| Summary: | External CSS style-sheets default to wrong charset | ||
|---|---|---|---|
| Product: | [Applications] konqueror | Reporter: | Thiago Macieira <thiago> |
| Component: | khtml parsing | Assignee: | Konqueror Developers <konq-bugs> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | illogical1 |
| Priority: | NOR | ||
| Version: | unspecified | ||
| Target Milestone: | --- | ||
| Platform: | unspecified | ||
| OS: | Linux | ||
| Latest Commit: | Version Fixed In: | ||
| Sentry Crash Report: | |||
| Attachments: |
Test page, HTML 4.01 Transitional, UTF-8
Test CSS stylesheet, UTF-8 encoded Attempt at fixing the problem Second attempt at fixing |
||
Created attachment 5531 [details]
Test page, HTML 4.01 Transitional, UTF-8
Created attachment 5532 [details]
Test CSS stylesheet, UTF-8 encoded
The functions at fault are: CachedObject::codecForBuffer (khtml/misc/loader.cpp) DocLoader::requestStyleSheet (same) Nowhere in misc/loader.cpp does it try and get the charset from the KIO metadata. Created attachment 5535 [details]
Attempt at fixing the problem
The attached patch fixes the problem for me, both for remote files and local
ones. It does:
- move the m_charset member from khtml::CachedCSSStyleSheet and
khtml::CachedScript into khtml::CachedObject. It won't be used, of course, for
images (khtml::CachedImage).
- in khtml::Loader::slotFinished, query the metadata from the job before
calling r->object->data. In case of local files, use the charset from
QTextCodec::codecForLocale
Created attachment 5632 [details]
Second attempt at fixing
The previous patch made the server charset parameter override the user's. This
one inverts that logic.
*** Bug 100993 has been marked as a duplicate of this bug. *** CVS commit by carewolf: Make charset in <link> actually mean something. Patch is simplified version of one by Thiago Maciera BUG: 79065 M +4 -0 ChangeLog 1.408 M +2 -2 misc/loader.cpp 1.181 --- kdelibs/khtml/ChangeLog #1.407:1.408 @@ -1,2 +1,6 @@ +2005-03-22 Allan Sandfeld Jensen <kde@carewolf.com> + + * misc/loader.cpp: Do not override existing charset with an empty one. + 2005-03-21 Allan Sandfeld Jensen <kde@carewolf.com> --- kdelibs/khtml/misc/loader.cpp #1.180:1.181 @@ -968,5 +968,5 @@ CachedCSSStyleSheet *DocLoader::requestS CachedCSSStyleSheet* s = Cache::requestObject<CachedCSSStyleSheet, CachedObject::CSSStyleSheet>( this, fullURL, accept ); - if ( s ) { + if ( s && !charset.isEmpty() ) { s->setCharset( charset ); } @@ -981,5 +981,5 @@ CachedScript *DocLoader::requestScript( CachedScript* s = Cache::requestObject<CachedScript, CachedObject::Script>( this, fullURL, 0 ); - if ( s ) + if ( s && !charset.isEmpty() ) s->setCharset( charset ); return s; |
Version: 3.2.0 (using KDE 3.2.90 (CVS >= 20040117), compiled sources) Compiler: gcc version 3.3.3 OS: Linux (i686) release 2.6.3 When a webpage (HTML and XHTML) references an external style-sheet through a <LINK> reference, the charset for the loaded file is incorrectly set: it defaults to ISO-8859-1 (Latin 1), even if metadata from the server specifies a different encoding. The attached testpage (valid HTML 4.01 Transitional) demonstrates this error. The external stylesheet when loaded like this: <link rel="StyleSheet" type="text/css" href="test.css"> Causes the text to appear in Konqueror: This should appear « quoted ». And this is a test of UTF-8: €. Changing the load line to the following: <link rel="StyleSheet" type="text/css" charset="utf-8" href="test.css"> Causes the text to appear as it should (and as it does in Mozilla): This should appear « quoted ». And this is a test of UTF-8: €. Note: my locale is UTF-8, so all files are supposed to be loaded UTF-8 (as the webpage showing the Euro symbol demonstrates). Also, when retrieving the webpage from a server, I get: kio_http: (918400) "Content-Type: text/css; charset=utf-8"