Add standalone script to parse Chrome Endure test results and dump to graph files.

Add standalone script to parse Chrome Endure test results and dump to graph files.

This script parses Chrome Endure test results by analyzing the "stdio" links
on the chromium.pyauto waterfall where the Chrome Endure bots run.  The
test results are then dumped into local text files that can be displayed
by the Chrome Endure graphing files.

BUG=chromium-os:32302
TEST=Verified the script runs on my local machine and result graphs can be viewed with the data files created by this script.


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=158728

[dennis_jeffrey, 2012-09-19 01:09:49]

Nirnimesh, could you please review this?

Ultimately this script should be removed once the functionality is integrated
with buildbot itself.  Right now the script is standalone, but that's how it
currently runs.  I'm checking it in now because dmikurube@ needs to work with it
to account for his recent changes integrating the deep memory profiler with
Chrome Endure.

[Dai Mikurube (NOT FULLTIME), 2012-09-20 04:26:08]

Just nits.  Let's use valid JSON like http://crrev.com/155120.

http://codereview.chromium.org/10944003/diff/2001/chrome/test/functional/perf...
File chrome/test/functional/perf/endure_result_parser.py (right):

http://codereview.chromium.org/10944003/diff/2001/chrome/test/functional/perf...
chrome/test/functional/perf/endure_result_parser.py:106: line_dict = eval(line)
eval => simplejson.loads ?

http://codereview.chromium.org/10944003/diff/2001/chrome/test/functional/perf...
chrome/test/functional/perf/endure_result_parser.py:108: existing_lines[i] =
str(new_line)
str => simplejson.dumps ?

http://codereview.chromium.org/10944003/diff/2001/chrome/test/functional/perf...
chrome/test/functional/perf/endure_result_parser.py:114:
existing_lines.insert(0, str(new_line))
str => simplejson.dumps ?

[Dai Mikurube (NOT FULLTIME), 2012-09-20 06:58:48]

Hi Dennis,

I tried accepting multiple performance values in endure_result_parser.py in
addition to the comment below.

Please take a look at http://codereview.chromium.org/10960009.  If it's ok, I'll
commit it after your CL is landed.  Or, it's ok to merge this CL to your CL.

http://codereview.chromium.org/10944003/diff/2001/chrome/test/functional/perf...
File chrome/test/functional/perf/endure_result_parser.py (right):

http://codereview.chromium.org/10944003/diff/2001/chrome/test/functional/perf...
chrome/test/functional/perf/endure_result_parser.py:108: existing_lines[i] =
str(new_line)
It looks overwriting completely, right?

I remember _OutputPerfForStandaloneGraphing in perf.py was merging new results
and existing lines.  But, it doesn't do that.

Does this script assume all RESULT lines are given for one revision run?  In
other words, can we assume that no new data are inserted for an existing
revision?

[dennis_jeffrey, 2012-09-25 01:21:21]

http://codereview.chromium.org/10944003/diff/2001/chrome/test/functional/perf...
File chrome/test/functional/perf/endure_result_parser.py (right):

http://codereview.chromium.org/10944003/diff/2001/chrome/test/functional/perf...
chrome/test/functional/perf/endure_result_parser.py:106: line_dict = eval(line)
On 2012/09/20 04:26:08, Dai Mikurube wrote:
> eval => simplejson.loads ?

Done.

http://codereview.chromium.org/10944003/diff/2001/chrome/test/functional/perf...
chrome/test/functional/perf/endure_result_parser.py:108: existing_lines[i] =
str(new_line)
On 2012/09/20 04:26:08, Dai Mikurube wrote:
> str => simplejson.dumps ?

Done.

http://codereview.chromium.org/10944003/diff/2001/chrome/test/functional/perf...
chrome/test/functional/perf/endure_result_parser.py:108: existing_lines[i] =
str(new_line)
On 2012/09/20 06:58:48, Dai Mikurube wrote:
> It looks overwriting completely, right?
> 
> I remember _OutputPerfForStandaloneGraphing in perf.py was merging new results
> and existing lines.  But, it doesn't do that.
> 
> Does this script assume all RESULT lines are given for one revision run?  In
> other words, can we assume that no new data are inserted for an existing
> revision?

You're right that the current script should be merging new results into existing
lines in the same way it's done for _OutputPerfForStandaloneGraphing.  This
should be taken care of though.  See line 355 below where it's appending new
results to an already-seen perf line in the same test execution.

http://codereview.chromium.org/10944003/diff/2001/chrome/test/functional/perf...
chrome/test/functional/perf/endure_result_parser.py:114:
existing_lines.insert(0, str(new_line))
On 2012/09/20 04:26:08, Dai Mikurube wrote:
> str => simplejson.dumps ?

Done.

[Nirnimesh, 2012-09-25 20:31:28]

Sorry for the delay; I somehow missed reviewing this CL

http://codereview.chromium.org/10944003/diff/9001/chrome/test/functional/perf...
File chrome/test/functional/perf/endure_result_parser.py (right):

http://codereview.chromium.org/10944003/diff/9001/chrome/test/functional/perf...
chrome/test/functional/perf/endure_result_parser.py:45: BUILDER_URL_BASE =
'http://chromegw.corp.google.com/i/chromium.pyauto/builders/'
we should not refer to internal URLs and paths in public scripts

http://codereview.chromium.org/10944003/diff/9001/chrome/test/functional/perf...
chrome/test/functional/perf/endure_result_parser.py:77: command = 'ln -s
../../endure_plotter.html ' + link_file
use os.symlink

[dennis_jeffrey, 2012-09-25 21:47:40]

Thanks!

http://codereview.chromium.org/10944003/diff/9001/chrome/test/functional/perf...
File chrome/test/functional/perf/endure_result_parser.py (right):

http://codereview.chromium.org/10944003/diff/9001/chrome/test/functional/perf...
chrome/test/functional/perf/endure_result_parser.py:45: BUILDER_URL_BASE =
'http://chromegw.corp.google.com/i/chromium.pyauto/builders/'
On 2012/09/25 20:31:28, Nirnimesh wrote:
> we should not refer to internal URLs and paths in public scripts

Changed to the external URL.

http://codereview.chromium.org/10944003/diff/9001/chrome/test/functional/perf...
chrome/test/functional/perf/endure_result_parser.py:77: command = 'ln -s
../../endure_plotter.html ' + link_file
On 2012/09/25 20:31:28, Nirnimesh wrote:
> use os.symlink

Done.

[Nirnimesh, 2012-09-25 22:33:03]

LGTM

[dennis_jeffrey, 2012-09-25 22:47:06]

Thanks, Nirnimesh.

Dai, does this CL look good to you?  I'll hold off on submitting it since you
had left some comments and I just want to make sure they were adequately
addressed before I commit.  If it looks good to you without any additional
changes, feel free to click the "commit" checkbox on this CL to commit it for
me.

[Dai Mikurube (NOT FULLTIME), 2012-09-26 00:13:20]

On 2012/09/25 22:47:06, dennis_jeffrey wrote:
> Thanks, Nirnimesh.
> 
> Dai, does this CL look good to you?  I'll hold off on submitting it since you
> had left some comments and I just want to make sure they were adequately
> addressed before I commit.  If it looks good to you without any additional
> changes, feel free to click the "commit" checkbox on this CL to commit it for
> me.

lgtm. Ok, I check the "Commit" box. :)

[commit-bot: I haz the power, 2012-09-26 00:18:23]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/dennisjeffrey@chromium.org/10944003/14001

[commit-bot: I haz the power, 2012-09-26 02:53:44]

Change committed as 158728