Translate: browser test for Chrome Translate in an isolated world

Translate: browser test for Chrome Translate in an isolated world

An browser test which cover whole translate process was disabled tentatively.
This is because now Translate scripts runs inside a v8 isolated world and
it make impossible to run with local testing server.
This change introduce a flag to change the allowed server for XHR, and
make the test enabled again.
Note that this flag is warned by infobar when it is enabled.

BUG=268785
TEST=browser_tests --gtest_filter='Translate*.*'
[email protected]

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=215912

[Takashi Toyoshima, 2013-04-22 06:31:27]

Hi Mike, and Mad,

I split browser test from the change for CSP support.
Now, the CSP change needs reviews from many persons.
To keep main CL as simple as possible will make reviews effective.

[Takashi Toyoshima, 2013-04-23 07:32:47]

Since the main change looks difficult to land soon and another change by Hajime
depends on this browser test, I made non-csp version of this test.

Mike:
Please take a look on main change to support csp,
https://codereview.chromium.org/14022005/

Mad:
Please take a look non-csp version test firstly
https://codereview.chromium.org/13825006/ .

Thanks.

[Takashi Toyoshima, 2013-07-30 07:09:05]

Hey, now is the time to enable browser tests in the isolated world.
Can you take a look?

[hajimehoshi, 2013-07-30 08:29:05]

https://chromiumcodereview.appspot.com/14180010/diff/25001/chrome/test/data/t...
File chrome/test/data/translate/fr_test.html (right):

https://chromiumcodereview.appspot.com/14180010/diff/25001/chrome/test/data/t...
chrome/test/data/translate/fr_test.html:12: <script>world='ng';</script> <!--
This line doesn't run because of CSP. -->
The string 'ng' seems not to be checked anywhere. Maybe the test should exist at
pseudo_main.js?

[Takashi Toyoshima, 2013-07-30 09:31:22]

I remember what this test does.
global variable 'world' checks two features working correctly.

1. content security policy blocks inline JavaScript executions
2. translate script injection happens in the isolated world, and the main world
and the isolated world can not affect each other.

world='ng'; in the fr_test.html doesn't run because of CSP restriction. Once it
runs, pseudo_main.js catches it because typeof world becomes 'string'.
But, once it runs inside the isolated world, this check doesn't work because
world is assigned in the main world, and it doesn't affect the isolated world.
So, I should check it in fr_test.js now.

On the other hand, world='main' of fr_test.js happens in the main world. Once
translate runs in the main world, pseudo_main.js catches it and fails.

[hajimehoshi, 2013-07-31 06:59:22]

https://codereview.chromium.org/14180010/diff/41001/chrome/browser/translate/...
File chrome/browser/translate/translate_browsertest.cc (right):

https://codereview.chromium.org/14180010/diff/41001/chrome/browser/translate/...
chrome/browser/translate/translate_browsertest.cc:101: EXPECT_TRUE(confirm);
Please add comment to explain why the confirm inforbar appears.

[hajimehoshi, 2013-07-31 07:04:00]

I think the comment should exist on each test.

https://codereview.chromium.org/14180010/diff/41001/chrome/browser/translate/...
File chrome/browser/translate/translate_browsertest.cc (right):

https://codereview.chromium.org/14180010/diff/41001/chrome/browser/translate/...
chrome/browser/translate/translate_browsertest.cc:221: EXPECT_TRUE(confirm);
ditto

https://codereview.chromium.org/14180010/diff/41001/chrome/browser/translate/...
chrome/browser/translate/translate_browsertest.cc:262: EXPECT_TRUE(confirm);
ditto

https://codereview.chromium.org/14180010/diff/41001/chrome/browser/translate/...
chrome/browser/translate/translate_browsertest.cc:303: EXPECT_TRUE(confirm);
ditto

https://codereview.chromium.org/14180010/diff/41001/chrome/browser/translate/...
chrome/browser/translate/translate_browsertest.cc:344: EXPECT_TRUE(confirm);
ditto

[hajimehoshi, 2013-07-31 07:55:50]

chrome/*/translate/* lgtm

On 2013/07/31 07:04:00, hajimehoshi wrote:
> I think the comment should exist on each test.
> 
>
https://codereview.chromium.org/14180010/diff/41001/chrome/browser/translate/...
> File chrome/browser/translate/translate_browsertest.cc (right):
> 
>
https://codereview.chromium.org/14180010/diff/41001/chrome/browser/translate/...
> chrome/browser/translate/translate_browsertest.cc:221: EXPECT_TRUE(confirm);
> ditto
> 
>
https://codereview.chromium.org/14180010/diff/41001/chrome/browser/translate/...
> chrome/browser/translate/translate_browsertest.cc:262: EXPECT_TRUE(confirm);
> ditto
> 
>
https://codereview.chromium.org/14180010/diff/41001/chrome/browser/translate/...
> chrome/browser/translate/translate_browsertest.cc:303: EXPECT_TRUE(confirm);
> ditto
> 
>
https://codereview.chromium.org/14180010/diff/41001/chrome/browser/translate/...
> chrome/browser/translate/translate_browsertest.cc:344: EXPECT_TRUE(confirm);
> ditto

[Takashi Toyoshima, 2013-07-31 13:19:32]

As we discussed offline, I factor out infobar check as a class method so to omit
boilerplate comments.

[Takashi Toyoshima, 2013-07-31 13:22:58]

+sky for chrome/ since this change introduces new *bad* flag and passes it to
renderers.
Note that this flag is not only for browser_tests, but needed to verify
experimental server implementations.

[sky, 2013-07-31 16:26:14]

LGTM - but I get nervous when I see 'security' in the name of something. Do the
security folks need to review this?

[sky, 2013-07-31 16:26:34]

Also, the description makes no sense to me. Please update it.

On Wed, Jul 31, 2013 at 6:22 AM,  <[email protected]> wrote:
> +sky for chrome/ since this change introduces new *bad* flag and passes it
> to
> renderers.
> Note that this flag is not only for browser_tests, but needed to verify
> experimental server implementations.
>
> https://chromiumcodereview.appspot.com/14180010/

[Takashi Toyoshima, 2013-08-01 02:57:36]

Add Palmer for security concern. He was looped in the discussion for Translate
CSP support.

Hi Palmer,
I'm going to add a flag to specify security origin for the Translate isolated
world.
This allows translate scripts in the isolated world to use XHR between localhost
instead of translate.googleapis.com.

Please let me know if you have any question.

[palmer, 2013-08-01 21:22:38]

This looks reasonable to me, but tsepez is a better reviewer for CSP things than
I am.

https://chromiumcodereview.appspot.com/14180010/diff/53001/chrome/browser/tra...
File chrome/browser/translate/translate_browsertest.cc (right):

https://chromiumcodereview.appspot.com/14180010/diff/53001/chrome/browser/tra...
chrome/browser/translate/translate_browsertest.cc:54: // Setup altenate
security-origin for testing in order to allow XHR against
Typos: "alternate", "security origin" (or just "origin")

https://chromiumcodereview.appspot.com/14180010/diff/53001/chrome/browser/tra...
chrome/browser/translate/translate_browsertest.cc:257: // TODO(toyoshim):
Refactoring this test, too.
Typo: "Refactor" (it's a command to yourself)

[Tom Sepez, 2013-08-01 21:31:32]

Looks reasonable, but seems redundant to have both

extern const char kTranslateScriptURL[];
extern const char kTranslateSecurityOrigin[];

Would you ever set one without the other to match?  If not, then perhaps extract
the second from the first.

[Takashi Toyoshima, 2013-08-02 04:18:31]

Palmer and Tom,
Thank you for your checking this.

Tom:
In this browser test, these two have the same host:port pair. But for server
side testing, We need to use different servers since one hosts scripts and the
other provides translating API over JSON/XHR.
I guess we can use the same name if I revise the server, but I'm not sure for
now. I may try it later, but I want to keep these two options for now.

[Takashi Toyoshima, 2013-08-02 04:40:19]

Thanks.
Now, I push this change to CQ.

https://chromiumcodereview.appspot.com/14180010/diff/53001/chrome/browser/tra...
File chrome/browser/translate/translate_browsertest.cc (right):

https://chromiumcodereview.appspot.com/14180010/diff/53001/chrome/browser/tra...
chrome/browser/translate/translate_browsertest.cc:54: // Setup altenate
security-origin for testing in order to allow XHR against
On 2013/08/01 21:22:38, Chromium Palmer wrote:
> Typos: "alternate", "security origin" (or just "origin")

Done.

https://chromiumcodereview.appspot.com/14180010/diff/53001/chrome/browser/tra...
chrome/browser/translate/translate_browsertest.cc:54: // Setup altenate
security-origin for testing in order to allow XHR against
On 2013/08/01 21:22:38, Chromium Palmer wrote:
> Typos: "alternate", "security origin" (or just "origin")

Done.

https://chromiumcodereview.appspot.com/14180010/diff/53001/chrome/browser/tra...
chrome/browser/translate/translate_browsertest.cc:257: // TODO(toyoshim):
Refactoring this test, too.
Oops. This was a memo when I'm doing this change.
So this is not just a type, but a bug which should be removed...
Now, I removed this TODO line,

[commit-bot: I haz the power, 2013-08-02 04:40:46]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/14180010/67001

[commit-bot: I haz the power, 2013-08-02 08:05:13]

Retried try job too often on win_rel for step(s) browser_tests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win_rel&nu...

[commit-bot: I haz the power, 2013-08-02 09:07:58]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/14180010/67001

[commit-bot: I haz the power, 2013-08-02 11:23:45]

Retried try job too often on win7_aura for step(s) browser_tests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win7_aura&...

[commit-bot: I haz the power, 2013-08-05 05:41:37]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/14180010/67001

[commit-bot: I haz the power, 2013-08-05 08:39:18]

Retried try job too often on win7_aura for step(s) browser_tests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=win7_aura&...

[Takashi Toyoshima, 2013-08-06 13:12:10]

Two memory bugs caused win7_aura and win_rel build failures.
I fixed them in Patch Set 14.

[commit-bot: I haz the power, 2013-08-06 13:13:17]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/14180010/107001

[commit-bot: I haz the power, 2013-08-06 17:14:23]

Change committed as 215912