Merge pull request #236 from stephenharris/bugfix/rfc2616-chunk-decod…

…ing-bug

Fixes bug in rfc2616 #3.6.1 implementation.

library/Requests.php

@@ -749,15 +749,17 @@ public static function parse_multiple(&$response, $request) {
 	 * @return string Decoded body
 	 */
 	protected static function decode_chunked($data) {
-		if (!preg_match('/^([0-9a-f]+)[^\r\n]*\r\n/i', trim($data))) {
+		if (!preg_match('/^([0-9a-f]+)(?:;(?:[\w-]*)(?:=(?:(?:[\w-]*)*|"(?:[^\r\n])*"))?)*\r\n/i', trim($data))) {
 			return $data;
 		}
 
+
+
 		$decoded = '';
 		$encoded = $data;
 
 		while (true) {
-			$is_chunked = (bool) preg_match('/^([0-9a-f]+)[^\r\n]*\r\n/i', $encoded, $matches);
+			$is_chunked = (bool) preg_match('/^([0-9a-f]+)(?:;(?:[\w-]*)(?:=(?:(?:[\w-]*)*|"(?:[^\r\n])*"))?)*\r\n/i', $encoded, $matches);
 			if (!$is_chunked) {
 				// Looks like it's not chunked after all
 				return $data;

tests/ChunkedEncoding.php

@@ -15,6 +15,18 @@ public static function chunkedProvider() {
 				"02\r\nab\r\n04\r\nra\nc\r\n06\r\nadabra\r\n0c\r\n\nall we got\n",
 				"abra\ncadabra\nall we got\n"
 			),
+			array(
+				"02;foo=bar;hello=world\r\nab\r\n04;foo=baz\r\nra\nc\r\n06;justfoo\r\nadabra\r\n0c\r\n\nall we got\n",
+				"abra\ncadabra\nall we got\n"
+			),
+			array(
+				"02;foo=\"quoted value\"\r\nab\r\n04\r\nra\nc\r\n06\r\nadabra\r\n0c\r\n\nall we got\n",
+				"abra\ncadabra\nall we got\n"
+			),
+			array(
+				"02;foo-bar=baz\r\nab\r\n04\r\nra\nc\r\n06\r\nadabra\r\n0c\r\n\nall we got\n",
+				"abra\ncadabra\nall we got\n"
+			),
 		);
 	}
 
@@ -34,12 +46,24 @@ public function testChunked($body, $expected){
 		$this->assertEquals($expected, $response->body);
 	}
 
+	public static function notChunkedProvider() {
+		return array(
+			'invalid chunk size' => array( 'Hello! This is a non-chunked response!' ),
+			'invalid chunk extension' => array( '1BNot chunked\r\nLooks chunked but it is not\r\n' ),
+			'unquoted chunk-ext-val with space' => array( "02;foo=unquoted with space\r\nab\r\n04\r\nra\nc\r\n06\r\nadabra\r\n0c\r\n\nall we got\n" ),
+			'unquoted chunk-ext-val with forbidden character' => array( "02;foo={unquoted}\r\nab\r\n04\r\nra\nc\r\n06\r\nadabra\r\n0c\r\n\nall we got\n" ),
+			'invalid chunk-ext-name' => array( "02;{foo}=bar\r\nab\r\n04\r\nra\nc\r\n06\r\nadabra\r\n0c\r\n\nall we got\n" ),
+			'incomplete quote for chunk-ext-value' => array( "02;foo=\"no end quote\r\nab\r\n04\r\nra\nc\r\n06\r\nadabra\r\n0c\r\n\nall we got\n" ),
+		);
+	}
+
 	/**
 	 * Response says it's chunked, but actually isn't
+	 * @dataProvider notChunkedProvider
 	 */
-	public function testNotActuallyChunked() {
+	public function testNotActuallyChunked($body) {
 		$transport = new MockTransport();
-		$transport->body = 'Hello! This is a non-chunked response!';
+		$transport->body = $body;
 		$transport->chunked = true;
 
 		$options = array(
@@ -50,6 +74,7 @@ public function testNotActuallyChunked() {
 		$this->assertEquals($transport->body, $response->body);
 	}
 
+
 	/**
 	 * Response says it's chunked and starts looking like it is, but turns out
 	 * that they're lying to us