Enhancement: Allow XmlTypeCheck for strings

physikerwelt · Gerrit Code Review · commit c6b37dbee572 · 2013-06-07T18:11:49.000Z
This change allows that the class XmlTypeCheck can be used to
check strings as well as files. Therefore the constructor has
been extended to one further element that indicates if the
input to be checked is a file or a string.

The functionality is tested with php unit tests.

Change-Id: I7432a754a512e0324c171667c3eac9d8e7838e0e
diff --git a/includes/XmlTypeCheck.php b/includes/XmlTypeCheck.php
@@ -40,16 +40,52 @@ class XmlTypeCheck {
 	public $rootElement = '';
 
 	/**
-	 * @param string $file filename
-	 * @param $filterCallback callable (optional)
+	 * @param string $input a filename or string containing the XML element
+	 * @param callable $filterCallback (optional)
 	 *        Function to call to do additional custom validity checks from the
 	 *        SAX element handler event. This gives you access to the element
 	 *        namespace, name, and attributes, but not to text contents.
 	 *        Filter should return 'true' to toggle on $this->filterMatch
+	 * @param boolean $isFile (optional) indicates if the first parameter is a
+	 *        filename (default, true) or if it is a string (false)
 	 */
-	function __construct( $file, $filterCallback = null ) {
+	function __construct( $input, $filterCallback = null, $isFile = true ) {
 		$this->filterCallback = $filterCallback;
-		$this->run( $file );
+		if ( $isFile ) {
+			$this->validateFromFile( $input );
+		} else {
+			$this->validateFromString( $input );
+		}
+	}
+
+	/**
+	 * Alternative constructor: from filename
+	 *
+	 * @param string $input a filename or string containing the XML element
+	 * @param callable $filterCallback (optional)
+	 *        Function to call to do additional custom validity checks from the
+	 *        SAX element handler event. This gives you access to the element
+	 *        namespace, name, and attributes, but not to text contents.
+	 *        Filter should return 'true' to toggle on $this->filterMatch
+	 * @return XmlTypeCheck
+	 */
+	public static function newFromFilename( $fname, $filterCallback = null ) {
+		return new self( $fname, $filterCallback, true );
+	}
+
+	/**
+	 * Alternative constructor: from string
+	 *
+	 * @param string $input a filename or string containing the XML element
+	 * @param callable $filterCallback (optional)
+	 *        Function to call to do additional custom validity checks from the
+	 *        SAX element handler event. This gives you access to the element
+	 *        namespace, name, and attributes, but not to text contents.
+	 *        Filter should return 'true' to toggle on $this->filterMatch
+	 * @return XmlTypeCheck
+	 */
+	public static function newFromString( $string, $filterCallback = null ) {
+		return new self( $string, $filterCallback, false );
 	}
 
 	/**
@@ -62,15 +98,23 @@ public function getRootElement() {
 	}
 
 	/**
-	 * @param $fname
+	 * Get an XML parser with the root element handler.
+	 * @see XmlTypeCheck::rootElementOpen()
+	 * @return resource a resource handle for the XML parser
 	 */
-	private function run( $fname ) {
+	private function getParser() {
 		$parser = xml_parser_create_ns( 'UTF-8' );
-
 		// case folding violates XML standard, turn it off
 		xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
-
 		xml_set_element_handler( $parser, array( $this, 'rootElementOpen' ), false );
+		return $parser;
+	}
+
+	/**
+	 * @param string $fname the filename
+	 */
+	private function validateFromFile( $fname ) {
+		$parser = $this->getParser();
 
 		if ( file_exists( $fname ) ) {
 			$file = fopen( $fname, "rb" );
@@ -79,7 +123,7 @@ private function run( $fname ) {
 					$chunk = fread( $file, 32768 );
 					$ret = xml_parse( $parser, $chunk, feof( $file ) );
 					if ( $ret == 0 ) {
-						// XML isn't well-formed!
+						$this->wellFormed = false;
 						fclose( $file );
 						xml_parser_free( $parser );
 						return;
@@ -89,12 +133,26 @@ private function run( $fname ) {
 				fclose( $file );
 			}
 		}
-
 		$this->wellFormed = true;
 
 		xml_parser_free( $parser );
 	}
 
+	/**
+	 *
+	 * @param string $string the XML-input-string to be checked.
+	 */
+	private function validateFromString( $string ) {
+		$parser = $this->getParser();
+		$ret = xml_parse( $parser, $string, true );
+		xml_parser_free( $parser );
+		if ( $ret == 0 ) {
+			$this->wellFormed = false;
+			return;
+		}
+		$this->wellFormed = true;
+	}
+
 	/**
 	 * @param $parser
 	 * @param $name
diff --git a/tests/phpunit/includes/XmlTypeCheckTest.php b/tests/phpunit/includes/XmlTypeCheckTest.php
@@ -0,0 +1,30 @@
+<?php
+/**
+ * PHPUnit tests for XMLTypeCheck.
+ * @author physikerwelt
+ * @group ?
+ * @covers XMLTypeCheck
+ */
+class XmlTypeCheckTest extends MediaWikiTestCase {
+	const WELL_FORMED_XML = "<root><child /></root>";
+	const MAL_FORMED_XML = "<root><child /></error>";
+
+	/**
+	 * @covers XMLTypeCheck::newFromString
+	 * @covers XMLTypeCheck::getRootElement
+	 */
+	public function testWellFormedXML() {
+		$testXML = XmlTypeCheck::newFromString( self::WELL_FORMED_XML );
+		$this->assertTrue( $testXML->wellFormed );
+		$this->assertEquals( 'root', $testXML->getRootElement() );
+	}
+
+	/**
+	 * @covers XMLTypeCheck::newFromString
+	 */
+	public function testMalFormedXML() {
+		$testXML = XmlTypeCheck::newFromString( self::MAL_FORMED_XML );
+		$this->assertFalse( $testXML->wellFormed );
+	}
+
+}
