Commite0b3824

Edward Z. Yang ext:(%22)

committed

Don't lump whitespace and regular text together for data.

1 parentd50aff7 commite0b3824Copy full SHA for e0b3824

File tree

+10

-4

lines changed

+10

-4

lines changed

Lines changed: 10 additions & 4 deletions

Original file line number	Diff line number	Diff line change
`@@ -85,6 +85,7 @@ class HTML5_Tokenizer {`
`85`	`85`	`constLOWER_ALPHA ='abcdefghijklmnopqrstuvwxyz';`
`86`	`86`	`constDIGIT ='0123456789';`
`87`	`87`	`constHEX ='0123456789ABCDEFabcdef';`
	`88`	`+constWHITESPACE ="\t\n\x0c";`
`88`	`89`
`89`	`90`	`/**`
`90`	`91`	`* @param $data Data to parse`
`@@ -217,12 +218,17 @@ public function parse() {`
`217`	`218`	`THIS IS AN OPTIMIZATION: Get as many character that`
`218`	`219`	`otherwise would also be treated as a character token and emit it`
`219`	`220`	`as a single character token. Stay in the data state. */`
	`221`	`+$chars ='';`
`220`	`222`
`221`		`-$mask ='->';`
`222`		`-if ($amp_cond)$mask .='&';`
`223`		`-if ($lt_cond)$mask .='<';`
	`223`	`+// XSKETCHY: introduced three more fails (at least)`
	`224`	`+if ($char !=='' &&$char !=="\n" &&$char !=="\r" &&`
	`225`	`+$char !=="\t" &&$char !=="\x0c") {`
	`226`	`+$mask ='->' .self::WHITESPACE;`
	`227`	`+if ($amp_cond)$mask .='&';`
	`228`	`+if ($lt_cond)$mask .='<';`
`224`	`229`
`225`		`-$chars =$this->stream->charsUntil($mask);`
	`230`	`+$chars =$this->stream->charsUntil($mask);`
	`231`	`+ }`
`226`	`232`
`227`	`233`	`$this->emitToken(array(`
`228`	`234`	`'type' =>self::CHARACTER,`

Comments

(0)