220581 - Bookmarks manager - problem interpreting ampersands in imported bookmarks

Status: RESOLVED WORKSFORME

(Firefox :: Bookmarks & History, defect)

Description

[Jeff Walden [:Waldo]]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6a) Gecko/20030926 Firebird/0.7+
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6a) Gecko/20030926 Firebird/0.7+

Whenever I delete my Firebird main and profile folders and reinstall Firebird, I
have to import my old bookmarks to be able to access them.  When this happens,
it appears Firebird wants to edit my bookmarks.html file and replace all & that
are not followed by character entity codes with & to make the code more
valid.  For example this:
http://forums.devshed.com/forumdisplay.php?s=&forumid=5

would be converted to this:
http://forums.devshed.com/forumdisplay.php?s=&forumid=5

The problem comes when I try to open these bookmarks.  For some reason the &
is not parsed out of the URL into a & and the second URL of above is sent
through.  This often breaks the CGI scripts the query string is sent to.

Reproducible: Always

Steps to Reproduce:
1.  On a clean install, import a bookmarks.html file with a bookmark whose URL
contains an ampersand.
2.  Open the bookmark.

Actual Results:  
The location bar shows the original URL with all "&" replaced by "&".

Expected Results:  
The location bar should show the original unadulterated URL with all "&" still
intact.

It's worth noting this is only a problem with bookmarks and not hyperlinks in
downloaded files.

Comment 1

[Boris Zbarsky [:bzbarsky]]

.

Comment 2

[Mike Connor [:mconnor]]

can you please attach a reduced version of the old bookmarks file you are
importing to use for testing?

Comment 3

[Jeff Walden [:Waldo]]

Attachment: Testcase - one line only

Testcase consists in its entirety of this line:

<A HREF="http://www.google.com/search?&amp;q=firebird">Bookmark</A>

The &amp; is not converted into a & in the URL, so the URL loaded is actually
<http://www.google.com/search?&amp;q=firebird> (with &amp; not changed into &).
 The actual URL loaded should be <http://www.google.com/search?&q=firebird>
(which does work as intended).	Tested using completely new setup (deled old
install directory, both Mozilla and Phoenix directories in my appdata directory
on WinXP) of latest Firebird 0.8 branch nightly:

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6b) Gecko/20031217
Firebird/0.7+

Comment 4

[Jeff Walden [:Waldo]]

See also bug 223762, which deals with a similar problem.  That bug mentions that
because the charset is UTF-8, HTML entities shouldn't be necessary.  It then
goes on to say that because the file is text/html, HTML entities should probably
be used.

This obviously creates a conflict.  Is a "&" an entity beginner or just a
regular character?  This begets the question: is it possible the code is being
interpreted as UTF-8 only?  If it is, perhaps "&" isn't being recognized as
such because of the charset, which causes "&" to be interpreted literally and
not as the beginning of an entity?

Comment 5

[Jeff Walden [:Waldo]]

Adding testcase keyword, tweaking summary, changing to NEW.  I'll give this a 
test again later today to see if it still exists, as it's been a few weeks 
since I last tried it out.

Comment 6

[Jeff Walden [:Waldo]]

Sorry about not rechecking this earlier, but it went off my internal radar.  A
check using:

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.6) Gecko/20040120 Firebird/0.7+

...still shows it in existence.  When the imported bookmark is run, the HTML
entity for the ampersand is not converted into a real ampersand.

Comment 7

[Jeff Walden [:Waldo]]

I can't reproduce this behavior any more.

-->WFM

Comment 8

[Mike Connor [:mconnor]]

sorry for bugspam, long-overdue mass reassign of ancient QA contact bugs, filter on "beltznerLovesGoats" to get rid of this mass change