Fix a minor regex bug.

Fix a minor regex bug.

Match against '.', rather than any character (how '.' is treated
in a regular expression).

Previous to this, xfermodes_8888*png matched the expression.

R=epoger@google.com

Committed: https://code.google.com/p/skia/source/detail?r=11714

[scroggo, 2013-10-10 00:01:04]

Not a problem, in practice, but I noticed it while investigation rebaseline.py

[bungeman-skia, 2013-10-10 14:38:15]

https://codereview.chromium.org/26709003/diff/1/gm/gm_json.py
File gm/gm_json.py (right):

https://codereview.chromium.org/26709003/diff/1/gm/gm_json.py#newcode86
gm/gm_json.py:86: IMAGE_FILENAME_PATTERN = '(\S+)_(\S+)\\.png'  # matches
(testname, config)
Wait, won't this match 'blah_blah\<anything>png'? I think you just want one '\'
here (otherwise the other '\S' are also wrong).

[epoger, 2013-10-10 14:41:31]

https://codereview.chromium.org/26709003/diff/1/gm/gm_json.py
File gm/gm_json.py (right):

https://codereview.chromium.org/26709003/diff/1/gm/gm_json.py#newcode86
gm/gm_json.py:86: IMAGE_FILENAME_PATTERN = '(\S+)_(\S+)\\.png'  # matches
(testname, config)
On 2013/10/10 14:38:16, bungeman1 wrote:
> Wait, won't this match 'blah_blah\<anything>png'? I think you just want one
'\'
> here (otherwise the other '\S' are also wrong).

I think Ben is right.

Either leave it alone, or change it to:
'(\S+)_(\S+)\.png'

[scroggo, 2013-10-10 14:59:02]

https://codereview.chromium.org/26709003/diff/1/gm/gm_json.py
File gm/gm_json.py (right):

https://codereview.chromium.org/26709003/diff/1/gm/gm_json.py#newcode86
gm/gm_json.py:86: IMAGE_FILENAME_PATTERN = '(\S+)_(\S+)\\.png'  # matches
(testname, config)
On 2013/10/10 14:41:32, epoger wrote:
> On 2013/10/10 14:38:16, bungeman1 wrote:
> > Wait, won't this match 'blah_blah\<anything>png'? I think you just want one
> '\'
> > here (otherwise the other '\S' are also wrong).
> 
> I think Ben is right.
> 
> Either leave it alone, or change it to:
> '(\S+)_(\S+)\.png'

It seems the two are equivalent, at least on my linux machine. Using my python
interpreter:

>>> import re
>>> pattern = '(\S+)_(\S+)\.png'
>>> matcher = re.compile(pattern)
>>> pattern2 = '(\S+)_(\S+)\\.png'
>>> matcher2 = re.compile(pattern2)
>>> matcher == matcher2
True

That said, I'm fine either way (and maybe the double \ will not work on
Windows?). I've uploaded a CL with only one \

[epoger, 2013-10-10 15:13:46]

LGTM at patchset 2

https://codereview.chromium.org/26709003/diff/1/gm/gm_json.py
File gm/gm_json.py (right):

https://codereview.chromium.org/26709003/diff/1/gm/gm_json.py#newcode86
gm/gm_json.py:86: IMAGE_FILENAME_PATTERN = '(\S+)_(\S+)\\.png'  # matches
(testname, config)
On 2013/10/10 14:59:02, scroggo wrote:
> It seems the two are equivalent, at least on my linux machine.

Wow, that's surprising.  I get the same results.  And here's some more detail on
how we get those results:

>>> import re
>>> pattern = '(\S+)_(\S+)\.png'
>>> pattern2 = '(\S+)_(\S+)\\.png'
>>> pattern
'(\\S+)_(\\S+)\\.png'
>>> pattern2
'(\\S+)_(\\S+)\\.png'
>>> pattern == pattern2
True

> That said, I'm fine either way (and maybe the double \ will not work on
> Windows?). I've uploaded a CL with only one \

I can't imagine that it would matter whether Unix/Windows for this.  But I've
been surprised before...

[scroggo, 2013-10-10 20:41:48]

Committed patchset #2 manually as r11714 (presubmit successful).