Setup ErrorCollector to maintain new error format, and output that HTML.

Also changed:
    - DirectLex keeps track of column numbers in context
    - New class HTMLPurifier_ErrorStruct

Signed-off-by: Edward Z. Yang <edwardzyang@thewritingpot.com>

configdoc/usage.xml

@@ -5,12 +5,12 @@
    <line>131</line>
   </file>
   <file name="HTMLPurifier/Lexer.php">
-   <line>85</line>
+   <line>81</line>
   </file>
   <file name="HTMLPurifier/Lexer/DirectLex.php">
-   <line>50</line>
+   <line>53</line>
-   <line>62</line>
+   <line>73</line>
-   <line>319</line>
+   <line>348</line>
   </file>
   <file name="HTMLPurifier/Strategy/RemoveForeignElements.php">
    <line>47</line>
@@ -83,17 +83,6 @@
    <line>304</line>
   </file>
  </directive>
- <directive id="Core.MaintainLineNumbers">
-  <file name="HTMLPurifier/ErrorCollector.php">
-   <line>148</line>
-  </file>
-  <file name="HTMLPurifier/Lexer.php">
-   <line>82</line>
-  </file>
-  <file name="HTMLPurifier/Lexer/DirectLex.php">
-   <line>45</line>
-  </file>
- </directive>
  <directive id="Output.CommentScriptContents">
   <file name="HTMLPurifier/Generator.php">
    <line>45</line>
@@ -154,13 +143,13 @@
    <line>202</line>
   </file>
   <file name="HTMLPurifier/Lexer.php">
-   <line>233</line>
+   <line>252</line>
   </file>
   <file name="HTMLPurifier/HTMLModule/Image.php">
    <line>27</line>
   </file>
   <file name="HTMLPurifier/Lexer/DirectLex.php">
-   <line>34</line>
+   <line>36</line>
   </file>
   <file name="HTMLPurifier/Strategy/RemoveForeignElements.php">
    <line>23</line>
@@ -178,17 +167,17 @@
  </directive>
  <directive id="HTML.Proprietary">
   <file name="HTMLPurifier/HTMLModuleManager.php">
-   <line>224</line>
+   <line>221</line>
   </file>
  </directive>
  <directive id="HTML.SafeObject">
   <file name="HTMLPurifier/HTMLModuleManager.php">
-   <line>229</line>
+   <line>226</line>
   </file>
  </directive>
  <directive id="HTML.SafeEmbed">
   <file name="HTMLPurifier/HTMLModuleManager.php">
-   <line>232</line>
+   <line>229</line>
   </file>
  </directive>
  <directive id="Attr.IDBlacklist">
@@ -203,12 +192,20 @@
  </directive>
  <directive id="Core.LexerImpl">
   <file name="HTMLPurifier/Lexer.php">
-   <line>70</line>
+   <line>76</line>
+  </file>
+ </directive>
+ <directive id="Core.MaintainLineNumbers">
+  <file name="HTMLPurifier/Lexer.php">
+   <line>80</line>
+  </file>
+  <file name="HTMLPurifier/Lexer/DirectLex.php">
+   <line>48</line>
   </file>
  </directive>
  <directive id="Core.ConvertDocumentToFragment">
   <file name="HTMLPurifier/Lexer.php">
-   <line>242</line>
+   <line>261</line>
   </file>
  </directive>
  <directive id="URI.Host">
@@ -361,7 +358,7 @@
  </directive>
  <directive id="Core.DirectLexLineNumberSyncInterval">
   <file name="HTMLPurifier/Lexer/DirectLex.php">
-   <line>59</line>
+   <line>70</line>
   </file>
  </directive>
  <directive id="Core.EscapeInvalidTags">

library/HTMLPurifier.includes.php

@@ -41,6 +41,7 @@ require 'HTMLPurifier/Encoder.php';
 require 'HTMLPurifier/EntityLookup.php';
 require 'HTMLPurifier/EntityParser.php';
 require 'HTMLPurifier/ErrorCollector.php';
+require 'HTMLPurifier/ErrorStruct.php';
 require 'HTMLPurifier/Exception.php';
 require 'HTMLPurifier/Filter.php';
 require 'HTMLPurifier/Generator.php';

library/HTMLPurifier.safe-includes.php

@@ -35,6 +35,7 @@ require_once $__dir . '/HTMLPurifier/Encoder.php';
 require_once $__dir . '/HTMLPurifier/EntityLookup.php';
 require_once $__dir . '/HTMLPurifier/EntityParser.php';
 require_once $__dir . '/HTMLPurifier/ErrorCollector.php';
+require_once $__dir . '/HTMLPurifier/ErrorStruct.php';
 require_once $__dir . '/HTMLPurifier/Exception.php';
 require_once $__dir . '/HTMLPurifier/Filter.php';
 require_once $__dir . '/HTMLPurifier/Generator.php';

library/HTMLPurifier/ErrorCollector.php

@@ -23,6 +23,8 @@ class HTMLPurifier_ErrorCollector
     protected $generator;
     protected $context;
     
+    protected $lines = array();
+    
     public function __construct($context) {
         $this->locale    =& $context->get('Locale');
         $this->context   = $context;
@@ -48,6 +50,7 @@ class HTMLPurifier_ErrorCollector
         
         $token = $this->context->get('CurrentToken', true);
         $line  = $token ? $token->line : $this->context->get('CurrentLine', true);
+        $col   = $token ? $token->col  : $this->context->get('CurrentCol',  true);
         $attr  = $this->context->get('CurrentAttr', true);
         
         // perform special substitutions, also add custom parameters
@@ -69,12 +72,58 @@ class HTMLPurifier_ErrorCollector
         if (!empty($subst)) $msg = strtr($msg, $subst);
         
         // (numerically indexed)
-        $this->_current[] = array(
+        $error = array(
             self::LINENO   => $line,
             self::SEVERITY => $severity,
             self::MESSAGE  => $msg,
             self::CHILDREN => array()
         );
+        $this->_current[] = $error;
+        
+        
+        // NEW CODE BELOW ...
+        
+        $struct = null;
+        // Top-level errors are either:
+        //  TOKEN type, if $value is set appropriately, or
+        //  "syntax" type, if $value is null
+        $new_struct = new HTMLPurifier_ErrorStruct();
+        $new_struct->type = HTMLPurifier_ErrorStruct::TOKEN;
+        if ($token) $new_struct->value = clone $token;
+        if (is_int($line) && is_int($col)) {
+            if (isset($this->lines[$line][$col])) {
+                $struct = $this->lines[$line][$col];
+            } else {
+                $struct = $this->lines[$line][$col] = $new_struct;
+            }
+            // These ksorts may present a performance problem
+            ksort($this->lines[$line], SORT_NUMERIC);
+        } else {
+            if (isset($this->lines[-1])) {
+                $struct = $this->lines[-1];
+            } else {
+                $struct = $this->lines[-1] = $new_struct;
+            }
+        }
+        ksort($this->lines, SORT_NUMERIC);
+        
+        // Now, check if we need to operate on a lower structure
+        if (!empty($attr)) {
+            $struct = $struct->getChild(HTMLPurifier_ErrorStruct::ATTR, $attr);
+            if (!$struct->value) {
+                $struct->value = array($attr, 'PUT VALUE HERE');
+            }
+        }
+        if (!empty($cssprop)) {
+            $struct = $struct->getChild(HTMLPurifier_ErrorStruct::CSSPROP, $cssprop);
+            if (!$struct->value) {
+                // if we tokenize CSS this might be a little more difficult to do
+                $struct->value = array($cssprop, 'PUT VALUE HERE');
+            }
+        }
+        
+        // Ok, structs are all setup, now time to register the error
+        $struct->addError($severity, $msg);
     }
     
     /**
@@ -95,38 +144,20 @@ class HTMLPurifier_ErrorCollector
     public function getHTMLFormatted($config, $errors = null) {
         $ret = array();
         
-        $generator = new HTMLPurifier_Generator($config, $this->context);
+        $this->generator = new HTMLPurifier_Generator($config, $this->context);
         if ($errors === null) $errors = $this->errors;
         
-        // sort error array by line
+        // 'At line' message needs to be removed
-        // line numbers are enabled if they aren't explicitly disabled
-        if ($config->get('Core', 'MaintainLineNumbers') !== false) {
-            $has_line       = array();
-            $lines          = array();
-            $original_order = array();
-            foreach ($errors as $i => $error) {
-                $has_line[] = (int) (bool) $error[self::LINENO];
-                $lines[] = $error[self::LINENO];
-                $original_order[] = $i;
-            }
-            array_multisort($has_line, SORT_DESC, $lines, SORT_ASC, $original_order, SORT_ASC, $errors);
-        }
         
-        foreach ($errors as $error) {
+        // generation code for new structure goes here. It needs to be recursive.
-            list($line, $severity, $msg, $children) = $error;
+        foreach ($this->lines as $line => $col_array) {
-            $string = '';
+            if ($line == -1) continue;
-            $string .= '<strong>' . $this->locale->getErrorName($severity) . '</strong>: ';
+            foreach ($col_array as $col => $struct) {
-            $string .= $generator->escape($msg); 
+                $this->_renderStruct($ret, $struct, $line, $col);
-            if ($line) {
-                // have javascript link generation that causes 
-                // textarea to skip to the specified line
-                $string .= $this->locale->formatMessage(
-                    'ErrorCollector: At line', array('line' => $line));
             }
-            if ($children) {
-                $string .= $this->getHTMLFormatted($config, $children);
         }
-            $ret[] = $string;
+        if (isset($this->lines[-1])) {
+            $this->_renderStruct($ret, $this->lines[-1]);
         }
         
         if (empty($errors)) {
@@ -137,5 +168,41 @@ class HTMLPurifier_ErrorCollector
         
     }
     
+    private function _renderStruct(&$ret, $struct, $line = null, $col = null) {
+        $stack = array($struct);
+        $context_stack = array(array());
+        while ($current = array_pop($stack)) {
+            $context = array_pop($context_stack);
+            foreach ($current->errors as $error) {
+                list($severity, $msg) = $error;
+                $string = '';
+                $string .= '<div>';
+                // W3C uses an icon to indicate the severity of the error.
+                $error = $this->locale->getErrorName($severity);
+                $string .= "<span class=\"error e$severity\"><strong>$error</strong></span> ";
+                if (!is_null($line) && !is_null($col)) {
+                    $string .= "<em class=\"location\">Line $line, Column $col: </em> ";
+                } else {
+                    $string .= '<em class="location">End of Document: </em> ';
+                }
+                $string .= '<strong class="description">' . $this->generator->escape($msg) . '</strong> ';
+                $string .= '</div>';
+                // Here, have a marker for the character on the column appropriate.
+                // Be sure to clip extremely long lines.
+                //$string .= '<pre>';
+                //$string .= '';
+                //$string .= '</pre>';
+                $ret[] = $string;
+            }
+            foreach ($current->children as $type => $array) {
+                $context[] = $current;
+                $stack = array_merge($stack, array_reverse($array, true));
+                for ($i = count($array); $i > 0; $i--) {
+                    $context_stack[] = $context;
+                }
+            }
+        }
+    }
+    
 }
 

library/HTMLPurifier/ErrorStruct.php

@@ -0,0 +1,58 @@
+<?php
+
+/**
+ * Records errors for particular segments of an HTML document such as tokens,
+ * attributes or CSS properties. They can contain error structs (which apply
+ * to components of what they represent), but their main purpose is to hold
+ * errors applying to whatever struct is being used.
+ */
+class HTMLPurifier_ErrorStruct
+{
+    
+    /**
+     * Possible values for $children first-key. Note that top-level structures
+     * are automatically token-level.
+     */
+    const TOKEN     = 0;
+    const ATTR      = 1;
+    const CSSPROP   = 2;
+    
+    /**
+     * Type of this struct.
+     */
+    public $type;
+    
+    /**
+     * Value of the struct we are recording errors for. There are various
+     * values for this:
+     *  - TOKEN: Instance of HTMLPurifier_Token
+     *  - ATTR: array('attr-name', 'value')
+     *  - CSSPROP: array('prop-name', 'value')
+     */
+    public $value;
+    
+    /**
+     * Errors registered for this structure.
+     */
+    public $errors = array();
+    
+    /**
+     * Child ErrorStructs that are from this structure. For example, a TOKEN
+     * ErrorStruct would contain ATTR ErrorStructs. This is a multi-dimensional
+     * array in structure: [TYPE]['identifier']
+     */
+    public $children = array();
+    
+    public function getChild($type, $id) {
+        if (!isset($this->children[$type][$id])) {
+            $this->children[$type][$id] = new HTMLPurifier_ErrorStruct();
+            $this->children[$type][$id]->type = $type;
+        }
+        return $this->children[$type][$id];
+    }
+    
+    public function addError($severity, $message) {
+        $this->errors[] = array($severity, $message);
+    }
+    
+}

library/HTMLPurifier/Lexer/DirectLex.php

@@ -63,6 +63,7 @@ class HTMLPurifier_Lexer_DirectLex extends HTMLPurifier_Lexer
             $length = false;
         }
         $context->register('CurrentLine', $current_line);
+        $context->register('CurrentCol',  $current_col);
         $nl = "\n";
         // how often to manually recalculate. This will ALWAYS be right,
         // but it's pretty wasteful. Set to 0 to turn off
@@ -312,6 +313,7 @@ class HTMLPurifier_Lexer_DirectLex extends HTMLPurifier_Lexer
         }
         
         $context->destroy('CurrentLine');
+        $context->destroy('CurrentCol');
         return $array;
     }
     

tests/HTMLPurifier/ErrorCollectorTest.php

@@ -1,5 +1,8 @@
 <?php
 
+/**
+ * @warning HTML output is in flux, but eventually needs to be stabilized.
+ */
 class HTMLPurifier_ErrorCollectorTest extends HTMLPurifier_Harness
 {
     
@@ -44,11 +47,13 @@ class HTMLPurifier_ErrorCollectorTest extends HTMLPurifier_Harness
         
         $this->assertIdentical($this->collector->getRaw(), $result);
         
+        /*
         $formatted_result = 
             '<ul><li><strong>Warning</strong>: Message 2 at line 3</li>'.
             '<li><strong>Error</strong>: Message 1 at line 23</li></ul>';
         
         $this->assertIdentical($this->collector->getHTMLFormatted($this->config), $formatted_result);
+        */
         
     }
     
@@ -75,10 +80,12 @@ class HTMLPurifier_ErrorCollectorTest extends HTMLPurifier_Harness
         );
         $this->assertIdentical($this->collector->getRaw(), $result);
         
+        /*
         $formatted_result = 
             '<ul><li><strong>Error</strong>: Message 1</li>'.
             '<li><strong>Error</strong>: Message 2</li></ul>';
         $this->assertIdentical($this->collector->getHTMLFormatted($this->config), $formatted_result);
+        */
     }
     
     function testContextSubstitutions() {