ninja&make/mac: Convert .strings files to UTF-16LE, not just UTF-16.

ninja&make/mac: Convert .strings files to UTF-16LE, not just UTF-16.

OS X iconv makes "UTF-16" write UTF-16BE data.  Just do the recoding in
Python, which lets "UTF-16" write UTF-16LE data (I guess it uses the host
byte order).  Also saves a subprocess invocation while here.

(Using "UTF-16LE" explicitly lets both iconv and python omit the BOM, so it's
shorter to use "UTF-16" than "UTF-16LE" and explicitly writing a BOM.)

This matches Xcode's behavior (at least Xcode 4.5).

BUG=chromium:280718
R=rsesek@chromium.org

Committed: https://code.google.com/p/gyp/source/detail?r=1707

[Nico, 2013-08-29 17:06:49]

ptal

https://codereview.chromium.org/23710010/diff/9001/pylib/gyp/mac_tool.py
File pylib/gyp/mac_tool.py (right):

https://codereview.chromium.org/23710010/diff/9001/pylib/gyp/mac_tool.py#newc...
pylib/gyp/mac_tool.py:97: fp.write(s.decode(input_code).encode('UTF-16'))
Do you remember why this used to shell out to iconv?

[Robert Sesek, 2013-08-29 17:48:23]

https://codereview.chromium.org/23710010/diff/9001/pylib/gyp/mac_tool.py
File pylib/gyp/mac_tool.py (right):

https://codereview.chromium.org/23710010/diff/9001/pylib/gyp/mac_tool.py#newc...
pylib/gyp/mac_tool.py:97: fp.write(s.decode(input_code).encode('UTF-16'))
On 2013/08/29 17:06:50, Nico wrote:
> Do you remember why this used to shell out to iconv?

It's what an older Xcode did specifically.

https://codereview.chromium.org/23710010/diff/12001/pylib/gyp/mac_tool.py
File pylib/gyp/mac_tool.py (right):

https://codereview.chromium.org/23710010/diff/12001/pylib/gyp/mac_tool.py#new...
pylib/gyp/mac_tool.py:97: fp.write(s.decode(input_code).encode('UTF-16'))
Does this specify LE order?

https://codereview.chromium.org/23710010/diff/12001/test/mac/gyptest-app.py
File test/mac/gyptest-app.py (right):

https://codereview.chromium.org/23710010/diff/12001/test/mac/gyptest-app.py#n...
test/mac/gyptest-app.py:49: test.must_contain(strings, '\xff\xfe' + '/*
Localized'.encode('utf-16le'))
Comment that this is the BOM?

[Nico, 2013-08-29 17:49:38]

https://codereview.chromium.org/23710010/diff/12001/pylib/gyp/mac_tool.py
File pylib/gyp/mac_tool.py (right):

https://codereview.chromium.org/23710010/diff/12001/pylib/gyp/mac_tool.py#new...
pylib/gyp/mac_tool.py:97: fp.write(s.decode(input_code).encode('UTF-16'))
On 2013/08/29 17:48:23, rsesek wrote:
> Does this specify LE order?

Not explicitly, but Python seems to go with host order, which is LE. I added a
test so we'll notice when this stops working. (I'm guessing xcodebuild uses host
order too, not always LE. But I don't know.)

[Nico, 2013-08-29 17:55:10]

https://codereview.chromium.org/23710010/diff/12001/test/mac/gyptest-app.py
File test/mac/gyptest-app.py (right):

https://codereview.chromium.org/23710010/diff/12001/test/mac/gyptest-app.py#n...
test/mac/gyptest-app.py:49: test.must_contain(strings, '\xff\xfe' + '/*
Localized'.encode('utf-16le'))
On 2013/08/29 17:48:23, rsesek wrote:
> Comment that this is the BOM?

Done.

[Robert Sesek, 2013-08-29 18:02:23]

LGTM

[Nico, 2013-08-29 18:14:54]

Committed patchset #6 manually as r1707 (presubmit successful).