mediawiki-extensions-Visual.../modules/ve-mw/dm/models/ve.dm.MWTransclusionModel.js
Máté Szabó ff4f382322 Use autogenerated parameter info for templates without TemplateData
When a template does not have user-provided TemplateData documentation,
the TemplateData API falls back to extracting possible parameters from the raw wikitext
to generate an API response with a list of potential parameters. However, it also
sets the "notemplatedata" field in the response, causing the VisualEditor to think
the response contains no useful information and ignore it. This appears to have been
an unintended side-effect of I97a1bfc9f9ead082a673a91b9d2053630a90309c.

This patch ensures that the VisualEditor will correctly consider such responses from
TemplateData by modifying ve.dm.MWTransclusionModel to check if the response contains
a parameter map. Some unit tests were added for the class to verify this behavior.

Bug: T243868
Change-Id: I72005880d9301a53224473900efe2917379e8708
2020-01-31 13:43:56 +01:00

572 lines
16 KiB
JavaScript

/*!
* VisualEditor DataModel MWTransclusionModel class.
*
* @copyright 2011-2020 VisualEditor Team and others; see AUTHORS.txt
* @license The MIT License (MIT); see LICENSE.txt
*/
( function () {
var hasOwn = Object.hasOwnProperty,
specCache = {};
/**
* MediaWiki transclusion model.
*
* @class
* @mixins OO.EventEmitter
*
* @constructor
* @param {ve.dm.Document} doc Document to use associate with API requests
*/
ve.dm.MWTransclusionModel = function VeDmMWTransclusionModel( doc ) {
// Mixin constructors
OO.EventEmitter.call( this );
// Properties
this.doc = doc;
this.parts = [];
this.uid = 0;
this.requests = [];
this.queue = [];
this.specCache = specCache;
};
/* Inheritance */
OO.mixinClass( ve.dm.MWTransclusionModel, OO.EventEmitter );
/* Events */
/**
* @event replace
* @param {ve.dm.MWTransclusionPartModel|null} removed Removed part
* @param {ve.dm.MWTransclusionPartModel|null} added Added part
*/
/**
* @event change
*/
/* Methods */
/**
* Insert transclusion at the end of a surface fragment.
*
* If forceType is not specified and this is used in async mode, users of this method
* should ensure the surface is not accessible while the type is being evaluated.
*
* @param {ve.dm.SurfaceFragment} surfaceFragment Surface fragment after which to insert.
* @param {boolean|undefined} [forceType] Force the type to 'inline' or 'block'. If not
* specified it will be evaluated asynchronously.
* @return {jQuery.Promise} Promise which resolves when the node has been inserted. If
* forceType was specified this will be instant.
*/
ve.dm.MWTransclusionModel.prototype.insertTransclusionNode = function ( surfaceFragment, forceType ) {
var model = this,
deferred = ve.createDeferred(),
baseNodeClass = ve.dm.MWTransclusionNode;
function insertNode( isInline, generatedContents ) {
var hash, store, nodeClass,
type = isInline ? baseNodeClass.static.inlineType : baseNodeClass.static.blockType,
data = [
{
type: type,
attributes: {
mw: model.getPlainObject()
}
},
{ type: '/' + type }
];
// If we just fetched the generated contents, put them in the store
// so we don't do a duplicate API call later.
if ( generatedContents ) {
nodeClass = ve.dm.modelRegistry.lookup( type );
store = surfaceFragment.getDocument().getStore();
hash = OO.getHash( [ nodeClass.static.getHashObjectForRendering( data[ 0 ] ), undefined ] );
store.hash( generatedContents, hash );
}
surfaceFragment.insertContent( data );
deferred.resolve();
}
if ( forceType ) {
insertNode( forceType === 'inline' );
} else {
ve.init.target.parseWikitextFragment(
baseNodeClass.static.getWikitext( this.getPlainObject() ),
true,
surfaceFragment.getDocument()
).then( function ( response ) {
var contentNodes;
if ( ve.getProp( response, 'visualeditor', 'result' ) === 'success' ) {
// This method is only ever run by a client, so it is okay to use jQuery
// eslint-disable-next-line no-undef
contentNodes = $.parseHTML( response.visualeditor.content, surfaceFragment.getDocument().getHtmlDocument() ) || [];
contentNodes = ve.ce.MWTransclusionNode.static.filterRendering( contentNodes );
insertNode(
baseNodeClass.static.isHybridInline( contentNodes, ve.dm.converter ),
contentNodes
);
} else {
// Request failed - just assume inline
insertNode( true );
}
}, function () {
insertNode( true );
} );
}
return deferred.promise();
};
/**
* Update transclusion node in a document.
*
* @param {ve.dm.Surface} surfaceModel Surface model of main document
* @param {ve.dm.MWTransclusionNode} node Transclusion node to update
*/
ve.dm.MWTransclusionModel.prototype.updateTransclusionNode = function ( surfaceModel, node ) {
var obj = this.getPlainObject();
if ( obj !== null ) {
surfaceModel.getLinearFragment( node.getOuterRange(), true )
.changeAttributes( { mw: obj } );
} else {
surfaceModel.getLinearFragment( node.getOuterRange(), true )
.removeContent();
}
};
/**
* Load from transclusion data, and fetch spec from server.
*
* @param {Object} data Transclusion data
* @return {jQuery.Promise} Promise, resolved when spec is loaded
*/
ve.dm.MWTransclusionModel.prototype.load = function ( data ) {
var i, len, part, deferred,
promises = [];
// Convert single part format to multi-part format
// Parsoid doesn't use this format any more, but we accept it for backwards compatibility
if ( data.params && data.target ) {
data = { parts: [ { template: data } ] };
}
if ( Array.isArray( data.parts ) ) {
for ( i = 0, len = data.parts.length; i < len; i++ ) {
part = data.parts[ i ];
if ( part.template ) {
deferred = ve.createDeferred();
promises.push( deferred.promise() );
this.queue.push( {
add: ve.dm.MWTemplateModel.newFromData( this, part.template ),
deferred: deferred
} );
} else if ( typeof part === 'string' ) {
deferred = ve.createDeferred();
promises.push( deferred.promise() );
this.queue.push( {
add: new ve.dm.MWTransclusionContentModel( this, part ),
deferred: deferred
} );
}
}
setTimeout( this.fetch.bind( this ) );
}
return ve.promiseAll( promises );
};
/**
* Process one or more queue items.
*
* @param {Object[]} queue List of objects containing parts to add and optionally indexes to add
* them at, if no index is given parts will be added at the end
* @fires replace For each item added
* @fires change
*/
ve.dm.MWTransclusionModel.prototype.process = function ( queue ) {
var i, len, item, title, index, remove, existing, resolveQueue = [];
for ( i = 0, len = queue.length; i < len; i++ ) {
remove = 0;
item = queue[ i ];
if ( item.add instanceof ve.dm.MWTemplateModel ) {
title = item.add.getTitle();
if ( hasOwn.call( specCache, title ) && specCache[ title ] ) {
item.add.getSpec().extend( specCache[ title ] );
}
}
// Use specified index
index = item.index;
// Auto-remove if already existing, preserving index
existing = this.parts.indexOf( item.add );
if ( existing !== -1 ) {
this.removePart( item.add );
if ( index && existing + 1 < index ) {
index--;
}
}
// Derive index from removal if given
if ( index === undefined && item.remove ) {
index = this.parts.indexOf( item.remove );
if ( index !== -1 ) {
remove = 1;
}
}
// Use last index as a last resort
if ( index === undefined || index === -1 ) {
index = this.parts.length;
}
this.parts.splice( index, remove, item.add );
if ( item.add ) {
item.add.connect( this, { change: [ 'emit', 'change' ] } );
}
if ( item.remove ) {
item.remove.disconnect( this );
}
this.emit( 'replace', item.remove || null, item.add );
// Resolve promises
if ( item.deferred ) {
resolveQueue.push( item.deferred );
}
}
this.emit( 'change' );
// We need to go back and resolve the deferreds after emitting change.
// Otherwise we get silly situations like a single change event being
// guaranteed after the transclusion loaded promise gets resolved.
for ( i = 0; i < resolveQueue.length; i++ ) {
resolveQueue[ i ].resolve();
}
};
/** */
ve.dm.MWTransclusionModel.prototype.fetch = function () {
var i, len, item, title, queue,
templateNamespaceId = mw.config.get( 'wgNamespaceIds' ).template,
titles = [],
specs = {};
if ( !this.queue.length ) {
return;
}
queue = this.queue.slice();
// Clear shared queue for future calls
this.queue.length = 0;
// Get unique list of template titles that aren't already loaded
for ( i = 0, len = queue.length; i < len; i++ ) {
item = queue[ i ];
if ( item.add instanceof ve.dm.MWTemplateModel ) {
title = item.add.getTitle();
if (
// Skip titles that don't have a resolvable href
title &&
// Skip titles outside the template namespace
mw.Title.newFromText(
title,
templateNamespaceId
).namespace === templateNamespaceId &&
// Skip already cached data
!hasOwn.call( specCache, title ) &&
// Skip duplicate titles in the same batch
titles.indexOf( title ) === -1
) {
titles.push( title );
}
}
}
// Bypass server for empty lists
if ( !titles.length ) {
setTimeout( this.process.bind( this, queue ) );
return;
}
this.requests.push( this.fetchRequest( titles, specs, queue ) );
};
ve.dm.MWTransclusionModel.prototype.fetchRequest = function ( titles, specs, queue ) {
var xhr = ve.init.target.getContentApi( this.doc ).get( {
action: 'templatedata',
titles: titles,
lang: mw.config.get( 'wgUserLanguage' ),
format: 'json',
doNotIgnoreMissingTitles: '1',
redirects: '1'
} ).done( this.fetchRequestDone.bind( this, titles, specs ) );
xhr.always( this.fetchRequestAlways.bind( this, queue, xhr ) );
return xhr;
};
ve.dm.MWTransclusionModel.prototype.fetchRequestDone = function ( titles, specs, data ) {
var i, len, id, title, missingTitle, aliasMap = [];
if ( data && data.pages ) {
// Keep spec data on hand for future use
for ( id in data.pages ) {
title = data.pages[ id ].title;
if ( data.pages[ id ].missing ) {
// Remmeber templates that don't exist in the link cache
// { title: { missing: true|false }
missingTitle = {};
missingTitle[ title ] = { missing: true };
ve.init.platform.linkCache.setMissing( missingTitle );
} else if ( data.pages[ id ].notemplatedata && !OO.isPlainObject( data.pages[ id ].params ) ) {
// (T243868) Prevent asking again for templates that have neither user-provided specs
// nor automatically detected params
specs[ title ] = null;
} else {
specs[ title ] = data.pages[ id ];
}
}
// Follow redirects
if ( data.redirects ) {
aliasMap = data.redirects;
}
// Follow MW's normalisation
if ( data.normalized ) {
ve.batchPush( aliasMap, data.normalized );
}
// Cross-reference aliased titles.
for ( i = 0, len = aliasMap.length; i < len; i++ ) {
// Only define the alias if the target exists, otherwise
// we create a new property with an invalid "undefined" value.
if ( hasOwn.call( specs, aliasMap[ i ].to ) ) {
specs[ aliasMap[ i ].from ] = specs[ aliasMap[ i ].to ];
}
}
ve.extendObject( specCache, specs );
}
};
ve.dm.MWTransclusionModel.prototype.fetchRequestAlways = function ( queue, apiPromise ) {
// Prune completed request
var index = this.requests.indexOf( apiPromise );
if ( index !== -1 ) {
this.requests.splice( index, 1 );
}
// Actually add queued items
this.process( queue );
};
/**
* Abort any pending requests.
*/
ve.dm.MWTransclusionModel.prototype.abortRequests = function () {
var i, len;
for ( i = 0, len = this.requests.length; i < len; i++ ) {
this.requests[ i ].abort();
}
this.requests.length = 0;
};
/**
* Get plain object representation of template transclusion.
*
* @return {Object|null} Plain object representation, or null if empty
*/
ve.dm.MWTransclusionModel.prototype.getPlainObject = function () {
var i, len, part, serialization,
obj = { parts: [] };
for ( i = 0, len = this.parts.length; i < len; i++ ) {
part = this.parts[ i ];
serialization = part.serialize();
if ( serialization !== undefined && serialization !== '' ) {
obj.parts.push( serialization );
}
}
if ( obj.parts.length === 0 ) {
return null;
}
return obj;
};
/**
* Get the wikitext for this transclusion.
*
* @return {string} Wikitext like `{{foo|1=bar|baz=quux}}`
*/
ve.dm.MWTransclusionModel.prototype.getWikitext = function () {
var i, len,
wikitext = '';
for ( i = 0, len = this.parts.length; i < len; i++ ) {
wikitext += this.parts[ i ].getWikitext();
}
return wikitext;
};
/**
* Get a unique ID for a part in the transclusion.
*
* This is used to give parts unique IDs, and returns a different value each time it's called.
*
* @return {number} Unique ID
*/
ve.dm.MWTransclusionModel.prototype.getUniquePartId = function () {
return this.uid++;
};
/**
* Replace part.
*
* Replace asynchronously.
*
* @param {ve.dm.MWTransclusionPartModel} remove Part to remove
* @param {ve.dm.MWTransclusionPartModel} add Part to add
* @throws {Error} If part to remove is not valid
* @throws {Error} If part to add is not valid
* @return {jQuery.Promise} Promise, resolved when part is added
*/
ve.dm.MWTransclusionModel.prototype.replacePart = function ( remove, add ) {
var deferred = ve.createDeferred();
if (
!( remove instanceof ve.dm.MWTransclusionPartModel ) ||
!( add instanceof ve.dm.MWTransclusionPartModel )
) {
throw new Error( 'Invalid transclusion part' );
}
this.queue.push( { remove: remove, add: add, deferred: deferred } );
// Fetch on next yield to process items in the queue together, subsequent calls to fetch will
// have no effect because the queue will be clear
setTimeout( this.fetch.bind( this ) );
return deferred.promise();
};
/**
* Add part.
*
* Added asynchronously, but order is preserved.
*
* @param {ve.dm.MWTransclusionPartModel} part Part to add
* @param {number} [index] Specific index to add content at, defaults to the end
* @throws {Error} If part is not valid
* @return {jQuery.Promise} Promise, resolved when part is added
*/
ve.dm.MWTransclusionModel.prototype.addPart = function ( part, index ) {
var deferred = ve.createDeferred();
if ( !( part instanceof ve.dm.MWTransclusionPartModel ) ) {
throw new Error( 'Invalid transclusion part' );
}
this.queue.push( { add: part, index: index, deferred: deferred } );
// Fetch on next yield to process items in the queue together, subsequent calls to fetch will
// have no effect because the queue will be clear
setTimeout( this.fetch.bind( this ) );
return deferred.promise();
};
/**
* Remove a part.
*
* @param {ve.dm.MWTransclusionPartModel} part Part to remove
* @fires replace
*/
ve.dm.MWTransclusionModel.prototype.removePart = function ( part ) {
var index = this.parts.indexOf( part );
if ( index !== -1 ) {
this.parts.splice( index, 1 );
part.disconnect( this );
this.emit( 'replace', part, null );
}
};
/**
* Get all parts.
*
* @return {ve.dm.MWTransclusionPartModel[]} Parts in transclusion
*/
ve.dm.MWTransclusionModel.prototype.getParts = function () {
return this.parts;
};
/**
* Get part by its ID.
*
* Matching is performed against the first section of the `id`, delimited by a '/'.
*
* @param {string} id Part ID
* @return {ve.dm.MWTransclusionPartModel|null} Part with matching ID, if found
*/
ve.dm.MWTransclusionModel.prototype.getPartFromId = function ( id ) {
var i, len,
// For ids from ve.dm.MWParameterModel, compare against the part id
// of the parameter instead of the entire model id (e.g. "part_1" instead of "part_1/foo").
partId = id.split( '/' )[ 0 ];
for ( i = 0, len = this.parts.length; i < len; i++ ) {
if ( this.parts[ i ].getId() === partId ) {
return this.parts[ i ];
}
}
return null;
};
/**
* Get the index of a part or parameter.
*
* Indexes are linear depth-first addresses in the transclusion tree.
*
* @param {ve.dm.MWTransclusionPartModel|ve.dm.MWParameterModel} model Part or parameter
* @return {number} Page index of model
*/
ve.dm.MWTransclusionModel.prototype.getIndex = function ( model ) {
var i, iLen, j, jLen, part, names,
parts = this.parts,
index = 0;
for ( i = 0, iLen = parts.length; i < iLen; i++ ) {
part = parts[ i ];
if ( part === model ) {
return index;
}
index++;
if ( part instanceof ve.dm.MWTemplateModel ) {
names = part.getParameterNames();
for ( j = 0, jLen = names.length; j < jLen; j++ ) {
if ( part.getParameter( names[ j ] ) === model ) {
return index;
}
index++;
}
}
}
return -1;
};
/*
* Add missing required and suggested parameters to each transclusion.
*/
ve.dm.MWTransclusionModel.prototype.addPromptedParameters = function () {
var i;
for ( i = 0; i < this.parts.length; i++ ) {
this.parts[ i ].addPromptedParameters();
}
};
}() );