Commit graph

3392 commits

Author SHA1 Message Date
Catrope 1a4c60ed17 Make translateRange() translate across wrapping transactions correctly
Previously, we would translate to the right of an insertion, but for
wrapping transactions that means we end up with something like
<ul><li>|<p>...</p></li></ul>|, which doesn't make any sense. This
change changes this to <ul><li>|<p>...</p>|</li></ul>.

* Add parameter to translateOffset() that toggles the behavior for
  the offset before an insertion
* In translateRange(), translate start and end differently
** In some cases this can map them across each other, fix that

Change-Id: Ia2197b08d9f6763be3f2db5a59546ddc3f74a281
2013-02-22 17:33:31 -08:00
Catrope 90114c7a48 Add ve.Range.prototype.isBackwards()
Change-Id: I6f6977cfc9d772239da22ceb779e5e37576c2e0c
2013-02-22 17:33:23 -08:00
Catrope 090e944dd9 Combine consecutive remove operations
We already combine consecutive operations of other types, but didn't do
that for remove. We have to do this, though, because translateOffset()
gets confused otherwise.

Change-Id: I4285a3efd9f2297398f57e2f24adb26adafdf465
2013-02-22 17:33:17 -08:00
jenkins-bot d00486a450 Merge "Cleanup selection displaying related methods in ve.ce.Surface." 2013-02-23 01:18:15 +00:00
jenkins-bot 951b3c78f0 Merge "Introduction of ve.ce.RangyRange class." 2013-02-23 01:16:46 +00:00
Inez Korczyński c3d10e965a Cleanup selection displaying related methods in ve.ce.Surface.
* Deleted blocks of code that are never executed.
* Deleted method (showCursor) that is never used.

Change-Id: I031585b40f384d3e0e5d4cb03f626c9229d97fcf
2013-02-22 17:05:54 -08:00
Inez Korczyński a5e56718cd Introduction of ve.ce.RangyRange class.
Change-Id: I44bcb79a5967c362774603ea27181b9c64d2ea96
2013-02-22 17:02:42 -08:00
Catrope 1463d03f44 Change one last .storeHTMLAttributes to .storeHtmlAttributes
Change-Id: I161b2e8bf22c3784ca660ab0961a528b23601022
2013-02-22 16:13:47 -08:00
jenkins-bot 065e9c8a67 Merge "Make the full document regex hack a bit more robust" 2013-02-22 23:26:38 +00:00
jenkins-bot 60b5c838ad Merge "Make string type matches override regex type matches" 2013-02-22 23:26:27 +00:00
jenkins-bot fa1f760deb Merge "Add MWMetaNode to clean up <meta>/<link> hack in the converter" 2013-02-22 23:26:24 +00:00
jenkins-bot 4e0ebfdba0 Merge "Support RDFa type regexes in ModelRegistry" 2013-02-22 23:26:21 +00:00
jenkins-bot fb11cc8a8c Merge "(bug 45062) Implement the new node API in the converter" 2013-02-22 23:26:18 +00:00
jenkins-bot 482c89a916 Merge "Change context.wrapping to context.inWrapper" 2013-02-22 23:26:15 +00:00
jenkins-bot 950ff37d9b Merge "Add .static.enableAboutGrouping" 2013-02-22 23:26:12 +00:00
jenkins-bot 78232847db Merge "Add .static.storeHtmlAttributes" 2013-02-22 23:26:09 +00:00
Catrope dbd3d2354a Make string type matches override regex type matches
Change-Id: Ic21e27c047d8cb1161a6613bb5232d0d19b37aa6
2013-02-22 15:21:40 -08:00
Catrope 04c72f6871 Add MWMetaNode to clean up <meta>/<link> hack in the converter
Change-Id: I4c69bff4981eef78415b43d31c3fd2ee271450ef
2013-02-22 15:21:40 -08:00
Catrope 314817925a Support RDFa type regexes in ModelRegistry
Change-Id: Id19c3a567140f9e27ba732b6e8d9e25bc7a28a4f
2013-02-22 15:21:40 -08:00
Catrope 2e36f1542b (bug 45062) Implement the new node API in the converter
This changes the node API to work with multiple elements, so we can
support about groups. Instead of passing in and returning single DOM
elements, we use arrays of DOM elements.

ve.dm.Converter:
* Pass modelRegistry into the constructor
* Remove onNodeRegister handler and its data
* Remove getDataElementFromDomElement() and
  getDataAnnotationFromDomElement(). Most logic moved into
  getDataFromDom(), some into createDataElement()
* Remove createAlien(), replaced with
  createDataElement( ve.dm.AlienNode, ... )
* Replace doAboutGrouping() (which wrapped about groups) with
  getAboutGroup() (which returns an array of the nodes in the group)
* Put in a hack so <meta>/<link> elements with an mw: property aren't
  alienated
* Remove about group wrapping behavior in favor of just outputting
  multiple nodes

ve.dm.AlienNode.js:
* For multi-element aliens, only choose inline if all elements are
  inline, not just the first one

ve.dm.example.js:
* Add html/0 stuff for meta nodes
* Fix test case to reflect new alien behavior

Change-Id: I40dcc27430f778bc00a44b91b7d824bfb2718be6
2013-02-22 15:21:40 -08:00
Catrope 5e16141750 Change context.wrapping to context.inWrapper
It's what the docs say it should be, so let's call it that :)

Change-Id: Ia10861ce77872243beb3d3a1886877824103f6c9
2013-02-22 15:21:40 -08:00
Catrope cf9e5dd953 Add .static.enableAboutGrouping
Defaults to false, nodes can opt into about grouping. Not being used
yet, will be used in the converter rewrite.

Change-Id: Ic2a529241e69fb07ac926c826ead0108193b3050
2013-02-22 15:21:40 -08:00
Catrope bbe3783d58 Add .static.storeHtmlAttributes
Defaults to true, but set to false, so we don't do redundant work for
aliens.

Change-Id: If35db3a67afd78124b4b2b46bb78ad60cbac46f5
2013-02-22 15:21:31 -08:00
jenkins-bot 26e623940a Merge "Implement SurfaceFragment.unwrapAllNodes and fix wrapAllNodes." 2013-02-22 22:36:03 +00:00
jenkins-bot 5c78ba9b6a Merge "Fix bug with replacements in translateOffset()" 2013-02-22 22:33:48 +00:00
jenkins-bot 8e0348c803 Merge "Fix ModelRegistry stupidities" 2013-02-22 22:31:03 +00:00
Ed Sanders 4f0ac479ed Implement SurfaceFragment.unwrapAllNodes and fix wrapAllNodes.
wrapAllNodes was calulating the new selection incorrectly. This has
been fixed and a test added.

unwrapAllNodes takes a depth as its argument and unwraps that many
elements from inside the selection.

Tests for wrap/unwrap apply also now check that applying a wrap
and then its inverse as an unwrap result in the document reverting
to its original state.

Change-Id: I7dcacdfb5894be59ffad69b369d7b32933a25b61
2013-02-22 13:37:42 -08:00
Catrope 8f84982e0d Fix bug with replacements in translateOffset()
In our test case, offset 12 was mapped to 16, but that should be 11.
The problem here was that the offset right before the removal was
mapped to right after the removal, but that's only valid if we're
dealing with a removal, not when we're dealing with a replacement
where we're both removing and inserting data.

Change-Id: Ibf3c1463c0de009578cd50736f19bae82669ced8
2013-02-22 13:35:33 -08:00
jenkins-bot eae2529064 Merge "(bug 43461) Hide Phantoms" 2013-02-22 21:19:37 +00:00
jenkins-bot 1ef5c07a93 Merge "(bug 45423) Create SurfaceFragment.isolate method" 2013-02-22 20:40:32 +00:00
Ed Sanders feaa422495 (bug 45423) Create SurfaceFragment.isolate method
This method will take a selection of siblings and ensure they
are the only chlidren in their first parent which can be placed anywhere
(for example the first parent of a tableCell which can be placed anywhere
is a table, and for a listItem is a list).

The method ensures no redundant empty tags are created, so if
the selection encompasses all siblings then no action is taken.

Also in this commit are two test cases run against ve.dm.example.isolationData.

Change-Id: I783bd5ecd9d43d61f9b2685985409b4d746cbe94
2013-02-22 12:38:06 -08:00
jenkins-bot ab718eef32 Merge "Create tests for ve.FormatAction.convert" 2013-02-22 19:55:54 +00:00
Ed Sanders c05b5d13cd Create tests for ve.FormatAction.convert
These 3 tests convert different selections of list items to headings.

Bug: 45216
Change-Id: I6cf042d9d5aa4afb68c0992f444b600ed69f0c04
2013-02-22 11:52:12 -08:00
Christian Williams 7e74f0aa70 (bug 43461) Hide Phantoms
This fixes a problem that phantoms would remain visible if moving the mouse quickly. Now binding to mousemove and also mouseout.

Change-Id: I4f5e3319038828f83b108e9f5a45f54ebfeb7027
2013-02-21 23:02:05 +00:00
Catrope 4b086b94d7 Make the full document regex hack a bit more robust
Anchor the regex and make it look for doctype too

Change-Id: I43abb6f214a28801642bde46b50f4ac12614d1e4
2013-02-21 14:59:19 -08:00
jenkins-bot 502a2a2e53 Merge "Change {Boolean} to {boolean} throughout" 2013-02-21 00:42:44 +00:00
Inez Korczyński 0c73aa012e Introduce ve.Keys based on DOM Level 3 KeyEvent
Change-Id: I6abee2c0c49e355aef39089c7c94e51aaee01929
2013-02-20 16:15:59 -08:00
jenkins-bot ddbc0343b7 Merge "Make ve.copyArray() / copyObject() work with booleans" 2013-02-20 18:24:31 +00:00
Catrope a095c0ab06 Change {Boolean} to {boolean} throughout
Change-Id: Ibbbaed3762efa6a6928ba6e4611b1f1fef91ad8f
2013-02-20 10:03:55 -08:00
jenkins-bot 1d0a383968 Merge "Very small cleanup of onCut method in ve.ce.Surface." 2013-02-20 17:36:48 +00:00
Inez Korczyński 2d9c5d5628 Very small cleanup of onCut method in ve.ce.Surface.
Use ve.bind in order to change context within which
callback is called - reduces a need of introducing
new local variable.

Change-Id: I75ece695548c87073dd22e5e7ec80057d7132d22
2013-02-20 17:35:30 +00:00
Catrope b553cb9f21 Fix ModelRegistry stupidities
* Register mw: as an extension-specific type
* Include 'property' as an RDFa type attribute

Change-Id: Ia70708b76294815eb8eb8132048c33b0636380e9
2013-02-20 09:30:15 -08:00
Catrope 477cec43d6 Make ve.copyArray() / copyObject() work with booleans
This should really be fixed to not enumerate all types but be nicer.
Timo said he'd do that.

Change-Id: I15807696a7324d3fb207ed41b442a83dc83a75ec
2013-02-20 09:30:09 -08:00
James D. Forrester 82114467f1 Bump copyright notice year range to -2013 over -2012
199 files touched. Whee!

Change-Id: Id82ce4a32f833406db4a1cc585674f2bdb39ba0d
2013-02-19 15:37:34 -08:00
jenkins-bot a6fc5ab0ac Merge "Upstream: Update QUnit from v1.10.0 to v1.11.0" 2013-02-19 21:52:30 +00:00
Timo Tijhof 0538841366 Upstream: Update QUnit from v1.10.0 to v1.11.0
* Code:
  http://code.jquery.com/qunit/qunit-1.11.0.js
  http://code.jquery.com/qunit/qunit-1.11.0.css
* Tree:
  https://github.com/jquery/qunit/tree/v1.11.0
* Changes:
  https://github.com/jquery/qunit/compare/v1.10.0...v1.11.0
* Release notes:
  https://github.com/jquery/qunit/blob/v1.11.0/History.md

Change-Id: I3591455f0096471d073c94eaab424d151203b297
2013-02-19 19:50:02 +01:00
Catrope 3035c311ec Make the converter work with full HTML documents rather than fragments
The Parsoid output will also be expected to be a full HTML document. For
backwards compatibility, we allow for the Parsoid output to be a
document fragment as well. We don't send a full document back yet, also
for b/c -- we'll change this later once Parsoid has been updated in
production.

ve.dm.Converter.js:
* Make getDataFromDom() accept a document rather than a node
** Split off the recursion (which does use nodes) into its own function
** For now we just convert the <body>. In the future, we'll want to do
   things with the <head> as well
* Pass the document around so we can use it when creating elements
* Make getDomFromData() return a document rather than a <div>

ve.init.mw.Target.js:
* Store a document (this.doc) rather than a DOM node (this.dom)
* Pass around documents rather than DOM nodes
* Detect whether the Parsoid output is an HTML document or a fragment
  using a hacky regex
* When submitting to Parsoid, submit the innerHTML of the <body>

ve.init.mw.ViewPageTarget.js:
* s/dom/doc/
* Store body.innerHTML in this.originalHtml

ve.Surface.js:
* s/dom/doc/

demos/ve/index.php:
* Don't wrap HTML in <div>
* Pass HTML document rather than DOM node to ve.Surface

ve.dm.Converter.test.js:
* Construct a document from the test HTML, rather than a <div>

ve.dm.example.js:
* Wrap the HTML in the converter test cases in <body> tags to prevent
  misinterpretation (HTML fragments starting with comments, <meta>,
  <link> and whitespace are problematic)

Change-Id: I82fdad0a099febc5e658486cbf8becfcdbc85a2d
2013-02-19 10:38:39 -08:00
Catrope d36e6851d5 Add ve.createDocumentFromHTML()
Converts an HTML string to a brand new document using an iframe hack
(proper ways to do this exist, but don't work cross-browser).

Parsoid will serve us full HTML documents rather than document fragments
soon, so we'll need this functionality (along with some other changes
that I'm working on now) to deal with that change.

This doesn't currently work quite right in IE8 (although we have lots of
other issues in IE8) as well. It's not that hard to fix up though: we
just have to leave the iframe attached to the main document (should
probably provide a destroy function in that case) and the tests have to
deal with the fact that IE normalizes <head></head> to
<head><title></title></head> .

(For backwards compatibility, we'll have to deal with document fragments
as well; this will be implemented as an MW-specific hack in the
integration in the next commit.)

Change-Id: I15f877583c39124ba1c5e8e22585297ff3bac8d6
2013-02-19 10:30:25 -08:00
Translation updater bot 028f7a411a Localisation updates from http://translatewiki.net.
Change-Id: I4a09f45475098c28a8862cbd419d4f7b6cafb0c9
2013-02-18 21:01:53 +00:00
Translation updater bot 7fc8644b90 Localisation updates from http://translatewiki.net.
Change-Id: I20aaef5aad0f74f46dd50cfd7cbd4ed00a153f11
2013-02-17 20:53:26 +00:00