Turns out this was only in place because the test was
(intentionally) incomplete. But it's dead code in
production. Let's get rid of it.
Change-Id: Ieeb145b6972dceb7eeda3a167d907b680d5c3ce4
As well as make some more test code more readable without
changing what the test does.
Bug: T281698
Change-Id: Ia153981d9196b47099ef3880ac334718895fb0fc
I was not able to reproduce the issue locally, but I believe
this might be the reason for the failure we see. It can't hurt
and needs to change anyway, with or without the failure.
Bug: T281324
Change-Id: I0002f952438d7afe2fb6c2100bf25760f7dda10d
Use mw.track instead of dedicated tracker for VirtualPageViews.
This way we can migrate it to the new Event Platform client.
The new client does not observe DNT, which was the reason this
instrument was moved to a dedicated tracker on the first place.
Bug: T279382
Change-Id: I8bb515eab337ffed686ba7522bc6153cfdd8ca8d
JUnit reporter configuration changed since WebdriverIO v6. Before this
patch, the reporter was outputing to terminal, instead of outputing to
file.
Bug: T214686
Change-Id: Ib315b0261b3724cded17c85519ac28842d7ab071
This is the smallest possible patch for a backport.
While it seems there is zero technical reason to limit user
option keys to lowercase with dashes, it's best-practice.
Same as message keys.
Bug: T281235
Change-Id: Ia4a45cf4459543c81b23b757ae9c2cfaf9676894
* Change more places to not hard-code the popup types, but use
loops and such.
* Change many `function ()` headers to use the more streamlined
ES6 sytnax.
Bug: T277639
Bug: T277640
Change-Id: Ifece87d51012e0e069286453b27f5c9ae273710e
We added reference preview as a checkbox the the
anonymous user settings. To handle both popup types
(pages and references), we changed the usage of
preview.enabled. We pass on all types as a map
inside preview.enabled. The footer link to edit the
settings will appear for anonymous users if at least
one type is disabled.
Bug: T277639
Change-Id: I860a1b35ac7749d8d0884575f6acb7186ad8e4d0
We still have 2 different mechanisms in place, maybe even 3:
* We simplify the CSS selector when we know a popup type is
disabled, and it's impossible an anonymous user can enable it
at run-time.
* We create that "initiallyEnabled" map that allows anonymous
users to toggle the individual popup types at run-time.
* This map is also used to check if the footer link should be
shown.
* There is also a wgPopupsReferencePreviews global that acts as
a "kill switch". However, this is not a pure feature flag,
but incorporates the user setting for registered users. This
is currently partly redundant (checking
`mw.user.options.get( 'popupsreferencepreviews' )` does the
same) and can be removed later when the feature flag is not
needed any more.
The footer link currently acts odd because anonymous users are
unable to enable ReferencePreviews, but get the footer link.
This patch introduces a 3-state model:
* `true` acts as before.
* `false` means a popup type is disabled, but anonymous users
can enable it (i.e. this is the opt-out behavior for anonymous
users).
* `null` means a popup type is not available at run-time, for
nobody. Anonymous users can't do anything about this.
Registered users must leave the page and change a setting.
Bug: T277640
Change-Id: Id8d1396c09cf0f706034a66f9cd3c880a8b33df8
* That `this.user` is unused.
* Some tests missed a module name. This means they are reported
as part of the previous module. While this is purely cosmetic,
it's confusing to see the wrong module name in the report.
Change-Id: I73915d3c4fd9a03bda1ddc8dff6dd5539113c3cd
… as well as in one place in production code. The motivation
for this change is to make the code easier and faster to read.
There is a little bit of duplication in the test setup now.
But I would like to argue this is a good thing. The values are
rather trivial. The difference (or absense of a difference) is
much easier to see now.
Change-Id: I9aa95b59f0c45ea7c9257970e2fcdba3a000d234
This patch does nothing but rename a pair of variables:
"prevState/state" becomes "oldState/newState". Reasoning:
1. The abbreviated "prev" is confusing, especially because we
are in a codebase that is all about "previews".
2. We are in a context that is all about a state **change**.
Change listeners get notified about the change from one state
to another. While it would be possible to stick to the already
mentioned "previous/current" terminology, I find the word
"current" confusing. What is "current" in this context? Did
the state already change? Am I notified about a change that is
**going** to happen or already happened? Is this even relevant?
I don't think it is. Therefor "old/new".
Another possibility is "previous/next".
Change-Id: Id886e1a095967fe86fb9021f59e335c62da8994e
The nextState() function was not able to understand updates that
are deeper than a single level. Example:
nextState( state, { pagePreviews: { enabled: true } } )
Before, this would replace whatever was in the "pagePreviews"
property with { enabled: true }, but not merge it. In some places
a single level of recursion was done manually because of this.
This can be removed now.
This is done in preparation for splitting the "enabled" flag into
separate ones for each popup type.
Bug: T277639
Change-Id: I35911c18018ba7cd1633a4c882b978656c3fee36
Note how getIsEnabled() is documented: "if the user hasn't
previously enabled or disabled Page Previews […] then they
are treated as if they have enabled them."
In other words: The idea that the default should be true is
encoded twice in this code. This is just not necessary. We can
remove one without loosing anything.
Motivtion: Simplifying the code and reducing the package size.
Since the code fundamentally depends on this default value
anyway, we can clear the users localStorage when they decide
to go back to the default – instead of storing a "1" which
does the same as the default.
Change-Id: I2814a1e9269979918609162a508eeee6944d9e52
The main motivation here is to dramatically reduce the number
of places that use the same property name "enabled" for values
on different objects (e.g. "state", "actions", and "updates"
are all different things) with slightly different meanings. I
tried hard to come up with names that reflect better what each
meaning is.
Bug: T277639
Change-Id: Ie766259793f716262e3d4622ca55156d11f4842c
Add message, description, extension for title. Update createPagePreview, renderPagePreview methods to add title attribute to settings gear icon. Add test for title attribute. Increase maxSize, maxAssetSize, maxEntrypointSize. Add compiled js files.
Bug: T274887
Change-Id: Ibb29deb3418569d8283b954b4b22074423e78bda
To reduce size of code added to the <head> and increase performance.
The increased bundlesize is still less than the size spared bytes in
ResourceLoaderGetConfigVars. - But nevertheless the main gain is loading
less in the <head> anyways.
To avoid further complexity in the code, the bitmask is converted to
the according config setting early on instead of adding checks on the
bitmask all over the place.
Tests will be added in follow ups.
Bug: T276716
Change-Id: Ib4f82bed58295b25f0a41cb37e36244e45f16317
Update copy and remove unnecessary reference preview preference
in favor of using the default preference. It seems there is no
stable method to link to the subsections on the preference page
for gadgets. So in all cases does the link just point to the
gadgets pref page.
These changes should only be visible when reference previews
are no longer marked as a beta feature.
Bug: T265709
Change-Id: I7b8ab91331092ada04b230315373548673b9272c
* When there are multiple <cite> elements, the first that contains a
known class is used. We assume the earlier one must be the relevant
one.
* When there are multiple known classes in e.g. <cite class="web news">,
the last one is used. This follows how CSS works.
* There is extra validation if the corresponding message exists, just
to be sure. This is cheap.
Bug: T274979
Change-Id: Iee3481ea16af96b40faf978e254718e5a48917f3
I added the common styling for the setting icon to the popup.less
and removed the now empty pagePreview.less.
Bug: T234205
Change-Id: I2a82831bc71a4208c4b66c18e2a4613127c43e1f
It turns out the TextExtracts extension is build in a way so
that the parameter plaintext=1 alone is not enough. It does
not really mean "please return plain text" but "please don't
return HTML". It can still return one of three formats:
* Wikitext.
* An intermediate parser format where headlines are not
`== this ==` but `��2��this`.
* Actual flat plain text.
The default is wikitext. That's why we see `== wikitext
syntax ==` in certain edge-case situations. Forcing it to
"plain" fixes this.
Bug: T271439
Change-Id: I3035fb3df99680af8bbd10c4513aed730013c344
This reduces a lot of churn in creating the SVG
element and related helpers.
When IE11 is dropped, the SVG code-path can also be dropped.
Bug: T269336
Change-Id: I7f91192dedc2a78f1c7c84179cff1687593177c0
This reverts commit 6bc2077ed5.
The change causes issues with various popular gadgets on Wikimedia
wikis. The 'title' attributes have been the easiest way to determine
the target pages of links, and many gadgets have come to rely on that.
Bug: T269297
Bug: T269873
Change-Id: I49d315a13c327a1c5af51d3de887c0c17642e9fe
Reduce layout/style thrashing by measuring all required geometries
at event handler, not waiting for delays/redux/style changes.
Use CSS bottom instead of top, to avoid having to measure the popup
before positioning it, if it's placed above the link ("flippedY").
Disable some test cases that relied on implementation detail of using
"top" CSS.
Change-Id: Id0cbf506009b824d0fb6af4d6fe220e2f69aaaad
This is a performance optimization - removing
all the titles when initializing the popup extension
reduces DOM manipulation during hover, removing/reinstating the title
attribute.
When the popup extension is loaded, the default "title" behvior is unnecessary.
Change-Id: I1a85394b6b67eabee50a8d554bfd9b62de2a24d3
The tests were skipped for some time already because the beta
feature mode is taken into account on the test systems. The tests
need to login and enable the beta feature mode to be executed. This
is fixed with that patch.
The patch also fixes broken tests due to changes that made one test
obsolete and another that needed adjustments.
There are also comments added to places where code can be removed or
altered if the feature gets out of beta status.
Bug: T268134
Change-Id: Ib96d23f3cb6c6130fd5880a78fafd252bf706475
Now that the footnote label links to the references section, we don't
need this additional link.
Bug: T265482
Change-Id: Ib9b2939eb49e7b826c7699a5b7fa0e8255fa9da1
I played around with a lot of options, and settled with:
* When there is nothing but text, but the text is all
whitespace, don't show it.
* Make sure an <img> with no text is still shown. This is done
by checking for child elements.
A possible future enhancement could be to check the visible
width of the element as well. Unfortunately this fails in
tests. Everything is 0 in tests.
Bug: T240543
Depends-On: I2929a86b6a09f3b72e5e2f4151cb13f52446897d
Change-Id: I94ed575abcd69241c82480ade07017e61cc26c9c
We no longer intercept reference clicks, now clicking on a reference
label will scroll to the reference definition in the same way as when
reference previews is disabled.
Metrics about clicking on the label are collected by the Cite
extension, are are unaffected by this change.
Bug: T265482
Change-Id: I2929a86b6a09f3b72e5e2f4151cb13f52446897d
Elements that are marked as collapsible (often tables, but can
actually be anything) are most certainly marked as such because
they are big and don't fit in a popup.
Another plugin makes tables sortable.
In both cases non-functional UI elements appear in the popup.
We decided:
* Hide collapsible elements (no matter if currently collapsed
or not), and show a placeholder text instead.
* Remove sortable arrows.
This is a baseline patch that solves everything, except the
(i) icon is missing. This will be added in the next patch.
Bug: T220208
Change-Id: I58f3036bf4988d0ebe5716b0a54506446fca10c3
There are 3 types of links:
* Links with no class are links to other pages in the same wiki.
* Links with "extiw" are interlanguage as well as interwiki
links to other known wikis (e.g. from Wikipedia to Wikidata).
* Links with "external" point to somewhere else on the web.
Bug: T215419
Bug: T239230
Change-Id: Ia25db94d02670a919fc003f3e3562725de2e8eae
The rule no-shadow gets enabled in the version eslint-config-wikimedia
0.17.0.
In prepration for the automatic update with libup this needs to be done
in a separate patch set due to T262450
Change-Id: I66d405aef6d2b777e9a7f63734f2b5a5649d2080
Update NPM packages: webdriverio, wdio-mediawiki.
Replace NPM packages:
- wdio-mocha-framework with @wdio/mocha-framework.
- wdio-spec-reporter with @wdio/spec-reporter.
- wdio-junit-reporter with @wdio/junit-reporter.
New NPM packages: @wdio/cli, @wdio/local-runner, @wdio/sync.
Commands that return data no longer wrapped in `{ value: … }`.
Values are now returned directly.
Replace:
- `browser.element` with `$`.
- `chromeOptions` with `'goog:chromeOptions'`.
- `password` with `mwPwd`.
- `username` with `mwUser`.
- `moveToObject()` with `moveTo()`.
- `isVisible()` with `isDisplayed()`.
Bug: T258620
Change-Id: If339ff08b5e87c91905a65927ccd7e1b72b9f8e4
In this patch we introduce a new config variable and update its default
to be the same as the production value of $wgPopupsPageBlacklist.
When this code is deployed everywhere, we will remove wgPopupsPageBlacklist
from mediawiki config and defer to the Popups extension as the source of
truth from now on.
Bug: T254676
Change-Id: Ifebdcf8f5eec854a2b947dc390eaf47704a5c5eb
AVoid the term if possible, in all internal code. The only remaining
use of the word is in the public $wgPopupsPageBlacklist config
variable.
Change-Id: Ib238731270f473ad44fcff13df617a70433f2452
Popups is enabled on the Wikipedias but disabled by default, which makes
it more difficult to enable on third-party wikis.
Things should be enabled by default, and them WMF configs should disable
features when they are not used/available.
Supporting changes:
- Rename PopupsContextTest#testAnonUserHasDisabledPagePreviews and
remove the associated data provider, which provided only one datum
- Set $wgPopupsReferencePreviews to false in
PopupsContextTest#testShouldSendModuleToUser in order to isolate the
code under test
Change-Id: Ib8cc7041d792bed0a19d18e14506627099a69bef
This is a direct follow up for I4d31805. Note the previous patch was
only testing the code path when Beta is disabled. But the bug was about
what happens when Beta is enabled. This is now properly covered by
these tests.
Bug: T240947
Change-Id: I54c5e9751e14a94808e85d935e1972eea0395002
It appears like we accidentially copy-pasted the behavior for
PagePreviews and made ReferencePreviews behave the same, not taking
into account that the later feature is in Beta mode.
Bug: T240947
Change-Id: I4d31805ee9b2045c49c7ab4179c5a4adbcba0394
Notice how this actually reduces the size of the final, compiled index.js.
It's not much, but still.
One issue I noticed is that the coverage reports for the JS code stopped
working. I have no idea why.
Bug: T208951
Change-Id: I2fe92579574b3b1ba4d2dd064899eee944045a96
The signature mw.RegExp.escape() is deprecated since MediaWiki 1.34.
By the way, this is the 4th or even 5th time in my short career this
tiny, single line (!) helper function is moved around and I need to
update all my code. This felt already ridiculous when it happened the
2nd time.
Change-Id: I4d05a49120aff64ebc316d0af7736c62385d9307
This was motivated by two things:
1. Notice the bad class name in UserPreferencesChangeHandlerTest. This
was possible because the default mock builder does not check if a class
exists, *and* there was no type hint in place. createMock() *does* check
if a class exists.
2. We are free to use createMock() from PHPUnit 5+ now. It's not only
shorter but more strict and reliable.
Change-Id: I301a171d587026eab0a62575ab2fdbfd7733c661
Logs events to the ReferencePreviewsPopups EventLogging schema, in
order to understand whether Reference Previews is helpful for
end-users.
This will be removed along with the older tracking, as soon as our analysis
phase is finished.
Incidentally disables a lint rule for the generated JS, it's about
readability so irrelevant to the minified code.
Bug: T214493
Change-Id: I2638611ba67b785338f7e98a1c4b08a5e829812d
A method that is expected to return, for example, a boolean true
should not return an other value that PHPs loose == comparison might
also consider true.
Same for all other loose equality checks.
Change-Id: If729c6e97d5337eee10b717da76dad428218ff69
The eventlogging `isEnabled` function determines when to sample,
this patch removes the `isAnon` conditional so that we can measure
reference reading habits for all users.
Also guards against a non-function navigator.sendBeacon, which was
previously intended but incorrectly tested.
Bug: T214493
Change-Id: I42cb3082fb85c7900426a2055dfa3c2f6ecfd968
I was able to track the issue down to the changes made in this patch:
https://gerrit.wikimedia.org/r/331563
This patch is mostly a simplification of ce8a2d4
(I9a73b3086fc8fb0edd897a347b5497d5362e20ef):
- Don't make wait#wait() abortable. This adds complexity and isn't
needed since the token is rechecked in the .then() of
actions#linkDwell(). The request is permitted to continue and fetch if
that token still matches and is never issued otherwise.
Once a request has been issued, that request is still abortable.
However, note that calling XHR.abort() is just a request to abort and
may not be granted. Whether or not XHR.catch() is invoked is what
dispatches the FETCH_ABORTED action (or doesn't if the request to
abort was denied).
- Remove the abort tests for wait#wait() since the functionality is no
longer provided.
- Pass the preview token in the FETCH_ABORTED action and reduce
FETCH_ABORTED the same way as ABANDON_COMPLETE in reducers/preview.
The follow-up patch, I3597b8025f7a12db0cf5d83cce5a77abace9bae3, adds
integration tests for the specific bug fix. Note that these Selenium
tests are incompatible with the content proxy, so it is probably best to
simply unset $wgPopupsRestGatewayEndpoint and $wgPopupsGateway and allow
the defaults to be used. Also note that the tests are incompatible with
recent versions of Node.js (so use NVM) and emit many deprecation
warnings (so set deprecationWarnings to false in
tests/selenium/wdio.conf.js) An example run of the tests looks something
like:
chromedriver --url-base=wd/hub --port=4444 &
# If any changes are made locally, also run `npm -s start &`.
MW_SERVER=http://localhost:8181 \
MEDIAWIKI_USER=foo \
MEDIAWIKI_PASSWORD=bar \
DISPLAY= \
npm -s run selenium-test
Live testing may be performed as well. Remember that RESTBase requests
are incompatible with MobileFrontend's content proxy hack so ensure to
comment it out if $wgPopupsGateway is configured for RESTBase (see
T218159).
1. Open the DevTools network tab.
2. Disable the browser cache. Chromium, at least, won't abort requests
coming form the cache.
3. Hover back and forth quickly over a preview. In Chromium, canceled
requests are labeled and appear red. This is a good scenario to test.
With the patch, a preview should always be shown when ultimately
resting on a link. Without the patch, it is possible to rest upon the
link with no preview showing. This may require several attempts.
Bug: T219434
Change-Id: I9da84b0296dd14e9ce69cb35f1ca475272fb249a
As mentioned in T205744, EventLogging schema ResourceLoader modules have
been deprecated. This removes those modules from loader calls.
Bug: T221281
Change-Id: I1b7355c69e09756f50ccd1c1955b45cae4a64b9e
This fixes page previews appearing when there is self link to a page
that has a file extension in it's name.
Bug: T222869
Change-Id: I5caf610fa76986026948a5b7b55537723752b755
Introducing the REFERENCE_CLICK action that will fetch and show the
preview for the clicked reference right away without any delay.
The main goal of the new chain of events introduced with the reference
click is showing the reference preview right away. The actions triggered
by the dwelling include delays in multiple parts of the process.
If there's a dwell action-chain in progress when the click action is
executed, the related promise ( that might still include steps with
delays ) and the reference preview is retrieved and shown right away
re-using the token.
In the case where there's no dwell action running ( e.g. when the click
was triggered via touch ) we create a new token and start from scratch.
In either case we want to avoid, that multiple clicks trigger multiple
actions and abort early when there's already a click action in progress.
Bug: T218765
Change-Id: I073a93be2d17a21178aebe12267765f60a2811b9
Pretty much all usages of this function do *not* use the second
parameter to pass a page name prefixed with a namespace, but pass the
page name only.
This patch here removes the redundancy. The namespace prefix is now a
generated one, saying "Namespace <number>:…". It turns out no test
relies on this so far.
Bug: T220097
Change-Id: Ibd45d49c91e86a2647afe676a5e3bb07dfeab6ed
config.get( 'wgTitle' ) returns the unnormalized title of the current
page while title.getName() gets the normalized title (e.g. with underscores ).
On pages with spaces in the check failed before this patch.
Bug: T220097
Change-Id: I58a532627bb27be030cbc553f1181a89109edd80
Named references may include non-ascii characters, so we decode the fragment before matching against reference IDs.
Bug: T220196
Change-Id: I63bba59fa8f0f6aa95aeadbb1f85745d480988bd
It is set based on the same conditional that loads the code,
thus checking it inside the loaded code is a no-op and adds
extra HTML to the <head> that blocks text/layout rendering and
delays fetching of Popups JS.
Bug: T219342
Change-Id: I9c1f4b3861ce2cecb654eb0a78469a616730a40b
I just realized on another test set, that this is actually not implemented
in a way where it is working correctly. The return value of the browser.exectue()
is returned as part of an object and not directly. So the condition was always
true and the wait until did not really wait for anything.
As a result I'm quite confident the pause is not necessary.
Change-Id: I274bdee0b3c39c418a2b61881d56f89889c53485
The test would lead to an exception when 'mediaWiki' is not defined. The
exception would then also abort the whole execution so in this context it's
safer to use typeof with 'undefined'.
When the mediaWiki js base is loaded though, loader.getState() is guaranteed
to be available since it is part of the root module.
This also uses "mw" instead of "mediaWiki" for consistency in test.
Change-Id: I1262d0b5c4a1136f4d2294f125336e72118c6e2c
- test page loading and waiting for the scripts moved to beforeEach
- removed unnecessary abondonLink call
- removed unnecessary browser.pause()
Change-Id: I28eb7b9b48f105315bf41f7a41e5a1e6ec21cb2b
