Implement working linkification, now, the real challenge is to get it to play nice with auto-paragraphing.

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@1216 48356398-32a2-884e-a903-53898d9a118a

library/HTMLPurifier/Injector/Linkify.php

@@ -0,0 +1,56 @@
+<?php
+
+require_once 'HTMLPurifier/Injector.php';
+
+/**
+ * Injector that converts http, https and ftp text URLs to actual links.
+ */
+class HTMLPurifier_Injector_Linkify extends HTMLPurifier_Injector
+{
+    
+    function handleText(&$token, $config, &$context) {
+        $current_nesting =& $context->get('CurrentNesting');
+        // this snippet could be factored out
+        $definition = $config->getHTMLDefinition();
+        if (!empty($current_nesting)) {
+            $parent_token = array_pop($current_nesting);
+            $current_nesting[] = $parent_token;
+            $parent = $definition->info[$parent_token->name];
+        } else {
+            $parent = $definition->info_parent_def;
+        }
+        if (!isset($parent->child->elements['a'])) {
+            // parent element does not allow link elements, don't bother
+            return;
+        }
+        if (strpos($token->data, '://') === false) {
+            // our really quick heuristic failed, abort
+            // this may not work so well if we want to match things like
+            // "google.com"
+            return;
+        }
+        // there is/are URL(s). Let's split the string:
+        
+        $bits = preg_split('#((?:https?|ftp)://[^\s\'"<>()]+)#S', $token->data, -1, PREG_SPLIT_DELIM_CAPTURE);
+        
+        $token = array();
+        
+        // $i = index
+        // $c = count
+        // $l = is link
+        for ($i = 0, $c = count($bits), $l = false; $i < $c; $i++, $l = !$l) {
+            if (!$l) {
+                if ($bits[$i] === '') continue;
+                $token[] = new HTMLPurifier_Token_Text($bits[$i]);
+            } else {
+                $token[] = new HTMLPurifier_Token_Start('a', array('href' => $bits[$i]));
+                $token[] = new HTMLPurifier_Token_Text($bits[$i]);
+                $token[] = new HTMLPurifier_Token_End('a');
+            }
+        }
+        
+    }
+    
+}
+
+?>

library/HTMLPurifier/Strategy/MakeWellFormed.php

@@ -5,6 +5,7 @@ require_once 'HTMLPurifier/HTMLDefinition.php';
 require_once 'HTMLPurifier/Generator.php';
 
 require_once 'HTMLPurifier/Injector/AutoParagraph.php';
+require_once 'HTMLPurifier/Injector/Linkify.php';
 
 HTMLPurifier_ConfigSchema::define(
     'Core', 'AutoParagraph', false, 'bool', '
@@ -16,6 +17,16 @@ HTMLPurifier_ConfigSchema::define(
 '
 );
 
+HTMLPurifier_ConfigSchema::define(
+    'Core', 'AutoLinkify', false, 'bool', '
+<p>
+  This directive will cause HTML Purifier to automatically linkify
+  text that looks like URLs. This directive has been available since
+  2.0.1.
+</p>
+'
+);
+
 /**
  * Takes tokens makes them well-formed (balance end tags, etc.)
  */
@@ -49,6 +60,12 @@ class HTMLPurifier_Strategy_MakeWellFormed extends HTMLPurifier_Strategy
             $injector_disabled[] = false;
         }
         
+        if ($config->get('Core', 'AutoLinkify')) {
+            $injector[] = new HTMLPurifier_Injector_Linkify();
+            $injector_skip[] = 0;
+            $injector_disabled[] = false;
+        }
+        
         $current_injector = 0;
         
         $context->register('Injector', $injector);
@@ -60,13 +77,12 @@ class HTMLPurifier_Strategy_MakeWellFormed extends HTMLPurifier_Strategy
             // if all goes well, this token will be passed through unharmed
             $token = $tokens[$tokens_index];
             
-            // this will be more complicated
+            foreach ($injector as $i => $x) {
-            if (isset($injector[$current_injector])) {
+                if ($injector_skip[$i] > 0) {
-                if ($injector_skip[$current_injector] > 0) {
+                    $injector_skip[$i]--;
-                    $injector_skip[$current_injector]--;
+                    $injector_disabled[$i] = true;
-                    $injector_disabled[$current_injector] = true;
                 } else {
-                    $injector_disabled[$current_injector] = false;
+                    $injector_disabled[$i] = false;
                 }
             }
             
@@ -74,8 +90,14 @@ class HTMLPurifier_Strategy_MakeWellFormed extends HTMLPurifier_Strategy
             if (empty( $token->is_tag )) {
                 
                 if ($token->type === 'text') {
-                     if (isset($injector[$current_injector]) && !$injector_disabled[$current_injector]) {
+                     foreach ($injector as $i => $x) {
-                         $injector[$current_injector]->handleText($token, $config, $context);
+                         if (!$injector_disabled[$i]) {
+                             $x->handleText($token, $config, $context);
+                         }
+                         if (is_array($token)) {
+                             $current_injector = $i;
+                             break;
+                         }
                      }
                 }
                 
@@ -127,8 +149,14 @@ class HTMLPurifier_Strategy_MakeWellFormed extends HTMLPurifier_Strategy
                     $current_nesting[] = $parent; // undo the pop
                 }
                 
-                if (isset($injector[$current_injector]) && !$injector_disabled[$current_injector]) {
+                foreach ($injector as $i => $x) {
-                    $injector[$current_injector]->handleStart($token, $config, $context);
+                    if (!$injector_disabled[$i]) {
+                        $x->handleStart($token, $config, $context);
+                    }
+                    if (is_array($token)) {
+                        $current_injector = $i;
+                        break;
+                    }
                 }
                 
                 $this->processToken($token, $config, $context);