106059 - Javascript return Unicode if link are designed with java Script

Status: VERIFIED FIXED

(Core :: Internationalization, defect)

Description

[slava]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.5+) Gecko/20011018
BuildID:    2001101822

The URL is now Under Constructing and its impossible to use it outside now :( I
describe problem as more detaly as can. And after project complete, write URL 
to this page. (Possibly, i rewrite code to work around this bug) Or writing
special test page later.

In construction, such
parent.Chat.location=parent.curlocation+0+"&channel="+this[this.selectedIndex].value;
parent.first=1; Where channel is in cyrilic, the result link has channel
encoding in 2byte Unicode, not in actual Character Set. This problem i see with
cyrillic. 
The construction like writeln(channel) or forms[0].value=channel work well.
Problem only then result is not for display at the screen.

MORE DETAIL:

<select name="on" size="1"
onChange='parent.Chat.location=parent.curlocation+0+"&channel="+this[this.selectedIndex].value;
parent.first=1;'>
  <option value="Wellcome" >Wellcome</option>
  <option value="Privat" >Privat</option>
  <option value="Тест" >Privat</option>
</select>
After change this list, it may be request with string, where cyrillic encoded in
current code table (MS IE work with this without problem), but Mozzila, make
link with cyrillic in 2 byte unicode in this case.

As i see, this is result of internal unicode, witch "deocded" then write to
screen or to forms[].X.value, according to current codetable, but in case of
writing to .location, using as is...

Did it possible to fix it or its to hardly ? :(

Comment 1

[slava]

The curlocation in this example is string such:
/cgi-bin/chat?workid=123&workpass=qwerty&dealer=321&nik=nikname&action=
The end of query may be generated by Java Script. But its impossible in Mozilla
now then channel value is in cyrillic :(  
As i see, it's may be too hard to fix such bug with recoding :( 

Comment 2

[Boris Zbarsky [:bzbarsky]]

DOM 0

Comment 3

[Johnny Stenback (:jst)]

Over to internatinalization. A simplified testcase would really help.

Comment 4

[Johnny Stenback (:jst)]

Over to internatinalization. A simplified testcase would really help.

Comment 5

[slava]

Attachment: Example with bug. Use /cgi-bin/printenv from apache distribution

Comment 6

[slava]

I added testpage with demo of bug at http://dealers.ambernet.kiev.ua/mozbug.html. 
It use the standart printenv demo cgi from apache.
Source of demo i attached to shis page too.

Comment 7

[slava]

Attachment: Demo page with use /cgi-bin/printenv from apache distribution

Comment 8

[slava]

The same bug is in Galeon too :(

Comment 9

[Roy Yokoyama]

ftang: can you take a look at this?
-> ftang

Comment 10

[Frank Tang]

nhotta, can you take a look at this one? Do we encoded it as % encoded UTF-8 
instead of % encoded KOI8-R?

Comment 11

[slava]

Excuse me...I check now... I was see that its 2 byte encoding and was think its
Unicode. But Unicode cyrillic is 0400-04ff but in our case the cyrillic
translate to  some codetable into range %D0XX-%D1XX - I cant find such codetable
:( Sorry :(
Now i can say that cyrillic in this case translate to unknown (for me) 2byte
codetable into range D0XX-D1XX instead KOI8 or CP1251 and it's not the UTF8 :(

Comment 12

[slava]

Possibly its UTF8 but not cyrrillic from UTF8...

Comment 13

[slava]

Attachment: The same example with some adition - its demonstrating writing the same string to screen with use document.write

Then we write the same string to screen all encoded well without problem. As i
see, the internal store cyrillic into variable is work ok, and problem only
with recode it before it write in construction such
onClick='window.location=result', but all ok in construction such
document.write(result). Possibly it internal representation of cyrillic in
mozilla witch use without translation in this case ? Or wrong translation from
internal in this case... And translation is not to cyrillic 2 byte UNICODE - it
translate it to some unknown for me 2byte code table :( Into range D0XX-D1XX
not into range of cyrillic in UNICODE code table.

Comment 14

[nhottanscp]

This migth be dup of bug 40469.

Comment 15

[nhottanscp]

I checked in a fix for bug 40469.
[e-mail], could you try again with the next available build
and see if the check in also fixes this bug, thanks?

Comment 16

[nhottanscp]

IQA, could you help me to try latest build to see if this is still reproducible?

Comment 17

[Teruko Kobayashi]

Naoki, I tested this with 2001-11-16 Win32 trunk build.  

http://bugzilla.mozilla.org/showattachment.cgi?attach_id=54625
I got the same results on both #2 and #3 in the test cases.

The result is 
REQUEST_URI="/cgi-bin/printenv?eng=English&cyr=%EB%C9%D2%C9%CC%C9%C3%D1"

Comment 18

[slava]

I try 2001111515. I check it with simple test case and in real system where i
find this bug. This bug fixed & All ok with it. Thank you for your work with
cyrillic bugs. And sorry for my dealy with checking :( To many work :( Thank
again for fixing this cyrillic bug.  Cyrillic is a big problem in Linux with
many programs, and in most cases its troubles is to hardly to fix :( More 50% of
problems with using linux based system in Ukraine is cyrillic :( Thank you for
your work and for help.

Comment 19

[nhottanscp]

Mark as fixed based on the reporter's comment.

Comment 20

[Teruko Kobayashi]

Verified.