fix bug #209: titles with colon bad parsed

1 files changed, 46 insertions, 0 deletions

diff --git a/inc/poche/PocheReadability.php b/inc/poche/PocheReadability.php
new file mode 100644
index 00000000..48ae90d0
--- /dev/null
+++ b/inc/poche/PocheReadability.php
@@ -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;
+    }
+}
\ No newline at end of file