feedprocessor.js

从国外的站点上淘的个人财务管理系统．开源．

      
      if (bagHasKey(guid, "isPermaLink"))
        isPermaLink = new Boolean(guid.getProperty("isPermaLink"));
      
      if (guid && isPermaLink)
        this.link = strToURI(guid.getProperty("guid"));
    }

    if (this.updated)
      this.updated = dateParse(this.updated);
    if (this.published)
      this.published = dateParse(this.published);

    this._resetBagMembersToRawText([this.searchLists.content, 
                                    this.searchLists.summary, 
                                    this.searchLists.title]);
  },
  
  QueryInterface: function(iid) {
    if (iid.equals(Ci.nsIFeedEntry) ||
        iid.equals(Ci.nsIFeedContainer) ||
        iid.equals(Ci.nsISupports))
    return this;

    throw Cr.NS_ERROR_NOINTERFACE;
  }
}

Entry.prototype._atomLinksToURI = Feed.prototype._atomLinksToURI;
Entry.prototype._resetBagMembersToRawText = 
   Feed.prototype._resetBagMembersToRawText;

// TextConstruct represents and element that could contain (X)HTML
function TextConstruct() {
  this.lang = null;
  this.base = null;
  this.type = "text";
  this.text = null;
}

TextConstruct.prototype = {
  plainText: function TC_plainText() {
    if (this.type != "text") {
      return gUnescapeHTML.unescape(stripTags(this.text));
    }
    return this.text;
  },

  createDocumentFragment: function TC_createDocumentFragment(element) {
    if (this.type == "text") {
      var doc = element.ownerDocument;
      var docFragment = doc.createDocumentFragment();
      var node = doc.createTextNode(this.text);
      docFragment.appendChild(node);
      return docFragment;
    }
    var isXML;
    if (this.type == "xhtml")
      isXML = true
    else if (this.type == "html")
      isXML = false;
    else
      return null;

    return gUnescapeHTML.parseFragment(this.text, isXML, this.base, element);
  },
 
  QueryInterface: function(iid) {
    if (iid.equals(Ci.nsIFeedTextConstruct) ||
        iid.equals(Ci.nsISupports))
    return this;

    throw Cr.NS_ERROR_NOINTERFACE;
  }
}

// Generator represents the software that produced the feed
function Generator() {
  this.lang = null;
  this.agent = null;
  this.version = null;
  this.uri = null;

  // nsIFeedElementBase
  this._attributes = null;
  this.baseURI = null;
}

Generator.prototype = {

  get attributes() {
    return this._attributes;
  },

  set attributes(value) {
    this._attributes = value;
    this.version = this._attributes.getValueFromName("","version");
    var uriAttribute = this._attributes.getValueFromName("","uri") ||
                       this._attributes.getValueFromName("","url");
    this.uri = strToURI(uriAttribute, this.baseURI);

    // RSS1
    uriAttribute = this._attributes.getValueFromName(RDF_NS,"resource");
    if (uriAttribute) {
      this.agent = uriAttribute;
      this.uri = strToURI(uriAttribute, this.baseURI);
    }
  },

  QueryInterface: function(iid) {
    if (iid.equals(Ci.nsIFeedGenerator) ||
        iid.equals(Ci.nsIFeedElementBase) ||
        iid.equals(Ci.nsISupports))
    return this;

    throw Cr.NS_ERROR_NOINTERFACE;
  }
}

function Person() {
  this.name = null;
  this.uri = null;
  this.email = null;

  // nsIFeedElementBase
  this.attributes = null;
  this.baseURI = null;
}

Person.prototype = {
  QueryInterface: function(iid) {
    if (iid.equals(Ci.nsIFeedPerson) ||
        iid.equals(Ci.nsIFeedElementBase) ||
        iid.equals(Ci.nsISupports))
    return this;

    throw Cr.NS_ERROR_NOINTERFACE;
  }
}

/** 
 * Map a list of fields into properties on a container.
 *
 * @param container An nsIFeedContainer
 * @param fields A list of fields to search for. List members can
 *               be a list, in which case the second member is 
 *               transformation function (like parseInt).
 */
function fieldsToObj(container, fields) {
  var props,prop,field,searchList;
  for (var key in fields) {
    searchList = fields[key];
    for (var i=0; i < searchList.length; ++i) {
      props = searchList[i];
      prop = null;
      field = isArray(props) ? props[0] : props;
      try {
        prop = container.fields.getProperty(field);
      } 
      catch(e) { 
      }
      if (prop) {
        prop = isArray(props) ? props[1](prop) : prop;
        container[key] = prop;
      }
    }
  }
}

/**
 * Lower cases an element's localName property
 * @param   element A DOM element.
 *
 * @returns The lower case localName property of the specified element
 */
function LC(element) {
  return element.localName.toLowerCase();
}

// TODO move these post-processor functions
// create a generator element
function atomGenerator(s, generator) {
  generator.QueryInterface(Ci.nsIFeedGenerator);
  generator.agent = trimString(s);
  return generator;
} 

// post-process an RSS category, map it to the Atom fields.
function rssCatTerm(s, cat) {
  // add slash handling?
  cat.setPropertyAsAString("term", trimString(s));
  return cat;
} 

// post-process a GUID 
function rssGuid(s, guid) {
  guid.setPropertyAsAString("guid", trimString(s));
  return guid;
}

// post-process an RSS author element
//
// It can contain a field like this:
// 
//  <author>[e-mail] (Lawyer Boyer)</author>
//
// or, delightfully, a field like this:
//
//  <dc:creator>Simon St.Laurent (mailto:[e-mail])</dc:creator>
//
// We want to split this up and assign it to corresponding Atom
// fields.
//
function rssAuthor(s,author) {
  author.QueryInterface(Ci.nsIFeedPerson);
  // check for RSS2 string format
  var chars = trimString(s);
  var matches = chars.match(/(.*)\((.*)\)/);
  var emailCheck = 
    /^([a-zA-Z0-9_\.\-])+\@(([a-zA-Z0-9\-])+\.)+([a-zA-Z0-9]{2,4})+$/;
  if (matches) {
    var match1 = trimString(matches[1]);
    var match2 = trimString(matches[2]);
    if (match2.indexOf("mailto:") == 0)
      match2 = match2.substring(7);
    if (emailCheck.test(match1)) {
      author.email = match1;
      author.name = match2;
    }
    else if (emailCheck.test(match2)) {
      author.email = match2;
      author.name = match1;
    }
    else {
      // put it back together
      author.name = match1 + " (" + match2 + ")";
    }
  }
  else {
    author.name = chars;
    if (chars.indexOf('@'))
      author.email = chars;
  }
  return author;
}

//
// skipHours and skipDays map to arrays, so we need to change the
// string to an nsISupports in order to stick it in there.
//
function rssArrayElement(s) {
  var str = Cc["@mozilla.org/supports-string;1"].
              createInstance(Ci.nsISupportsString);
  str.data = s;
  str.QueryInterface(Ci.nsISupportsString);
  return str;
}

/***** Some feed utils from TBird *****/

/**
 * Tests a RFC822 date against a regex.
 * @param aDateStr A string to test as an RFC822 date.
 *
 * @returns A boolean indicating whether the string is a valid RFC822 date.
 */
function isValidRFC822Date(aDateStr) {
  var regex = new RegExp(RFC822_RE);
  return regex.test(aDateStr);
}

/**
 * Removes leading and trailing whitespace from a string.
 * @param s The string to trim.
 *
 * @returns A new string with whitespace stripped.
 */
function trimString(s) {
  return(s.replace(/^\s+/, "").replace(/\s+$/, ""));
}

// Regular expression matching RFC822 dates 
const RFC822_RE = "^(((Mon)|(Tue)|(Wed)|(Thu)|(Fri)|(Sat)|(Sun)), *)?\\d\\d?"
+ " +((Jan)|(Feb)|(Mar)|(Apr)|(May)|(Jun)|(Jul)|(Aug)|(Sep)|(Oct)|(Nov)|(Dec))"
+ " +\\d\\d(\\d\\d)? +\\d\\d:\\d\\d(:\\d\\d)? +(([+-]?\\d\\d\\d\\d)|(UT)|(GMT)"
+ "|(EST)|(EDT)|(CST)|(CDT)|(MST)|(MDT)|(PST)|(PDT)|\\w)$";

/**
 * XXX -- need to decide what this should return. 
 * XXX -- Is there a Date class usable from C++?
 *
 * Tries tries parsing various date formats.
 * @param dateString
 *        A string that is supposedly an RFC822 or RFC3339 date.
 * @returns A Date.toString XXX--fixme
 */
function dateParse(dateString) {
  var date = trimString(dateString);

  if (date.search(/^\d\d\d\d/) != -1) //Could be a ISO8601/W3C date
    return W3CToIETFDate(dateString);

  if (isValidRFC822Date(date))
    return date; 
  
  if (!isNaN(parseInt(date, 10))) { 
    //It's an integer, so maybe it's a timestamp
    var d = new Date(parseInt(date, 10) * 1000);
    var now = new Date();
    var yeardiff = now.getFullYear() - d.getFullYear();
    if ((yeardiff >= 0) && (yeardiff < 3)) {
      // it's quite likely the correct date. 3 years is an arbitrary cutoff,
      // but this is an invalid date format, and there's no way to verify
      // its correctness.
      return d.toString();
    }
  }
  // Can't help.
  return null;
} 


const XHTML_NS = "http://www.w3.org/1999/xhtml";

// The XHTMLHandler handles inline XHTML found in things like atom:summary
function XHTMLHandler(processor, isAtom) {
  this._buf = "";
  this._processor = processor;
  this._depth = 0;
  this._isAtom = isAtom;
}

// The fidelity can be improved here, to allow handling of stuff like
// SVG and MathML. XXX
XHTMLHandler.prototype = {
  startDocument: function XH_startDocument() {
  },
  endDocument: function XH_endDocument() {
  },
  startElement: function XH_startElement(uri, localName, qName, attributes) {
    ++this._depth;

    // RFC4287 requires XHTML to be wrapped in a div that is *not* part of 
    // the content. This prevents people from screwing up namespaces, but
    // we need to skip it here.
    if (this._isAtom && this._depth == 1 && localName == "div")
      return;

    // If it's an XHTML element, record it. Otherwise, it's ignored.
    if (uri == XHTML_NS) {
      this._buf += "<" + localName;
      for (var i=0; i < attributes.length; ++i) {
        // XHTML attributes aren't in a namespace
        if (attributes.getURI(i) == "") { 
          this._buf += (" " + attributes.getLocalName(i) + "='" +
                        xmlEscape(attributes.getValue(i)) + "'");
        }
      }
      this._buf += ">";
    }
  },
  endElement: function XH_endElement(uri, localName, qName) {
    --this._depth;
    
    // We need to skip outer divs in Atom. See comment in startElement.
    if (this._isAtom && this._depth == 0 && localName == "div")
      return;

    // When we peek too far, go back to the main processor
    if (this._depth < 0) {
      this._processor.returnFromXHTMLHandler(trimString(this._buf),
                                             uri, localName, qName);
      return;
    }
    // If it's an XHTML element, record it. Otherwise, it's ignored.
    if (uri == XHTML_NS) {
      this._buf += "</" + localName + ">";
    }
  },
  characters: function XH_characters(data) {
    this._buf += xmlEscape(data);
  },
  startPrefixMapping: function XH_startPrefixMapping() {
  },
  endPrefixMapping: function XH_endPrefixMapping() {
  },
  processingInstruction: function XH_processingInstruction() {
  }, 
}

/**
 * The ExtensionHandler deals with elements we haven't explicitly
 * added to our transition table in the FeedProcessor.
 */
function ExtensionHandler(processor) {
  this._buf = "";
  this._depth = 0;
  this._hasChildElements = false;

  // The FeedProcessor
  this._processor = processor;

  // Fields of the outermost extension element.
  this._localName = null;
  this._uri = null;
  this._qName = null;
  this._attrs = null;
}

ExtensionHandler.prototype = {
  startDocument: function EH_startDocument() {
  },
  endDocument: function EH_endDocument() {
  },
  startElement: function EH_startElement(uri, localName, qName, attrs) {
    ++this._depth;
    var prefix = gNamespaces[uri] ? gNamespaces[uri] + ":" : "";
    var key =  prefix + localName;
    
    if (this._depth == 1) {
      this._uri = uri;
      this._localName = localName;
      this._qName = qName;
      this._attrs = attrs;
    }
    
    // if we descend into another element, we won't send text
    this._hasChildElements = (this._depth > 1);
    
  },
  endElement: function EH_endElement(uri, localName, qName) {
    --this._depth;
    if (this._depth == 0) {
      var text = this._hasChildElements ? null : trimString(this._buf);
      this._processor.returnFromExtHandler(this._uri, this._localName, 
                                           text, this._attrs);
    }
  },
  characters: function EH_characters(data) {
    if (!this._hasChildElements)
      this._buf += data;
  },
  startPrefixMapping: function EH_startPrefixMapping() {
  },
  endPrefixMapping: function EH_endPrefixMapping() {
  },
  processingInstruction: function EH_processingInstruction() {
  }, 
};