184848 - Wrong font choice for japanese in Unicode

Status: VERIFIED FIXED

(Core :: Internationalization, defect)

Description

[Erwan Loisant]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.3a) Gecko/20021207 Phoenix/0.5
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.3a) Gecko/20021207 Phoenix/0.5

It concerns any page displaying Japanese using unicode. I don't know how Mozilla
chooses the font used for Japanese when displaying Unicode, but it usually
chooses a fantasy font I installed that is very hard to read on a webpage
(imagine an english webpage using a "script" font with size 10).

In the prefs, I choosed non-japanese fonts for Unicode. In fact I need to
display other languages so choosing a japanese font will not change anything.

Maybe the Unicode font selection should be richer, like choosing the font for
each language.

I noticed the same problem in the messenger, and with Phoenix (version 0.5).


Reproducible: Always

Steps to Reproduce:
1. Install a bunch of fantasy japanese font
2. Open a japanese unicode page, like http://www.google.co.jp

Actual Results:  
A fantasy font is used.

Expected Results:  
Use a more common font, doing like most windows app do (I don't know how they
choose the font) or by making richer the unicode font selection preference box.

Comment 1

[Erwan Loisant]

Attachment: Google on Mozilla

The font is not appropriate for this size, it is a fantasy font. It is ugly and
hard to read.

Comment 2

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

language group stuff -> i18n

Comment 3

[Erwan Loisant]

Attachment: Google on IE

IE choosed an appropriate font (probably MS Gothic) and the rendering is nice
and readable. This is what Mozilla was supposed to display.

Comment 4

[Erwan Loisant]

Note: I filled the bug with Phoenix so this is the Phoenix string displayed.
However, I noticed the bug on the following Mozilla build:

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.2.1) Gecko/20021130

Note that the bug is present in Phoenix also (the build string being precised in
the report), as well as olders Mozilla versions.

Comment 5

[Simon Montagu :smontagu]

Do you have the same problem with sites that declare that they are in Japanese?
Try http://www.unicode.org/unicode/standard/translations/japanese.html, which
has <meta http-equiv="Content-Language" content="ja">, and
http://www.unicode.org/iuc/iuc10/x-utf8.html, which has a Japanese section with
lang="ja".

Comment 6

[Yuying Long]

Which system language locale are you using?

Comment 7

[Erwan Loisant]

- I notice the same problem with websites that precise that they are in Japanese
- Although my Windows is an English version, the locale is set to Japanese (to
use non-unicode english applications).

Comment 8

[Erwan Loisant]

(Sorry, I mean: it is set to Japanese to use Japanese non-unicode applications,
of course...)

Comment 9

[Erwan Loisant]

Still present with:

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.3) Gecko/20030312

Comment 10

[rbs]

The page has the CSS:

body,td,a,p,.h{font-family:arial,sans-serif;}

As far as I can tell, 'Arial' includes Japanese glyphs, and Mozilla may be
picking the glyphs from there.

Try to UNCHECK 'Allow documents to use other fonts', and choose another font of
your liking. (remember to CHECK it back...)

Comment 11

[Erwan Loisant]

Unchecking "allow document to use other fonts" works, I can display the page
with the font I choosed.

I don't think Arial includes Japanese glyphs, but Mozilla/Windows automatically
uses another font to display Japanese. I can identify the fantasy font used (in
the screenshot it is a "stamp-like" font), if I delete this font another one is
choosed.

MS software as well as OpenOffice.org 1.0 choose appriopriate alternate font to
display Japanese when the glyphs are not available. However I have no idea how
the "choose alternate font to display when glyph is not available" is implemented.

Comment 12

[rbs]

The prefs below (which you can see with about:confing or in winpref.js) are the
default fonts used for Japanese. Unfortunately, Google is not using lang="ja"
and the page isn't being detected as a Japanese page. It is instead detected as
a vanilla Unicode page (see View -> Character Coding). You will probably get a
better result if you save the page locally and add:

<html lang="ja">
...
</html>

pref("font.name.serif.ja", "ＭＳ Ｐ明朝"); // "MS PMincho"
pref("font.name.sans-serif.ja", "ＭＳ Ｐゴシック"); // "MS PGothic"
pref("font.name.monospace.ja", "ＭＳ ゴシック"); // "MS Gothic"
pref("font.name-list.serif.ja", "MS PMincho, MS PGothic, MS Mincho, MS Gothic");
pref("font.name-list.sans-serif.ja", "MS PGothic, MS PMincho, MS Gothic, MS
Mincho");
pref("font.name-list.monospace.ja", "MS Gothic, MS Mincho, MS PGothic, MS PMincho");

Comment 13

[rbs]

In fact, Mozilla's font code tries hard to detect the language from the
character itself. Indeed bug 116030 was aimed at identifying the language group
for a unicode char, and to use that to find the preferred fonts. I wonder if/why
that code isn't kicking in here.

Comment 14

[Erwan Loisant]

Right, adding lang="ja" in the body of the page solves the problem.

However, if we can blame Google for not precising the language for this Japanese
page, a unicode page may be mostly in English and have only some strings in
Japanese. Make multilingual pages is the main advantage of unicode :). In this
case, a recognition of Japanese characters would be appropriate.

Comment 15

[rbs]

> a recognition of Japanese characters would be appropriate.

Indeed, and that's what bug 116030 intended to do. Which is why I am wondering
if there isn't a bug somewhere.

Comment 16

[rbs]

Attachment: fix

There was a little oversight when addressing bug 116030 comment 12. C.f. the
'<' in the first line of that comment.

Comment 17

[rbs]

Taking to checkin.

Comment 18

[Simon Montagu :smontagu]

Comment on attachment 120757 [details] [diff] [review]
fix

r=smontagu

Comment 19

[rbs]

Checked in.

Comment 20

[rbs]

Marking fixed.

Comment 21

[Yuying Long]

Tried this on 05-08 trunk build, seems it won't use fantasy font.  Mark as
verified as fix, please re-open if still see the problem.