| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
| |
Bug: 10427705
Allows Browser to compile even when WebViewClassic is no longer in
frameworks/base. The deleted functionality was already disabled at
runtime when using non-classic webview.
Change-Id: I21b58ec52ac059c699b49a2dd80008d9a9e6791a
|
|
|
|
|
|
| |
Interdepends on https://android-git.corp.google.com/g/#/c/198193
Change-Id: I9f44c0f4f84674965b7afc652420986314e2fc8e
|
|
|
|
| |
Change-Id: I40cae53d7ae724c44c40fb7f21649e807416ae54
|
|
|
|
| |
Change-Id: I5447e251a4b1b75561ae9a9a2174330505a6912b
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Receive the value for the new "capture" attribute on HTML
file pickers, and do the right thing with it.
Requires changes in WebKit
(I0a921be31fda79a43c05da4fe22d9c808d92709c)
and the framework (I494adc1274ca21ce8fe52a6c7b6b758217927e66).
Bug: 5771207
Change-Id: I38dfe2df043fdba1388384dbd3b5370737eb38e5
|
|
|
|
|
|
|
|
| |
Has to escape out to WebViewClassic whenever it needs a hidden API.
Interdepends on https://android-git.corp.google.com/g/165608
Change-Id: Ic61dd7f57f75381864075605d5901f2688648cbd
|
|\
| |
| |
| |
| | |
* commit '3e5e5fada69d74bb50b14ef3588e7beadaedfd69':
Remove the psychic labs feature.
|
| |
| |
| |
| |
| |
| |
| |
| | |
Reverts the UrlInputView to a standard autocomplete
that causes less trouble for IMEs.
bug:5598837
Change-Id: If469d76f4feda88e4a122264a39d317737404578
|
|/
|
|
|
|
|
| |
This is superseded by a new set of tests in frameworks/base/tests/WebViewTests.
Bug: 5140673
Change-Id: I2874afa5a52ad9ac77bad0aa07ec84cde117c54e
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously we passed a null Context into the WebStorageSizeManager
from the unit test. After the refactoring to get the
application context from the passed in Context, this no longer
works. So pass in a valid Context.
(See Ib0be5f8c for the regression)
Bug: 5319628
Change-Id: I691e2edb0a4174076c7599525569d104a83d3442
|
|
|
|
|
|
|
| |
This reverts commit faed5798f2ca3320e0b6afdb197feda25052f79a.
Bug: 5252155
Change-Id: Ibb3a84cd353fc63a7378d7106b1a6f44a9eb9768
|
|
|
|
|
|
|
| |
injected Java objects
Bug: 5222742
Change-Id: Id6222aebce7570b0649cd9cb7f4f16b0f0b640a8
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Bug: 5142434
For testIsValidParentWithAccount, the test would remove the account*
from a bookmark it was inserting with a valid parent. The provider
would correctly identify that a parent was set, but no account info and
thus used the account info of the parent. This is not what the test was
expecting, however, leading to this error. Fix the tests to explicitly
check for child account* not matching parent account* (in which case the
parent is discarded, and the bookmark is inserted at the root folder of
the child's account*), and for testing the lack of a specified account*.
Change-Id: I15b318ee1e846b18ffa187720814c3761d19933a
|
|
|
|
|
|
|
|
|
| |
Bug: 5113381
Fix case of using the wrong context (BP1 would use the app context
explicitly, bypassing the mock context)
Clean up test databases after runs
Change-Id: Ibac09d08d4ab7235d5763c209d84be2562e6fae3
|
|
|
|
|
|
|
| |
Bug: 5092056
Turns out the test was wrong, not the provider. Oops
Change-Id: I46439a342c810035c8dba0423c75a5daa882759c
|
|
|
|
|
|
| |
Bug: 5087957
Change-Id: Idf70c488baa515222ce8ed905aa5c80a529fd15f
|
|
|
|
|
|
|
|
| |
append an EXTRA_APPLICATION_ID field in the intent to navigate to
a new website, so that the browser won't open a new tab for every
new site
Change-Id: I69dddabeef83e124c43adab7bf3ed1e064b396ab
|
|
|
|
|
|
|
| |
WebView is complaining that some calls used in test app is not from
UI thread.
Change-Id: I7ef61694e3988a663eeedf03463580a93e57fe46
|
|
|
|
|
|
|
| |
Bug: 4473779
Plus, TESTS! Huzzah!
Change-Id: I043e100d99d4653b7fb7885217d7fb305930a137
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Following the example of onReceivedSslError, implement onReceivedClientCertRequest
ERROR CASE CLIENT CERT CASE
<... From frameworks/base ...> <... From frameworks/base ...>
Tab.SubWindowClient.onReceivedSslError Tab.SubWindowClient.onReceivedClientCertRequest
Tab.mWebViewClient.onReceivedSslError Tab.mWebViewClient.onReceivedClientCertRequest
<... ssl_warnings dialog ...> <... KeyChain.choosePrivateKeyAlias/KeyChainLookup ...>
SslErrorHandler.proceed (with SslCertLookupTable) ClientCertRequestHandler.proceed (with SslClientCertLookupTable)
<... To frameworks/base ...> <... To frameworks/base ...>
Change-Id: I3ed3789c4efc97c87ab4773cdaed3e654a1fd1e3
|
|
|
|
|
|
| |
Also move BP1 to provider package
Change-Id: I082df8f675d32e2b9072914eab4ee1d50543e169
|
|
|
|
|
|
|
|
|
| |
This is useful for testing, where crash recovery is handled by the
tests and also dialog interaction in undesirabls and also dialog
interaction in undesirable. Also included a minor unrelated change
to test app to handle/disable pop windows.
Change-Id: Iab337b31c803d50b74ba24ca80cc614fa752a846
|
|
|
|
|
|
|
|
| |
Bug: 3506038
Lookup the parent itself, not the children of the parent in
isValidParent check. Also, adds testing to watch for regressions
Change-Id: I59f00094f862f2e9cd4b4f9fd92d370b87e612e5
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Code and tests based on the google search app. A lot
less code would be duplicated if we could somehow
override AutoCompleteTextView but that is made impossible
by it calling a bunch of stuff in its constructor. To do
so would require changes to the existing API.
I've verified that the unit test passes, but other browser
tests appear to fail - even on a clean branch with none
of my changes.
Also fixes a minor bug in SearchEngines.getSearchableInfo( ).
Change-Id: Ic61bc6b8fa27cd210a45dc181ebf15accf503244
|
|
|
|
|
|
|
|
|
| |
currently, if a page load ended up in Java crash, the crash will be
intercepted by test harness and during the test tearDown step, the
test status file will be removed, and the harness would think all
sites have been loaded successfully
Change-Id: Ifc02d7f4b7e76c8e8aad06fff273f9d61060874d
|
|\ |
|
| |
| |
| |
| |
| |
| |
| | |
Bug: 3466212
Also added a test case to check for regressions
Change-Id: I9a8979a0ced165f1c94945c62687b0bf3b7076fe
|
|/
|
|
|
|
|
| |
Bug: 3466250
URL query parameters are imporant, don't strip them
Change-Id: I9c43598f56150470658485364aadfd0425bfa8cd
|
|
|
|
|
|
|
|
|
| |
looks like these tests has no external dependency and runs
reasonably fast. marking them as "SmallTest" will make them
automatically picked up by continuous tests on dev branch &
monitor failures.
Change-Id: I4addee3ed369f00cb52e828f146a6958b9cb3a09
|
|
|
|
|
|
| |
Bug: 3431373
Change-Id: I0ce30213150c6dd51b128723f044bd32eb2b46a3
|
|
|
|
|
|
|
|
| |
This is because of webkit change:
http://trac.webkit.org/changeset/72974
which will treat JS numbers as doubles.
Change-Id: I45a9966bcd435778457dbd5ca01953696c6c2d84
|
|
|
|
|
|
|
|
| |
this situation usually happens when WebView is spinning or
deadlocked on something, throwing a RuntimeException will
make this kind of error more noticeable
Change-Id: Id2f495494f67206830fd9d9e8a4bc2b175e82614
|
|
|
|
|
|
|
|
| |
some websites may trigger the browser to launch a non-http url.
typical case is that some website saw the client is Android, then
launch a market search urln to promote their native app.
Change-Id: I7785f66c8516a95597fd264b88f130008c309618
|
|
|
|
|
|
|
|
| |
* intercept JavaScript onUnload confirmation, no dialog shown
* using null download manager to avoid actual downloading during
random url test
Change-Id: I41758a87a0c5de5b42d34db8a403bb6d6ca053b2
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
There are two reasons to use hardcoded external path as website
list: 1) the InstrumentationTestRunner did not provide a
mechanism for passing extra parameters down to test class, so
it's hard to provide test class with a location of test list file
in a parameter 2) if we include the test list file as part of the
asset and loaded via context during test, then it become hard for
test harness to inject a list at test execution time because test
list file is essentially compiled in when the apk is generated.
So I'm reverting the test list loading part back to previous
approach.
Change-Id: I5532de8d153172771dba160d6ad89092f81ec6e4
|
|
|
|
|
|
|
|
| |
Bug: 3248213
Fixes errors in the tests that were caused by missing files, proguard,
and the refactor.
Change-Id: I503dfc7ee42d0173e3a5ad032f58a6f4f310588c
|
|
|
|
|
|
|
|
|
| |
The JNI bindings test was broken after BrowserFrame.documentAsText
started to take a parameter to determine if it should capture
the main or child frames. Update the call here to pass the correct
parameter.
Change-Id: I5142214164db371d9106d799bfa4e5bd7293dad3
|
|\
| |
| |
| |
| |
| |
| |
| |
| | |
status file
Merge commit 'f3c6c24d6ba7dd448c67fc064a7becc029c5981f'
* commit 'f3c6c24d6ba7dd448c67fc064a7becc029c5981f':
improvement test app's handling of irregular status file
|
| |
| |
| |
| |
| |
| |
| | |
if it reads an empty line, or have reached end of file, it
will default to the start from beginning scenario.
Change-Id: Ife2e49bf8adf70d6caba7a1d65094401990325a2
|
| |
| |
| |
| |
| |
| |
| |
| | |
Currently the test does not increment page count after relaunched
after a crash. This means the crashed page will be loaded again,
causing a dead loop.
Change-Id: If4591fbd05521f0454151affafec09788e9208d9
|
| |
| |
| |
| |
| |
| |
| |
| | |
Currently the test does not increment page count after relaunched
after a crash. This means the crashed page will be loaded again,
causing a dead loop.
Change-Id: I4eb0255e2bdf8eea0d3c16797cbb21e1711c3d5b
|
|\ \
| |/
| |
| |
| |
| |
| | |
Merge commit '1c5fe40bbea43d4a2d8de74ce169f9f705896f26'
* commit '1c5fe40bbea43d4a2d8de74ce169f9f705896f26':
various fixes for browser test harness
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
* the definitive way to determine page load finish seems to be
onPageLoadFinished AND onProgressChanged==100, changing test
harness logic to reflect that
* webview does not seem to be happen if ClearCache is called too
often (or for certain pages?). It may complain that failed to
delete blahblah, then the test harness goes crazy. So changing
cache clear to after an entire iteration of web pages have been
loaded
* browser will load default home page, which may be disruptive
to the page load process. So for the test, the browser is now
started with initial intent to visit about:blank, which does
not trigger any page load at all
Change-Id: I1cfbd8d0196a05caea6a1c5137f45f1f59cc4c5f
|
|\ \
| |/
| |
| |
| |
| |
| | |
Merge commit '6cb16461e1524eb3272b4e9440db6c041feacbb3'
* commit '6cb16461e1524eb3272b4e9440db6c041feacbb3':
add check for FileNotFoundException
|
| |
| |
| |
| |
| |
| |
| | |
make the test fail gracefully instead of throwing exception. because
otherwise it interrupts test run.
Change-Id: Ia2b92355db83cd13c55263f0699b29c10e00ea96
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
* the name "testComplete" will cause test runner to actually execute it as a
test case, which is not the intention here, changed to "notifyComplete"
* the test depends on "/sdcard/bindings_test.html". since the same file is
already being carried in the apk as asset, added a setup/teardown step to
extract the file. this removes manual setup step for the tests.
Change-Id: I74dc42a7ca0442a84910441e535d950c049efc98
|
| |
| |
| |
| |
| |
| |
| | |
Determine whether to show the camera based on the accept type associated with
the input element.
Change-Id: I48fea76fbd7a1f3df3c87288f28e01a32d9e955e
|
|/
|
|
|
|
|
|
|
|
| |
* the name "testComplete" will cause test runner to actually execute it as a
test case, which is not the intention here, changed to "notifyComplete"
* the test depends on "/sdcard/bindings_test.html". since the same file is
already being carried in the apk as asset, added a setup/teardown step to
extract the file. this removes manual setup step for the tests.
Change-Id: I45e1e9755ec829086fae6a36885153874f3a94a2
|
|\ |
|
| |
| |
| |
| |
| |
| | |
this is part of a larger effort to collect app launch performance for the platform.
Change-Id: Id7159c67b0d746c9a109988e05f53be6a10921af
|