Add chrome frame tests for common navigation cases to ensure IE is not broken.

Add chrome frame tests for common navigation cases to ensure IE is not broken.
BUG=none
TEST=none

[kkania, 2010-05-25 14:41:26]

I just added a couple test cases to make sure you were ok with the approach.  I
ended up just performing clicks for interacting with all the menu items.  Let me
know if you know of a better approach. I was planning to send a WM_COMMAND with
the menu item id, but I could not figure out what window to send it to.

[amit, 2010-05-25 21:20:08]

Very nice and cool new tests. We'll need to create chrome frame counterparts for
the redirect tests.

http://codereview.chromium.org/2173002/diff/6001/7002
File chrome_frame/test/chrome_frame_test_utils.cc (right):

http://codereview.chromium.org/2173002/diff/6001/7002#newcode47
chrome_frame/test/chrome_frame_test_utils.cc:47: const int MenuItemIds::kOpen =
2136;
These constants are defined in mshtmcid.h Look for IDM_FOLLOWLINKC,
IDM_FOLLOWLINKT etc. 
http://msdn.microsoft.com/en-us/library/aa741315(v=VS.85).aspx

http://codereview.chromium.org/2173002/diff/6001/7002#newcode876
chrome_frame/test/chrome_frame_test_utils.cc:876: bool
WebBrowserEventSink::ExecuteJavascript(const std::wstring& script,
Looking at the following functions, it's somewhat complicated.  Another way do
do this is to use the following method to invoke menus.

- Click on the mshtml content to give it input focus.
- Tab to highlight next link.
- Shift + F10 to pop up menu.
- Arrow keys + enter to select.

Further advantage of using 'arrow keys + enter' is that it works even if the
menu is not yet displayed.

http://codereview.chromium.org/2173002/diff/6001/7007
File chrome_frame/test/no_interference_test.cc (right):

http://codereview.chromium.org/2173002/diff/6001/7007#newcode153
chrome_frame/test/no_interference_test.cc:153: TEST_F(NoInterferenceTest,
JavascriptRedirect) {
Nice tests!

http://codereview.chromium.org/2173002/diff/6001/7007#newcode159
chrome_frame/test/no_interference_test.cc:159:
.WillOnce(ExecuteJavascript(&mock_, redirect_script));
Executing script from the automation is different from the script execution on
the page. This test should load the page containing the script for
'window.location=...'

http://codereview.chromium.org/2173002/diff/6001/7007#newcode171
chrome_frame/test/no_interference_test.cc:171:
.WillOnce(ClickIEDOMElement(&mock_, L"document.links[0]",
might be easier to simulate this by tab+enter

http://codereview.chromium.org/2173002/diff/6001/7007#newcode174
chrome_frame/test/no_interference_test.cc:174:
mock_.ExpectNavigationSequenceForAnchors(empty_page_url());
Ideally, you should be using ExpectNavigation flavor of expection here.
ExpectNavigation (and especially ExpectNavigationAndSwitch) itself is not
entirely correct since it has vague expectations (testing::Anynumber). Others
are quite chrome frame specific. Fee free to create a new ExpectNavigationInIE
that suits your needs.

[kkania, 2010-05-26 22:12:32]

http://codereview.chromium.org/2173002/diff/6001/7002
File chrome_frame/test/chrome_frame_test_utils.cc (right):

http://codereview.chromium.org/2173002/diff/6001/7002#newcode47
chrome_frame/test/chrome_frame_test_utils.cc:47: const int MenuItemIds::kOpen =
2136;
On 2010/05/25 21:20:08, amit wrote:
> These constants are defined in mshtmcid.h Look for IDM_FOLLOWLINKC,
> IDM_FOLLOWLINKT etc. 
> http://msdn.microsoft.com/en-us/library/aa741315%28v=VS.85%29.aspx

Done.

http://codereview.chromium.org/2173002/diff/6001/7002#newcode876
chrome_frame/test/chrome_frame_test_utils.cc:876: bool
WebBrowserEventSink::ExecuteJavascript(const std::wstring& script,
On 2010/05/25 21:20:08, amit wrote:
> Looking at the following functions, it's somewhat complicated.  Another way do
> do this is to use the following method to invoke menus.
> 
> - Click on the mshtml content to give it input focus.
> - Tab to highlight next link.
> - Shift + F10 to pop up menu.
> - Arrow keys + enter to select.
> 
> Further advantage of using 'arrow keys + enter' is that it works even if the
> menu is not yet displayed. 
> 

I switched to using the keyboard method, although I think the tests themselves
were simpler with the previous methods. Also, the new way requires delays.

http://codereview.chromium.org/2173002/diff/6001/7007
File chrome_frame/test/no_interference_test.cc (right):

http://codereview.chromium.org/2173002/diff/6001/7007#newcode159
chrome_frame/test/no_interference_test.cc:159:
.WillOnce(ExecuteJavascript(&mock_, redirect_script));
On 2010/05/25 21:20:08, amit wrote:
> Executing script from the automation is different from the script execution on
> the page. This test should load the page containing the script for
> 'window.location=...'

Done.

http://codereview.chromium.org/2173002/diff/6001/7007#newcode171
chrome_frame/test/no_interference_test.cc:171:
.WillOnce(ClickIEDOMElement(&mock_, L"document.links[0]",
On 2010/05/25 21:20:08, amit wrote:
> might be easier to simulate this by tab+enter

Done.

http://codereview.chromium.org/2173002/diff/6001/7007#newcode174
chrome_frame/test/no_interference_test.cc:174:
mock_.ExpectNavigationSequenceForAnchors(empty_page_url());
On 2010/05/25 21:20:08, amit wrote:
> Ideally, you should be using ExpectNavigation flavor of expection here.
> ExpectNavigation (and especially ExpectNavigationAndSwitch) itself is not
> entirely correct since it has vague expectations (testing::Anynumber). Others
> are quite chrome frame specific. Fee free to create a new ExpectNavigationInIE
> that suits your needs.

Done. The only change in my expectations was that there should be only one
navigate complete.

[amit, 2010-05-26 22:58:42]

Looks excellent. LGTM!

http://codereview.chromium.org/2173002/diff/16001/17006
File chrome_frame/test/no_interference_test.cc (right):

http://codereview.chromium.org/2173002/diff/16001/17006#newcode115
chrome_frame/test/no_interference_test.cc:115: ACTION_P3(SelectItem, loop,
delay, index) {
nice action :)

http://codereview.chromium.org/2173002/diff/16001/17006#newcode134
chrome_frame/test/no_interference_test.cc:134: const std::wstring kWindowOpenUrl
= GetTestUrl(L"window_open.html");
Is this the same one as used in other window open tests? AFAIK, it needed a
keyboard input to open new windows as a work around for popup blockers.
Does this work as expected?

http://codereview.chromium.org/2173002/diff/16001/17009
File chrome_frame/test/test_mock_with_web_server.cc (right):

http://codereview.chromium.org/2173002/diff/16001/17009#newcode158
chrome_frame/test/test_mock_with_web_server.cc:158:
.Times(testing::AnyNumber());
I think we need to be precise here about cardinality here.

[kkania, 2010-05-27 01:06:55]

Can you please review this new change? I was not exactly sure when/why
FileDownload occurs. It seems to happen only for the first page only.

[amit, 2010-05-27 18:25:32]

Ugg, looking at how you have to jump through hoops to work around FileDownload
notification, I think original scheme, any cardinility for Download, was
probably OK.
would you like to revert to that? sorry for the trouble.
Still lgtm :)

[kkania, 2013-02-28 21:50:54]

https://codereview.chromium.org/2173002/diff/16001/chrome_frame/test/no_inter...
File chrome_frame/test/no_interference_test.cc (right):

https://codereview.chromium.org/2173002/diff/16001/chrome_frame/test/no_inter...
chrome_frame/test/no_interference_test.cc:134: const std::wstring kWindowOpenUrl
= GetTestUrl(L"window_open.html");
On 2010/05/26 22:58:42, amit wrote:
> Is this the same one as used in other window open tests? AFAIK, it needed a
> keyboard input to open new windows as a work around for popup blockers.
> Does this work as expected?

It works on XP ie6 and win7 ie8. Perhaps it was Chrome that was blocking the
popup in the other test?

https://codereview.chromium.org/2173002/diff/16001/chrome_frame/test/test_moc...
File chrome_frame/test/test_mock_with_web_server.cc (right):

https://codereview.chromium.org/2173002/diff/16001/chrome_frame/test/test_moc...
chrome_frame/test/test_mock_with_web_server.cc:158:
.Times(testing::AnyNumber());
On 2010/05/26 22:58:42, amit wrote:
> I think we need to be precise here about cardinality here.

Done.