feedprocessor.js

从国外的站点上淘的个人财务管理系统．开源．

/**
 * ElementInfo is a simple container object that describes
 * some characteristics of a feed element. For example, it
 * says whether an element can be expected to appear more
 * than once inside a given entry or feed.
 */ 
function ElementInfo(fieldName, containerClass, closeFunc, isArray) {
  this.fieldName = fieldName;
  this.containerClass = containerClass;
  this.closeFunc = closeFunc;
  this.isArray = isArray;
  this.isWrapper = false;
}

/**
 * FeedElementInfo represents a feed element, usually the root.
 */
function FeedElementInfo(fieldName, feedVersion) {
  this.isWrapper = false;
  this.fieldName = fieldName;
  this.feedVersion = feedVersion;
}

/**
 * Some feed formats include vestigial wrapper elements that we don't
 * want to include in our object model, but we do need to keep track
 * of during parsing.
 */
function WrapperElementInfo(fieldName) {
  this.isWrapper = true;
  this.fieldName = fieldName;
}

/***** The Processor *****/
function FeedProcessor() {
  this._reader = Cc[SAX_CONTRACTID].createInstance(Ci.nsISAXXMLReader);
  this._buf =  "";
  this._feed = Cc[BAG_CONTRACTID].createInstance(Ci.nsIWritablePropertyBag2);
  this._handlerStack = [];
  this._xmlBaseStack = []; // sparse array keyed to nesting depth
  this._depth = 0;
  this._state = "START";
  this._result = null;
  this._extensionHandler = null;
  this._xhtmlHandler = null;

  // The nsIFeedResultListener waiting for the parse results
  this.listener = null;

  // These elements can contain (X)HTML or plain text.
  // We keep a table here that contains their default treatment
  this._textConstructs = {"atom:title":"text",
                          "atom:summary":"text",
                          "atom:rights":"text",
                          "atom:content":"text",
                          "atom:subtitle":"text",
                          "description":"html",
                          "rss1:description":"html",
                          "dc:description":"html",
                          "content:encoded":"html",
                          "title":"text",
                          "rss1:title":"text",
                          "atom03:title":"text",
                          "atom03:tagline":"text",
                          "atom03:summary":"text",
                          "atom03:content":"text"};
  this._stack = [];

  this._trans = {   
    "START": {
      //If we hit a root RSS element, treat as RSS2.
      "rss": new FeedElementInfo("RSS2", "rss2"),

      // If we hit an RDF element, if could be RSS1, but we can't
      // verify that until we hit a rss1:channel element.
      "rdf:RDF": new WrapperElementInfo("RDF"),

      // If we hit a Atom 1.0 element, treat as Atom 1.0.
      "atom:feed": new FeedElementInfo("Atom", "atom"),

      // Treat as Atom 0.3
      "atom03:feed": new FeedElementInfo("Atom03", "atom03"),
    },
    
    /********* RSS2 **********/
    "IN_RSS2": {
      "channel": new WrapperElementInfo("channel")
    },

    "IN_CHANNEL": {
      "item": new ElementInfo("items", Cc[ENTRY_CONTRACTID], null, true),
      "managingEditor": new ElementInfo("authors", Cc[PERSON_CONTRACTID],
                                        rssAuthor, true),
      "dc:creator": new ElementInfo("authors", Cc[PERSON_CONTRACTID],
                                    rssAuthor, true),
      "dc:author": new ElementInfo("authors", Cc[PERSON_CONTRACTID],
                                   rssAuthor, true),
      "dc:contributor": new ElementInfo("contributors", Cc[PERSON_CONTRACTID],
                                         rssAuthor, true),
      "category": new ElementInfo("categories", null, rssCatTerm, true),
      "cloud": new ElementInfo("cloud", null, null, false),
      "image": new ElementInfo("image", null, null, false),
      "textInput": new ElementInfo("textInput", null, null, false),
      "skipDays": new ElementInfo("skipDays", null, null, false),
      "skipHours": new ElementInfo("skipHours", null, null, false),
      "generator": new ElementInfo("generator", Cc[GENERATOR_CONTRACTID],
                                   atomGenerator, false),
    },

    "IN_ITEMS": {
      "author": new ElementInfo("authors", Cc[PERSON_CONTRACTID],
                                rssAuthor, true),
      "dc:creator": new ElementInfo("authors", Cc[PERSON_CONTRACTID],
                                    rssAuthor, true),
      "dc:author": new ElementInfo("authors", Cc[PERSON_CONTRACTID],
                                   rssAuthor, true),
      "dc:contributor": new ElementInfo("contributors", Cc[PERSON_CONTRACTID],
                                         rssAuthor, true),
      "category": new ElementInfo("categories", null, rssCatTerm, true),
      "enclosure": new ElementInfo("enclosure", null, null, true),
      "guid": new ElementInfo("guid", null, rssGuid, false)
    },

    "IN_SKIPDAYS": {
      "day": new ElementInfo("days", null, rssArrayElement, true)
    },

    "IN_SKIPHOURS":{
      "hour": new ElementInfo("hours", null, rssArrayElement, true)
    },

    /********* RSS1 **********/
    "IN_RDF": {
      // If we hit a rss1:channel, we can verify that we have RSS1
      "rss1:channel": new FeedElementInfo("rdf_channel", "rss1"),
      "rss1:image": new ElementInfo("image", null, null, false),
      "rss1:textinput": new ElementInfo("textInput", null, null, false),
      "rss1:item": new ElementInfo("items", Cc[ENTRY_CONTRACTID], null, true),
    },

    "IN_RDF_CHANNEL": {
      "admin:generatorAgent": new ElementInfo("generator",
                                              Cc[GENERATOR_CONTRACTID],
                                              null, false),
      "dc:creator": new ElementInfo("authors", Cc[PERSON_CONTRACTID],
                                    rssAuthor, true),
      "dc:author": new ElementInfo("authors", Cc[PERSON_CONTRACTID],
                                   rssAuthor, true),
      "dc:contributor": new ElementInfo("contributors", Cc[PERSON_CONTRACTID],
                                         rssAuthor, true),
    },

    /********* ATOM 1.0 **********/
    "IN_ATOM": {
      "atom:author": new ElementInfo("authors", Cc[PERSON_CONTRACTID],
                                     null, true),
      "atom:generator": new ElementInfo("generator", Cc[GENERATOR_CONTRACTID],
                                        atomGenerator, false),
      "atom:contributor": new ElementInfo("contributors",  Cc[PERSON_CONTRACTID],
                                          null, true),
      "atom:link": new ElementInfo("links", null, null, true),
      "atom:entry": new ElementInfo("entries", Cc[ENTRY_CONTRACTID],
                                    null, true)
    },

    "IN_ENTRIES": {
      "atom:author": new ElementInfo("authors", Cc[PERSON_CONTRACTID],
                                     null, true),
      "atom:contributor": new ElementInfo("contributors", Cc[PERSON_CONTRACTID],
                                          null, true),
      "atom:link": new ElementInfo("links", null, null, true),
    },

    /********* ATOM 0.3 **********/
    "IN_ATOM03": {
      "atom03:author": new ElementInfo("authors", Cc[PERSON_CONTRACTID],
                                       null, true),
      "atom03:contributor": new ElementInfo("contributors",
                                            Cc[PERSON_CONTRACTID],
                                            null, true),
      "atom03:link": new ElementInfo("links", null, null, true),
      "atom03:entry": new ElementInfo("atom03_entries", Cc[ENTRY_CONTRACTID],
                                      null, true),
      "atom03:generator": new ElementInfo("generator", Cc[GENERATOR_CONTRACTID],
                                          atomGenerator, false),
    },

    "IN_ATOM03_ENTRIES": {
      "atom03:author": new ElementInfo("authors", Cc[PERSON_CONTRACTID],
                                       null, true),
      "atom03:contributor": new ElementInfo("contributors",
                                            Cc[PERSON_CONTRACTID],
                                            null, true),
      "atom03:link": new ElementInfo("links", null, null, true),
      "atom03:entry": new ElementInfo("atom03_entries", Cc[ENTRY_CONTRACTID],
                                      null, true)
    }
  }
}

// See startElement for a long description of how feeds are processed.
FeedProcessor.prototype = { 
  
  // Set ourselves as the SAX handler, and set the base URI
  _init: function FP_init(uri) {
    this._reader.contentHandler = this;
    this._reader.errorHandler = this;
    this._result = Cc[FR_CONTRACTID].createInstance(Ci.nsIFeedResult);
    if (uri) {
      this._result.uri = uri;
      this._reader.baseURI = uri;
      this._xmlBaseStack[0] = uri;
    }
  },

  // This function is called once we figure out what type of feed
  // we're dealing with. Some feed types require digging a bit further
  // than the root.
  _docVerified: function FP_docVerified(version) {
    this._result.doc = Cc[FEED_CONTRACTID].createInstance(Ci.nsIFeed);
    this._result.doc.baseURI = 
      this._xmlBaseStack[this._xmlBaseStack.length - 1];
    this._result.doc.fields = this._feed;
    this._result.version = version;
  },

  // When we're done with the feed, let the listener know what
  // happened.
  _sendResult: function FP_sendResult() {
    try {
      // Can be null when a non-feed is fed to us
      if (this._result.doc)
        this._result.doc.normalize();
    }
    catch (e) {
      LOG("FIXME: " + e);
    }

    try {
      if (this.listener != null)
        this.listener.handleResult(this._result);
    }
    finally {
      this._result = null;
      this._reader = null;
    }
  },

  // Parsing functions
  parseFromStream: function FP_parseFromStream(stream, uri) {
    this._init(uri);
    this._reader.parseFromStream(stream, null, stream.available(), 
                                 "application/xml");
    this._reader = null;
  },

  parseFromString: function FP_parseFromString(inputString, uri) {
    this._init(uri);
    this._reader.parseFromString(inputString, "application/xml");
    this._reader = null;
  },

  parseAsync: function FP_parseAsync(requestObserver, uri) {
    this._init(uri);
    this._reader.parseAsync(requestObserver);
  },

  // nsIStreamListener 

  // The XMLReader will throw sensible exceptions if these get called
  // out of order.
  onStartRequest: function FP_onStartRequest(request, context) {
    this._reader.onStartRequest(request, context);
  },

  onStopRequest: function FP_onStopRequest(request, context, statusCode) {
    this._reader.onStopRequest(request, context, statusCode);
  },

  onDataAvailable:
  function FP_onDataAvailable(request, context, inputStream, offset, count) {
    this._reader.onDataAvailable(request, context, inputStream, offset, count);
  },

  // nsISAXErrorHandler

  // We only care about fatal errors. When this happens, we may have
  // parsed through the feed metadata and some number of entries. The
  // listener can still show some of that data if it wants, and we'll
  // set the bozo bit to indicate we were unable to parse all the way
  // through.
  fatalError: function FP_reportError() {
    this._result.bozo = true;
    //XXX need to QI to FeedProgressListener
    this._sendResult();
  },

  // nsISAXContentHandler

  startDocument: function FP_startDocument() {
    //LOG("----------");
  },

  endDocument: function FP_endDocument() {
    this._sendResult();
  },

  // The transitions defined above identify elements that contain more
  // than just text. For example RSS items contain many fields, and so
  // do Atom authors. The only commonly used elements that contain
  // mixed content are Atom Text Constructs of type="xhtml", which we
  // delegate to another handler for cleaning. That leaves a couple
  // different types of elements to deal with: those that should occur
  // only once, such as title elements, and those that can occur
  // multiple times, such as the RSS category element and the Atom
  // link element. Most of the RSS1/DC elements can occur multiple
  // times in theory, but in practice, the only ones that do have
  // analogues in Atom. 
  //
  // Some elements are also groups of attributes or sub-elements,
  // while others are simple text fields. For the most part, we don't
  // have to pay explicit attention to the simple text elements,
  // unless we want to post-process the resulting string to transform
  // it into some richer object like a Date or URI.
  //
  // Elements that have more sophisticated content models still end up
  // being dictionaries, whether they are based on attributes like RSS
  // cloud, sub-elements like Atom author, or even items and
  // entries. These elements are treated as "containers". It's
  // theoretically possible for a container to have an attribute with 
  // the same universal name as a sub-element, but none of the feed
  // formats allow this by default, and I don't of any extension that
  // works this way.
  //
  startElement: function FP_startElement(uri, localName, qName, attributes) {
    this._buf = "";
    ++this._depth;
    var elementInfo;

    //LOG("<" + localName + ">");

    // Check for xml:base
    var base = attributes.getValueFromName(XMLNS, "base");
    if (base) {
      this._xmlBaseStack[this._depth] =
        strToURI(base, this._xmlBaseStack[this._xmlBaseStack.length - 1]);
    }

    // To identify the element we're dealing with, we look up the
    // namespace URI in our gNamespaces dictionary, which will give us
    // a "canonical" prefix for a namespace URI. For example, this
    // allows Dublin Core "creator" elements to be consistently mapped
    // to "dc:creator", for easy field access by consumer code. This
    // strategy also happens to shorten up our state table.
    var key =  this._prefixForNS(uri) + localName;

    // Check to see if we need to hand this off to our XHTML handler.
    // The elements we're dealing with will look like this:
    // 
    // <title type="xhtml">
    //   <div xmlns="http://www.w3.org/1999/xhtml">
    //     A title with <b>bold</b> and <i>italics</i>.
    //   </div>
    // </title>
    //
    // When it returns in returnFromXHTMLHandler, the handler should
    // give us back a string like this: 
    // 
    //    "A title with <b>bold</b> and <i>italics</i>."
    //
    // The Atom spec explicitly says the div is not part of the content,
    // and explicitly allows whitespace collapsing.
    // 
    if ((this._result.version == "atom" || this._result.version == "atom03") &&
        this._textConstructs[key] != null) {
      var type = attributes.getValueFromName("","type");
      if (type != null && type.indexOf("xhtml") >= 0) {
        this._xhtmlHandler = 
          new XHTMLHandler(this, (this._result.version == "atom"));
        this._reader.contentHandler = this._xhtmlHandler;
        return;
      }
    }

    // Check our current state, and see if that state has a defined
    // transition. For example, this._trans["atom:entry"]["atom:author"]
    // will have one, and it tells us to add an item to our authors array.
    if (this._trans[this._state] && this._trans[this._state][key]) {
      elementInfo = this._trans[this._state][key];
    }
    else {
      // If we don't have a transition, hand off to extension handler
      this._extensionHandler = new ExtensionHandler(this);
      this._reader.contentHandler = this._extensionHandler;
      this._extensionHandler.startElement(uri, localName, qName, attributes);
      return;
    }
      
    // This distinguishes wrappers like 'channel' from elements
    // we'd actually like to do something with (which will test true).
    this._handlerStack[this._depth] = elementInfo; 
    if (elementInfo.isWrapper) {
      this._state = "IN_" + elementInfo.fieldName.toUpperCase();
      this._stack.push([this._feed, this._state]);
    } 
    else if (elementInfo.feedVersion) {
      this._state = "IN_" + elementInfo.fieldName.toUpperCase();

      // Check for the older RSS2 variants
      if (elementInfo.feedVersion == "rss2")
        elementInfo.feedVersion = this._findRSSVersion(attributes);
      else if (uri == RSS090NS)
        elementInfo.feedVersion = "rss090";

      this._docVerified(elementInfo.feedVersion);
      this._stack.push([this._feed, this._state]);
      this._mapAttributes(this._feed, attributes);
    }
    else {
      this._state = this._processComplexElement(elementInfo, attributes);
    }
  },

  // In the endElement handler, we decrement the stack and look
  // for cleanup/transition functions to execute. The second part
  // of the state transition works as above in startElement, but
  // the state we're looking for is prefixed with an underscore
  // to distinguish endElement events from startElement events.
  endElement:  function FP_endElement(uri, localName, qName) {
    var elementInfo = this._handlerStack[this._depth];
    //LOG("</" + localName + ">");
    if (elementInfo && !elementInfo.isWrapper)
      this._closeComplexElement(elementInfo);
  
    // cut down xml:base context
    if (this._xmlBaseStack.length == this._depth + 1)
      this._xmlBaseStack = this._xmlBaseStack.slice(0, this._depth);

    // our new state is whatever is at the top of the stack now
    if (this._stack.length > 0)
      this._state = this._stack[this._stack.length - 1][1];
    this._handlerStack = this._handlerStack.slice(0, this._depth);
    --this._depth;
  },

  // Buffer up character data. The buffer is cleared with every
  // opening element.
  characters: function FP_characters(data) {
    this._buf += data;