michael@0: /* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*- */
michael@0: /* This Source Code Form is subject to the terms of the Mozilla Public
michael@0:  * License, v. 2.0. If a copy of the MPL was not distributed with this
michael@0:  * file, You can obtain one at http://mozilla.org/MPL/2.0/. */
michael@0: 
michael@0: #include "domstubs.idl"
michael@0: 
michael@0: interface nsIAtom;
michael@0: interface nsIArray;
michael@0: interface nsISimpleEnumerator;
michael@0: interface nsIXULTemplateResult;
michael@0: interface nsIXULTemplateRuleFilter;
michael@0: interface nsIXULTemplateBuilder;
michael@0: 
michael@0: /**
michael@0:  * A query processor takes a template query and generates results for it given
michael@0:  * a datasource and a reference point. There is a one-to-one relationship
michael@0:  * between a template builder and a query processor. The template builder
michael@0:  * creates the query processor, and there is no other means to retrieve it.
michael@0:  *
michael@0:  * A template query is the contents inside a <query> element within the
michael@0:  * template. The actual syntax is opaque to the template builder and defined
michael@0:  * by a query processor. The query is expected to consist of either text or
michael@0:  * DOM nodes that, when executed by a call to the generateResults method, will
michael@0:  * allow the generation of a list of results.
michael@0:  *
michael@0:  * The template builder will supply two variables, the reference variable and
michael@0:  * the member variable to further indicate what part of the datasource is to
michael@0:  * be examined in addition to the query itself. The reference is always
michael@0:  * a placeholder for the starting point and the member is always a placeholder
michael@0:  * for the end points (the results).
michael@0:  *
michael@0:  * The reference point is important when generating output recursively, as
michael@0:  * the query will be the same for each iteration, however, the reference point
michael@0:  * will differ.
michael@0:  *
michael@0:  * For instance, when examining an XML source, an XML query processor might
michael@0:  * begin at the node referred by the reference variable and end at a list of
michael@0:  * that node's children.
michael@0:  *
michael@0:  * Some queries may not need the reference variable if the syntax or the form
michael@0:  * of the data implies the value. For instance, a datasource that holds a
michael@0:  * table that can only produce one set of results.
michael@0:  *
michael@0:  * The reference variable may be specified in a template by setting the
michael@0:  * "container" attribute on the <template> element to the variable to use. The
michael@0:  * member variable may be specified in a similar way using the "member"
michael@0:  * attribute, or it may be specified in the first <action> body in the
michael@0:  * template as the value of a uri attribute on an element. A breadth-first
michael@0:  * search of the first action is performed to find this element.
michael@0:  *
michael@0:  * If unspecified, the default value of the reference variable is ?uri.
michael@0:  *
michael@0:  * For example, a query might have the following syntax:
michael@0:  *
michael@0:  * (?id, ?name, ?url) from Bookmarks where parentfolder = ?start
michael@0:  *
michael@0:  * This query might generate a result for each bookmark within a given folder.
michael@0:  * The variable ?start would be the reference variable, while the variable ?id
michael@0:  * would be the member variable, since it is the unique value that identifies
michael@0:  * a result. Each result will have the four variables referred to defined for
michael@0:  * it and the values may be retrieved using the result's getBindingFor and
michael@0:  * getBindingObjectFor methods.
michael@0:  *
michael@0:  * The template builder must call initializeForBuilding before the other
michael@0:  * methods, except for translateRef. The builder will then call compileQuery
michael@0:  * for each query in the template to compile the queries. When results need
michael@0:  * to be generated, the builder will call generateResults. The
michael@0:  * initializeForBuilding, compileQuery and addBinding methods may not be
michael@0:  * called after generateResults has been called until the builder indicates
michael@0:  * that the generated output is being removed by calling the done method.
michael@0:  *
michael@0:  * Currently, the datasource supplied to the methods will always be an
michael@0:  * nsIRDFDataSource or a DOM node, and will always be the same one in between
michael@0:  * calls to initializeForBuilding and done.
michael@0:  */
michael@0: [scriptable, uuid(C257573F-444F-468A-BA27-DE979DC55FE4)]
michael@0: interface nsIXULTemplateQueryProcessor : nsISupports
michael@0: {
michael@0:   /**
michael@0:    * Retrieve the datasource to use for the query processor. The list of
michael@0:    * datasources in a template is specified using the datasources attribute as
michael@0:    * a space separated list of URIs. This list is processed by the builder and
michael@0:    * supplied to the query processor in the aDataSources array as a list of
michael@0:    * nsIURI objects or nsIDOMNode objects. This method may return an object
michael@0:    * corresponding to these URIs and the builder will supply this object to
michael@0:    * other query processor methods. For example, for an XML source, the
michael@0:    * datasource might be an nsIDOMNode.
michael@0:    *
michael@0:    * All of these URIs are checked by the builder so it is safe to use them,
michael@0:    * however note that a URI that redirects may still needs to be checked to
michael@0:    * ensure that the document containing aRootNode may access it. This is the
michael@0:    * responsibility of the query processor if it needs to load the content of
michael@0:    * the URI.
michael@0:    *
michael@0:    * If the query processor needs to load the datasource asynchronously, it
michael@0:    * may set the aShouldDelayBuilding returned parameter to true to delay
michael@0:    * building the template content, and call the builder's Rebuild method when
michael@0:    * the data is available.
michael@0:    *
michael@0:    * @param aDataSources  the list of nsIURI objects and/or nsIDOMNode objects
michael@0:    * @param aRootNode     the root node the builder is attached to
michael@0:    * @param aIsTrusted    true if the template is in a trusted document
michael@0:    * @param aBuilder      the template builder
michael@0:    * @param aShouldDelayBuilding [out] whether the builder should wait to
michael@0:    *                                   build the content or not
michael@0:    * @returns a datasource object
michael@0:    */
michael@0:   nsISupports getDatasource(in nsIArray aDataSources,
michael@0:                             in nsIDOMNode aRootNode,
michael@0:                             in boolean aIsTrusted,
michael@0:                             in nsIXULTemplateBuilder aBuilder,
michael@0:                             out boolean aShouldDelayBuilding);
michael@0: 
michael@0:   /**
michael@0:    * Initialize for query generation. This will be called before the rules are
michael@0:    * processed and whenever the template is rebuilt. This method must be
michael@0:    * called once before any of the other query processor methods except for
michael@0:    * translateRef.
michael@0:    *
michael@0:    * @param aDatasource datasource for the data
michael@0:    * @param aBuilder the template builder
michael@0:    * @param aRootNode the root node the builder is attached to
michael@0:    *
michael@0:    * @throws NS_ERROR_INVALID_ARG if the datasource is not supported or
michael@0:    *         NS_ERROR_UNEXPECTED if generateResults has already been called.
michael@0:    */
michael@0:   void initializeForBuilding(in nsISupports aDatasource,
michael@0:                              in nsIXULTemplateBuilder aBuilder,
michael@0:                              in nsIDOMNode aRootNode);
michael@0: 
michael@0:   /**
michael@0:    * Called when the template builder is being destroyed so that the query
michael@0:    * processor can clean up any state. The query processor should remove as
michael@0:    * much state as possible, such as results or references to the builder.
michael@0:    * This method will also be called when the template is going to be rebuilt.
michael@0:    */
michael@0:   void done();
michael@0: 
michael@0:   /**
michael@0:    * Compile a query from a node. The result of this function will later be
michael@0:    * passed to generateResults for result generation. If null is returned,
michael@0:    * the query will be ignored.
michael@0:    *
michael@0:    * The template builder will call this method once for each query within
michael@0:    * the template, before any results can be generated using generateResults,
michael@0:    * but after initializeForBuilding has been called. This method should not
michael@0:    * be called again for the same query unless the template is rebuilt.
michael@0:    *
michael@0:    * The reference variable may be used by the query processor as a
michael@0:    * placeholder for the reference point, or starting point in the query.
michael@0:    *
michael@0:    * The member variable is determined from the member attribute on the
michael@0:    * template, or from the uri in the first action's rule if that attribute is
michael@0:    * not present. A rule processor may use the member variable as a hint to
michael@0:    * indicate what variable is expected to contain the results.
michael@0:    *
michael@0:    * @param aBuilder the template builder
michael@0:    * @param aQuery <query> node to compile
michael@0:    * @param aRefVariable the reference variable
michael@0:    * @param aMemberVariable the member variable
michael@0:    *
michael@0:    * @returns a compiled query object
michael@0:    */
michael@0:   nsISupports compileQuery(in nsIXULTemplateBuilder aBuilder,
michael@0:                            in nsIDOMNode aQuery,
michael@0:                            in nsIAtom aRefVariable,
michael@0:                            in nsIAtom aMemberVariable);
michael@0: 
michael@0:   /**
michael@0:    * Generate the results of a query and return them in an enumerator. The
michael@0:    * enumerator must contain nsIXULTemplateResult objects. If there are no
michael@0:    * results, an empty enumerator must be returned.
michael@0:    *
michael@0:    * The datasource will be the same as the one passed to the earlier
michael@0:    * initializeForBuilding method. The context reference (aRef) is a reference
michael@0:    * point used when calculating results.
michael@0:    *
michael@0:    * The value of aQuery must be the result of a previous call to compileQuery
michael@0:    * from this query processor. This method may be called multiple times,
michael@0:    * typically with different values for aRef.
michael@0:    *
michael@0:    * @param aDatasource datasource for the data
michael@0:    * @param aRef context reference value used as a starting point
michael@0:    * @param aQuery the compiled query returned from query compilation
michael@0:    *
michael@0:    * @returns an enumerator of nsIXULTemplateResult objects as the results
michael@0:    *
michael@0:    * @throws NS_ERROR_INVALID_ARG if aQuery is invalid
michael@0:    */
michael@0:   nsISimpleEnumerator generateResults(in nsISupports aDatasource,
michael@0:                                       in nsIXULTemplateResult aRef,
michael@0:                                       in nsISupports aQuery);
michael@0: 
michael@0:   /**
michael@0:    * Add a variable binding for a particular rule. A binding allows an
michael@0:    * additional variable to be set for a result, outside of those defined
michael@0:    * within the query. These bindings are always optional, in that they will
michael@0:    * never affect the results generated.
michael@0:    *
michael@0:    * This function will never be called after generateResults. Any bindings
michael@0:    * that were added should be applied to each result when the result's
michael@0:    * ruleMatched method is called, since the bindings are different for each
michael@0:    * rule.
michael@0:    *
michael@0:    * The reference aRef may be used to determine the reference when
michael@0:    * calculating the value for the binding, for example when a value should
michael@0:    * depend on the value of another variable.
michael@0:    *
michael@0:    * The syntax of the expression aExpr is defined by the query processor. If
michael@0:    * the syntax is invalid, the binding should be ignored. Only fatal errors
michael@0:    * should be thrown, or NS_ERROR_UNEXPECTED if generateResults has already
michael@0:    * been called.
michael@0:    *
michael@0:    * As an example, if the reference aRef is the variable '?count' which
michael@0:    * holds the value 5, and the expression aExpr is the string '+2', the value
michael@0:    * of the variable aVar would be 7, assuming the query processor considers
michael@0:    * the syntax '+2' to mean add two to the reference.
michael@0:    *
michael@0:    * @param aRuleNode rule to add the binding to
michael@0:    * @param aVar variable that will be bound
michael@0:    * @param aRef variable that holds reference value
michael@0:    * @param aExpr expression used to compute the value to assign
michael@0:    */
michael@0:   void addBinding(in nsIDOMNode aRuleNode,
michael@0:                   in nsIAtom aVar,
michael@0:                   in nsIAtom aRef,
michael@0:                   in AString aExpr);
michael@0: 
michael@0:   /**
michael@0:    * Translate a ref attribute string into a result. This is used as the
michael@0:    * reference point by the template builder when generating the first level
michael@0:    * of content. For recursive generation, the result from the parent
michael@0:    * generation phase will be used directly as the reference so a translation
michael@0:    * is not needed. This allows all levels to be generated using objects that
michael@0:    * all implement the nsIXULTemplateResult interface.
michael@0:    *
michael@0:    * This method may be called before initializeForBuilding, so the
michael@0:    * implementation may use the supplied datasource if it is needed to
michael@0:    * translate the reference.
michael@0:    *
michael@0:    * @param aDatasource datasource for the data
michael@0:    * @param aRefString the ref attribute string
michael@0:    *
michael@0:    * @return the translated ref
michael@0:    */
michael@0:   nsIXULTemplateResult translateRef(in nsISupports aDatasource,
michael@0:                                     in AString aRefString);
michael@0: 
michael@0:   /**
michael@0:    * Compare two results to determine their order, used when sorting results.
michael@0:    * This method should return -1 when the left result is less than the right,
michael@0:    * 0 if both are equivalent, and 1 if the left is greater than the right.
michael@0:    * The comparison should only consider the values for the specified
michael@0:    * variable.
michael@0:    *
michael@0:    * If the comparison variable is null, the results may be
michael@0:    * sorted in a natural order, for instance, based on the order the data in
michael@0:    * stored in the datasource.
michael@0:    *
michael@0:    * The sort hints are the flags in nsIXULSortService.
michael@0:    *
michael@0:    * This method must only be called with results that were created by this
michael@0:    * query processor.
michael@0:    *
michael@0:    * @param aLeft the left result to compare
michael@0:    * @param aRight the right result to compare
michael@0:    * @param aVar variable to compare
michael@0:    *
michael@0:    * @param returns -1 if less, 0 if equal, or 1 if greater
michael@0:    */
michael@0:    int32_t compareResults(in nsIXULTemplateResult aLeft,
michael@0:                           in nsIXULTemplateResult aRight,
michael@0:                           in nsIAtom aVar,
michael@0:                           in unsigned long aSortHints);
michael@0: };