oss/wallabag
oss/wallabag
1
0
Fork 0
mirror of https://github.com/wallabag/wallabag.git synced 2025-07-27 17:28:39 +00:00
Code Wiki Activity

ContentProxy: replace ignoreUrl with new RuleBasedIgnoreOriginProcessor

Browse source 
Signed-off-by: Kevin Decherf <kevin@kdecherf.com>
This commit is contained in:
Kevin Decherf 2019-08-11 23:55:52 +02:00
parent 2495b19761
commit b22eb27623
2 changed files with 108 additions and 60 deletions
Download patch file Download diff file Expand all files Collapse all files

41
src/Wallabag/CoreBundle/Helper/ContentProxy.php
View file

@ -19,6 +19,7 @@ class ContentProxy
{
protected $graby;
protected $tagger;
protected $ignoreOriginProcessor;
protected $validator;
protected $logger;
protected $mimeGuesser;
@ -26,10 +27,11 @@ class ContentProxy
protected $eventDispatcher;
protected $storeArticleHeaders;
public function __construct(Graby $graby, RuleBasedTagger $tagger, ValidatorInterface $validator, LoggerInterface $logger, $fetchingErrorMessage, $storeArticleHeaders = false)
public function __construct(Graby $graby, RuleBasedTagger $tagger, RuleBasedIgnoreOriginProcessor $ignoreOriginProcessor, ValidatorInterface $validator, LoggerInterface $logger, $fetchingErrorMessage, $storeArticleHeaders = false)
{
$this->graby = $graby;
$this->tagger = $tagger;
$this->ignoreOriginProcessor = $ignoreOriginProcessor;
$this->validator = $validator;
$this->logger = $logger;
$this->mimeGuesser = new MimeTypeExtensionGuesser();
@ -356,7 +358,7 @@ class ContentProxy
$diff_keys = array_keys($diff);
sort($diff_keys);
if ($this->ignoreUrl($entry->getUrl())) {
if ($this->ignoreOriginProcessor->process($entry)) {
$entry->setUrl($url);
return false;
@ -395,41 +397,6 @@ class ContentProxy
}
}
/**
* Check entry url against an ignore list to replace with content url.
*
* XXX: move the ignore list in the database to let users handle it
*
* @param string $url url to test
*
* @return bool true if url matches ignore list otherwise false
*/
private function ignoreUrl($url)
{
$ignored_hosts = ['feedproxy.google.com', 'feeds.reuters.com'];
$ignored_patterns = ['https?://www\.lemonde\.fr/tiny.*'];
$parsed_url = parse_url($url);
$filtered = array_filter($ignored_hosts, function ($var) use ($parsed_url) {
return $var === $parsed_url['host'];
});
if ([] !== $filtered) {
return true;
}
$filtered = array_filter($ignored_patterns, function ($var) use ($url) {
return preg_match("`$var`i", $url);
});
if ([] !== $filtered) {
return true;
}
return false;
}
/**
* Validate that the given content has at least a title, an html and a url.
*