Skip to content
This repository has been archived by the owner on Mar 19, 2020. It is now read-only.
/ pymarc Public archive
forked from edsu/pymarc

process MARC records from Python

Notifications You must be signed in to change notification settings

Wooble/pymarc

 
 

Repository files navigation

_|_|_|    _|    _|  _|_|_|  _|_|      _|_|_|  _|  _|_|    _|_|_|
_|    _|  _|    _|  _|    _|    _|  _|    _|  _|_|      _|
_|    _|  _|    _|  _|    _|    _|  _|    _|  _|        _|
_|_|_|      _|_|_|  _|    _|    _|    _|_|_|  _|          _|_|_|
_|              _|
_|          _|_|

Build Status Coverage Status

pymarc is a python library for working with bibliographic data encoded in MARC21. It should work under python 2.x and 3.x. It provides an API for reading, writing and modifying MARC records. It was mostly designed to be an emergency eject seat, for getting your data assets out of MARC and into some kind of saner representation. However over the years it has been used to create and modify MARC records, since despite repeated calls for it to die as a format, MARC seems to be living quite happily as a zombie.

Below are some common examples of how you might want to use pymarc. If you run across an example that you think should be here please send a pull request.

Installation

You'll probably just want to use pip to install pymarc:

pip install pymarc

If you'd like to download and install the latest source you'll need git:

git clone git://github.com/edsu/pymarc.git

You'll also need setuptools. Once you have the source and setuptools run the pymarc test suite to make sure things are in order with the distribution:

python setup.py test

And then install:

python setup.py install

Reading

Most often you will have some MARC data and will want to extract data from it. Here's an example of reading a batch of records and printing out the title. If you are curious this example uses the batch file available here in pymarc repository:

from pymarc import MARCReader
with open('test/marc.dat', 'rb') as fh:
    reader = MARCReader(fh)
    for record in reader:
        print(record.title())
The pragmatic programmer : from journeyman to master /
Programming Python /
Learning Python /
Python cookbook /
Python programming for the absolute beginner /
Web programming : techniques for integrating Python, Linux, Apache, and MySQL /
Python programming on Win32 /
Python programming : an introduction to computer science /
Python Web programming /
Core python programming /
Python and Tkinter programming /
Game programming with Python, Lua, and Ruby /
Python programming patterns /
Python programming with the Java class libraries : a tutorial for building Web
and Enterprise applications /
Learn to program using Python : a tutorial for hobbyists, self-starters, and all
who want to learn the art of computer programming /
Programming with Python /
BSD Sockets programming from a multi-language perspective /
Design patterns : elements of reusable object-oriented software /
Introduction to algorithms /
ANSI Common Lisp /

A pymarc.Record object has a few handy methods like title for getting at bits of a bibliographic record, others include: author, isbn, subjects, location, notes, physicaldescription, publisher, pubyear. But really, to work with MARC data you need to understand the numeric field tags and subfield codes that are used to designate various bits of information. There is a lot more hiding in a MARC record than these methods provide access to. For example the title method extracts the information from the 245 field, subfields a and b. You can access 245a like so:

print(record['245']['a'])

Some fields like subjects can repeat. In cases like that you will want to use get_fields to get all of them as pymarc.Field objects, which you can then interact with further:

for f in record.get_fields('650'):
    print(f)

If you are new to MARC fields Understanding MARC is a pretty good primer, and the MARC 21 Formats page at the Library of Congress is a good reference once you understand the basics.

Writing

Here's an example of creating a record and writing it out to a file.

from pymarc import Record, Field
record = Record()
record.add_field(
    Field(
        tag = '245',
        indicators = ['0','1'],
        subfields = [
            'a', 'The pragmatic programmer : ',
            'b', 'from journeyman to master /',
            'c', 'Andrew Hunt, David Thomas.'
        ]))
with open('file.dat', 'wb') as out:
    out.write(record.as_marc())

Updating

Updating works the same way, you read it in, modify it, and then write it out again:

from pymarc import MARCReader
with open('test/marc.dat', 'rb') as fh:
    reader = MARCReader(fh)
    record = next(reader)
    record['245']['a'] = 'The Zombie Programmer'
with open('file.dat', 'wb') as out:
    out.write(record.as_marc())

JSON and XML

If you find yourself using MARC data a fair bit, and distributing it, you may make other developers a bit happier by using the JSON or XML serializations. The main benefit to using XML or JSON is that the UTF8 character encoding is used, rather than the frustratingly archaic MARC8 encoding. Also they will be able to use standard JSON and XML reading/writing tools to get at the data they want instead of some crazy MARC processing library like, ahem, pymarc.

pymarc's support for JSON and XML is currently a bit lopsided and ad hoc. pymarc allows you to read XML in a variety of ways, but not write it. On the other hand pymarc allows you to write JSON, but not read it. Part of the reason for this unevenness is that the functionality was added to solve a particular need at a particular time. If you are interested in providing a more holistic solution pull requests (with unit tests) are always welcome.

XML

To parse a file of MARCXML records you can:

from pymarc import parse_xml_to_array

records = parse_xml_to_array('test/batch.xml')

If you have a large XML file and would rather not read them all into memory you can:

from pymarc import map_xml

def print_title(r):
    print(r.title())

map_xml(print_title, 'test/batch.xml')

Also, if you prefer you can pass in a file like object in addition to the path to both map_xml and parse_xml_to_array:

records = parse_xml_to_array(open('batch.xml'))

JSON

JSON support is fairly minimal in that you can call a pymarc.Record's as_json() method to return JSON for a given MARC Record:

from pymarc import MARCReader

record = MARCReader('test/one.dat')
print(record.as_json(indent=2))
{
  "leader": "01060cam  22002894a 4500",
  "fields": [
    {
      "001": "11778504"
    }, 
    {
      "010": {
        "ind1": " ", 
        "subfields": [
          {
            "a": "   99043581 "
          }
        ], 
        "ind2": " "
      }
    }, 
    {
      "100": {
        "ind1": "1", 
        "subfields": [
          {
            "a": "Hunt, Andrew,"
          }, 
          {
            "d": "1964-"
          }
        ], 
        "ind2": " "
      }
    }, 
    {
      "245": {
        "ind1": "1", 
        "subfields": [
          {
            "a": "The pragmatic programmer :"
          }, 
          {
            "b": "from journeyman to master /"
          }, 
          {
            "c": "Andrew Hunt, David Thomas."
          }
        ], 
        "ind2": "4"
      }
    }, 
    {
      "260": {
        "ind1": " ", 
        "subfields": [
          {
            "a": "Reading, Mass :"
          }, 
          {
            "b": "Addison-Wesley,"
          }, 
          {
            "c": "2000."
          }
        ], 
        "ind2": " "
      }
    }, 
    {
      "300": {
        "ind1": " ", 
        "subfields": [
          {
            "a": "xxiv, 321 p. ;"
          }, 
          {
            "c": "24 cm."
          }
        ], 
        "ind2": " "
      }
    }, 
    {
      "504": {
        "ind1": " ", 
        "subfields": [
          {
            "a": "Includes bibliographical references."
          }
        ], 
        "ind2": " "
      }
    }, 
    {
      "650": {
        "ind1": " ", 
        "subfields": [
          {
            "a": "Computer programming."
          }
        ], 
        "ind2": "0"
      }
    }, 
    {
      "700": {
        "ind1": "1", 
        "subfields": [
          {
            "a": "Thomas, David,"
          }, 
          {
            "d": "1956-"
          }
        ], 
        "ind2": " "
      }
    }
  ]
}

Support

The pymarc developers encourage you to join the pymarc Google Group if you need help. Also, please feel free to use issue tracking on Github to submit feature requests or bug reports. If you've got an itch to scratch, please scratch it, and send merge requests on Github.

If you start working with MARC you may feel like you need moral support in addition to technical support. The #code4lib channel on Freenode is a good place for both.

Copyright

Copyright (c) 2005-2016 Gabriel Farrell, Mark Matienzo, Geoffrey Spear, Ed Summers

License

BSD

About

process MARC records from Python

Resources

Stars

Watchers

Forks

Packages

No packages published

Languages

  • Python 100.0%