Changeset 43c40f0 in github

Timestamp:

Feb 15, 2012 8:24:04 AM (15 months ago)

Author:

alecpl <alec@…>

Branches:

master, HEAD, courier-fix, dev-browser-capabilities, pdo, release-0.8

Children:

b138a9b

Parents:

3ed9e80

Message:

• Don't list mailto: and anchor links
• List only unique URLs

File:

• program/lib/html2text.php (modified) (4 diffs)

program/lib/html2text.php

@@ -318 +318 @@
      *  Contains URL addresses from links to be rendered in plain text.
      *
-     *  @var string $_link_list
+     *  @var array $_link_list
      *  @access private
      *  @see _build_link_list()
      */
-    var $_link_list = '';
-
-    /**
-     *  Number of valid links detected in the text, used for plain text
-     *  display (rendered similar to footnotes).
-     *
-     *  @var integer $_link_count
-     *  @access private
-     *  @see _build_link_list()
-     */
-    var $_link_count = 0;
+    var $_link_list = array();
 
     /**
@@ -473 +463 @@
     {
         // Variables used for building the link list
-        $this->_link_count = 0;
-        $this->_link_list = '';
+        $this->_link_list = array();
 
         $text = trim(stripslashes($this->html));
@@ -482 +471 @@
 
         // Add link list
-        if ( !empty($this->_link_list) ) {
-            $text .= "\n\nLinks:\n------\n" . $this->_link_list;
+        if (!empty($this->_link_list)) {
+            $text .= "\n\nLinks:\n------\n";
+            foreach ($this->_link_list as $idx => $url) {
+                $text .= '[' . ($idx+1) . '] ' . $url . "\n";
+            }
         }
 
@@ -564 +556 @@
     function _build_link_list( $link, $display )
     {
-            if ( !$this->_do_links )
+            if (!$this->_do_links || empty($link)) {
                 return $display;
-
-            if ( preg_match('!^(https?://|mailto:)!', $link) ) {
-            $this->_link_count++;
-            $this->_link_list .= '[' . $this->_link_count . "] $link\n";
-            $additional = ' [' . $this->_link_count . ']';
-            } elseif ( substr($link, 0, 11) == 'javascript:' ) {
-                    // Don't count the link; ignore it
-                    $additional = '';
-                // what about href="#anchor" ?
-        } else {
-            $this->_link_count++;
-            $this->_link_list .= '[' . $this->_link_count . '] ' . $this->url;
-            if ( substr($link, 0, 1) != '/' ) {
-                $this->_link_list .= '/';
+            }
+
+        // Ignored link types
+            if (preg_match('!^(javascript|mailto|#):!i', $link)) {
+                    return $display;
+        }
+
+            if (preg_match('!^(https?://)!i', $link)) {
+            $url = $link;
+        }
+        else {
+            $url = $this->url;
+            if (substr($link, 0, 1) != '/') {
+                $url .= '/';
             }
-            $this->_link_list .= "$link\n";
-            $additional = ' [' . $this->_link_count . ']';
-        }
-
-        return $display . $additional;
+            $url .= "$link";
+        }
+
+        if (($index = array_search($url, $this->_link_list)) === false) {
+            $this->_link_list[] = $url;
+            $index = count($this->_link_list);
+        }
+
+        return $display . ' [' . ($index+1) . ']';
     }