[3.1.0] Add multi-parse capability for StringHash

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@1624 48356398-32a2-884e-a903-53898d9a118a

library/HTMLPurifier/StringHashParser.php

@@ -21,21 +21,60 @@
  *
  * We use this as an easy to use file-format for configuration schema
  * files, but the class itself is usage agnostic.
+ *
+ * You can use ---- to forcibly terminate parsing of a single string-hash;
+ * this marker is used in multi string-hashes to delimit boundaries.
  */
 class HTMLPurifier_StringHashParser
 {
     
     public $default = 'ID';
     
+    /**
+     * Parses a file that contains a single string-hash.
+     */
     public function parseFile($file) {
         if (!file_exists($file)) return false;
         $fh = fopen($file, 'r');
+        if (!$fh) return false;
+        $ret = $this->parseHandle($fh);
+        fclose($fh);
+        return $ret;
+    }
+    
+    /**
+     * Parses a file that contains multiple string-hashes delimited by '----'
+     */
+    public function parseMultiFile($file) {
+        if (!file_exists($file)) return false;
+        $ret = array();
+        $fh = fopen($file, 'r');
+        if (!$fh) return false;
+        while (!feof($fh)) {
+            $ret[] = $this->parseHandle($fh);
+        }
+        fclose($fh);
+        return $ret;
+    }
+    
+    /**
+     * Internal parser that acepts a file handle.
+     * @note While it's possible to simulate in-memory parsing by using
+     *       custom stream wrappers, if such a use-case arises we should
+     *       factor out the file handle into its own class.
+     * @param $fh File handle with pointer at start of valid string-hash
+     *            block.
+     */
+    protected function parseHandle($fh) {
         $state   = false;
         $single  = false;
         $ret     = array();
-        while (($line = fgets($fh)) !== false) {
+        do {
+            $line = fgets($fh);
+            if ($line === false) break;
             $line = rtrim($line, "\n\r");
             if (!$state && $line === '') continue;
+            if ($line === '----') break;
             if (strncmp('--', $line, 2) === 0) {
                 // Multiline declaration
                 $state = trim($line, '- ');
@@ -58,8 +97,7 @@ class HTMLPurifier_StringHashParser
                 if (!isset($ret[$state])) $ret[$state] = '';
                 $ret[$state] .= "$line\n";
             }
-        }
-        fclose($fh);
+        } while (!feof($fh));
         return $ret;
     }
     

tests/HTMLPurifier/StringHashParser/Multi.txt

@@ -0,0 +1,18 @@
+Namespace.Directive
+TYPE: string
+CHAIN-ME: 2
+--DESCRIPTION--
+Multiline
+stuff
+--FOR-WHO--
+Single multiline
+----
+
+Namespace.Directive2
+TYPE: integer
+CHAIN-ME: 3
+--DESCRIPTION--
+M
+stuff
+--FOR-WHO--
+Single multiline2

tests/HTMLPurifier/StringHashParserTest.php

@@ -11,19 +11,19 @@ class HTMLPurifier_StringHashParserTest extends UnitTestCase
      */
     protected $parser;
     
-    function setup() {
+    public function setup() {
         $this->parser = new HTMLPurifier_StringHashParser();
     }
     
     /**
      * Assert that $file gets parsed into the form of $expect
      */
-    function assertParse($file, $expect) {
+    public function assertParse($file, $expect) {
         $result = $this->parser->parseFile(dirname(__FILE__) . '/StringHashParser/' . $file);
         $this->assertIdentical($result, $expect);
     }
     
-    function testSimple() {
+    public function testSimple() {
         $this->assertParse('Simple.txt', array(
             'ID' => 'Namespace.Directive',
             'TYPE' => 'string',
@@ -33,26 +33,26 @@ class HTMLPurifier_StringHashParserTest extends UnitTestCase
         ));
     }
     
-    function testOverrideSingle() {
+    public function testOverrideSingle() {
         $this->assertParse('OverrideSingle.txt', array(
             'KEY' => 'New',
         ));
     }
     
-    function testAppendMultiline() {
+    public function testAppendMultiline() {
         $this->assertParse('AppendMultiline.txt', array(
             'KEY' => "Line1\nLine2\n",
         ));
     }
     
-    function testDefault() {
+    public function testDefault() {
         $this->parser->default = 'NEW-ID';
         $this->assertParse('Default.txt', array(
             'NEW-ID' => 'DefaultValue',
         ));
     }
     
-    function testError() {
+    public function testError() {
         try {
             $this->parser->parseFile('NoExist.txt');
         } catch (HTMLPurifier_ConfigSchema_Exception $e) {
@@ -60,4 +60,27 @@ class HTMLPurifier_StringHashParserTest extends UnitTestCase
         }
     }
     
+    public function testParseMultiple() {
+        $result = $this->parser->parseMultiFile(dirname(__FILE__) . '/StringHashParser/Multi.txt');
+        $this->assertIdentical(
+            $result,
+            array(
+                array(
+                    'ID' => 'Namespace.Directive',
+                    'TYPE' => 'string',
+                    'CHAIN-ME' => '2',
+                    'DESCRIPTION' => "Multiline\nstuff\n",
+                    'FOR-WHO' => "Single multiline\n",
+                ),
+                array(
+                    'ID' => 'Namespace.Directive2',
+                    'TYPE' => 'integer',
+                    'CHAIN-ME' => '3',
+                    'DESCRIPTION' => "M\nstuff\n",
+                    'FOR-WHO' => "Single multiline2\n",
+                )
+            )
+        );
+    }
+    
 }