Earlgrey test for webScenarioBrowsingLinkPreventDefaultOverridesHref

ios/web/shell/test/web_shell_navigation_egtest.mm

Index: ios/web/shell/test/web_shell_navigation_egtest.mm
diff --git a/ios/web/shell/test/web_shell_navigation_egtest.mm b/ios/web/shell/test/web_shell_navigation_egtest.mm
index e3a58281a5b9764af9cfc3d648af87073c3298d2..23851f01fd9787ba1f2befa633c6ccfc7e75ef86 100644
--- a/ios/web/shell/test/web_shell_navigation_egtest.mm
+++ b/ios/web/shell/test/web_shell_navigation_egtest.mm
@@ -139,4 +139,26 @@
       assertWithMatcher:grey_notNil()];
 }
 
+// Tests tapping a link with onclick="event.preventDefault()" and verifies that
+// the URL didn't change..
+- (void)testNavigationLinkPreventDefaultOverridesHref {
+  // Create map of canned responses and set up the test HTML server.
+  std::map<GURL, std::string> responses;
+  const GURL URL = web::test::HttpServer::MakeUrl("http://overridesHrefLink");
+  const std::string response =
+      "<a href='#hash' id='overridesHref' "

[Eugene But (OOO till 7-30), 2016/06/23 22:31:50]

NIT: s/overridesHref/overrides-href Per JS Style Guide

[huangml1, 2016/06/23 23:25:31]

Done.

+      "onclick='event.preventDefault()'>linkPreventDefaultOverridesHref</a>";

[Eugene But (OOO till 7-30), 2016/06/23 22:31:50]

Maybe you should add something to the page (like text marker) and check if
marker is present? This way you will make sure that navigation did not happen
because of |preventDefault| and not because there was a bug in the test or so.

[huangml1, 2016/06/23 23:25:31]

Done.

+  responses[URL] = response;
+
+  web::test::SetUpSimpleHttpServer(responses);
+  web::shell_test_util::LoadUrl(URL);
+  [[EarlGrey selectElementWithMatcher:web::addressFieldText(URL.spec())]
+      assertWithMatcher:grey_notNil()];
+
+  web::shell_test_util::TapWebViewElementWithId("overridesHref");
+
+  [[EarlGrey selectElementWithMatcher:web::addressFieldText(URL.spec())]

[huangml1, 2016/06/23 22:21:39]

I only checked no url change here for this test.  Is it necessary to check the
tab counts as well?

[Eugene But (OOO till 7-30), 2016/06/23 22:31:50]

Well you can not check tab counts here. So maybe this test should be downstream?
Alternatively we can have 2 tests one upstream (this one) and one downstream
(with Tab count check). In any case I do see a value in landing this test as it
is.

[Eugene But (OOO till 7-30), 2016/06/24 00:41:25]

So this does not really protect from the broken test. Imagine that there is a
typo in preventDefault() and it throws an exception. This test will pass and
will keep passing even if there will be a bug in software. My proposal was to
add something to the page and make sure that added thing is there.

+      assertWithMatcher:grey_notNil()];
+}
+
 @end