This makes hasOption stable and usable even after getOption.
Also getOption can now be called twice with different defaults.
Strictly speaking this is a breaking change, but the actual
behaviour is now closer to the expected/documented behaviour.
Bug: T275619
Change-Id: I65e32a7e1bc253f4b29378be6980c42e43f93032
Sometimes the special page can’t be created – for example, the list of
registered pages could have changed between the data provider and the
test function, due to extensions conditionally registering their special
pages. Skip the test in this case instead of crashing.
Bug: T275661
Change-Id: I573967ac22ed3a596b4f80b76d395ac948f802e8
Replaces calls directly to PermissionManager with calls to
the Authority object available from Context or the
GroupPermissionLookup service.
This patch does not address use of PermissionManager for
blocks.
Deprecations:
- ApiBase::checkUserRightsAny deprecated passing optional
User parameter
- ApiBase::checkTitleUserPermissions deprecated passing
LinkTarget as first parameter, takes PageIdentity instead
Bug: T271462
Bug: T271854
Change-Id: I5d7cac1c28a37e074750c46cda03283980a07fca
Cut the explicit dependency on PermissionManager
and rely on the fact that User implements Authority.
There's still a few ends we need to cut before converting
the method parameters to Authority, but that can be done
in followup steps.
Bug: T273510
Change-Id: I69617e624f47944f8a383d0b28885678c1249f23
This introduces a specialized SelectQueryBuilder for
queries related to actor table. So far it's fairly basic,
but in future it is intended to grow.
This patch only introduces the new abstraction, using
it will be separated into separate patches.
Bug: T274223
Change-Id: Ibb1c58016e3956a46bfaf1a9f51f265114bb9ace
ActorNormalization needs to have separate wiring if we want to be able
to access it by name. This is needed to allow access to the service in
declarative wiring, e.g. via ObjectFactory specs.
As an aside, this changes SpecialPageFatalTest to not construct special
pages in a data provider. Accessing services in data providers should be
avoided. This change does not eliminate all such access, but reduces it
by deferring service lookups via ObjectFactory.
Change-Id: I5792cdcc0e5b0d1681f55dbe3e1f70c34a0cb775
When this code parses a string that looks like this:
*
** A
** B
It creates an option group with 2 options, but the name
of the group is an empty string. This makes the OOUI
DropdownInputWidget fail later.
Since the empty group name is useless anyway, drop it. This
makes the code behave as if the parsed string looked like
this:
* A
* B
Live example for the issue:
https://el.wikipedia.org/wiki/MediaWiki:Protect-dropdown
Bug: T275125
Change-Id: I780c1be27740b0ed3b35aa569b5a528112d7238f
There is nothing the updates can do,
LinksUpdate also fails if there is no id for the given title on master
Bug: T271356
Change-Id: I7fb690bec8dbfadf71eb9caacb4edbfe6b785e73
Mostly copied from UzConverter.
This is a very simple converter, with bidirectional one to one
correspondence: for every Latin letter there is a corresponding
Cyrillic letter and vice versa. There are no digraphs or punctuation
to convert.
The Latin alphabet is the primary one used for this language today,
and will probably remain so for the foreseeable future, so "tly" remains
the usual code, and "tly-cyrl" is added for Cyrillic.
Language name is changed:
* The main language name is now Latin.
* The word "language" ("зывон") is removed.
* The spelling of the word "Talysh" is based on the Pireyko dictionary.
Bug: T258975
Change-Id: I552e07967ea82e03c413a0b10b129a846aa007c7
For the optional message in the "Account data" section, introduced in
d4357d621f.
Passes the user name escaped with underscores, so that links can be
built correctly in the message translation.
This means we have to expand yet further the list of User methods called
in ApiOptionsTest, but so be it.
Bug: T272412
Change-Id: Ia1c135144763db9e92f1b555f009229bd1a9d3c8
This also means we don't need to take special care for French spacing in
attributes, since it's no longer applied there.
Adds a test that captures this change.
Note that the test "Nowiki and french spacing" wonders whether this
escaping should be applied to nowiki content.
Bug: T255007
Change-Id: Ic8965e81882d7cf024bdced437f684064a30ac86
I keep fixing type hints in my local dev environment whenever I
see something that is worth updating. This is what I collected
over the past weeks.
Change-Id: Ia4f1b4bee2019abe4ab0fb0df8164e1b446229e7
The timestamp in the year 3030 was presumably chosen so as to
always be in the future (since expiry in the past would be ignored).
However, while seemingly clever, this backfires because it gets
shrunk by the default wgWatchlistExpiryMaxDuration value to some
time in August 2021, around the "current" time of day.
And "current" generally is thought of as moving forward at a rate
of 1s/s, bending of spacetime notwithstanding. Thus the test would
sometimes fail in ways such as:
> 1) ApiQueryInfoTest::testExecute
> Failed asserting that two strings are identical.
> --- Expected
> +++ Actual
> @@ @@
> -'2021-08-13T01:23:19Z'
> +'2021-08-13T01:23:18Z'
Fix this by using a mock current time, paired with a suitable
max duration and expiry value.
Also fix some of the other asserted values to not be reflections
of the source code but actually explicit assertions (at least the
details that we know to be constant, given the environment ensured
by ApiTestCase and MediaWikiLangTestCase). This reduces the chances
of false positives from looking at itself too much, and also speeds
up the test a little.
Change-Id: If361aa98bef789b1f841741a7c83bcb2ac9edf05
