Changeset 16355

Timestamp:

05/09/08 16:34:27 (2 months ago)

Author:

xfade

Message:

Add more intelligent word count truncation to social news. Based on algo by rambo. Also add configurable html tag filter.

Files:

• branches/MidCOM_2_8/org.maemo.socialnews/config/config.inc (modified) (1 diff)
• branches/MidCOM_2_8/org.maemo.socialnews/config/manifest.inc (modified) (1 diff)
• branches/MidCOM_2_8/org.maemo.socialnews/documentation/CHANGES (modified) (1 diff)
• branches/MidCOM_2_8/org.maemo.socialnews/handler/index.php (modified) (2 diffs)

branches/MidCOM_2_8/org.maemo.socialnews/config/config.inc

@@ -9 +9 @@
-'frontpage_show_main_items' => 3,
-'frontpage_show_abstract_length' => 1000,
+'frontpage_show_abstract_length_words' => 0,
+'frontpage_allowed_tags' => '<p><a><img><b><i><strong><pre><ul><li><br>',
-
-'frontpage_show_secondary_items' => 7,

branches/MidCOM_2_8/org.maemo.socialnews/config/manifest.inc

@@ -1 +1 @@
-'name' => 'org.maemo.socialnews',
-'purecode' => false,
-4
+5
-'state' => 'stable',
-'privileges' => array(),

branches/MidCOM_2_8/org.maemo.socialnews/documentation/CHANGES

@@ -10 +10 @@
-2008-05-09 xfade
-  + RSS feed for latest items
+  + Add more intelligent word count truncater
-
-2007-08-29 bergie

branches/MidCOM_2_8/org.maemo.socialnews/handler/index.php

@@ -147 +147 @@
-    }
-
+    private function word_count_truncater(&$content, &$url)
+    {
+        // Normalize newlines to \n
+        $content = preg_replace("/\n\r|\r\n|\r/", "\n", $content);
+        $content = strip_tags($content, $this->_config->get('frontpage_allowed_tags'));
+
+        $word_limit = $this->_config->get('frontpage_show_abstract_length_words');
+        // Naive but works
+        $words = count(preg_split('/\b/', strip_tags($content)));
+        if ($words <= $word_limit)
+        {
+            return $content;
+        }
+        // The difficult part starts here
+        // In stead of trying parse the HTML tree to get N words while correctly closing all tags we get first paragraph
+
+        $content_matches = array();
+        $ret = "";
+        if (preg_match("%((^\s+.*?)|(^.*?))(</p>|(\n\s*){2,}|(<br\s*/?>\s*){2,}|\w+\s*<p>)%ms", $content, $content_matches))
+        {
+            $first_paragraph =& $content_matches[1];
+            $first_paragraph_words = count(preg_split('/\b/', strip_tags($first_paragraph)));
+
+            // Check that our "first paragraph" is of sane size
+            if ($first_paragraph_words <= $word_limit)
+            {
+                $remaining_words = $words - $first_paragraph_words;
+                $ret = $first_paragraph;
+                if ($remaining_words < 1)
+                {
+                    return;
+                }
+                $ret .= "<div class='entry-truncated'><a href='{$url}'>Click to read {$remaining_words} more words</a></div>\n";
+                return $ret;
+            }
+        }
+        else 
+        {
+            //This post doesn't have nice <p> paragraphs
+            $ret = substr($content,0,strpos($content,"\n"));
+            $words = count(preg_split('/\b/', strip_tags($ret)));
+            if ($words < $word_limit)
+            {
+                $ret .= "<div class='entry-truncated'><a href='{$url}'>Click to read {$remaining_words} more words</a></div>\n";
+                return $ret;
+            }
+        }
+
+        // Fallback, show only "xxx words, click here to read"
+        $ret .= "<div class='entry-truncated'><a href='{$url}'>Click to read {$words} words</a></div>\n";
+        return $ret;
+    }
+
-    private function generate_caption($data, $getCnt)
-    {
@@ -221 +274 @@
-            if (empty($article->abstract))
-            {
-
+
+
+
+
+
+
+
+
-            }
-            else
-            {
-
+
+
+
+
+
+
+
+
-            }
-