Bug 99880

Summary:

Always parse pasted fragments as HTML even on XHTML pages

Product:

WebKit

Reporter:

Adam Klein <adamk>

Component:

HTML Editing

Assignee:

Adam Klein <adamk>

Status:

RESOLVED FIXED

Severity:

Normal

CC:

ap, enrica, esprehn, jochen, mifenton, ojan, rniwa, webkit.review.bot

Priority:

Version:

528+ (Nightly build)

Hardware:

Unspecified

OS:

Unspecified

Bug Depends on:

Bug Blocks:

99607

Attachments:

Description	Flags
Patch	none
Patch	none
Patch for landing	none
Test case	none

Adam Klein

Reported 2012-10-19 15:27:37 PDT

Always parse pasted fragments as HTML even on XHTML pages

Attachments
Patch (10.30 KB, patch) 2012-10-19 15:36 PDT, Adam Klein	no flags	Details Formatted Diff Diff
Patch (10.42 KB, patch) 2012-10-19 16:09 PDT, Adam Klein	no flags	Details Formatted Diff Diff
Patch for landing (6.82 KB, patch) 2012-10-19 16:53 PDT, Adam Klein	no flags	Details Formatted Diff Diff
Test case (191 bytes, application/xhtml+xml) 2012-10-19 18:54 PDT, Adam Klein	no flags	Details
Show Obsolete (2) View All Add attachment proposed patch, testcase, etc.

Adam Klein

Comment 1 2012-10-19 15:36:29 PDT

Created attachment 169713 [details] Patch

Elliott Sprehn

Comment 2 2012-10-19 15:43:07 PDT

Comment on attachment 169713 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=169713&action=review > Source/WebCore/editing/markup.cpp:670 > + element->removeChild(child.get()); Is there a reason to unwrap the RefPtr here but pass it below? > Source/WebCore/editing/markup.cpp:681 > + if (node->hasTagName(htmlTag) || node->hasTagName(headTag) || node->hasTagName(bodyTag)) { What's the purpose of this method? I can just put <head><head></head></head> into the pasteboard and you'd leave the inner one behind.

Adam Klein

Comment 3 2012-10-19 16:07:18 PDT

Comment on attachment 169713 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=169713&action=review >> Source/WebCore/editing/markup.cpp:670 >> + element->removeChild(child.get()); > > Is there a reason to unwrap the RefPtr here but pass it below? removeChild() takes a raw pointer, insertBefore() takes a PassRefPtr. So I'm using .get() as minimally as I can. >> Source/WebCore/editing/markup.cpp:681 >> + if (node->hasTagName(htmlTag) || node->hasTagName(headTag) || node->hasTagName(bodyTag)) { > > What's the purpose of this method? I can just put <head><head></head></head> into the pasteboard and you'd leave the inner one behind. Added a short comment, see if that helps.

Adam Klein

Comment 4 2012-10-19 16:09:33 PDT

Created attachment 169715 [details] Patch

Ojan Vafai

Comment 5 2012-10-19 16:25:38 PDT

Comment on attachment 169715 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=169715&action=review > LayoutTests/ChangeLog:17 > + * platform/mac/editing/pasteboard/paste-xml-expected.txt: Now that we > + use the HTML parser, parsing the paste succeeds and so we insert DOM > + instead of plain text. Similar rebaselines may be needed on other > + platforms. maybe point to the pretty-diff bug to explain this diff?

Adam Klein

Comment 6 2012-10-19 16:53:13 PDT

Created attachment 169727 [details] Patch for landing

Adam Klein

Comment 7 2012-10-19 16:56:47 PDT

Patch for landing is greatly simplified, since it turns out the special code to handle <html>, <body>, and <head> is not needed at all (because we use the HTML5 fragment parsing algorithm with <body> as the context element, which means we ignore all three of those tags.

Adam Klein

Comment 8 2012-10-19 16:59:05 PDT

Comment on attachment 169727 [details] Patch for landing Waiting to cq+ so I can watch the waterfalls in case rebaselining is needed.

Alexey Proskuryakov

Comment 9 2012-10-19 17:08:09 PDT

+ The Mac port previously worked around this problem by falling back to plain text + when parsing failed, but switching to HTML seems like a clear improvement. Are you aware of any way to reproduce this in Safari on Mac in practice? I briefly tried, but couldn't get in this code path.

Adam Klein

Comment 10 2012-10-19 17:13:28 PDT

(In reply to comment #9) > + The Mac port previously worked around this problem by falling back to plain text > + when parsing failed, but switching to HTML seems like a clear improvement. > > Are you aware of any way to reproduce this in Safari on Mac in practice? I briefly tried, but couldn't get in this code path. All you should need is an xhtml page with a contentEditable element in it. Try to copy some html and paste it into that element. Fwiw, The relevant code is in EditorMac.mm, Editor::documentFragment(). Under DRT, editing/pasteboard/paste-xml.xhtml walks through this "try to parse as XML, fail, fall back to text" case.

Alexey Proskuryakov

Comment 11 2012-10-19 17:20:30 PDT

I tried that, and got styled text pasted. Possibly because there were additional flavors on the pasteboard, but this is why I'm asking for steps to reproduce.

Adam Klein

Comment 12 2012-10-19 18:54:58 PDT

Created attachment 169742 [details] Test case Note that the ".xhtml" suffix is necessary for Safari to treat this as an XHTML document. When I select the first line and paste it onto the second line, the bolding disappears.

Alexey Proskuryakov

Comment 13 2012-10-19 22:04:07 PDT

Thank you, I can reproduce now. I experimented with a data:application/xhtml+xml URL, and looks like these do not make XHTML documents in Safari.

WebKit Review Bot

Comment 14 2012-10-23 05:37:11 PDT

Comment on attachment 169727 [details] Patch for landing Rejecting attachment 169727 [details] from commit-queue. Failed to run "['/mnt/git/webkit-commit-queue/Tools/Scripts/webkit-patch', '--status-host=queues.webkit.org', '-..." exit_code: 2 Last 500 characters of output: ent): Merge conflict in Source/WebKit2/ChangeLog Failed to merge in the changes. Patch failed at 0001 [EFL] Make plugin process debugging easier (PLUGIN_PROCESS_COMMAND_PREFIX) When you have resolved this problem run "git rebase --continue". If you would prefer to skip this patch, instead run "git rebase --skip". To restore the original branch and stop rebasing run "git rebase --abort". rebase refs/remotes/origin/master: command returned error: 1 Died at Tools/Scripts/update-webkit line 164. Full output: http://queues.webkit.org/results/14518146

WebKit Review Bot

Comment 15 2012-10-23 06:06:43 PDT

Comment on attachment 169727 [details] Patch for landing Clearing flags on attachment: 169727 Committed r132211: <http://trac.webkit.org/changeset/132211>

WebKit Review Bot

Comment 16 2012-10-23 06:06:48 PDT

All reviewed patches have been landed. Closing bug.

Note You need to log in before you can comment on or make changes to this bug.