122077 - Angle brackets in tags within NOFRAMES content escaped on save as complete page

Status: VERIFIED FIXED

(Core Graveyard :: File Handling, defect)

Description

[Dan Tobias]

Try using the "Save Page As" on this URL (to a frameset), using "Save as
complete page" option.  The NOFRAMES section includes a hyperlink, but in the
saved version the angle brackets used to delineate the A tag are escaped into
ampersand sequences, meaning that the link doesn't work as expected when the
file is used in a non-frame browser -- the literal text of the tag is shown
instead.  This is an unexpected and undesired alteration of the original HTML
code.  (I'm using build 2002012403 in Win98.)

Comment 1

[Dan Tobias]

My attempts to create a simple testcase for this turned up another problem with
saving framesets... see bug 122078.

Comment 2

[Boris Zbarsky [:bzbarsky]]

To adamlock.  Seeing this on current linux too.

Comment 3

[Adam Lock]

Attachment: A simpler example of the problem

Original sample URL is pretty nasty, so here is a simpler example of the
problem.

My guess is that the HTML parser is simply turning everything inside a
<NOFRAMES> element into text (since Moz never looks at it anyway). When the
document is saved the encoder spews out the text converting angle brackets to
entities as it does so.

This line in the parser suggests some other elements which will get treated the
same way.

http://lxr.mozilla.org/seamonkey/source/htmlparser/src/nsHTMLTokenizer.cpp#755

Comment 4

[Adam Lock]

Attachment: A simpler example of the problem

Original sample URL is pretty nasty, so here is a simpler example of the
problem.

My guess is that the HTML parser is simply turning everything inside a
<NOFRAMES> element into text (since Moz never looks at it anyway). When the
document is saved the encoder is spews out the text converting angle brackets
to entities as it does so.

This line in the parser suggests some other elements which will get treated the
same way.

http://lxr.mozilla.org/seamonkey/source/htmlparser/src/nsHTMLTokenizer.cpp#755

Comment 5

[Adam Lock]

*** Bug 125392 has been marked as a duplicate of this bug. ***

Comment 6

[Johnny Stenback (:jst)]

Harish would know all about that...

Comment 7

[Adam Lock]

Harish is this your bug?

Comment 8

[harishd]

>My guess is that the HTML parser is simply turning everything inside a
><NOFRAMES> element into text (since Moz never looks at it anyway). When the
>document is saved the encoder is spews out the text converting angle brackets
>to entities as it does so.

You're right when FRAMES are enabled NOFRAMES content would get collected as
text ( mainly to avoid residual style from doing nasty stuff to the NOFRAMES
content and to the rest of the content ). I don't think this is a parser
problem. May be the encoder should not do any kind of conversion to the NOFRAMES
content.

Comment 9

[Adam Lock]

Attachment: Simple fix

Patch tells HTML serializer to treat noframes content as CDATA so it doesn't
replace angle brackets with entities any more.

Reviews?

Comment 10

[harishd]

Comment on attachment 70381 [details] [diff] [review]
Simple fix

r=harishd

Comment 11

[Johnny Stenback (:jst)]

Comment on attachment 70381 [details] [diff] [review]
Simple fix

sr=jst

Comment 12

[Adam Lock]

Fix is in

Comment 13

[sairuh (rarely reading bugmail)]

vrfy'd fixed using 2002.03.13 comm bits on linux rh7.2, win2k and mac 10.1.3.
tested with http://hopey.mcom.com/tests/simple-frames-2.html