oss/wallabag
oss/wallabag
1
0
Fork 0
mirror of https://github.com/wallabag/wallabag.git synced 2025-08-01 17:38:38 +00:00
Code Wiki Activity

fix bug #209: titles with colon bad parsed

Browse source
This commit is contained in:
Nicolas Lœuillet 2013-09-12 19:28:59 +02:00
parent 084ec2a63d
commit b9523a0ba0
3 changed files with 49 additions and 9 deletions
Download patch file Download diff file Expand all files Collapse all files

46
inc/poche/PocheReadability.php Normal file
View file

@ -0,0 +1,46 @@
<?php
class PocheReadability extends Readability
{
/**
* Get the article title as an H1.
*
* @return DOMElement
*/
protected function getArticleTitle() {
$curTitle = '';
$origTitle = '';
try {
$curTitle = $origTitle = $this->getInnerText($this->dom->getElementsByTagName('title')->item(0));
} catch(Exception $e) {}
if (preg_match('/ [\|\-] /', $curTitle))
{
$curTitle = preg_replace('/(.*)[\|\-] .*/i', '$1', $origTitle);
if (count(explode(' ', $curTitle)) < 3) {
$curTitle = preg_replace('/[^\|\-]*[\|\-](.*)/i', '$1', $origTitle);
}
}
else if(strlen($curTitle) > 150 || strlen($curTitle) < 15)
{
$hOnes = $this->dom->getElementsByTagName('h1');
if($hOnes->length == 1)
{
$curTitle = $this->getInnerText($hOnes->item(0));
}
}
$curTitle = trim($curTitle);
if (count(explode(' ', $curTitle)) <= 4) {
$curTitle = $origTitle;
}
$articleTitle = $this->dom->createElement('h1');
$articleTitle->innerHTML = $curTitle;
return $articleTitle;
}
}