Integration tests for WebLite to Lo-Fi fallback

Integration tests for WebLite to Lo-Fi fallback

Test that the DataCompressionProxyLitePageFallback field trial requests
Lo-Fi images when a Lite Page is not returned. Also check that when the
user is not in the field trial, Lo-Fi images are not requested.

Seperate the Lo-Fi and WebLite test files.

BUG=660212
Review-Url: https://codereview.chromium.org/2726423003
Cr-Commit-Position: refs/heads/master@{#454724}
Committed: https://chromium.googlesource.com/chromium/src/+/a9a7fa9290bc1d58586c6047980951ab7917afd6

[megjablon, 2017-03-03 20:29:00]

megjablon@chromium.org changed reviewers:
+ tbansal@chromium.org

[megjablon, 2017-03-03 20:29:01]

PTAL, thanks!

[tbansal1, 2017-03-03 21:17:56]

https://codereview.chromium.org/2726423003/diff/1/tools/chrome_proxy/webdrive...
File tools/chrome_proxy/webdriver/lite_page.py (right):

https://codereview.chromium.org/2726423003/diff/1/tools/chrome_proxy/webdrive...
tools/chrome_proxy/webdriver/lite_page.py:44:
test_driver.AddChromeArg('--data-reduction-proxy-lo-fi=always-on')
Can you add a comment that setting these two flags turns on the lite page
fallback mechanism.

Also, why not set the field trial instead of the flags to make it consistent
with the test below?

https://codereview.chromium.org/2726423003/diff/1/tools/chrome_proxy/webdrive...
tools/chrome_proxy/webdriver/lite_page.py:83: 'effective_connection_type/4G')
I think it is safer to also force NQE to return ECT of say 2G.

--force-net-effective-connection-type=2g

This flag is not there in Chromium yet, but I will add code for it soon.

https://codereview.chromium.org/2726423003/diff/1/tools/chrome_proxy/webdrive...
tools/chrome_proxy/webdriver/lite_page.py:102: print response.url
Why print?

[megjablon, 2017-03-03 22:05:03]

https://codereview.chromium.org/2726423003/diff/1/tools/chrome_proxy/webdrive...
File tools/chrome_proxy/webdriver/lite_page.py (right):

https://codereview.chromium.org/2726423003/diff/1/tools/chrome_proxy/webdrive...
tools/chrome_proxy/webdriver/lite_page.py:44:
test_driver.AddChromeArg('--data-reduction-proxy-lo-fi=always-on')
On 2017/03/03 21:17:56, tbansal1 wrote:
> Can you add a comment that setting these two flags turns on the lite page
> fallback mechanism.
> 
> Also, why not set the field trial instead of the flags to make it consistent
> with the test below?

Made it consistent with the test below.

https://codereview.chromium.org/2726423003/diff/1/tools/chrome_proxy/webdrive...
tools/chrome_proxy/webdriver/lite_page.py:83: 'effective_connection_type/4G')
On 2017/03/03 21:17:56, tbansal1 wrote:
> I think it is safer to also force NQE to return ECT of say 2G.
> 
> --force-net-effective-connection-type=2g
> 
> This flag is not there in Chromium yet, but I will add code for it soon.

Done.

https://codereview.chromium.org/2726423003/diff/1/tools/chrome_proxy/webdrive...
tools/chrome_proxy/webdriver/lite_page.py:102: print response.url
On 2017/03/03 21:17:56, tbansal1 wrote:
> Why print?

Ah my bad, was using this when .png was getting lite page and forgot to remove
when it succeeded because it only prints when the test fails.

[tbansal1, 2017-03-03 22:40:37]

lgtm % comment. Thanks for the tests.

https://codereview.chromium.org/2726423003/diff/20001/tools/chrome_proxy/webd...
File tools/chrome_proxy/webdriver/lite_page.py (right):

https://codereview.chromium.org/2726423003/diff/20001/tools/chrome_proxy/webd...
tools/chrome_proxy/webdriver/lite_page.py:44:
test_driver.AddChromeArg('--force-fieldtrials='
Since we are not using the flags anymore, it means
"exp=ignore_preview_blacklist" would not be set. I think that should be okay,
but please double check.

[megjablon, 2017-03-03 22:48:52]

https://codereview.chromium.org/2726423003/diff/20001/tools/chrome_proxy/webd...
File tools/chrome_proxy/webdriver/lite_page.py (right):

https://codereview.chromium.org/2726423003/diff/20001/tools/chrome_proxy/webd...
tools/chrome_proxy/webdriver/lite_page.py:44:
test_driver.AddChromeArg('--force-fieldtrials='
On 2017/03/03 22:40:36, tbansal1 wrote:
> Since we are not using the flags anymore, it means
> "exp=ignore_preview_blacklist" would not be set. I think that should be okay,
> but please double check.

Yes this won't be set anymore, but we manually "blacklisted" Lite Pages for the
test page so this is the behavior we want (i.e. request a Lite Page and don't
receive one).

[megjablon, 2017-03-04 00:02:28]

The CQ bit was checked by megjablon@chromium.org

[commit-bot: I haz the power, 2017-03-04 00:03:06]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-03-04 00:33:23]

CQ is committing da patch.

Bot data: {"patchset_id": 20001, "attempt_start_ts": 1488585748975630,
"parent_rev": "0f799230c0f27e3e8882868f46d9d5cefe9c8c6d", "commit_rev":
"a9a7fa9290bc1d58586c6047980951ab7917afd6"}

[commit-bot: I haz the power, 2017-03-04 00:34:10]

Description was changed from

==========
Integration tests for WebLite to Lo-Fi fallback

Test that the DataCompressionProxyLitePageFallback field trial requests
Lo-Fi images when a Lite Page is not returned. Also check that when the
user is not in the field trial, Lo-Fi images are not requested.

Seperate the Lo-Fi and WebLite test files.

BUG=660212
==========

to

==========
Integration tests for WebLite to Lo-Fi fallback

Test that the DataCompressionProxyLitePageFallback field trial requests
Lo-Fi images when a Lite Page is not returned. Also check that when the
user is not in the field trial, Lo-Fi images are not requested.

Seperate the Lo-Fi and WebLite test files.

BUG=660212

Review-Url: https://codereview.chromium.org/2726423003
Cr-Commit-Position: refs/heads/master@{#454724}
Committed:
https://chromium.googlesource.com/chromium/src/+/a9a7fa9290bc1d58586c60479809...
==========

[commit-bot: I haz the power, 2017-03-04 00:34:12]

Committed patchset #2 (id:20001) as
https://chromium.googlesource.com/chromium/src/+/a9a7fa9290bc1d58586c60479809...