diff options
-rw-r--r-- | src/HTML5/Parser/DOMTreeBuilder.php | 12 | ||||
-rw-r--r-- | test/HTML5/Parser/DOMTreeBuilderTest.php | 4 |
2 files changed, 7 insertions, 9 deletions
diff --git a/src/HTML5/Parser/DOMTreeBuilder.php b/src/HTML5/Parser/DOMTreeBuilder.php index b0e2e11..fe22eac 100644 --- a/src/HTML5/Parser/DOMTreeBuilder.php +++ b/src/HTML5/Parser/DOMTreeBuilder.php @@ -323,18 +323,16 @@ class DOMTreeBuilder implements EventHandler { public function text($data) { // XXX: Hmmm.... should we really be this strict? if ($this->insertMode < self::IM_IN_HEAD) { - // Per '8.2.5.4.3 The "before head" insertion mode' we are supposed to - // ignore " \t\n\r\f" characters and throw a parse error for other strings. - // In this case we are throwing a parse error for other strings while - // passing " \t\n\r\f" through to the DOM. Since this parser is not creating - // a DOM that will be used for rendering a display and the DOM may be - // turned back into html these characters are passed along to the DOM. + // Per '8.2.5.4.3 The "before head" insertion mode' the characters + // " \t\n\r\f" should be ignored but no mention of a parse error. This is + // practical as most documents contain these characters. Other text is not + // expected here so recording a parse error is necessary. $dataTmp = trim($data, " \t\n\r\f"); if (!empty($dataTmp)) { //fprintf(STDOUT, "Unexpected insert mode: %d", $this->insertMode); $this->parseError("Unexpected text. Ignoring: " . $dataTmp); - $data = str_replace($dataTmp, '', $data); } + return; } //fprintf(STDOUT, "Appending text %s.", $data); $node = $this->doc->createTextNode($data); diff --git a/test/HTML5/Parser/DOMTreeBuilderTest.php b/test/HTML5/Parser/DOMTreeBuilderTest.php index a6c6a9c..1cb4a8c 100644 --- a/test/HTML5/Parser/DOMTreeBuilderTest.php +++ b/test/HTML5/Parser/DOMTreeBuilderTest.php @@ -200,8 +200,8 @@ class DOMTreeBuilderTest extends \HTML5\Tests\TestCase { Foo<head></head><body></body></html>"; $doc = $this->parse($html); $this->assertEquals('Line 0, Col 0: Unexpected text. Ignoring: Foo', $doc->errors[0]); - $whiteSpace = $doc->documentElement->firstChild; - $this->assertEquals("\n ", $whiteSpace->data); + $headElement = $doc->documentElement->firstChild; + $this->assertEquals('head', $headElement->tagName); } public function testParseErrors() { |