]> git.immae.eu Git - github/wallabag/wallabag.git/blame - inc/poche/PocheReadability.php
[add] add RSS feed for archive
[github/wallabag/wallabag.git] / inc / poche / PocheReadability.php
CommitLineData
b9523a0b
NL
1<?php
2
3class PocheReadability extends Readability
4{
5 /**
6 * Get the article title as an H1.
7 *
8 * @return DOMElement
9 */
10 protected function getArticleTitle() {
11 $curTitle = '';
12 $origTitle = '';
13
14 try {
15 $curTitle = $origTitle = $this->getInnerText($this->dom->getElementsByTagName('title')->item(0));
16 } catch(Exception $e) {}
17
18 if (preg_match('/ [\|\-] /', $curTitle))
19 {
20 $curTitle = preg_replace('/(.*)[\|\-] .*/i', '$1', $origTitle);
21
22 if (count(explode(' ', $curTitle)) < 3) {
23 $curTitle = preg_replace('/[^\|\-]*[\|\-](.*)/i', '$1', $origTitle);
24 }
25 }
26 else if(strlen($curTitle) > 150 || strlen($curTitle) < 15)
27 {
28 $hOnes = $this->dom->getElementsByTagName('h1');
29 if($hOnes->length == 1)
30 {
31 $curTitle = $this->getInnerText($hOnes->item(0));
32 }
33 }
34
35 $curTitle = trim($curTitle);
36
37 if (count(explode(' ', $curTitle)) <= 4) {
38 $curTitle = $origTitle;
39 }
40
41 $articleTitle = $this->dom->createElement('h1');
42 $articleTitle->innerHTML = $curTitle;
43
44 return $articleTitle;
45 }
46}