Changeset 138

Timestamp:

03/07/08 19:01:55 (9 months ago)

Author:

Slynderdale

Message:

Fixes #32 and adds transliteration support to Phergie's base as well as making URL and FeedTicker? use the new transliterate function.

Files:

• trunk/Phergie/Plugin/Abstract/Base.php (modified) (1 diff)
• trunk/Phergie/Plugin/FeedTicker.php (modified) (6 diffs)
• trunk/Phergie/Plugin/Url.php (modified) (2 diffs)

trunk/Phergie/Plugin/Abstract/Base.php

@@ -214 +214 @@
-
-    /**
+    * Decodes the entities of a given string and
+    * transliterates the UTF-8 string into corresponding ASCII characters.
+    *
+    * @param string $text The text to decode.
+    * @param string $charSetFrom The charset used as the base chrset to convert from
+    * @param string $charSetTo The charset used to convert to
+    * @return string
+    */
+    public function decodeTranslit($text, $charSetFrom = 'UTF-8', $charSetTo = 'ISO-8859-1') {
+        $text = html_entity_decode($text, ENT_QUOTES, $charSetFrom);
+        $text = preg_replace('/&#0*([0-9]+);/me', "$this->codeToUtf(\\1)", $text);
+        $text = preg_replace('/&#x0*([a-f0-9]+);/mei', "$this->codeToUtf(hexdec(\\1))", $text);
+
+        // Use the translit extension if installed else fallback on to basic transliteration
+        if (extension_loaded('iconv')) {
+                  $text = iconv($charSetFrom, $charSetTo.'//TRANSLIT', $text);
+        // Transliteration supprt via the translit extension is still experimental
+        } else if (false && extension_loaded('translit')) {
+            $text = transliterate($text, array('han_transliterate', 'diacritical_remove'), $charSetFrom, $charSetTo);
+        } else {
+            $text = strtr($text, '��������������������������',
+                                 'AAAAAAACEEEEIIIIDNOOOOOOUUUUYPYaaaaaaaceeeeiiiidnoooooouuuuypy');
+            $text = preg_replace('{[^a-z0-9&|"#\'\{\}()�^!�\[\]$*���%�`~=+:/;.,?><\\ _-]}i', '', $text);
+            $text = utf8_decode($text);
+        }
+
+        return $text;
+    }
+
+    /**
+    * Converts a given unicode to its UTF-8 equivalent.
+    *
+    * @param int $code The code to decode to UTF-8.
+    * @return string
+    */
+    public function codeToUtf($code) {
+        $code = intval($code);
+        switch ($code) {
+            // 1 byte, 7 bits
+            case 0:
+                return chr(0);
+                    case ($code & 0x7F):
+                        return chr($code);
+
+                    // 2 bytes, 11 bits
+                    case ($code & 0x7FF):
+                        return chr(0xC0 | (($code >> 6) & 0x1F)) .
+                       chr(0x80 | ($code & 0x3F));
+
+                    // 3 bytes, 16 bits
+                    case ($code & 0xFFFF):
+                        return chr(0xE0 | (($code >> 12) & 0x0F)) .
+                       chr(0x80 | (($code >> 6) & 0x3F)) .
+                       chr(0x80 | ($code & 0x3F));
+
+                    // 4 bytes, 21 bits
+                    case ($code & 0x1FFFFF):
+                        return chr(0xF0 | ($code >> 18)) .
+                       chr(0x80 | (($code >> 12) & 0x3F)) .
+                       chr(0x80 | (($code >> 6) & 0x3F)) .
+                       chr(0x80 | ($code & 0x3F));
+          }
+    }
+
+    /**
-    * Returns whether or not the current environment meets the requirements
-    * of the plugin in order for it to be run, including the PHP version,

trunk/Phergie/Plugin/FeedTicker.php

@@ -93 +93 @@
-        // Global Feed Title, Feed URL and Article Title Filters
-        $globalTitle = trim($this->getPluginIni('filter_title'));
-               
+        
-        $globalArticle = trim($this->getPluginIni('filter_article'));
-
+
-        $i = 0;
-        $this->feeds = array();
@@ -101 +101 @@
-                // Feed and Chan data
-            $feed = $this->getPluginIni('feed' . $i);
-      
+
-            // Feed Title, Feed URL and Article Title Filter Data
-            $filterTitle = $this->getPluginIni('filter_title' . $i);
-            $filterUrl = $this->getPluginIni('filter_url' . $i);
-            $filterArticle = $this->getPluginIni('filter_article' . $i);
-
+
-            if (!empty($feed) && !empty($chans)) {
-                $this->feeds[] = array($feed, preg_split('#[\s\r\n,]+#', $chans));
-                // Feed Title, Feed URL and Article Title Filters
-
-
-
+
+
+
+
-            }
-        } while (++$i < 10);
@@ -194 +195 @@
-            $content = @file_get_contents($url, null, $context);
-            if (empty ($content)) {
- 
+
-                continue;
-            }
@@ -213 +214 @@
-            } else { // Trouble
-                $this->debug('Feed format unrecognized: ' . $url);
- 
+
-            }
-        }
@@ -248 +249 @@
-          // Feed Title, Feed URL and Article Title Filters
-        $filterTitle = $this->filterTitle[$id];
-    
-   
-
+
+
+
-        // Check against the filters if any are set
-/'.$filterTitle.'/
-/'.$filterUrl.'/
-/'.$filterArticle.'/
+{'.$filterTitle.'}
+{'.$filterUrl.'}
+{'.$filterArticle.'}
-              return false;
-          }
-          return true;
-    }
-
+
-    /**
-    * Transliterates a UTF-8 string into corresponding ASCII characters and
@@ -269 +270 @@
-    * @return string
-    */
-
-
-
-
-
-
-
+
+
-        if($trim > 0) {
-            $out = substr($out, 0, $trim) . (strlen($out) > $trim ? '...' : '');

trunk/Phergie/Plugin/Url.php

@@ -117 +117 @@
-                $this->cache[] = $url;
-                $this->cache[] = $tinyUrl;
-
-
-
-
+        
+  
+  
+        
-
-                unset($title, $tinyUrl, $title);
@@ -136 +136 @@
-    * @return string
-    */
-
-
-
-
-
-
-
-
+
+
-        if($trim > 0) {
-            $out = substr($out, 0, $trim) . (strlen($out) > $trim ? '...' : '');