Bug 20088 – parse CSS Animations

• Component: CSS
• Status: RESOLVED FIXED

[Dean Jackson]

This patch updates the CSS parser to accept the CSS Animation properties, including the keyframe syntax.

[Dean Jackson]

Created attachment 22354 [details]
patch to provide parsing for css animation properties

[Dean Jackson]

Created attachment 22395 [details]
updated patch for css parsing

Removed an old keyframes @rule from the grammar. I'd also left out the keyword in the tokenizer so keyframes were not being noticed.

[Dave Hyatt]

Comment on attachment 22395 [details]
updated patch for css parsing

CSSComputedStyleDeclaration should not be part of this patch.

Need a followup bug that keyframes should support a comma-separated list of percentages and not just a single percentage.

The way keyframe rules are added is wrong.  You need to add keyframe styles as part of the normal addRulesFromSheet process.  You can't add them in a separate walk or you'll miss @import and @media behavior.

Make a test with a <style> element that @imports a sheet with keyframes and you'll see that it doesn't work.  Similarly try putting a @keyframes rule inside @media print { } and watch it incorrectly apply to screen.

You already have the hashtable from name -> keyframes rule held by the CSSStyleSelector.  Just build it up during the normal addRulesFromSheet process instead.

The code that aggressively constructs RenderStyles at keyframe-buildup time in the style selector is wrong, but that's the part I'm looking the other way on.

This patch duplicates stuff from 19938.... the Node/Document/Element changes.  Pick one patch to put them in. :)

[Dean Jackson]

Created attachment 22646 [details]
Updated patch that adds keyframes correctly

This (hopefully) addresses the previous review comments about adding Keyframes at the wrong time during parsing. They should now be picked up within @media rules, for example.

[Dave Hyatt]

Comment on attachment 22646 [details]
Updated patch that adds keyframes correctly

This looks good.  

I think we need tests of error recovery from messed up keyframes, since I'm a bit worried that malformed keyframes rules aren't going to error recover properly, but I think that could be a followup bug.

[Dean Jackson]

r35579 = 684ed79bb9f8a782bb5f4d789adef25c161ec0e7 (trunk)
	M	WebCore/GNUmakefile.am
	M	WebCore/WebCore.xcodeproj/project.pbxproj
	M	WebCore/WebCore.pro
	M	WebCore/ChangeLog
	M	WebCore/WebCore.vcproj/WebCore.vcproj
	M	WebCore/css/CSSValueKeywords.in
	M	WebCore/css/CSSStyleSelector.cpp
	M	WebCore/css/CSSStyleSelector.h
	M	WebCore/css/CSSRule.h
	A	WebCore/css/WebKitCSSKeyframesRule.cpp
	M	WebCore/css/CSSParser.cpp
	M	WebCore/css/CSSGrammar.y
	A	WebCore/css/WebKitCSSKeyframeRule.cpp
	M	WebCore/css/tokenizer.flex
	M	WebCore/css/StyleBase.h
	M	WebCore/css/CSSPropertyNames.in
	A	WebCore/css/WebKitCSSKeyframeRule.h
	A	WebCore/css/WebKitCSSKeyframesRule.h
	M	WebCore/css/CSSParser.h
	M	WebCore/WebCoreSources.bkl