Bug 22466 – REGRESSION (35867): Many resources missing when saving webarchive of webkit.org

• Status: RESOLVED
• Resolution: FIXED
• Component: WebCore Misc.

[David Kilzer (:ddkilzer)]

* SUMMARY
When saving a webarchive of <http://webkit.org/> with WebKit nightly build r38707, there are many resources missing that are linked from the main web page.  (See Bug 11850 for images referenced within CSS that are not saved.)

* STEPS TO REPRODUCE
1. Launch Safari.
2. Open URL:  http://webkit.org/
3. Select "Save As..." from the File menu.
4. Change Format: to "Web Archive".
5. Save webkit.webarchive.
6. Open webkit.webarchive in "Property List Editor" application (installed with Xcode).

* RESULTS
Only 5 subresources are saved from the main web page.  Some of the <color>.css files are missing from the archive.

* REGRESSION
I think this is a regression, but I don't have the ability to verify this now.

[David Kilzer (:ddkilzer)]

Created attachment 25450 [details]
WebArchive saved from nightly build r38707

This is the webarchive file saved from WebKit nightly build r38707.

[David Kilzer (:ddkilzer)]

Created attachment 25451 [details]
Patch v1 (naïve fix)

This patch fixes the bug, but I don't understand why DocumentLoader::cachedResource() doesn't have a reference to resources loaded on the page.  Are we violating a contract, or should we just be checking for all resources in the WebCore cache instead?

[David Kilzer (:ddkilzer)]

Comment on attachment 25451 [details]
Patch v1 (naïve fix)

Clearing review flag and obsoleting since we need a ChangeLog and a layout test.

[Antti Koivisto]

http://trac.webkit.org/changeset/35867

[David Kilzer (:ddkilzer)]

Created attachment 25479 [details]
Patch v2

Patch with layout test and changelog.

[Alexey Proskuryakov]

See also: bug 17151.

[David Kilzer (:ddkilzer)]

Created attachment 25497 [details]
Patch v3

Same as Patch v2, except we make sure "Etag" and "Last-Modified" headers are normalized as well.

[David Kilzer (:ddkilzer)]

<rdar://problem/6403593>

[David Kilzer (:ddkilzer)]

Created attachment 25551 [details]
Patch v4

Null check ArchiveResource after creating it from a CacheResource.  Rearranged the logic in LegacyWebArchive::create() to use more continue statements.

[David Kilzer (:ddkilzer)]

Created attachment 25599 [details]
Patch v5

Changes since Patch v4:
- Also normalize the "Server" HTTP response header.
- Added radar bug number to ChangeLogs.

[Brady Eidson]

Comment on attachment 25599 [details]
Patch v5

Nice, r+

[David Kilzer (:ddkilzer)]

$ git svn dcommit
Committing to http://svn.webkit.org/repository/webkit/trunk ...
        M       LayoutTests/ChangeLog
        A       LayoutTests/http/tests/webarchive/resources/test-preload-resources.css
        A       LayoutTests/http/tests/webarchive/test-preload-resources-expected.webarchive
        A       LayoutTests/http/tests/webarchive/test-preload-resources.html
        M       LayoutTests/platform/qt/Skipped
        M       LayoutTests/platform/win/Skipped
        M       WebCore/ChangeLog
        M       WebCore/loader/archive/cf/LegacyWebArchive.cpp
        M       WebKitTools/ChangeLog
        M       WebKitTools/DumpRenderTree/mac/DumpRenderTree.mm
Committed r38884

http://trac.webkit.org/changeset/38884

[David Kilzer (:ddkilzer)]

Follow-up fix:

$ git svn dcommit
Committing to http://svn.webkit.org/repository/webkit/trunk ...
        M       WebCore/ChangeLog
        M       WebCore/loader/archive/cf/LegacyWebArchive.cpp
Committed r38885

http://trac.webkit.org/changeset/38885