195035 - google.com - bidirectional text displayed wrongly in source view

Status: RESOLVED WORKSFORME

(Tech Evangelism Graveyard :: Other, defect)

Description

[punya]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.3b) Gecko/20030210
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.3b) Gecko/20030210

In the source view for http://www.google.com/intl/ur/, parts of the text which
should be displayed left-to-right are displayed right-to-left instead, for
instance - </html> appears as <lmth/>. This happens at the end of the file,
immediately after a large chunk of Urdu (right-to-left) text. There is no
corresponding error in what is displayed by the browser.

Reproducible: Always

Steps to Reproduce:
1. Navigate to http://www.google.com/intl/ur/
2. View the source using View > Page Source
3. Read the last line, scrolling to the right if necessary

Actual Results:  
The source view displays left-to-right text in right-to-left order (e.g.
<lmth/><ydob/>, in the middle of the last line of an html file)

Expected Results:  
It should display the left-to-right text in its correct order (e.g. </body></html>)

Comment 1

[Tavis Ormandy]

WFM Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3b) Gecko/20030224

Possibly fixed by bug 192919 ?

Comment 2

[punya]

Bug 192919 is about misrendered html (rtl rendered as ltr) in the browser
component (gecko), while this is about misrendered html source (i.e. plain text)
in the source viewer (in this case, ltr rendered as rtl). It doesn't seem that
it's been fixed.

Comment 3

[Jesiah S]

I also see part of the source going the wrong way.
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3b) Gecko/20030224 (nightly)

Comment 4

[Thamer Mahmoud]

Confirmed using Mozilla 20030210 on WinXP

Comment 5

[Simon Montagu :smontagu]

Actually, view source *is* html rendered in the browser component, but this is
not the same issue as bug 192919.

Comment 6

[Simon Montagu :smontagu]

What is happening is this: the page source contains a single RLO (right-to-left
override) character with no corresponding PDF (pop directional format). 

When viewing the HTML in the normal fashion, this doesn't matter because the
override ends at the end of the block element.

However, in the generated HTML that we use for view source, the elements that
were block elements originally are no longer block elements, so the Bidi
Algorithm produces incorrect results.

Much the same thing happens viewing the source from IE, and for the same reason.

Comment 7

[Hixie (not reading bugmail)]

I recommend adding LRO...PDF characters around tags in the source view.

Comment 8

[Simon Montagu :smontagu]

What about adding 

span {
 unicode-bidi: embed;
}

to viewsource.css?

Comment 9

[Simon Montagu :smontagu]

To evangelism for now. The problem would be solved if that RLO wasn't there.
It's unnecessary since the text in the element is right-to-left anyway, and
embedding raw Bidi formatting characters in HTML is deprecated.

Comment 10

[Boris Zbarsky [:bzbarsky]]

Ian, Simon, could you please explain to me the costs, benefits, and implications
of adding directionality chars to the source view?  Either in a separate bug
filed for this purpose or offline.  I'm looking at things like: when is it
needed, what does it do, what is the perf hit, what is the ram hit, etc.

Comment 11

[Bob Clary [:bc] (inactive)]

-> evang500

Comment 12

[Katsuhiko Momoi]

This is going to Other category. I will keep this bug, however.

Comment 13

[Bob Clary [:bc] (inactive)]

Kat, can you explain to me what I should say to Google to get them to change
their content?

Comment 14

[Simon Montagu :smontagu]

(In reply to comment #13)
> Kat, can you explain to me what I should say to Google to get them to change
> their content?

An expanded version of what I said in comment 6 and comment 9 :)

The Urdu translation of "I'm feeling lucky" appears in the HTML source with an
RLO (right-to-left override) character, at the beginning but without the correct
corresponding PDF (pop directional format) character at the end.

The override character is unnecessary, because the Urdu text has right-to-left
directionality already, and even if it were necessary, Unicode directional
characters are deprecated in markup languages according to
http://www.w3.org/TR/2003/NOTE-unicode-xml-20030613/#Bidi, and it would be more
correct to use a <bdo> element to achieve the same effect.

Since the RLO is only used in this one text on the whole page, I assume it's
unintentional anyway.

Comment 15

[Felix Miata]

Conforming summary to TFM item 10 at 
http://www.mozilla.org/projects/tech-evangelism/site/procedures.html#file-new