infong/php-readability
1
0
Fork 0
mirror of https://github.com/j0k3r/php-readability.git synced 2026-09-26 22:16:12 +00:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #16 from j0k3r/info

Browse Source 
“info” is too agressive in unlikelyCandidates
This commit is contained in:
Jeremy Benoist
2016-10-03 09:13:35 +02:00
committed by GitHub
parent fb9810a827 3de4e918b4
commit 47ce4fb7f9
1 changed files with 7 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/Readability.php
+7 -4
View File
@@ -66,7 +66,7 @@ class Readability implements LoggerAwareInterface
public $tidied = false;
// article domain regexp for calibration
protected $domainRegExp = null;
protected $body = null; //
protected $body = null;
// Cache the body HTML in case we need to re-use it later
protected $bodyCache = null;
// 1 | 2 | 4; // Start with all processing flags set.
@@ -83,11 +83,11 @@ class Readability implements LoggerAwareInterface
* Defined up here so we don't instantiate them repeatedly in loops.
*/
public $regexps = array(
'unlikelyCandidates' => '/display\s*:\s*none|ignore|\binfo|annoy|clock|date|time|author|intro|links|hidd?e|about|archive|\bprint|bookmark|tags|tag-list|share|search|social|robot|published|combx|comment|mast(?:head)|subscri|community|category|disqus|extra|head|head(?:er|note)|floor|foot(?:er|note)|menu|tool|function|nav|remark|rss|shoutbox|tool|widget|meta|banner|sponsor|adsense|inner-?ad|ad-|sponsor|\badv\b|\bads\b|agr?egate?|pager|sidebar|popup|tweet|twitter/i',
'unlikelyCandidates' => '/display\s*:\s*none|ignore|\binfos?\b|annoy|clock|date|time|author|intro|links|hidd?e|about|archive|\bprint|bookmark|tags|tag-list|share|search|social|robot|published|combx|comment|mast(?:head)|subscri|community|category|disqus|extra|head|head(?:er|note)|floor|foot(?:er|note)|menu|tool|function|nav|remark|rss|shoutbox|tool|widget|meta|banner|sponsor|adsense|inner-?ad|ad-|sponsor|\badv\b|\bads\b|agr?egate?|pager|sidebar|popup|tweet|twitter/i',
'okMaybeItsACandidate' => '/article\b|contain|\bcontent|column|general|detail|shadow|lightbox|blog|body|entry|main|page/i',
'positive' => '/read|full|article|body|\bcontent|contain|entry|main|markdown|page|attach|pagination|post|text|blog|story/i',
'negative' => '/bottom|stat|info|discuss|e[\-]?mail|comment|reply|log.{2}(n|ed)|sign|single|combx|com-|contact|_nav|link|media|\bout|promo|\bad-|related|scroll|shoutbox|sidebar|sponsor|shopping|teaser|recommend/i',
'divToPElements' => '/<(?:blockquote|code|div|article|footer|aside|img|p|pre|dl|ol|ul)/mi',
'divToPElements' => '/<(?:blockquote|header|section|code|div|article|footer|aside|img|p|pre|dl|ol|ul)/mi',
'killBreaks' => '/(<br\s*\/?>([ \r\n\s]|&nbsp;?)*)+/',
'media' => '!//(?:[^\.\?/]+\.)?(?:youtu(?:be)?|soundcloud|dailymotion|vimeo|pornhub|xvideos|twitvid|rutube|viddler)\.(?:com|be|org|net)/!i',
'skipFootnoteLink' => '/^\s*(\[?[a-z0-9]{1,2}\]?|^|edit|citation needed)\s*$/i',
@@ -376,6 +376,7 @@ class Readability implements LoggerAwareInterface
* Debug.
*
* @deprecated use $this->logger->debug() instead
* @codeCoverageIgnore
*/
protected function dbg($msg)
{
@@ -386,6 +387,7 @@ class Readability implements LoggerAwareInterface
* Dump debug info.
*
* @deprecated since Monolog gather log, we don't need it
* @codeCoverageIgnore
*/
protected function dump_dbg()
{
@@ -730,8 +732,9 @@ class Readability implements LoggerAwareInterface
for ($nodeIndex = 0; ($node = $allElements->item($nodeIndex)); ++$nodeIndex) {
$tagName = $node->tagName;
// Some well known site uses sections as paragraphs.
if (strcasecmp($tagName, 'p') === 0 || strcasecmp($tagName, 'td') === 0 || strcasecmp($tagName, 'section') === 0) {
if (strcasecmp($tagName, 'p') === 0 || strcasecmp($tagName, 'td') === 0 || strcasecmp($tagName, 'pre') === 0 || strcasecmp($tagName, 'section') === 0) {
$nodesToScore[] = $node;
}