwf> to add file descriptions to the list. 
The B<-lang> option may be combined with B<-listx> to output descriptions in
a single language, and the B<-sort> and/or B<-lang> options may be combined
with B<-listgeo>.  Also, the API GeolocMinPop, GeolocFeature and
GeolocAltNames options apply to the B<-listgeo> output.  Here are some
examples:

    -list               # list all tag names
    -list -EXIF:All     # list all EXIF tags
    -list -xmp:time:all # list all XMP tags relating to time
    -listw -XMP-dc:All  # list all writable XMP-dc tags
    -listf              # list all supported file extensions
    -listr              # list all recognized file extensions
    -listwf             # list all writable file extensions
    -listg1             # list all groups in family 1
    -listd              # list all deletable groups
    -listx -EXIF:All    # list database of EXIF tags in XML format
    -listx -XMP:All -s  # list short XML database of XMP tags
    -listgeo -lang de   # list geolocation database in German

When combined with B<-listx>, the B<-s> option shortens the output by
omitting the descriptions and values (as in the last example above), and
B<-f> adds 'flags' and 'struct' attributes if applicable.  The flags are
formatted as a comma-separated list of the following possible values: 
Avoid, Binary, List, Mandatory, Permanent, Protected, Unknown and Unsafe
(see the L<Tag Name documentation|Image::ExifTool::TagNames>).  For XMP List
tags, the list type (Alt, Bag or Seq) is added to the flags, and flattened
structure tags are indicated by a Flattened flag with 'struct' giving the ID
of the parent structure.

Note that none of the B<-list> options require an input I<FILE>.

=item B<-ver>

Print exiftool version number.  The B<-v> option may be added to print
addition system information (see the README file of the full distribution
for more details about optional libraries), or B<-v2> to also list the Perl
include directories.

=item B<-->

Indicates the end of options.  Any remaining arguments are treated as file
names, even if they begin with a dash (C<->).

=back

=head3 Special features

=over 5

=item B<-geotag> I<TRKFILE>

Geotag images from the specified GPS track log file.  Using the B<-geotag>
option is equivalent to writing a value to the C<Geotag> tag.  The GPS
position is interpolated from the track at a time specified by the value
written to the C<Geotime> tag.  If C<Geotime> is not specified, the value is
copied from C<DateTimeOriginal#> (the C<#> is added to copy the unformatted
value, avoiding potential conflicts with the B<-d> option).  For example,
the following two commands are equivalent:

    exiftool -geotag trk.log image.jpg
    exiftool -geotag trk.log "-Geotime<DateTimeOriginal#" image.jpg

When the C<Geotime> value is converted to UTC, the local system timezone is
assumed unless the date/time value contains a timezone.  Writing C<Geotime>
causes the following tags to be written (provided they can be calculated
from the track log, and they are supported by the destination metadata
format):  GPSLatitude, GPSLatitudeRef, GPSLongitude, GPSLongitudeRef,
GPSAltitude, GPSAltitudeRef, GPSDateStamp, GPSTimeStamp, GPSDateTime,
GPSTrack, GPSTrackRef, GPSSpeed, GPSSpeedRef, GPSImgDirection,
GPSImgDirectionRef, GPSPitch, GPSRoll, GPSCoordinates, AmbientTemperature
and CameraElevationAngle.  By default, in image files tags are created in
EXIF, and updated in XMP only if they already exist.  In QuickTime-format
files GPSCoordinates is created in the preferred location (ItemList by
default) as well as in XMP.  However, C<EXIF:Geotime>, C<XMP:Geotime> or
C<QuickTime:Geotime> may be specified to write to write only to one group. 
Also, C<ItemList:Geotime>, C<Keys:Geotime> or C<UserData:Geotime> may be
used to write to a specific location in QuickTime-format files.  Note that
GPSPitch and GPSRoll are non-standard, and require user-defined tags in
order to be written.

The C<Geosync> tag may be used to specify a time correction which is applied
to each C<Geotime> value for synchronization with GPS time.  For example,
the following command compensates for image times which are 1 minute and 20
seconds behind GPS:

    exiftool -geosync=+1:20 -geotag a.log DIR

Advanced C<Geosync> features allow a piecewise linear time drift correction
and synchronization from previously geotagged images.  See "geotag.html" in
the full ExifTool distribution for more information.

Multiple B<-geotag> options may be used to concatenate GPS track log data.
Also, a single B<-geotag> option may be used to load multiple track log
files by using wildcards in the I<TRKFILE> name, but note that in this case
I<TRKFILE> must be quoted on most systems (with the notable exception of
Windows) to prevent filename expansion.  For example:

    exiftool -geotag "TRACKDIR/*.log" IMAGEDIR

Currently supported track file formats are GPX, NMEA RMC/GGA/GLL, KML, IGC,
Garmin XML and TCX, Magellan PMGNTRK, Honeywell PTNTHPR, Bramor gEO, Winplus
Beacon TXT, and GPS/IMU CSV files.  See L</GEOTAGGING EXAMPLES> for
examples. Also see "geotag.html" in the full ExifTool distribution and the
L<Image::ExifTool Options|Image::ExifTool/Options> for more details and for
information about geotag configuration options.

The API Geolocation option may be set to the value "geotag" to also write
the name, province/state and country of the nearest city while geotagging.
See L<https://exiftool.org/geolocation.html> for details.

=item B<-globalTimeShift> I<SHIFT>

Shift all formatted date/time values by the specified amount when reading. 
Does not apply to unformatted (B<-n>) output.  I<SHIFT> takes the same form
as the date/time shift when writing (see
L<Image::ExifTool::Shift.pl|Image::ExifTool::Shift.pl> for details), with a
negative shift being indicated with a minus sign (C<->) at the start of the
I<SHIFT> string.  For example:

    # return all date/times, shifted back by 1 hour
    exiftool -globalTimeShift -1 -time:all a.jpg

    # set the file name from the shifted CreateDate (-1 day) for
    # all images in a directory
    exiftool "-filename<createdate" -globaltimeshift "-0:0:1 0:0:0" \
        -d %Y%m%d-%H%M%S.%%e dir

=item B<-use> I<MODULE>

Add features from specified plug-in I<MODULE>.  Currently, the MWG module is
the only plug-in module distributed with exiftool.  This module adds
read/write support for tags as recommended by the Metadata Working Group. As
a convenience, C<-use MWG> is assumed if the group name prefix starts with
C<MWG:> exactly for any requested tag.  See the
L<MWG Tags documentation|Image::ExifTool::TagNames/MWG Tags> for more
details.  Note that this option is not reversible, and remains in effect
until the application terminates, even across the B<-execute> option.

=back

=head3 Utilities

=over 5

=item B<-restore_original>

=item B<-delete_original>[!]

These utility options automate the maintenance of the C<_original> files
created by exiftool.  They have no effect on files without an C<_original>
copy.  The B<-restore_original> option restores the specified files from
their original copies by renaming the C<_original> files to replace the
edited versions.  For example, the following command restores the originals
of all JPG images in directory C<DIR>:

    exiftool -restore_original -ext jpg DIR

The B<-delete_original> option deletes the C<_original> copies of all files
specified on the command line.  Without a trailing C<!> this option prompts
for confirmation before continuing.  For example, the following command
deletes C<a.jpg_original> if it exists, after asking "Are you sure?":

    exiftool -delete_original a.jpg

These options may not be used with other options to read or write tag values
in the same command, but may be combined with options such B<-ext>, B<-if>,
B<-r>, B<-q> and B<-v>.

=back

=head3 Advanced options

Among other things, the advanced options allow complex processing to be
performed from a single command without the need for additional scripting.
This may be particularly useful for implementations such as Windows
drag-and-drop applications.  These options may also be used to improve
performance in multi-pass processing by reducing the overhead required to
load exiftool for each invocation.

=over 5

=item B<-api> [I<OPT[[^]=[VAL]]>]

Set ExifTool API option.  I<OPT> is an API option name.  The option value is
set to 1 if I<=VAL> is omitted.  If I<VAL> is omitted, the option value is
set to undef if C<=> is used, or an empty string with C<^=>.  If I<OPT> is
not specified a list of available options is returned.  The option name is
not case senstive, but the option values are.  See
L<Image::ExifTool Options|Image::ExifTool/Options> for option details.  This
overrides API options set via the config file.  Note that the exiftool app
sets some API options internally, and attempts to change these via the
command line will have no effect.

=item B<-common_args>

Specifies that all arguments following this option are common to all
executed commands when B<-execute> is used.  This and the B<-config> option
are the only options that may not be used inside a B<-@> I<ARGFILE>.  Note
that by definition this option and its arguments MUST come after all other
options on the command line.

=item B<-config> I<CFGFILE>

Load specified configuration file instead of the default ".ExifTool_config".
If used, this option must come before all other arguments on the command
line and applies to all B<-execute>'d commands.  This file is used to create
user-defined tags as well as set default ExifTool options.  The I<CFGFILE>
must exist relative to the current working directory or the exiftool
application directory unless an absolute path is specified.  Loading of the
default config file may be disabled by setting I<CFGFILE> to an empty string
(ie. "").  See L<https://exiftool.org/config.html> and
config_files/example.config in the full ExifTool distribution for details
about the configuration file syntax.

=item B<-echo>[I<NUM>] I<TEXT>

Echo I<TEXT> to stdout (B<-echo> or B<-echo1>) or stderr (B<-echo2>).  Text
is output as the command line is parsed, before the processing of any input
files.  I<NUM> may also be 3 or 4 to output text (to stdout or stderr
respectively) after processing is complete.  For B<-echo3> and B<-echo4>,
"${status}" may be used in the I<TEXT> string to represent the numerical
exit status of the command (see L</EXIT STATUS>).

=item B<-efile>[I<NUM>][!] I<TXTFILE>

Save the names of files giving errors (I<NUM> missing or 1), files that were
unchanged (I<NUM> is 2), files that fail the B<-if> condition (I<NUM> is 4),
files that were updated (I<NUM> is 8), files that were created (I<NUM> is
16), or any combination thereof by summing I<NUM> (eg. B<-efile3> is the
same has having both B<-efile> and B<-efile2> options with the same
I<TXTFILE>). By default, file names are appended to any existing I<TXTFILE>,
but I<TXTFILE> is overwritten if an exclamation point is added to the option
(eg. B<-efile!>).  Saves the name of the file specified by the B<-srcfile>
option if applicable.

=item B<-execute>[I<NUM>]

Execute command for all arguments up to this point on the command line (plus
any arguments specified by B<-common_args>).  The result is as if the
commands were executed as separate command lines (with the exception of the
B<-config> and B<-use> options which remain in effect for subsequent
commands).  Allows multiple commands to be executed from a single command
line.  I<NUM> is an optional number that is echoed in the "{ready}" message
when using the B<-stay_open> feature.  If a I<NUM> is specified, the B<-q>
option no longer suppresses the output "{readyNUM}" message.

=item B<-file>I<NUM> I<ALTFILE>

Read tags from an alternate source file.  Among other things, this allows
tags from different files to be compared and combined using the B<-if> and
B<-p> options.  I<NUM> is any string of digits.  Tags from alternate files
are accessed via the corresponding family 8 group name (eg. C<File1:TAG> for
the B<-file1> option, C<File2:TAG> for B<-file2>, etc).  I<ALTFILE> may
contain filename formatting codes like the B<-w> option (%d, %f, etc),
and/or tag names with a leading C<$> symbol to access tags from the source
file in the same way as the B<-p> option (so any other dollar symbol in the
file name must be doubled, eg. C<money$$.jpg>).  For example, assuming that
the OriginalFileName tag has been set in the edited file, a command to copy
Rights from the original file could look like this:

    exiftool -file1 '$originalfilename' '-rights<file1:rights' edited.jpg

Subtle note:  If a B<-tagsFromFile> option is used, tags in the I<ALTFILE>
argument come from the I<SRCFILE> that applies to the first argument
accessing tags from the corresponding C<FileNUM> group.

User-defined Composite tags may access tags from alternate files using the
appropriate (case-sensitive) family 8 group name.

=item B<-list_dir>

List directories themselves instead of their contents.  This option
effectively causes directories to be treated as normal files when reading
and writing.  For example, with this option the output of the C<ls -la>
command on Mac/Linux may be approximated by this exiftool command:

    exiftool -list_dir -T -ls-l -api systemtags -fast5 .* *

(The B<-T> option formats the output in tab-separated columns, B<-ls-l> is a
L<shortcut tag|Image::ExifTool::Shortcuts>, the API SystemTags option is
required to extract some necessary tags, and the B<-fast5> option is added
for speed since only system tags are being extracted.)

=item B<-srcfile> I<FMT>

Specify a different source file to be processed based on the name of the
original I<FILE>.  This may be useful in some special situations for
processing related preview images or sidecar files.  See the B<-w> option
for a description of the I<FMT> syntax.  Note that file name I<FMT> strings
for all options are based on the original I<FILE> specified from the command
line, not the name of the source file specified by B<-srcfile>.

For example, to copy metadata from NEF files to the corresponding JPG
previews in a directory where other JPG images may exist:

    exiftool -ext nef -tagsfromfile @ -srcfile %d%f.jpg dir

If more than one B<-srcfile> option is specified, the files are tested in
order and the first existing source file is processed.  If none of the
source files already exist, then exiftool uses the first B<-srcfile>
specified.

A I<FMT> of C<@> may be used to represent the original I<FILE>, which may be
useful when specifying multiple B<-srcfile> options (eg. to fall back to
processing the original I<FILE> if no sidecar exists).

When this option is used, two special UserParam tags (OriginalFileName and
OriginalDirectory) are generated to allow access to the original I<FILE>
name and directory.

=item B<-stay_open> I<FLAG>

If I<FLAG> is C<1> or C<True> (case insensitive), causes exiftool keep
reading from the B<-@> I<ARGFILE> even after reaching the end of file.  This
feature allows calling applications to pre-load exiftool, thus avoiding the
overhead of loading exiftool for each command.  The procedure is as follows:

1) Execute C<exiftool -stay_open True -@ I<ARGFILE>>, where I<ARGFILE> is the
name of an existing (possibly empty) argument file or C<-> to pipe arguments
from the standard input.

2) Write exiftool command-line arguments to I<ARGFILE>, one argument per
line (see the B<-@> option for details).

3) Write C<-execute\n> to I<ARGFILE>, where C<\n> represents a newline
sequence.  (Note: You may need to flush your write buffers here if using
buffered output.)  ExifTool will then execute the command with the arguments
received up to this point, send a "{ready}" message to stdout when done
(unless the B<-q> or B<-T> option is used), and continue trying to read
arguments for the next command from I<ARGFILE>.  To aid in command/response
synchronization, any number appended to the B<-execute> option is echoed in
the "{ready}" message.  For example, C<-execute613> results in "{ready613}". 
When this number is added, B<-q> no longer suppresses the "{ready}" message.
(Also, see the B<-echo3> and B<-echo4> options for additional ways to pass
signals back to your application.)

4) Repeat steps 2 and 3 for each command.

5) Write C<-stay_open\nFalse\n> (or C<-stay_open\n0\n>) to I<ARGFILE> when
done.  This will cause exiftool to process any remaining command-line
arguments then exit normally.

The input I<ARGFILE> may be changed at any time before step 5 above by
writing the following lines to the currently open I<ARGFILE>:

    -stay_open
    True
    -@
    NEWARGFILE

This causes I<ARGFILE> to be closed, and I<NEWARGFILE> to be kept open. 
(Without the B<-stay_open> here, exiftool would have returned to reading
arguments from I<ARGFILE> after reaching the end of I<NEWARGFILE>.)

Note:  When writing arguments to a disk file there is a delay of up to 0.01
seconds after writing C<-execute\n> before exiftool starts processing the
command.  This delay may be avoided by sending a CONT signal to the exiftool
process immediately after writing C<-execute\n>.  (There is no associated
delay when writing arguments via a pipe with C<-@ ->, so the signal is not
necessary when using this technique.)

=item B<-userParam> I<PARAM[[^]=[VAL]]>

Set user parameter.  I<PARAM> is an arbitrary user parameter name.  This is
an interface to the API UserParam option (see the
L<Image::ExifTool Options|Image::ExifTool/Options> documentation), and
provides a method to access user-defined parameters in arguments to the
B<-if> and B<-p> options as if they were any other tag.  Appending a hash
tag (C<#>) to I<PARAM> (eg. C<-userParam MyTag#=yes>) also causes the
parameter to be extracted as a normal tag in the UserParam group.  Similar
to the B<-api> option, the parameter value is set to 1 if I<=VAL> is
omitted, undef if just I<VAL> is omitted with C<=>, or an empty string if
I<VAL> is omitted with C<^=>.

    exiftool -p '$test from $filename' -userparam test=Hello FILE

=back

=head3 Advanced formatting feature

An advanced formatting feature allows modification of the value of any tag
interpolated within a B<-if> or B<-p> option argument, or a B<-tagsFromFile>
redirection string.  Tag names within these strings are prefixed by a C<$>
symbol, and an arbitrary Perl expression may be applied to the tag value by
placing braces around the tag name and inserting the expression after the
name, separated by a semicolon (ie. C<${TAG;EXPR}>).  The expression acts on
the value of the tag through the default input variable (C<$_>), and has
access to the full ExifTool API through the current ExifTool object
(C<$self>) and the tag key (C<$tag>).  It may contain any valid Perl code,
including translation (C<tr///>) and substitution (C<s///>) operations, but
note that braces within the expression must be balanced.  The example below
prints the camera Make with spaces translated to underlines, and multiple
consecutive underlines replaced by a single underline:

    exiftool -p '${make;tr/ /_/;s/__+/_/g}' image.jpg

An C<@> may be added after the tag name to make the expression act on
individual list items for list-type tags, simplifying list processing.  Set
C<$_> to undef to remove an item from the list.  As an example, the
following command returns all subjects not containing the string "xxx":

    exiftool -p '${subject@;$_=undef if /xxx/}' image.jpg

A default expression of C<tr(/\\?*:|"E<lt>E<gt>\0)()d> is assumed if the
expression is empty (ie. C<${TAG;}>).  This removes the characters / \ ? * :
| E<lt> E<gt> and null from the printed value.  (These characters are
illegal in Windows file names, so this feature is useful if tag values are
used in file names.)

=head4 Helper functions

C<DateFmt>

Simplifies reformatting of individual date/time values.  This function acts
on a standard EXIF-formatted date/time value in C<$_> and formats it
according to the specified format string (see the B<-d> option).  To avoid
trying to reformat an already-formatted date/time value, a C<#> must be
added to the tag name (as in the example below) if the B<-d> option is also
used.  For example:

    exiftool -p '${createdate#;DateFmt("%Y-%m-%d_%H%M%S")}' a.jpg

C<ShiftTime>

Shifts EXIF-formatted date/time string by a specified amount.  Start with a
leading minus sign to shift backwards in time.  See
L<Image::ExifTool::Shift.pl|Image::ExifTool::Shift.pl> for details about
shift syntax.  For example, to shift a date/time value back by one year:

    exiftool -p '${createdate;ShiftTime("-1:0:0 0")}' a.jpg

C<NoDups>

Removes duplicate items from a list with a separator specified by the
B<-sep> option.  This function is most useful when copying list-type tags. 
For example, the following command may be used to remove duplicate Keywords:

    exiftool -sep '##' '-keywords<${keywords;NoDups}' a.jpg

The B<-sep> option is necessary to split the string back into individual
list items when writing to a list-type tag.

An optional flag argument may be set to 1 to cause C<NoDups> to set C<$_> to
undef if no duplicates existed, thus preventing the file from being
rewritten unnecessarily:

    exiftool -sep '##' '-keywords<${keywords;NoDups(1)}' a.jpg

Note that function names are case sensitive.

ExifTool 12.64 adds an API NoDups option which makes the NoDups helper
function largely redundant, with all the functionality except the ability to
avoid rewriting the file if there are no duplicates, but with the advantage
the duplicates may be removed when accumulating list items from multiple
sources.  An equivalent to the above commands using this feature would be:

    exiftool -tagsfromfile @ -keywords -api nodups a.jpg

C<SetTags>

Used to set tags in extracted images.  With no arguments, copies all tags
from the source file to the embedded image:

    exiftool -p '${previewimage;SetTags}' -b a.arw > preview.jpg

Arguments may be added to copy or set specific tags.  Arguments take exactly
the same form as those on the command line when copying or writing tags,
but without the leading dash.  For example:

    exiftool -p '${previewimage;SetTags("comment=test","title<filename")}' ...

=head1 WINDOWS UNICODE FILE NAMES

In Windows, command-line arguments are specified using the current code page
and are recoded automatically to the system code page.  This recoding is not
done for arguments in ExifTool arg files, so by default filenames in arg
files use the system code page.  Unfortunately, these code pages are not
complete character sets, so not all file names may be represented.

ExifTool 9.79 and later allow the file name encoding to be specified with
C<-charset filename=CHARSET>, where C<CHARSET> is the name of a valid
ExifTool character set, preferably C<UTF8> (see the B<-charset> option for a
complete list).  Setting this triggers the use of Windows wide-character i/o
routines, thus providing support for most Unicode file names (see note 4). 
But note that it is not trivial to pass properly encoded file names on the
Windows command line (see L<https://exiftool.org/faq.html#Q18> for details),
so placing them in a UTF-8 encoded B<-@> argfile and using
C<-charset filename=utf8> is recommended if possible.

A warning is issued if a specified filename contains special characters and
the filename character set was not provided.  However, the warning may be
disabled by setting C<-charset filename="">, and ExifTool may still function
correctly if the system code page matches the character set used for the
file names.

When a directory name is provided, the file name encoding need not be
specified (unless the directory name contains special characters), and
ExifTool will automatically use wide-character routines to scan the
directory.

The filename character set applies to the I<FILE> arguments as well as
filename arguments of B<-@>, B<-geotag>, B<-o>, B<-p>, B<-srcfile>,
B<-tagsFromFile>, B<-csv>=, B<-j>= and B<->I<TAG>E<lt>=.  However, it does
not apply to the B<-config> filename, which always uses the system character
set.  The C<-charset filename=> option must come before the B<-@> option to
be effective, but the order doesn't matter with respect to other options.

Notes:

1) FileName and Directory tag values still use the same encoding as other
tag values, and are converted to/from the filename character set when
writing/reading if specified.

2) Unicode support is not yet implemented for other Windows-based systems
like Cygwin.

3) See L</WRITING READ-ONLY FILES> below for a note about editing read-only
files with Unicode names.

4) Unicode file names with surrogate pairs (code points over U+FFFF) still
cause problems.

=head1 WRITING READ-ONLY FILES

In general, ExifTool may be used to write metadata to read-only files
provided that the user has write permission in the directory.  However,
there are three cases where file write permission is also required:

1) When using the B<-overwrite_original_in_place> option.

2) When writing only pseudo System tags (eg. FileModifyDate).

3) On Windows if the file has Unicode characters in its name, and a) the
B<-overwrite_original> option is used, or b) the C<_original> backup already
exists.

Hidden files in Windows behave as read-only files when attempting to write
any real tags to the file -- an error is generated when using the
B<-overwrite_original_in_place>, otherwise writing should be successful and
the hidden attribute will be removed.  But the B<-if> option may be used to
avoid processing hidden files (provided Win32API::File is available):

    exiftool -if "$fileattributes !~ /Hidden/" ...

=head1 READING EXAMPLES

B<Note>: Beware when cutting and pasting these examples into your terminal!
Some characters such as single and double quotes and hyphens may have been
changed into similar-looking yet functionally-different characters by the
text formatter used to display this documentation.  Also note that Windows
users must use double quotes instead of single quotes as below around
arguments containing special characters.

=over 5

=item exiftool -a -u -g1 a.jpg

Print all meta information in an image, including duplicate and unknown
tags, sorted by group (for family 1).  For performance reasons, this command
may not extract all available metadata.  (Metadata in embedded documents,
metadata extracted by external utilities, and metadata requiring excessive
processing time may not be extracted).  Add C<-ee3> and C<-api RequestAll=3>
to the command to extract absolutely everything available.

=item exiftool -common dir

Print common meta information for all images in C<dir>.  C<-common> is a
L<shortcut tag|Image::ExifTool::Shortcuts> representing common EXIF meta
information.

=item exiftool -T -createdate -aperture -shutterspeed -iso dir > out.txt

List specified meta information in tab-delimited column form for all images
in C<dir> to an output text file named "out.txt".

=item exiftool -s -ImageSize -ExposureTime b.jpg

Print ImageSize and ExposureTime tag names and values.

=item exiftool -l -canon c.jpg d.jpg

Print standard Canon information from two image files.

=item exiftool -r -w .txt -common pictures

Recursively extract common meta information from files in C<pictures>
directory, writing text output to C<.txt> files with the same names.

=item exiftool -b -ThumbnailImage image.jpg > thumbnail.jpg

Save thumbnail image from C<image.jpg> to a file called C<thumbnail.jpg>.

=item exiftool -b -JpgFromRaw -w _JFR.JPG -ext NEF -r .

Recursively extract JPG image from all Nikon NEF files in the current
directory, adding C<_JFR.JPG> for the name of the output JPG files.

=item exiftool -a -b -W %d%f_%t%-c.%s -preview:all dir

Extract all types of preview images (ThumbnailImage, PreviewImage,
JpgFromRaw, etc.) from files in directory "dir", adding the tag name to the
output preview image file names.

=item exiftool -d '%r %a, %B %e, %Y' -DateTimeOriginal -S -s -ext jpg .

Print formatted date/time for all JPG files in the current directory.

=item exiftool -IFD1:XResolution -IFD1:YResolution image.jpg

Extract image resolution from EXIF IFD1 information (thumbnail image IFD).

=item exiftool '-*resolution*' image.jpg

Extract all tags with names containing the word "Resolution" from an image.

=item exiftool -xmp:author:all -a image.jpg

Extract all author-related XMP information from an image.

=item exiftool -xmp -b a.jpg > out.xmp

Extract complete XMP data record intact from C<a.jpg> and write it to
C<out.xmp> using the special C<XMP> tag (see the Extra tags in
L<Image::ExifTool::TagNames|Image::ExifTool::TagNames>).

=item exiftool -p '$filename has date $dateTimeOriginal' -q -f dir

Print one line of output containing the file name and DateTimeOriginal for
each image in directory C<dir>.

=item exiftool -ee3 -p '$gpslatitude, $gpslongitude, $gpstimestamp' a.m2ts

Extract all GPS positions from an AVCHD video.

=item exiftool -icc_profile -b -w icc image.jpg

Save complete ICC_Profile from an image to an output file with the same name
and an extension of C<.icc>.

=item exiftool -htmldump -w tmp/%f_%e.html t/images

Generate HTML pages from a hex dump of EXIF information in all images from
the C<t/images> directory.  The output HTML files are written to the C<tmp>
directory (which is created if it didn't exist), with names of the form
'FILENAME_EXT.html'.

=item exiftool -a -b -ee -embeddedimage -W Image_%.3g3.%s file.pdf

Extract embedded JPG and JP2 images from a PDF file.  The output images will
have file names like "Image_#.jpg" or "Image_#.jp2", where "#" is the
ExifTool family 3 embedded document number for the image.

=back

=head1 WRITING EXAMPLES

Note that quotes are necessary around arguments which contain certain
special characters such as C<E<gt>>, C<E<lt>> or any white space.  These
quoting techniques are shell dependent, but the examples below will work for
most Unix shells.  With the Windows cmd shell however, double quotes should
be used (eg. -Comment=E<34>This is a new commentE<34>).

=over 5

=item exiftool -Comment='This is a new comment' dst.jpg

Write new comment to a JPG image (replaces any existing comment).

=item exiftool -comment= -o newdir -ext jpg .

Remove comment from all JPG images in the current directory, writing the
modified images to a new directory.

=item exiftool -keywords=EXIF -keywords=editor dst.jpg

Replace existing keyword list with two new keywords (C<EXIF> and C<editor>).

=item exiftool -Keywords+=word -o newfile.jpg src.jpg

Copy a source image to a new file, and add a keyword (C<word>) to the
current list of keywords.

=item exiftool -exposurecompensation+=-0.5 a.jpg

Decrement the value of ExposureCompensation by 0.5 EV.  Note that += with a
negative value is used for decrementing because the -= operator is used for
conditional deletion (see next example).

=item exiftool -credit-=xxx dir

Delete Credit information from all files in a directory where the Credit
value was C<xxx>.

=item exiftool -xmp:description-de='k&uuml;hl' -E dst.jpg

Write alternate language for XMP:Description, using HTML character escaping
to input special characters.

=item exiftool -all= dst.jpg

Delete all meta information from an image.  Note: You should NOT do this to
RAW images (except DNG) since proprietary RAW image formats often contain
information in the makernotes that is necessary for converting the image.

=item exiftool -all= -comment='lonely' dst.jpg

Delete all meta information from an image and add a comment back in.  (Note
that the order is important: C<-comment='lonely' -all=> would also delete
the new comment.)

=item exiftool -all= --jfif:all dst.jpg

Delete all meta information except JFIF group from an image.

=item exiftool -Photoshop:All= dst.jpg

Delete Photoshop meta information from an image (note that the Photoshop
information also includes IPTC).

=item exiftool -r -XMP-crss:all= DIR

Recursively delete all XMP-crss information from images in a directory.

=item exiftool '-ThumbnailImageE<lt>=thumb.jpg' dst.jpg

Set the thumbnail image from specified file (Note: The quotes are necessary
to prevent shell redirection).

=item exiftool '-JpgFromRawE<lt>=%d%f_JFR.JPG' -ext NEF -r .

Recursively write JPEG images with filenames ending in C<_JFR.JPG> to the
JpgFromRaw tag of like-named files with extension C<.NEF> in the current
directory.  (This is the inverse of the C<-JpgFromRaw> command of the
L</READING EXAMPLES> section above.)

=item exiftool -DateTimeOriginal-='0:0:0 1:30:0' dir

Adjust original date/time of all images in directory C<dir> by subtracting
one hour and 30 minutes.  (This is equivalent to C<-DateTimeOriginal-=1.5>.
See L<Image::ExifTool::Shift.pl|Image::ExifTool::Shift.pl> for details.)

=item exiftool -createdate+=3 -modifydate+=3 a.jpg b.jpg

Add 3 hours to the CreateDate and ModifyDate timestamps of two images.

=item exiftool -AllDates+=1:30 -if '$make eq E<34>CanonE<34>' dir

Shift the values of DateTimeOriginal, CreateDate and ModifyDate forward by 1
hour and 30 minutes for all Canon images in a directory.  (The AllDates tag
is provided as a shortcut for these three tags, allowing them to be accessed
via a single tag.)

=item exiftool -xmp:city=Kingston image1.jpg image2.nef

Write a tag to the XMP group of two images.  (Without the C<xmp:> this tag
would get written to the IPTC group since C<City> exists in both, and IPTC
is preferred by default.)

=item exiftool -LightSource-='Unknown (0)' dst.tiff

Delete C<LightSource> tag only if it is unknown with a value of 0.

=item exiftool -whitebalance-=auto -WhiteBalance=tung dst.jpg

Set C<WhiteBalance> to C<Tungsten> only if it was previously C<Auto>.

=item exiftool -comment-= -comment='new comment' a.jpg

Write a new comment only if the image doesn't have one already.

=item exiftool -o %d%f.xmp dir

Create XMP meta information data files for all images in C<dir>.

=item exiftool -o test.xmp -owner=Phil -title='XMP File'

Create an XMP data file only from tags defined on the command line.

=item exiftool '-ICC_Profile<=%d%f.icc' image.jpg

Write ICC_Profile to an image from a C<.icc> file of the same name.

=item exiftool -hierarchicalkeywords='{keyword=one,children={keyword=B}}'

Write structured XMP information.  See L<https://exiftool.org/struct.html>
for more details.

=item exiftool -trailer:all= image.jpg

Delete any trailer found after the end of image (EOI) in a JPEG file.  A
number of digital cameras store a large PreviewImage after the JPEG EOI, and
the file size may be reduced significantly by deleting this trailer.  See
the L<JPEG Tags documentation|Image::ExifTool::TagNames/JPEG Tags> for a
list of recognized JPEG trailers.

=back

=head1 COPYING EXAMPLES

These examples demonstrate the ability to copy tag values between files.

=over 5

=item exiftool -tagsFromFile src.cr2 dst.jpg

Copy the values of all writable tags from C<src.cr2> to C<dst.jpg>, writing
the information to same-named tags in the preferred groups.

=item exiftool -TagsFromFile src.jpg -all:all dst.jpg

Copy the values of all writable tags from C<src.jpg> to C<dst.jpg>,
preserving the original tag groups.

=item exiftool -all= -tagsfromfile src.jpg -exif:all dst.jpg

Erase all meta information from C<dst.jpg> image, then copy EXIF tags from
C<src.jpg>.

=item exiftool -exif:all= -tagsfromfile @ -all:all -unsafe bad.jpg

Rebuild all EXIF meta information from scratch in an image.  This technique
can be used in JPEG images to repair corrupted EXIF information which
otherwise could not be written due to errors.  The C<Unsafe> tag is a
shortcut for unsafe EXIF tags in JPEG images which are not normally copied. 
See the L<tag name documentation|Image::ExifTool::TagNames> for more details
about unsafe tags.

=item exiftool -Tagsfromfile a.jpg out.xmp

Copy meta information from C<a.jpg> to an XMP data file.  If the XMP data
file C<out.xmp> already exists, it will be updated with the new information. 
Otherwise the XMP data file will be created.  Only metadata-only files may
be created like this (files containing images may be edited but not
created).  See L</WRITING EXAMPLES> above for another technique to generate
XMP files.

=item exiftool -tagsFromFile a.jpg -XMP:All= -ThumbnailImage= -m b.jpg

Copy all meta information from C<a.jpg> to C<b.jpg>, deleting all XMP
information and the thumbnail image from the destination.

=item exiftool -TagsFromFile src.jpg -title -author=Phil dst.jpg

Copy title from one image to another and set a new author name.

=item exiftool -TagsFromFile a.jpg -ISO -TagsFromFile b.jpg -comment
dst.jpg

Copy ISO from one image and Comment from another image to a destination
image.

=item exiftool -tagsfromfile src.jpg -exif:all --subifd:all dst.jpg

Copy only the EXIF information from one image to another, excluding SubIFD
tags.

=item exiftool '-FileModifyDateE<lt>DateTimeOriginal' dir

Use the original date from the meta information to set the same file's
filesystem modification date for all images in a directory.  (Note that
C<-TagsFromFile @> is assumed if no other B<-TagsFromFile> is specified when
redirecting information as in this example.)

=item exiftool -TagsFromFile src.jpg '-xmp:allE<lt>all' dst.jpg

Copy all possible information from C<src.jpg> and write in XMP format to
C<dst.jpg>.

=item exiftool '-Description<${FileName;s/\.[^.]*$//}' dir

Set the image Description from the file name after removing the extension. 
This example uses the L</Advanced formatting feature> to perform a
substitution operation to remove the last dot and subsequent characters from
the file name.

=item exiftool -@ iptc2xmp.args -iptc:all= a.jpg

Translate IPTC information to XMP with appropriate tag name conversions, and
delete the original IPTC information from an image.  This example uses
iptc2xmp.args, which is a file included with the ExifTool distribution that
contains the required arguments to convert IPTC information to XMP format.
Also included with the distribution are xmp2iptc.args (which performs the
inverse conversion) and a few more .args files for other conversions between
EXIF, IPTC and XMP.

=item exiftool -tagsfromfile %d%f.CR2 -r -ext JPG dir

Recursively rewrite all C<JPG> images in C<dir> with information copied from
the corresponding C<CR2> images in the same directories.

=item exiftool '-keywords+E<lt>make' image.jpg

Add camera make to list of keywords.

=item exiftool '-commentE<lt>ISO=$exif:iso Exposure=${shutterspeed}' dir

Set the Comment tag of all images in C<dir> from the values of the EXIF:ISO
and ShutterSpeed tags.  The resulting comment will be in the form "ISO=100
Exposure=1/60".

=item exiftool -TagsFromFile src.jpg -icc_profile dst.jpg

Copy ICC_Profile from one image to another.

=item exiftool -TagsFromFile src.jpg -all:all dst.mie

Copy all meta information in its original form from a JPEG image to a MIE
file.  The MIE file will be created if it doesn't exist.  This technique can
be used to store the metadata of an image so it can be inserted back into
the image (with the inverse command) later in a workflow.

=item exiftool -o dst.mie -all:all src.jpg

This command performs exactly the same task as the command above, except
that the B<-o> option will not write to an output file that already exists.

=item exiftool -b -jpgfromraw -w %d%f_%ue.jpg -execute -b -previewimage -w
%d%f_%ue.jpg -execute -tagsfromfile @ -srcfile %d%f_%ue.jpg
-overwrite_original -common_args --ext jpg DIR

[Advanced] Extract JpgFromRaw or PreviewImage from all but JPG files in DIR,
saving them with file names like C<image_EXT.jpg>, then add all meta
information from the original files to the extracted images.  Here, the
command line is broken into three sections (separated by B<-execute>
options), and each is executed as if it were a separate command.  The
B<-common_args> option causes the C<--ext jpg DIR> arguments to be applied
to all three commands, and the B<-srcfile> option allows the extracted JPG
image to be the source file for the third command (whereas the RAW files are
the source files for the other two commands).

=back

=head1 RENAMING EXAMPLES

By writing the C<FileName> and C<Directory> tags, files are renamed and/or
moved to new directories.  This can be particularly useful and powerful for
organizing files by date when combined with the B<-d> option.  New
directories are created as necessary, but existing files will not be
overwritten.  The format codes %d, %f and %e may be used in the new file
name to represent the directory, name and extension of the original file,
and %c may be used to add a copy number if the file already exists (see the
B<-w> option for details).  Note that if used within a date format string,
an extra '%' must be added to pass these codes through the date/time parser.
(And further note that in a Windows batch file, all '%' characters must also
be escaped, so in this extreme case '%%%%f' is necessary to pass a simple
'%f' through the two levels of parsing.)  See
L<https://exiftool.org/filename.html> for additional documentation and
examples.

=over 5

=item exiftool -filename=new.jpg dir/old.jpg

Rename C<old.jpg> to C<new.jpg> in directory C<dir>.

=item exiftool -directory=%e dir

Move all files from directory C<dir> into directories named by the original
file extensions.

=item exiftool '-Directory<DateTimeOriginal' -d %Y/%m/%d dir

Move all files in C<dir> into a directory hierarchy based on year, month and
day of C<DateTimeOriginal>.  eg) This command would move the file
C<dir/image.jpg> with a C<DateTimeOriginal> of C<2005:10:12 16:05:56> to
C<2005/10/12/image.jpg>.

=item exiftool -o . '-Directory<DateTimeOriginal' -d %Y/%m/%d dir

Same effect as above except files are copied instead of moved.

=item exiftool '-filename<%f_${model;}.%e' dir

Rename all files in C<dir> by adding the camera model name to the file name.
The semicolon after the tag name inside the braces causes characters which
are invalid in Windows file names to be deleted from the tag value (see the
L</Advanced formatting feature> for an explanation).

=item exiftool '-FileName<CreateDate' -d %Y%m%d_%H%M%S%%-c.%%e dir

Rename all images in C<dir> according to the C<CreateDate> date and time,
adding a copy number with leading '-' if the file already exists (C<%-c>),
and preserving the original file extension (C<%e>).  Note the extra '%'
necessary to escape the filename codes (C<%c> and C<%e>) in the date format
string.

=item exiftool -r '-FileName<CreateDate' -d %Y-%m-%d/%H%M_%%f.%%e dir

Both the directory and the filename may be changed together via the
C<FileName> tag if the new C<FileName> contains a '/'.  The example above
recursively renames all images in a directory by adding a C<CreateDate>
timestamp to the start of the filename, then moves them into new directories
named by date.

=item exiftool '-FileName<${CreateDate}_$filenumber.jpg' -d %Y%m%d -ext jpg .

Set the filename of all JPG images in the current directory from the
CreateDate and FileNumber tags, in the form "20060507_118-1861.jpg".

=back

=head1 GEOTAGGING EXAMPLES

ExifTool implements geotagging from GPS log files via 3 special tags: Geotag
(which for convenience is also implemented as an exiftool option), Geosync
and Geotime. The examples below highlight some geotagging features.  See
L<https://exiftool.org/geotag.html> for additional documentation.  (Note
that geotagging from known GPS coordinates is done by writing the
L<GPS tags|Image::ExifTool::TagNames/GPS Tags> directly rather than using
the B<-geotag> option.)

=over 5

=item exiftool -geotag track.log a.jpg

Geotag an image (C<a.jpg>) from position information in a GPS track log
(C<track.log>).  Since the C<Geotime> tag is not specified, the value of
DateTimeOriginal is used for geotagging.  Local system time is assumed
unless DateTimeOriginal contains a timezone.

=item exiftool -geotag track.log -geolocate=geotag a.jpg

Geotag an image and also write geolocation information of the nearest city
(city name, state/province and country).  Read here for more details about
the Geolocation feature: L<https://exiftool.org/geolocation.html#Write>

=item exiftool -geotag t.log -geotime='2009:04:02 13:41:12-05:00' a.jpg

Geotag an image with the GPS position for a specific time.

=item exiftool -geotag log.gpx '-xmp:geotimeE<lt>createdate' dir

Geotag all images in directory C<dir> with XMP tags instead of EXIF tags,
based on the image CreateDate.

=item exiftool -geotag a.log -geosync=-20 dir

Geotag images in directory C<dir>, accounting for image timestamps which
were 20 seconds ahead of GPS.

=item exiftool -geotag a.log -geosync=1.jpg -geosync=2.jpg dir

Geotag images using time synchronization from two previously geotagged images
(1.jpg and 2.jpg), synchronizing the image and GPS times using a linear time
drift correction.

=item exiftool -geotag a.log '-geotimeE<lt>${createdate}+01:00' dir

Geotag images in C<dir> using CreateDate with the specified timezone.  If
CreateDate already contained a timezone, then the timezone specified on the
command line is ignored.

=item exiftool -geotag= a.jpg

Delete GPS tags which may have been added by the geotag feature.  Note that
this does not remove all GPS tags -- to do this instead use C<-gps:all=>.

=item exiftool -xmp:geotag= a.jpg

Delete XMP GPS tags which were added by the geotag feature.

=item exiftool -xmp:geotag=track.log a.jpg

Geotag an image with XMP tags, using the time from DateTimeOriginal.

=item exiftool -geotag a.log -geotag b.log -r dir

Combine multiple track logs and geotag an entire directory tree of images.

=item exiftool -geotag 'tracks/*.log' -r dir

Read all track logs from the C<tracks> directory.

=item exiftool -p gpx.fmt dir > out.gpx

Generate a GPX track log from all images in directory C<dir>.  This example
uses the C<gpx.fmt> file included in the full ExifTool distribution package
and assumes that the images in C<dir> have all been previously geotagged.

=back

=head1 PIPING EXAMPLES

=over 5

=item cat a.jpg | exiftool -

Extract information from stdin.

=item exiftool image.jpg -thumbnailimage -b | exiftool -

Extract information from an embedded thumbnail image.

=item cat a.jpg | exiftool -iptc:keywords+=fantastic - > b.jpg

Add an IPTC keyword in a pipeline, saving output to a new file.

=item curl -s http://a.domain.com/bigfile.jpg | exiftool -fast -

Extract information from an image over the internet using the cURL utility. 
The B<-fast> option prevents exiftool from scanning for trailer information,
so only the meta information header is transferred.

=item exiftool a.jpg -thumbnailimage -b | exiftool -comment=wow - |
exiftool a.jpg -thumbnailimage'<=-'

Add a comment to an embedded thumbnail image.  (Why anyone would want to do
this I don't know, but I've included this as an example to illustrate the
flexibility of ExifTool.)

=back

=head1 INTERRUPTING EXIFTOOL

Interrupting exiftool with a CTRL-C or SIGINT will not result in partially
written files or temporary files remaining on the hard disk.  The exiftool
application traps SIGINT and defers it until the end of critical processes
if necessary, then does a proper cleanup before exiting.

=head1 EXIT STATUS

The exiftool application exits with a status of 0 on success, or 1 if an
error occurred, or 2 if all files failed the B<-if> condition (for any of
the commands if B<-execute> was used).

=head1 AUTHOR

Copyright 2003-2024, Phil Harvey

This is free software; you can redistribute it and/or modify it under the
same terms as Perl itself.

=head1 SEE ALSO

L<Image::ExifTool(3pm)|Image::ExifTool>,
L<Image::ExifTool::TagNames(3pm)|Image::ExifTool::TagNames>,
L<Image::ExifTool::Shortcuts(3pm)|Image::ExifTool::Shortcuts>,
L<Image::ExifTool::Shift.pl|Image::ExifTool::Shift.pl>

=cut

#------------------------------------------------------------------------------
# end