UI font localization part 2

For some Indian locales, the automatic font fallback by Windows UI components leads to too tiny glyphs for UI strings.  Therefore, this patch makes it possible to override the UI font family and UI font size localizable by adding two entries to locale_settings (IDS_UI_FONT_FAMILY and IDS_UI_FONT_SIZE_SCALER - percentile scale).

It's is also to fix a P1 bug for Chrome 2.0 final (so this patch needs to be merged back to the branch).

For most locales, the UI font family is set to 'default' and the UI font size scaler is set to 100, which indicates that the UI font (menu, message, etc) obtained from Windows will be used. 

For ml and bn, it's set to the 'kartica' and 'vrinda' (the default Windows fonts for those scripts) and the scaler is set to 150 and 160 respectively. For Hindi and Telugu, only the font size scaler is set to 150. 

When IDS_UI_FONT_FAMILY is 'default' and the scaler is 100, the behavior will remain the same. When it's not, their values are used to create ChromeFont (base and derived) and WindowsTitle font. In addition, menu will be drawn by 'owner' (to override the windows font) and the font for table view, tree and tooltip is also set to IDS_UI_FONT_FAMILY. 

While working on this, I replaced all the instances of 'static ChromeFont' with 'static ChromeFont*' and initialized them in a lazy manner.

The whole approach is still a hack necessary due to the size issue with the default fonts for some Indic scripts on Windows. We'd not need this on Linux and Mac. 

TEST=1. Run chrome with '--lang=bn' or '--lang=ml' and see UI strings are legible in menu, context menu, bookmark, bookmark manager, tooltips, and tab titles. With '--lang=hi' and '--lang=te', the difference is not dramatic but should be more readable. In other locales, it should remain the same. 
2.UI test in en-US locale should pass.
3. Running UI tests under Purify should not have any new leak.

BUG=7319
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=13773

[sky, 2009-04-09 15:50:39]

> Currently, ChromeFont (base) is initialized before main()
> because there are several static ChromeFont defined

We shouldn't be using static objects. I see some UI code doing this:
CustomFrameView, DialogClientView... These should all be updated to ChromeFont*
and lazily created. Are these the only places using static ChromeFont?

[jungshik at Google, 2009-04-09 17:24:21]

On 2009/04/09 15:50:39, sky wrote:
> > Currently, ChromeFont (base) is initialized before main()
> > because there are several static ChromeFont defined
> 
> We shouldn't be using static objects. I see some UI code doing this:
> CustomFrameView, DialogClientView... These should all be updated to
ChromeFont*

I think so. If there are any other places, my locale-based change for
ChromeFont(base) will not be effective because at static initialization time, we
don't have an access to the resource bundle. I feel stupid now. I'll replace
those static ChromeFont with ChromeFont* and do what you suggested below. 

> and lazily created. Are these the only places using static ChromeFont?

[jungshik at Google, 2009-04-10 00:47:35]

Uploaded a new CL with all the instances of 'static ChromeFont' replaced with
'static ChromeFont*'. Please, take another look. 


On 2009/04/09 17:24:21, Jungshik Shin wrote:
> On 2009/04/09 15:50:39, sky wrote:
> > > Currently, ChromeFont (base) is initialized before main()
> > > because there are several static ChromeFont defined
> > 
> > We shouldn't be using static objects. I see some UI code doing this:
> > CustomFrameView, DialogClientView... These should all be updated to
> ChromeFont*
> 
> I think so. If there are any other places, my locale-based change for
> ChromeFont(base) will not be effective because at static initialization time,
we
> don't have an access to the resource bundle. I feel stupid now. I'll replace
> those static ChromeFont with ChromeFont* and do what you suggested below. 
> 
> > and lazily created. Are these the only places using static ChromeFont?

[sky, 2009-04-10 18:02:58]

You should run the ui tests through Purify to make sure it doesn't show a slew
of new leaks after your change.

http://codereview.chromium.org/62064/diff/5078/5079
File chrome/browser/views/constrained_window_impl.cc (right):

http://codereview.chromium.org/62064/diff/5078/5079#newcode584
Line 584: *title_font_ = win_util::GetWindowTitleFont();
Use the copy constructor here like the rest of your code.

http://codereview.chromium.org/62064/diff/5078/5089
File chrome/common/gfx/chrome_font.h (right):

http://codereview.chromium.org/62064/diff/5078/5089#newcode104
Line 104: // Creates a font with the default name and style.
Is this necessary?

http://codereview.chromium.org/62064/diff/5078/5095
File chrome/common/l10n_util_win.cc (right):

http://codereview.chromium.org/62064/diff/5078/5095#newcode31
Line 31: memcpy(logfont->lfFaceName, font_family.data(), name_len *
sizeof(WORD));
nit: line is too long.

http://codereview.chromium.org/62064/diff/5078/5085
File chrome/views/controls/menu/menu.cc (right):

http://codereview.chromium.org/62064/diff/5078/5085#newcode115
Line 115: // If the label contains an accelerator, make a room for tab.
nit: "a room" -> "room"

http://codereview.chromium.org/62064/diff/5078/5085#newcode167
Line 167: // If an accelerator is specified (with a tab delimeting the rest
nit: delimeting -> delimiting.

http://codereview.chromium.org/62064/diff/5078/5087
File chrome/views/window/custom_frame_view.cc (right):

http://codereview.chromium.org/62064/diff/5078/5087#newcode109
Line 109: //static ChromeFont title_font_;
If this isn't needed, nuke it.

[jungshik at Google, 2009-04-10 23:08:42]

Thank you for the review. I addressed all your comments.

Somehow Purify, after applying the latest patch, does not start. I'm
reinstalling from the scratch and will run the UI test under it.

http://codereview.chromium.org/62064/diff/5078/5079
File chrome/browser/views/constrained_window_impl.cc (right):

http://codereview.chromium.org/62064/diff/5078/5079#newcode584
Line 584: *title_font_ = win_util::GetWindowTitleFont();
On 2009/04/10 18:02:58, sky wrote:
> Use the copy constructor here like the rest of your code.

done.

http://codereview.chromium.org/62064/diff/5078/5089
File chrome/common/gfx/chrome_font.h (right):

http://codereview.chromium.org/62064/diff/5078/5089#newcode104
Line 104: // Creates a font with the default name and style.
On 2009/04/10 18:02:58, sky wrote:
> Is this necessary?

You meant the comment is not necessary? It doesn't add much and I can delete it.

http://codereview.chromium.org/62064/diff/5078/5095
File chrome/common/l10n_util_win.cc (right):

http://codereview.chromium.org/62064/diff/5078/5095#newcode31
Line 31: memcpy(logfont->lfFaceName, font_family.data(), name_len *
sizeof(WORD));
On 2009/04/10 18:02:58, sky wrote:
> nit: line is too long.
done

http://codereview.chromium.org/62064/diff/5078/5085
File chrome/views/controls/menu/menu.cc (right):

http://codereview.chromium.org/62064/diff/5078/5085#newcode115
Line 115: // If the label contains an accelerator, make a room for tab.
On 2009/04/10 18:02:58, sky wrote:
> nit: "a room" -> "room"

done

http://codereview.chromium.org/62064/diff/5078/5085#newcode167
Line 167: // If an accelerator is specified (with a tab delimeting the rest
On 2009/04/10 18:02:58, sky wrote:
> nit: delimeting -> delimiting.
done

http://codereview.chromium.org/62064/diff/5078/5087
File chrome/views/window/custom_frame_view.cc (right):

http://codereview.chromium.org/62064/diff/5078/5087#newcode109
Line 109: //static ChromeFont title_font_;
On 2009/04/10 18:02:58, sky wrote:
> If this isn't needed, nuke it.
nuked.

[sky, 2009-04-10 23:15:54]

Assuming there is no purify problems, LGTM.

http://codereview.chromium.org/62064/diff/5078/5089
File chrome/common/gfx/chrome_font.h (right):

http://codereview.chromium.org/62064/diff/5078/5089#newcode104
Line 104: // Creates a font with the default name and style.
On 2009/04/10 23:08:42, Jungshik Shin wrote:
> On 2009/04/10 18:02:58, sky wrote:
> > Is this necessary?
> 
> You meant the comment is not necessary? It doesn't add much and I can delete
it.

No, I meant it looks like you haven't changed anything in this file. Could it be
reverted?

[jungshik at Google, 2009-04-14 00:14:18]

Thank you for the review. 

On 2009/04/10 23:15:54, sky wrote:
> Assuming there is no purify problems, LGTM.
> 
> http://codereview.chromium.org/62064/diff/5078/5089
> File chrome/common/gfx/chrome_font.h (right):
> 
> http://codereview.chromium.org/62064/diff/5078/5089#newcode104
> Line 104: // Creates a font with the default name and style.
> On 2009/04/10 23:08:42, Jungshik Shin wrote:
> > On 2009/04/10 18:02:58, sky wrote:
> > > Is this necessary?
> > 
> > You meant the comment is not necessary? It doesn't add much and I can delete
> it.
> 
> No, I meant it looks like you haven't changed anything in this file. Could it
be
> reverted?

Aha. I did and updated the CL. 

I ran ui_tests under purify (it took ~ 14 hrs on my machine), but was told that
it's not easy to compare it with what buildbot has (Erik restarted ui_test +
purify bot this morning and it'll take ~ 14 hours there, too) because ui test
purify results haven't been baselined properly, yet (according to Paul). I'm
about to start ui tests under purify once more *without* my CL. That way,
perhaps it's a bit easier to spot any new error. We'll see tomorrow :-)