Replace Locator and Content\Type\Sniffer with Content\Detector

src/Content/Detector.php

@@ -0,0 +1,52 @@
+<?php
+
+// SPDX-FileCopyrightText: 2004-2023 Ryan Parman, Sam Sneddon, Ryan McCue
+// SPDX-License-Identifier: BSD-3-Clause
+
+declare(strict_types=1);
+
+namespace SimplePie\Content;
+
+use SimplePie\HTTP\Response;
+use SimplePie\SimplePie;
+
+/**
+ * Helper for feed auto-discovery and type sniffing
+ *
+ *
+ * This interface replaces
+ * - \SimplePie\Locator and
+ * - \SimplePie\Content\Type\Sniffer
+ */
+interface Detector
+{
+ /**
+ * Discover possible feed urls from HTML response
+ *
+ * @see https://simplepie.org/wiki/reference/simplepie/set_autodiscovery_level
+ *
+ * Inspired by the Ultra-liberal RSS locator from Mark Pilgrim
+ * @link http://web.archive.org/web/20110607232437/http://diveintomark.org/archives/2002/08/15/ultraliberal_rss_locator
+ *
+ * @param SimplePie::LOCATOR_* $discovery_level
+ *
+ * @return string[] Array of possible feed urls. The urls are not requested or checked for containing a feed
+ */
+ public function discover_possible_feed_urls(Response $response, int $discovery_level = SimplePie::LOCATOR_ALL): array;
+
+ /**
+ * Check if the response contains a feed
+ *
+ * @return bool
+ */
+ public function contains_feed(Response $response): bool;
+
+ /**
+ * Get the IANA Media-Type of the provided response
+ *
+ * @link https://www.iana.org/assignments/media-types/media-types.xhtml
+ *
+ * @return string Actual Media-Type
+ */
+ public function detect_media_type(Response $response): string;
+}

src/Content/VerifiedFeedsDetector.php

@@ -0,0 +1,119 @@
+<?php
+
+// SPDX-FileCopyrightText: 2004-2023 Ryan Parman, Sam Sneddon, Ryan McCue
+// SPDX-License-Identifier: BSD-3-Clause
+
+declare(strict_types=1);
+
+namespace SimplePie\Content;
+
+use SimplePie\Content\Type\Sniffer;
+use SimplePie\File;
+use SimplePie\HTTP\Response;
+use SimplePie\Locator;
+use SimplePie\Registry;
+use SimplePie\RegistryAware;
+use SimplePie\SimplePie;
+
+/**
+ * BC helper for feed auto-discovery and type sniffing
+ *
+ *
+ * This class uses
+ * - \SimplePie\Locator and
+ * - \SimplePie\Content\Type\Sniffer
+ *
+ * @internal
+ */
+final class VerifiedFeedsDetector implements Detector, RegistryAware
+{
+ /**
+ * @var Registry $registry
+ */
+ private $registry;
+
+ public function set_registry(Registry $registry)
+ {
+ $this->registry = $registry;
+ }
+
+ /**
+ * Discover possible feed urls from HTML response
+ *
+ * @see https://simplepie.org/wiki/reference/simplepie/set_autodiscovery_level
+ *
+ * Inspired by the Ultra-liberal RSS locator from Mark Pilgrim
+ * @link http://web.archive.org/web/20110607232437/http://diveintomark.org/archives/2002/08/15/ultraliberal_rss_locator
+ *
+ * @param SimplePie::LOCATOR_* $discovery_level
+ *
+ * @return string[] Array of possible feed urls. The urls are not requested or checked for containing a feed
+ */
+ public function discover_possible_feed_urls(Response $response, int $discovery_level = SimplePie::LOCATOR_ALL): array
+ {
+ /** @var Locator */
+ $locator = $this->registry->create(
+ Locator::class,
+ [
+ (! $response instanceof File) ? File::fromResponse($response) : $response,
+ 10,
+ null,
+ 10,
+ false,
+ []
+ ]
+ );
+
+ $all_discovered_feeds = [];
+
+ /** @var File|null */
+ $result = $locator->find($discovery_level, $all_discovered_feeds);
+
+ if (is_object($result) && $result instanceof Response) {
+ return [$result->get_permanent_uri()];
+ }
+
+ return [];
+ }
+
+ /**
+ * Check if the response contains a feed
+ *
+ * @return bool
+ */
+ public function contains_feed(Response $response): bool
+ {
+ /** @var Locator */
+ $locator = $this->registry->create(
+ Locator::class,
+ [
+ (! $response instanceof File) ? File::fromResponse($response) : $response,
+ 10,
+ null,
+ 10,
+ false,
+ []
+ ]
+ );
+
+ return (bool) $locator->is_feed($response, false);
+ }
+
+ /**
+ * Get the IANA Media-Type of the provided response
+ *
+ * @link https://www.iana.org/assignments/media-types/media-types.xhtml
+ *
+ * @return string Actual Media-Type
+ */
+ public function detect_media_type(Response $response): string
+ {
+ /** @var Sniffer */
+ $sniffer = $this->registry->create(
+ Sniffer::class,
+ [$response]
+ );
+
+ return (string) $sniffer->get_type();
+ }
+}

src/Registry.php

@@ -7,7 +7,9 @@
 
 namespace SimplePie;
 
+use SimplePie\Content\Detector;
 use SimplePie\Content\Type\Sniffer;
+use SimplePie\Content\VerifiedFeedsDetector;
 use SimplePie\Parse\Date;
 use SimplePie\XML\Declaration\Parser as DeclarationParser;
 
@@ -45,6 +47,7 @@ class Registry
  Misc::class => Misc::class,
  DeclarationParser::class => DeclarationParser::class,
  Date::class => Date::class,
+ Detector::class => VerifiedFeedsDetector::class,
  ];
 
  /**

src/SimplePie.php

@@ -18,6 +18,7 @@
 use SimplePie\Cache\DataCache;
 use SimplePie\Cache\NameFilter;
 use SimplePie\Cache\Psr16;
+use SimplePie\Content\Detector;
 use SimplePie\Content\Type\Sniffer;
 use SimplePie\Exception as SimplePieException;
 use SimplePie\Exception\HttpException;
@@ -2008,7 +2009,10 @@ protected function fetch_data(&$cache)
  );
  }
 
- if (!$locate->is_feed($file)) {
+ /** @var Detector */
+ $detector = $this->registry->create(Detector::class);
+
+ if (! $detector->contains_feed($file)) {
  $copyStatusCode = $file->get_status_code();
  $copyContentType = $file->get_header_line('content-type');
  try {
@@ -2026,10 +2030,47 @@ protected function fetch_data(&$cache)
  }
  // Now also do feed discovery, but if microformats were found don't
  // overwrite the current value of file.
- $discovered = $locate->find(
- $this->autodiscovery,
- $this->all_discovered_feeds
+ $possible_feed_urls = $detector->discover_possible_feed_urls(
+ $file,
+ $this->autodiscovery
  );
+
+ $discovered = null;
+ $checked_feeds = 0;
+
+ foreach ($possible_feed_urls as $href) {
+ $checked_feeds++;
+
+ try {
+ //code...
+ $possible_feed = $http_client->request(
+ Client::METHOD_GET,
+ $href,
+ ['Accept' => SimplePie::DEFAULT_HTTP_ACCEPT_HEADER]
+ );
+ } catch (HttpException $th) {
+ continue;
+ }
+
+ if (
+ (
+ preg_match('/^http(s)?:\/\//i', $href)
+ || (
+ $possible_feed->get_status_code() === 200
+ || $possible_feed->get_status_code() > 206
+ && $possible_feed->get_status_code() < 300
+ )
+ )
+ && $detector->contains_feed($possible_feed)
+ ) {
+ $this->all_discovered_feeds[$href] = $possible_feed;
+ }
+ }
+
+ if (! empty($this->all_discovered_feeds)) {
+ $discovered = $this->all_discovered_feeds[0];
+ }
+
  if ($microformats) {
  if ($hub = $locate->get_rel_link('hub')) {
  $self = $locate->get_rel_link('self');