Reorganize VarParser; there may be multiple implementations.

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@1602 48356398-32a2-884e-a903-53898d9a118a

NEWS

@@ -22,7 +22,7 @@ NEWS ( CHANGELOG and HISTORY )                                     HTMLPurifier
   maintenance/generate-schema-cache.php generates the schema.ser file, which
   is now instantiated. Support for userland schema changes coming soon!
 ! Extra utility classes for testing and non-library operations can
-  be found in extras/. Specifically, these are FSTools and ConfigSchema.
+  be found in extras/. Specifically, these are FSTools and ConfigDoc.
   You may find a use for these in your own project, but right now they
   are highly experimental and volatile.
 ! Integration with PHPT allows for automated smoketests

library/HTMLPurifier.includes.php

@@ -200,3 +200,4 @@ require 'HTMLPurifier/URIScheme/https.php';
 require 'HTMLPurifier/URIScheme/mailto.php';
 require 'HTMLPurifier/URIScheme/news.php';
 require 'HTMLPurifier/URIScheme/nntp.php';
+require 'HTMLPurifier/VarParser/Flexible.php';

library/HTMLPurifier/Config.php

@@ -75,7 +75,7 @@ class HTMLPurifier_Config
     public function __construct(&$definition) {
         $this->conf = $definition->defaults; // set up, copy in defaults
         $this->def  = $definition; // keep a copy around for checking
-        $this->parser = new HTMLPurifier_VarParser();
+        $this->parser = new HTMLPurifier_VarParser_Flexible();
     }
     
     /**

library/HTMLPurifier/ConfigSchema.php

@@ -51,7 +51,7 @@ class HTMLPurifier_ConfigSchema {
     );
     
     public function __construct() {
-        $this->parser = new HTMLPurifier_VarParser();
+        $this->parser = new HTMLPurifier_VarParser_Flexible();
     }
     
     /**

library/HTMLPurifier/ConfigSchema/Interchange.php

@@ -8,23 +8,6 @@
 class HTMLPurifier_ConfigSchema_Interchange
 {
     
-    /**
-     * Hash table of allowed types.
-     */
-    public $types = array(
-        'string'    => 'String',
-        'istring'   => 'Case-insensitive string',
-        'text'      => 'Text',
-        'itext'     => 'Case-insensitive text',
-        'int'       => 'Integer',
-        'float'     => 'Float',
-        'bool'      => 'Boolean',
-        'lookup'    => 'Lookup array',
-        'list'      => 'Array list',
-        'hash'      => 'Associative array',
-        'mixed'     => 'Mixed'
-    );
-    
     /**
      * Array of Namespace ID => array(namespace info)
      */

library/HTMLPurifier/ConfigSchema/Validator/ParseType.php

@@ -8,7 +8,7 @@ class HTMLPurifier_ConfigSchema_Validator_ParseType extends HTMLPurifier_ConfigS
     
     public function validate(&$arr, $interchange) {
         $r = explode('/', $arr['TYPE'], 2);
-        if (!isset($interchange->types[$r[0]])) {
+        if (!isset(HTMLPurifier_VarParser::$types[$r[0]])) {
             $this->error('Invalid type ' . $r[0] . ' for configuration directive ' . $arr['ID']);
         }
         $arr['_TYPE'] = $r[0];

library/HTMLPurifier/VarParser.php

@@ -4,13 +4,13 @@
  * Parses string representations into their corresponding native PHP
  * variable type.
  */
-class HTMLPurifier_VarParser
+abstract class HTMLPurifier_VarParser
 {
     
     /**
      * Lookup table of allowed types.
      */
-    public $types = array(
+    static public $types = array(
         'string'    => true,
         'istring'   => true,
         'text'      => true,
@@ -25,100 +25,15 @@ class HTMLPurifier_VarParser
     );
     
     /**
-     * Validate a variable according to type. Throws exception if invalid.
+     * Validate a variable according to type. Throws
-     * It may return NULL as a valid type.
+     * HTMLPurifier_VarParserException if invalid.
+     * It may return NULL as a valid type if $allow_null is true.
+     *
+     * @param $var Variable to validate
+     * @param $type Type of variable, see HTMLPurifier_VarParser->types
+     * @param $allow_null Whether or not to permit null as a value
+     * @return Validated and type-coerced variable
      */
-    public function parse($var, $type, $allow_null = false) {
+    abstract public function parse($var, $type, $allow_null = false);
-        if (!isset($this->types[$type])) {
-            throw new HTMLPurifier_VarParserException("Invalid type $type");
-        }
-        if ($allow_null && $var === null) return null;
-        switch ($type) {
-            // Note: if code "breaks" from the switch, it triggers a generic
-            // exception to be thrown. Specific errors can be specifically
-            // done here.
-            case 'mixed':
-                //if (is_string($var)) $var = unserialize($var);
-                return $var;
-            case 'istring':
-            case 'string':
-            case 'text': // no difference, just is longer/multiple line string
-            case 'itext':
-                if (!is_string($var)) break;
-                if ($type === 'istring' || $type === 'itext') $var = strtolower($var);
-                return $var;
-            case 'int':
-                if (is_string($var) && ctype_digit($var)) $var = (int) $var;
-                elseif (!is_int($var)) break;
-                return $var;
-            case 'float':
-                if (is_string($var) && is_numeric($var)) $var = (float) $var;
-                elseif (!is_float($var)) break;
-                return $var;
-            case 'bool':
-                if (is_int($var) && ($var === 0 || $var === 1)) {
-                    $var = (bool) $var;
-                } elseif (is_string($var)) {
-                    if ($var == 'on' || $var == 'true' || $var == '1') {
-                        $var = true;
-                    } elseif ($var == 'off' || $var == 'false' || $var == '0') {
-                        $var = false;
-                    } else {
-                        throw new HTMLPurifier_VarParserException("Unrecognized value '$var' for $type");
-                    }
-                } elseif (!is_bool($var)) break;
-                return $var;
-            case 'list':
-            case 'hash':
-            case 'lookup':
-                if (is_string($var)) {
-                    // special case: technically, this is an array with
-                    // a single empty string item, but having an empty
-                    // array is more intuitive
-                    if ($var == '') return array();
-                    if (strpos($var, "\n") === false && strpos($var, "\r") === false) {
-                        // simplistic string to array method that only works
-                        // for simple lists of tag names or alphanumeric characters
-                        $var = explode(',',$var);
-                    } else {
-                        $var = preg_split('/(,|[\n\r]+)/', $var);
-                    }
-                    // remove spaces
-                    foreach ($var as $i => $j) $var[$i] = trim($j);
-                    if ($type === 'hash') {
-                        // key:value,key2:value2
-                        $nvar = array();
-                        foreach ($var as $keypair) {
-                            $c = explode(':', $keypair, 2);
-                            if (!isset($c[1])) continue;
-                            $nvar[$c[0]] = $c[1];
-                        }
-                        $var = $nvar;
-                    }
-                }
-                if (!is_array($var)) break;
-                $keys = array_keys($var);
-                if ($keys === array_keys($keys)) {
-                    if ($type == 'list') return $var;
-                    elseif ($type == 'lookup') {
-                        $new = array();
-                        foreach ($var as $key) {
-                            $new[$key] = true;
-                        }
-                        return $new;
-                    } else break;
-                }
-                if ($type === 'lookup') {
-                    foreach ($var as $key => $value) {
-                        $var[$key] = true;
-                    }
-                }
-                return $var;
-            default:
-                // This should not happen!
-                throw new HTMLPurifier_Exception("Inconsistency in HTMLPurifier_VarParser: $type is not implemented");
-        }
-        throw new HTMLPurifier_VarParserException("Invalid input for type $type");
-    }
     
 }

library/HTMLPurifier/VarParser/Flexible.php

@@ -0,0 +1,103 @@
+<?php
+
+/**
+ * Performs safe variable parsing based on types which can be used by
+ * users. This may not be able to represent all possible data inputs,
+ * however.
+ */
+class HTMLPurifier_VarParser_Flexible extends HTMLPurifier_VarParser
+{
+    
+    public function parse($var, $type, $allow_null = false) {
+        if (!isset(HTMLPurifier_VarParser::$types[$type])) {
+            throw new HTMLPurifier_VarParserException("Invalid type $type");
+        }
+        if ($allow_null && $var === null) return null;
+        switch ($type) {
+            // Note: if code "breaks" from the switch, it triggers a generic
+            // exception to be thrown. Specific errors can be specifically
+            // done here.
+            case 'mixed':
+                return $var;
+            case 'istring':
+            case 'string':
+            case 'text': // no difference, just is longer/multiple line string
+            case 'itext':
+                if (!is_string($var)) break;
+                if ($type === 'istring' || $type === 'itext') $var = strtolower($var);
+                return $var;
+            case 'int':
+                if (is_string($var) && ctype_digit($var)) $var = (int) $var;
+                elseif (!is_int($var)) break;
+                return $var;
+            case 'float':
+                if (is_string($var) && is_numeric($var)) $var = (float) $var;
+                elseif (!is_float($var)) break;
+                return $var;
+            case 'bool':
+                if (is_int($var) && ($var === 0 || $var === 1)) {
+                    $var = (bool) $var;
+                } elseif (is_string($var)) {
+                    if ($var == 'on' || $var == 'true' || $var == '1') {
+                        $var = true;
+                    } elseif ($var == 'off' || $var == 'false' || $var == '0') {
+                        $var = false;
+                    } else {
+                        throw new HTMLPurifier_VarParserException("Unrecognized value '$var' for $type");
+                    }
+                } elseif (!is_bool($var)) break;
+                return $var;
+            case 'list':
+            case 'hash':
+            case 'lookup':
+                if (is_string($var)) {
+                    // special case: technically, this is an array with
+                    // a single empty string item, but having an empty
+                    // array is more intuitive
+                    if ($var == '') return array();
+                    if (strpos($var, "\n") === false && strpos($var, "\r") === false) {
+                        // simplistic string to array method that only works
+                        // for simple lists of tag names or alphanumeric characters
+                        $var = explode(',',$var);
+                    } else {
+                        $var = preg_split('/(,|[\n\r]+)/', $var);
+                    }
+                    // remove spaces
+                    foreach ($var as $i => $j) $var[$i] = trim($j);
+                    if ($type === 'hash') {
+                        // key:value,key2:value2
+                        $nvar = array();
+                        foreach ($var as $keypair) {
+                            $c = explode(':', $keypair, 2);
+                            if (!isset($c[1])) continue;
+                            $nvar[$c[0]] = $c[1];
+                        }
+                        $var = $nvar;
+                    }
+                }
+                if (!is_array($var)) break;
+                $keys = array_keys($var);
+                if ($keys === array_keys($keys)) {
+                    if ($type == 'list') return $var;
+                    elseif ($type == 'lookup') {
+                        $new = array();
+                        foreach ($var as $key) {
+                            $new[$key] = true;
+                        }
+                        return $new;
+                    } else break;
+                }
+                if ($type === 'lookup') {
+                    foreach ($var as $key => $value) {
+                        $var[$key] = true;
+                    }
+                }
+                return $var;
+            default:
+                // This should not happen!
+                throw new HTMLPurifier_Exception("Inconsistency in HTMLPurifier_VarParser_Flexible: $type is not implemented");
+        }
+        throw new HTMLPurifier_VarParserException("Invalid input for type $type");
+    }
+    
+}

tests/HTMLPurifier/VarParser/FlexibleTest.php

@@ -1,32 +1,8 @@
 <?php
 
-class HTMLPurifier_VarParserTest extends UnitTestCase
+class HTMLPurifier_VarParser_FlexibleTest extends HTMLPurifier_VarParserHarness
 {
     
-    protected $parser;
-    
-    public function setup() {
-        $this->parser = new HTMLPurifier_VarParser();
-    }
-    
-    function assertValid($var, $type, $ret = null) {
-        $ret = ($ret === null) ? $var : $ret;
-        $this->assertIdentical($this->parser->parse($var, $type), $ret);
-    }
-    
-    function assertInvalid($var, $type, $msg = null) {
-        $caught = false;
-        try {
-            $this->parser->parse($var, $type);
-        } catch (HTMLPurifier_VarParserException $e) {
-            $caught = true;
-            if ($msg !== null) $this->assertIdentical($e->getMessage(), $msg);
-        }
-        if (!$caught) {
-            $this->fail('Did not catch expected error');
-        }
-    }
-    
     function testValidate() {
         
         $this->assertValid('foobar', 'string');

tests/HTMLPurifier/VarParserHarness.php

@@ -0,0 +1,31 @@
+<?php
+
+class HTMLPurifier_VarParserHarness extends UnitTestCase
+{
+    
+    protected $parser;
+    
+    public function setup() {
+        $class = substr(get_class($this), 0, -4);
+        $this->parser = new $class();
+    }
+    
+    function assertValid($var, $type, $ret = null) {
+        $ret = ($ret === null) ? $var : $ret;
+        $this->assertIdentical($this->parser->parse($var, $type), $ret);
+    }
+    
+    function assertInvalid($var, $type, $msg = null) {
+        $caught = false;
+        try {
+            $this->parser->parse($var, $type);
+        } catch (HTMLPurifier_VarParserException $e) {
+            $caught = true;
+            if ($msg !== null) $this->assertIdentical($e->getMessage(), $msg);
+        }
+        if (!$caught) {
+            $this->fail('Did not catch expected error');
+        }
+    }
+    
+}