248179 - Page broken if xsl:output encoding is different from frame src encoding

Status: RESOLVED FIXED

(Core :: XSLT, defect)

Description

[HARUNAGA Hirotoshi]

XSLT-generated output is displayed as garbage, if XSL style sheet
generates <frame> (or <iframe>, <object>) and the frame src (or embedded
html) encoding is different from xsl:output encoding.

testcase (XML):
  http://bugzilla.mozilla.gr.jp/attachment.cgi?id=2289&amp;action=view

XSL style sheet used by XML above
(generates frameset, xsl:output encoding=UTF-8):
  http://bugzilla.mozilla.gr.jp/attachment.cgi?id=2285&amp;action=view

frame src used by the testcase (EUC-JP):
  http://www.mozilla.gr.jp/

Comment 1

[Axel Hecht]

Attachment: html file, standards rendering mode, utf8 encoding

just to add to this, I cross checked this with an html iframe src, in standards

rendering mode. I'll attach it for the sake of completeness.

Comment 2

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Attachment: Patch to fix

The nsIParser.h change is what really fixes this bug. Transformiix was setting
the charsetsource so high that it overrode childframes. I thought it was safe
to change the value for kCharsetFromOtherComponent since it's only used by
transformiix nowadays (don't know what it was originally created for).

The change to txMozillaTextOutput and txMozillaXMLOutput is to fix a separate
bug where we're setting bad charsets in the result document.

Comment 3

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 173439 [details] [diff] [review]
Patch to fix

r+sr=bzbarsky, but could you document in nsIParser that
kCharsetFromParentForced and higher will force the charset on child frames too,
unless they're explicitly overridden?

Comment 4

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Checked in. I didn't include 'unless explicitly overridden' part because
overrides like <meta> and httpheaders won't actually override.

I wonder if the kCharsetFromPreviousLoading should be lowered too, though it
doesn't look like it's used at all.