Commit graph

3124 commits

Author SHA1 Message Date
Catrope 9611bf9f64 Send a full document back to Parsoid
To do this, we take the document generated by the converter and
transplant things from the original Parsoid document into it.

Change-Id: I2f5058220669526130a360cec3389c3f42b41771
2013-05-13 11:23:34 -07:00
Ed Sanders 634947198a Use consistent capitalisation of Html
If the previous commit properInnerHTML was renamed to
properInnerHtml, but its invocations weren't (a bug).
While DOM uses .innerHTML we use Html throughout the
rest of our code so we should be consistent either way.

Change-Id: If46bb256e938a097951c159b7a278667fd8e06a6
2013-05-11 13:04:00 +01:00
Catrope 606baacc1b Add ve.properOuterHtml
Like ve.properInnerHtml but for outerHTML. Factored out the shared code
into ve.fixupPreBug.

Change-Id: I5c5b6d017936c6c2e71631fb8859ba1dbd95615a
2013-05-10 17:03:04 -07:00
jenkins-bot db644e915f Merge "Fieldsets and default sort keys" 2013-05-10 23:58:17 +00:00
jenkins-bot 89aebd5112 Merge "Edit and Edit source tab were reversed in RTL in the Vector integration" 2013-05-10 23:49:30 +00:00
Trevor Parscal 8409d16f0f Fieldsets and default sort keys
Objective:
* Add default sort key field to meta dialog
* Replace PagePanelLayout with a generic panel containing one or more FieldsetLayout elements

Changes:

*.php
* Added/removed file links

ve.dm.MWDefaultSortMetaItem.js
* Added getContent method

ve.dm.MetaItem.js
* Added replaceWith method

ve.dm.MetaList.js
* Allow insertion at the end by omitting offset and index

ve.dm.MWMetaDialog.js
* Added default sort key field
* Put category and default sort fields inside fieldsets
* Added loading/saving default sort key

ve.ui.PagedLayout.js
* Changed class used for pages to generic panel layout

ve.ui.PagePanelLayout
* Moved title/icon stuff to field set

ve.ui.FieldsetLayout.js
* New class, adds fieldset with legend

ve.ui.StackPanelLayout.js
* Moved up to the layouts directory

ve.ui.Dialog.css
* Moved style for paged panel from layout stylesheet

ve.ui.Layout.css
* Added styles for fieldsets

ve.ui.Widget.css
* Adjusted margins of input label widgets

ve.ui.MWCategoryWidget.js, ve.ui.MWCategoryPopupWidget.js
* Added setDefaultSortKey method

Change-Id: I979f5e3f08a688790c9f54086206ed1999af13ea
2013-05-10 16:46:08 -07:00
Christian Williams 741fa76868 Switching back to Rangy 1.2
Rangy files are now 1.2 and I've added 1.3 alpha versions to make
it easier to switch between while testing.

Change-Id: I2f5d86b14bd4f87719d01e3255c28f0ae8764f4f
2013-05-10 16:44:32 -07:00
jenkins-bot 5f8db49a9f Merge "(bug 48022) Ignore certain function keys in key press handler" 2013-05-10 21:33:52 +00:00
Inez Korczyński b32b5702d4 (bug 48022) Ignore certain function keys in key press handler
This is a temporary solution that still may some function keys go through.

Change-Id: I7a8d6c474ecca5af02b72b6453c900ed61acea58
2013-05-10 13:51:23 -07:00
jenkins-bot 57b05aa433 Merge "Clean up variable initialization in setupSkinTabs()" 2013-05-10 20:31:03 +00:00
Catrope 2efd4f0061 Edit and Edit source tab were reversed in RTL in the Vector integration
If we're in RTL mode and the skin is Vector-based, we need to reverse
the order of the tabs in the DOM, because that's a weird thing that
Vector does to render tabs in RTL.

See https://bugzilla.wikimedia.org/show_bug.cgi?id=46947 for discussion
about the Vector behavior.

Bug: 48017
Change-Id: Ie1214b08450aefed893739a2b862cb1e9b23a2ef
2013-05-10 11:43:05 -07:00
Catrope b295dc8d99 Clean up variable initialization in setupSkinTabs()
Change-Id: Ic596f1c98162557f33f5ca398d2358e91a3d5cb0
2013-05-10 11:43:01 -07:00
Ed Sanders 634714bab9 Remove hack for old version of Parsoid
Removing check for parsoid returning document fragments.

Change-Id: I5a8729d93907d13c699a1f3cf43ffd3e0c45b003
2013-05-10 19:42:35 +01:00
jenkins-bot 326879b75b Merge "Use pageLanguage rather than pageViewLanguage for the surface's lang&dir" 2013-05-10 18:13:09 +00:00
Catrope a65ff4fdb9 Use pageLanguage rather than pageViewLanguage for the surface's lang&dir
Bug: 33175
Change-Id: Ica006404227dcd286c387de4f637036341b17eae
2013-05-10 11:04:31 -07:00
jenkins-bot fa73dbeca5 Merge "Make reference list nodes render contents again" 2013-05-10 17:49:30 +00:00
Trevor Parscal 8858ed932f Make reference list nodes render contents again
Really Roan? How dare you!

See I3f5116d7dfbaa5889b7c5cac01d7341272749c58

Change-Id: Ia0ae514fc07752edb7b7100716e09163d3bfacba
2013-05-10 10:47:10 -07:00
jenkins-bot 51388095e2 Merge "Category UI improvements" 2013-05-10 17:32:34 +00:00
jenkins-bot 8bac303893 Merge "When removing an empty list item, also remove the list if it's now empty" 2013-05-10 17:17:36 +00:00
jenkins-bot ff15fbf21b Merge "Whitespace between unwrapped inline nodes assigned to paragraph" 2013-05-10 17:13:15 +00:00
Ed Sanders 328dd8e21f Whitespace between unwrapped inline nodes assigned to paragraph
In stopWrapping we assign any left over whitespace to the paragraph
in position 3, however we weren't clearing this whitespace buffer
if an inline content node followed it.

Change-Id: I8b3ee3915044abd6bafda386430bf7f992ca4aa8
2013-05-10 16:40:11 +01:00
jenkins-bot f966e9c41d Merge "Clone template dom elements being sent to converter" 2013-05-10 15:37:24 +00:00
jenkins-bot d7c3dd8ec6 Merge "Reduction of renderContents calls" 2013-05-10 15:35:37 +00:00
Catrope 811721f1e1 When removing an empty list item, also remove the list if it's now empty
When the user presses Enter in an empty list item, we remove it. But if
the list item was the only child of the list, that leaves an empty list
which then gets a block slug, leading to all sorts of weird things, and
even pawns in Firefox. So check whether the list item was the last child
and if so, remove the list.

Bug: 48287
Change-Id: If22d9b904b8861e24d56944d791545635b2e4254
2013-05-09 17:47:36 -07:00
Inez Korczyński 1e85621bd9 Reduction of renderContents calls
There is no need to call renderContents in ContentBranchNode constructor
because it is going to be called anyways in onSplice

Change-Id: Id1ab983668299658ecd6e89a37667cc34c701689
2013-05-10 00:28:30 +00:00
Catrope 9372ca60ad Kill getOffsetFromNode() with fire
The way it operated was evil. It did a depth-first search from the root,
finding the node using reference equality. For documents with deep
structures, this could take a long time. Inez did some profiling and
found it was called tens of millions of times on a complex document.

Kill getOffsetFromNode() and move its functionality to getOffset().
The logic has been completely rewritten: getOffset() now traverses
up from the node rather than down from the root, and pretty much does
the reverse of what getNodeFromOffset() does. This should be much more
efficient even without offset caching in the node objects (which we may
still implement later).

Change-Id: I125f9fa423c40db6472e2c4a7c94214218ba3bc7
2013-05-09 17:26:22 -07:00
Trevor Parscal e888d7b985 Category UI improvements
Objectives:
* Ensure items don't get moved to the end when their sort-key is edited
* Add placeholder text and pending styling to input
* Auto-expand input to the end of the line
* Make the minimum input width smaller

Changes:

ve.ui.MWMetaDialog.js
* Added calls to fitInput on initialize
* Fixed sort key update and insert handlers to maintain item position when updating

ve.ui.GroupElement.js
* Added index argument to addItems, allowing items to be inserted at a specific location

ve.ui.PagePanelLayout.js
* Fixed CSS class name

ve.ui.StackPanelLayout.js, ve.ui.MenuWidget.js, ve.ui.SelectWidget.js
* Passed index argument through to group element

ve.ui.PanelLayout.js
* Fixed overflow direction for scrolling option

ve.ui.Inspector.css
* Moved border-box properties to text input widget class
* Set input widget within inspectors to be 100% by default

ve.ui.Layout.css
* Updated CSS class name
* Whitespace fixes

ve.ui.Widget.css
* Made text input widgets's wrapper default to 20em wide and the input inside it be 100%, using border-box to ensure proper sizing
* Adjusted category list item and input styles to make input appear more like a category item
* Whitespace fixes

ve.ui.MWCategoryInputWidget.js
* Made category input widget inherit text input widget, rather than just input widget

ve.ui.MWCategoryWidget.js
* Replaced group functionality by mixing in group element
* Added fitInput, which automatically make the input fill the rest of the line or take up the entire next line depending on how much space is left

VisualEditor.i18n.php
* Adjusted placeholder text for category input

Change-Id: I79a18a7b849804027473084a42c36133fdacad57
2013-05-10 00:19:46 +00:00
jenkins-bot 2cbc5045b6 Merge "Move completeHistory from dm.Surface to dm.Document" 2013-05-09 22:19:58 +00:00
Catrope 3d21609ca9 Move completeHistory from dm.Surface to dm.Document
Per IRL conversation with Trevor, that's where it belongs.

Change-Id: I9eae05380597ed83122f29864b77d2278d3dd863
2013-05-09 14:54:25 -07:00
Ed Sanders 3495c6f747 Clone template dom elements being sent to converter
The dom elements in the IV store are used for rendering, so if they
are sent by reference to the converter they get re-attached, causing
all templates to disappear from the page whenever you press 'review
and save'.

Fix is to run it through ve.copyArray, which clones all the nodes.

Change-Id: I1b03351a28ac82e0fdb7e94e761cf65d6548e501
2013-05-09 13:42:44 +01:00
Ed Sanders cced110c3e Mo' vectors (mo' problems)
Serve vectors to any browser which supports them, instead
of just high dpi devices.

Bug: 48148
Change-Id: Iafe8cd290fb195369776f51a96b6de4a51577b89
2013-05-09 11:22:32 +01:00
jenkins-bot cf99b23219 Merge "Fixing shields" 2013-05-08 19:54:06 +00:00
Christian Williams 30cfe6998f Fixing shields
Shields were being added twice. Switched the blank png to gif
because the red IE background couldn't be replicated. Changed
some styles from Alien to Protected.

Change-Id: I9c62665e4e0dc54b8511749b9d2a629db7990a16
2013-05-08 12:29:25 -07:00
Timo Tijhof 6467a5a0e4 ViewPageTarget: Bind callback in onSerialize like in onShowChanges
init.mw.Target#serialize is an odd one because it is the only
target method that has a callback parameter (instead of
emitting the event on Target and having ViewPageTarget get
the event through that).

This line from onShowChanges needed to be added to onSerialize

  // Invalidate the viewer wikitext on next change
  this.surface.getModel().connect( this, { 'transact': 'onSurfaceModelTransact' } );

Though onSerialize is currently bound manually from the
serialize callback, it should be refactored to be like the others.

Bug: 44446
Change-Id: I9eddebbdf9294ee3d46286bdf1b157e00252d300
2013-05-08 20:28:28 +02:00
jenkins-bot 5c9b24eb1f Merge "Also render attributes of the form html/i-j/attrName in CE" 2013-05-08 18:14:33 +00:00
jenkins-bot 3c9dd89f89 Merge "Preserve HTML attributes recursively" 2013-05-08 18:12:46 +00:00
Catrope 31c2165770 Also render attributes of the form html/i-j/attrName in CE
Factored the parsing of html/* attributes out into a static function.
Factored attribute (re)rendering out into ce.View, attribute updates
are much simpler now.

Change-Id: I4caa6d5e1e2c21c28ddff61c3c864e47f66cc6b2
2013-05-08 18:10:46 +00:00
Catrope 05828cc3f1 Preserve HTML attributes recursively
For nodes that handle their own children (as well as leaf nodes and
meta items), store the first child's attributes in html/0-0/*, the
second child's attributes in html/0-1/*, the second element's third
child's fourth child's attributes in html/1-2-3/* , etc.

This obsoletes the ad-hoc code that basically did the same thing in
MWInlineImageNode.

Change-Id: If5abd2d5d9c361b359617ff4b0f3d6ba4c9b0142
2013-05-08 11:10:07 -07:00
Ed Sanders 4a4cf34a86 Convert DOM elements to stringifiable objects for reporting
We already do this in unit test so moving getDomElementSummary
and convertDomElements from ve.qunit.js to ve.js.

Apply ve.convertDomElements to report data before serialising.

Bug: 47948
Change-Id: Id807ccc6ff31d063be815ed4988cb35684adb76a
2013-05-08 14:42:43 +01:00
Catrope 85d67ce44f Do proper HTML preservation in ve.dm.MWReferenceListNode
Store the HTML as a domElements array like everywhere else, rather than
as a string. Also disable HTML attribute preservation because there's
no point doing that when we're already preserving all of the HTML.

Also fixed a misnamed attribute (<li li="foo"> --> <li id="foo">) in
the test case.

Change-Id: I36bf8bade8118e07a75eb6f3a2427a00ef4915d7
2013-05-07 21:41:29 -07:00
jenkins-bot 9f06258b8e Merge "ve.test: Import K-JS test suite for object utilities" 2013-05-08 00:18:28 +00:00
Timo Tijhof 36e29c12f6 ve.test: Import K-JS test suite for object utilities
We've housed these utilities for a while and changed some of
them quite a bit. Importing them to avoid regressions and
ensure we keep them in sync with our fork of it.

Removed references back through `@source`. They are sufficiently
different that the reference no longer adds any value.

Imported from https://github.com/Krinkle/K-js/blob/v0.1.2/test/K.test.js.

Change-Id: I9e71297246b7c248c1f032ba6b6ae1123519f3c1
2013-05-08 02:13:16 +02:00
jenkins-bot 638bf98e98 Merge "Fix ContentBranchNode test broken by ve.ce.ProtectedNode" 2013-05-07 21:54:18 +00:00
jenkins-bot b0eca923bd Merge "Make .static.storeHtmlAttributes more versatile" 2013-05-07 21:53:46 +00:00
Catrope f6d28e6117 Fix ContentBranchNode test broken by ve.ce.ProtectedNode
Also make sure Jenkins actually runs that test suite

Change-Id: I72b10db3875370da95cb20e127ac8f1efee425bf
2013-05-07 14:48:23 -07:00
Catrope 317a404ece Make .static.storeHtmlAttributes more versatile
It now allows you to specify which attributes to preserve in various
ways rather than just setting true or false.

Removed unused factory methods that exposed the old value.

Change-Id: I914164adcf1f0e48fa3fa85277e68c72dbad393e
2013-05-07 14:45:26 -07:00
jenkins-bot ba8ffe629b Merge "ve.ce.FocusableNode: Fix documentation" 2013-05-07 20:01:44 +00:00
Timo Tijhof 13dd31c046 ve.ce.FocusableNode: Fix documentation
The constructor doesn't use any parameter named "$resizable".
And the one usage of it (ve.ce.ImageNode) doesn't pass any
arguments.

Follows-up 3fe3032.

Change-Id: I30530f4199a7c8383933be286b7b7a705c05c26c
2013-05-07 21:51:38 +02:00
Timo Tijhof a631af342d ve.isMixedIn: Fix bugs and add unit tests
* It returned undefined instead of false if the subject
  didn't have a 'mixins' property. This is because '&&'
  is a DEFAULT operator, not an AND operator, it returns the
  value, not a boolean per se.
* The logic of traversing to the constructor property was
  broken since all objects in javascript have a constructor
  property, and functions/constructors are also objects.

Follows-up Ic3e4472b9e694.

Change-Id: I462e7ce270c8cfc7e1970e359894ee4b7d90b881
2013-05-07 19:44:55 +00:00
jenkins-bot ab7f7ae915 Merge "ve.ce.ProtectedNode" 2013-05-07 01:33:43 +00:00