Some of the code encountered accessed mw.config directly,
I cleaned that up by migrating to mmv.Config, which is an
abstraction added to avoid peeking at mw.config directly.
Bug: T132064
Change-Id: I2a95ba703e6c7f46296f8e435bceec769dceebf9
On not-MMV -> not-MMV hash changes the bootstrap class has set
up and immediately torn down the overlay. Besides slowing things
down, this broke TOC navigation on Chrome in an ugly way.
Add a special case to skip the overlay loading when the new hash
does not contain #/media. This is a quick-and-dirty fix; the whole
loading and hash handling could use a rewrite.
Bug: T119854
Change-Id: I5494903dfe778e533773ff142fdf756475c2df32
This makes sure that MediaViewer works correctly when content
is changed via frontend code (such as a VisualEditor edit or
an AJAX preview).
Bug: T97010
Bug: T110493
Change-Id: I9c57af281ed7f416521d80512f95a79150ce13a2
When cookies are disabled in Firefox, merely referencing window.localStorage is
enough to cause Firefox to throw a SecurityError exception. (See the referenced
bug for reproduction instructions.) This causes MMV to fail to start.
Bug: T109110
Change-Id: I28399cf8b63328ee34316b785f2d091f6f052224
Currently, if VisualEditor saved an edit, it replaces the current content
with the new content, which removes all already added event listeners, e.g.
on thumbs, added by MMV.
Add a new hook handler to the postEdit JavaScript hook to re-add all event
handlers to the thumb images to cover this case and enable MMV on a page
after a finished VE edit.
Bug: T97010
Change-Id: Ic1ee0b52265363978640358510cab89db4de2bb8
Bug with history.pushState in Edge fixed in Win10 build 10240;
Dave Storey is asking us to remove the workaround to keep our
code pure and them honest ;)
Reverts I6ea4d367
Bug: T104381
Change-Id: I5f53416dda7c31662330dc5ec67e1bbea55dace7
When attempting to clear the hash state via history.pushState(),
MS Edge in Windows 10 build 10159 errors out with the helpful
"Unspecified Error".
Catching the exception and trying again with a "#" works around
the error and doesn't look any worse than the non-pushState code
path -- and should have no side effects when the bug is fixed.
Bug: T104381
Change-Id: I6ea4d367af64f85018b06b859ce688053a1caf0f
Added mmv.HtmlUtils.htmlToTextWithTags()
which is similar to htmlToTextWithLinks()
but allows <b> and <i>
Added test for mmv.HtmlUtils.htmlToTextWithTags()
Most text fields now use htmlToTextWithTags()
except Permission field which is not supposed to
have HTML
Bug: T69887
Change-Id: I16833f218e2f64586aa13925356fa2b8b7ec3100
Pass alt parameter from mmv.bootstrap.js to mmv.js and
set it as a parameter on the displayed lightbox image.
Include the alt text in the embed text.
Bug: T66519
Bug: T75923
Change-Id: I29503eb582ac2bc8cf89f737a3bcb787b660d918
MediaViewer now handles Template:Multiple_image. Instead of looking
for caption in whole thumbnail container, it tries to find the
closest one to the image.
Bug: T85354
Change-Id: I18d982a4bf245c4925213d83a3410274d499845e
Most special pages which list images generate a caption using
various information and MediaViewer displays this caption.It should
fall back to the file description instead.
Bug: T85234
Change-Id: I5448e5de7d6f8de9852a2a845400299b6b51b8ef
This is complete, but it would be better if the HEAD request
was actually aborted by Varnish when the viewDuration parameter is
present, or if the hit pointed to a script that does that.
Change-Id: I66cafd97427756411e967de1901324af2215e3ae
Mingle: https://wikimedia.mingle.thoughtworks.com/projects/multimedia/cards/1001
Images inside the current WLA banner are getting picked up, spamming
all Media Viewer users with large-scale couscous imagery.
Couscous is nice, but too much couscous will make you bloated.
Change-Id: Ie5726be6de97da13e8dc650031285f899c2d6440
Mingle: https://wikimedia.mingle.thoughtworks.com/projects/multimedia/cards/1011
Includes some small refactoring of the bootstrap code to open
MediaViewer so it can be used in a more controller-like fashion, and
removing the .mw-mmv-filepage-menu class which was added in the parent
commit but had no effect since the module was not loaded by default.
Change-Id: I2e7405e694af96e8eca4fcc839b60306232ead01
Mingle: https://wikimedia.mingle.thoughtworks.com/projects/multimedia/cards/719
Due to a jQuery bug, errors in local code (gadgets, user scripts)
can cause onready handlers to not be executed. For MMV this causes
catastrophic failure, with a black screen of death on exit.
This change makes sure that the setup code necessary for Media
Viewer to work is executed at latest when MV is invoked, even if some
onready handlers were skipped.
Opening MediaViewer via a hash-URL will still not work if the onready
handler fails, but that's hard to avoid and it is not a catastrophic
failure anymore. This change can be reverted when bug 70772 gets fixed.
Bug: 70756
Change-Id: Ida3b780791bc9dfec29303567d33e3aa4f44dd81
Mingle: https://wikimedia.mingle.thoughtworks.com/projects/multimedia/cards/891
This works because the title doesn't exist if there's no caption and we
won't get to this logic branch if the thumbnail is an explicit |thumb|
with a caption already.
Refactored caption-fetching a bit.
Change-Id: If84c890e7b71880db640a0993f8e3d6cd59951b8
Mingle: https://wikimedia.mingle.thoughtworks.com/projects/multimedia/cards/513
Categories are implemented via galleries, so they have the same
markup, but while galleries have human-written captions, category
captions just contain the file page.
Change-Id: I6a8548fe696418befc789e20b114778fc724c314
Changes document title (which is shown in history navigation)
to include the image name.
Bug: 67008
Change-Id: Id1b030f2b984571fb0877e35db2ca2ccc86f0130
This includes images where the <img> element has that class
(achievable with [[File:Foo.png|class=noviewer]]) and also those
where some parent element has it.
Change-Id: I666be026828ea9ecb6e8c93d3f5ad1e3c190f81e
Mingle: https://wikimedia.mingle.thoughtworks.com/projects/multimedia/cards/511
- Fix JS error on pushState
- Fix blur issue where blur(0px) filter would blur anyway
- Fix wrapper sizing issue where its size would be 0 when measured
Bug: 65225
Change-Id: If9279cd56f55f71f261ec54dda8228194988b9ae
Mingle: https://wikimedia.mingle.thoughtworks.com/projects/multimedia/cards/597
The code to replay clicks and clean up the handler was called after
processing each thumbnail, instead of just once at the end.
This might have caused many subtle issues such as clicks on any
but the first image not replaying correctly; more problematically,
of there were no MediaViewer-compatible thumbs, the handler was
never cleaned up and the clicks were never replayed.
Besides fixing that, this commit also adds a try..finally wrapper
so that the click replaying is not broken even if there is an error
in the thumb processing code. This might or might not be a good
idea. The internets say that try..finally without a catch causes
errors in IE8, but only if it is not wrapped in another try..catch,
so we are probably fine. (Adding a catch which just rethrows the
error would be an alternative, but it messes up stack traces.)
Change-Id: I2f645762103274c92c15a0d4b595d18d93b08415
Mingle: https://wikimedia.mingle.thoughtworks.com/projects/multimedia/cards/497
Bug: 64345
processThumb() is a mess. This change is kept small since we might
want to backport it per the bugzilla comments; I might follow up
with some refactoring in a separate commit.
Bug: 62518
Change-Id: I1f916b88fe1b667c6c7e51c9bdca186d4aa2ef2c
Mingle: https://wikimedia.mingle.thoughtworks.com/projects/multimedia/cards/338
* deduplicates URL generating/parsing code
* gets rid of spaces in URLs
* fixes error for file names with / in them (in case they exist;
current MediaWiki seems to disallow such names anyway)
Change-Id: I5aad43f6af1b99523c597c39befcc9db1ecab83a
Mingle: https://wikimedia.mingle.thoughtworks.com/projects/multimedia/cards/371
There used to be a CSS trick with the order we added things to the
page and removed them from it, but it doesn't seem possible anymore
with the new order of execution, with the overlay appearing
immediately and being taken care of inside bootstrap.
The main cause of the bug, however, was the hash reset happening
after the interface was closed.
Doing the scroll restore with jQuery.scrollTo is more future-proof
and testable in QUnit.
Additions were also made to the cucumber E2E test because QUnit
alone wouldn't have caught the hash issue.
This also cleans up custom events a little and reintroduces
pushState on browsers that support the history API.
Change-Id: I63187383b632a2e8793f05380c18db2713856865
Mingle: https://wikimedia.mingle.thoughtworks.com/projects/multimedia/cards/439
Bug: 63892
