microformats.js

this is Microformats.js for linux

        /* when contained in hCard are a special case */
        var parentnode;
        var node = propnodes[i];
        var xpathExpression = "";
        for (let j=0; j < Microformats.list.length; j++) {
          /* Don't treat adr or geo in an hCard as a microformat in this case */
          if ((mfname == "hCard") && ((Microformats.list[j] == "adr") || (Microformats.list[j] == "geo"))) {
            continue;
          }
          if (Microformats[Microformats.list[j]].className) {
            if (xpathExpression.length == 0) {
              xpathExpression = "ancestor::*[";
            } else {
              xpathExpression += " or ";
            }
            xpathExpression += "contains(concat(' ', @class, ' '), ' " + Microformats[Microformats.list[j]].className + " ')";
          }
        }
        xpathExpression += "][1]";
        var xpathResult = (node.ownerDocument || node).evaluate(xpathExpression, node, null,  Components.interfaces.nsIDOMXPathResult.FIRST_ORDERED_NODE_TYPE, null);
        if (xpathResult.singleNodeValue) {
          xpathResult.singleNodeValue.microformat = mfname;
          parentnode = xpathResult.singleNodeValue;
        }
        /* If the propnode is not a child of the microformat, and */
        /* the property belongs to the parent microformat as well, */
        /* remove it. */
        if (parentnode != mfnode) {
          var mfNameString = Microformats.getNamesFromNode(parentnode);
          var mfNames = mfNameString.split(" ");
          var j;
          for (j=0; j < mfNames.length; j++) {
            /* If this property is in the parent microformat, remove the node  */
            if (Microformats[mfNames[j]].properties[propname]) {
              propnodes.splice(i,1);;
              break;
            }
          }
        }
      }
      if (propnodes.length > 0) {
        var resultArray = [];
        for (let i = 0; i < propnodes.length; i++) {
          var subresult = Microformats.parser.getPropertyInternal(propnodes[i],
                                                                  mfnode,
                                                                  propobj,
                                                                  propname);
          if (subresult != undefined) {
            resultArray.push(subresult);
            /* If we're not a plural property, don't bother getting more */
            if (!propobj.plural) {
              return resultArray[0];
            }
          }
        }
        if (resultArray.length > 0) {
          return resultArray;
        }
      } else {
        /* If we didn't find any class nodes, check to see if this property */
        /* is virtual and if so, call getPropertyInternal again */
        if (propobj.virtual) {
          return Microformats.parser.getPropertyInternal(mfnode, null,
                                                         propobj, propname);
        }
      }
      return;
    },
    /**
     * Internal parser API used to resolve includes and headers. Includes are
     * resolved by simply cloning the node and replacing it in a clone of the
     * original DOM node. Headers are resolved by creating a span and then copying
     * the innerHTML and the class name.
     *
     * @param  in_mfnode The node to preProcess.
     * @return If the node had includes or headers, a cloned node otherwise
     *         the original node. You can check to see if the node was cloned
     *         by looking for .origNode in the new node.
     */
    preProcessMicroformat: function preProcessMicroformat(in_mfnode) {
      var mfnode;
      var includes = Microformats.getElementsByClassName(in_mfnode, "include");
      if ((includes.length > 0) || ((in_mfnode.nodeName.toLowerCase() == "td") && (in_mfnode.getAttribute("headers")))) {
        mfnode = in_mfnode.cloneNode(true);
        mfnode.origNode = in_mfnode;
        if (includes.length > 0) {
          includes = Microformats.getElementsByClassName(mfnode, "include");
          var includeId;
          var include_length = includes.length;
          for (let i = include_length -1; i >= 0; i--) {
            if (includes[i].nodeName.toLowerCase() == "a") {
              includeId = includes[i].getAttribute("href").substr(1);
            }
            if (includes[i].nodeName.toLowerCase() == "object") {
              includeId = includes[i].getAttribute("data").substr(1);
            }
            if (in_mfnode.ownerDocument.getElementById(includeId)) {
              includes[i].parentNode.replaceChild(in_mfnode.ownerDocument.getElementById(includeId).cloneNode(true), includes[i]);
            }
          }
        } else {
          var headers = in_mfnode.getAttribute("headers").split(" ");
          for (let i = 0; i < headers.length; i++) {
            var tempNode = in_mfnode.ownerDocument.createElement("span");
            var headerNode = in_mfnode.ownerDocument.getElementById(headers[i]);
            if (headerNode) {
              tempNode.innerHTML = headerNode.innerHTML;
              tempNode.className = headerNode.className;
              mfnode.appendChild(tempNode);
            }
          }
        }
      } else {
        mfnode = in_mfnode;
      }
      return mfnode;
    },
    validate: function validate(mfnode, mfname) {
      var error = "";
      if (Microformats[mfname].validate) {
        return Microformats[mfname].validate(mfnode);
      } else if (Microformats[mfname].required) {
        for (let i=0;i<Microformats[mfname].required.length;i++) {
          if (!Microformats.parser.getMicroformatProperty(mfnode, mfname, Microformats[mfname].required[i])) {
            error += "Required property " + Microformats[mfname].required[i] + " not specified\n";
          }
        }
        if (error.length > 0) {
          throw(error);
        }
        return true;
      }
    },
    /* This function normalizes an ISO8601 date by adding punctuation and */
    /* ensuring that hours and seconds have values */
    normalizeISO8601: function normalizeISO8601(string)
    {
      var dateArray = string.match(/(\d\d\d\d)(?:-?(\d\d)(?:-?(\d\d)(?:[T ](\d\d)(?::?(\d\d)(?::?(\d\d)(?:\.(\d+))?)?)?(?:([-+Z])(?:(\d\d)(?::?(\d\d))?)?)?)?)?)?/);
  
      var dateString;
      var tzOffset = 0;
      if (!dateArray) {
        return;
      }
      if (dateArray[1]) {
        dateString = dateArray[1];
        if (dateArray[2]) {
          dateString += "-" + dateArray[2];
          if (dateArray[3]) {
            dateString += "-" + dateArray[3];
            if (dateArray[4]) {
              dateString += "T" + dateArray[4];
              if (dateArray[5]) {
                dateString += ":" + dateArray[5];
              } else {
                dateString += ":" + "00";
              }
              if (dateArray[6]) {
                dateString += ":" + dateArray[6];
              } else {
                dateString += ":" + "00";
              }
              if (dateArray[7]) {
                dateString += "." + dateArray[7];
              }
              if (dateArray[8]) {
                dateString += dateArray[8];
                if ((dateArray[8] == "+") || (dateArray[8] == "-")) {
                  if (dateArray[9]) {
                    dateString += dateArray[9];
                    if (dateArray[10]) {
                      dateString += dateArray[10];
                    }
                  }
                }
              }
            }
          }
        }
      }
      return dateString;
    }
  },
  /**
   * Converts an ISO8601 date into a JavaScript date object, honoring the TZ
   * offset and Z if present to convert the date to local time
   * NOTE: I'm using an extra parameter on the date object for this function.
   * I set date.time to true if there is a date, otherwise date.time is false.
   * 
   * @param  string ISO8601 formatted date
   * @return JavaScript date object that represents the ISO date. 
   */
  dateFromISO8601: function dateFromISO8601(string) {
    var dateArray = string.match(/(\d\d\d\d)(?:-?(\d\d)(?:-?(\d\d)(?:[T ](\d\d)(?::?(\d\d)(?::?(\d\d)(?:\.(\d+))?)?)?(?:([-+Z])(?:(\d\d)(?::?(\d\d))?)?)?)?)?)?/);
  
    var date = new Date(dateArray[1], 0, 1);
    date.time = false;

    if (dateArray[2]) {
      date.setMonth(dateArray[2] - 1);
    }
    if (dateArray[3]) {
      date.setDate(dateArray[3]);
    }
    if (dateArray[4]) {
      date.setHours(dateArray[4]);
      date.time = true;
      if (dateArray[5]) {
        date.setMinutes(dateArray[5]);
        if (dateArray[6]) {
          date.setSeconds(dateArray[6]);
          if (dateArray[7]) {
            date.setMilliseconds(Number("0." + dateArray[7]) * 1000);
          }
        }
      }
    }
    if (dateArray[8]) {
      if (dateArray[8] == "-") {
        if (dateArray[9] && dateArray[10]) {
          date.setHours(date.getHours() + parseInt(dateArray[9], 10));
          date.setMinutes(date.getMinutes() + parseInt(dateArray[10], 10));
        }
      } else if (dateArray[8] == "+") {
        if (dateArray[9] && dateArray[10]) {
          date.setHours(date.getHours() - parseInt(dateArray[9], 10));
          date.setMinutes(date.getMinutes() - parseInt(dateArray[10], 10));
        }
      }
      /* at this point we have the time in gmt */
      /* convert to local if we had a Z - or + */
      if (dateArray[8]) {
        var tzOffset = date.getTimezoneOffset();
        if (tzOffset < 0) {
          date.setMinutes(date.getMinutes() + tzOffset); 
        } else if (tzOffset > 0) {
          date.setMinutes(date.getMinutes() - tzOffset); 
        }
      }
    }
    return date;
  },
  /**
   * Converts a Javascript date object into an ISO 8601 formatted date
   * NOTE: I'm using an extra parameter on the date object for this function.
   * If date.time is NOT true, this function only outputs the date.
   * 
   * @param  date        Javascript Date object
   * @param  punctuation true if the date should have -/:
   * @return string with the ISO date. 
   */
  iso8601FromDate: function iso8601FromDate(date, punctuation) {
    var string = date.getFullYear().toString();
    if (punctuation) {
      string += "-";
    }
    string += (date.getMonth() + 1).toString().replace(/\b(\d)\b/g, '0$1');
    if (punctuation) {
      string += "-";
    }
    string += date.getDate().toString().replace(/\b(\d)\b/g, '0$1');
    if (date.time) {
      string += "T";
      string += date.getHours().toString().replace(/\b(\d)\b/g, '0$1');
      if (punctuation) {
        string += ":";
      }
      string += date.getMinutes().toString().replace(/\b(\d)\b/g, '0$1');
      if (punctuation) {
        string += ":";
      }
      string += date.getSeconds().toString().replace(/\b(\d)\b/g, '0$1');
      if (date.getMilliseconds() > 0) {
        if (punctuation) {
          string += ".";
        }
        string += date.getMilliseconds().toString();
      }
    }
    return string;
  },
  simpleEscape: function simpleEscape(s)
  {
    s = s.replace(/\&/g, '%26');
    s = s.replace(/\#/g, '%23');
    s = s.replace(/\+/g, '%2B');
    s = s.replace(/\-/g, '%2D');
    s = s.replace(/\=/g, '%3D');
    s = s.replace(/\'/g, '%27');
    s = s.replace(/\,/g, '%2C');
//    s = s.replace(/\r/g, '%0D');
//    s = s.replace(/\n/g, '%0A');
    s = s.replace(/ /g, '+');
    return s;
  },
  /**
   * Not intended for external consumption. Microformat implementations might use it.
   *
   * Retrieve elements matching all classes listed in a space-separated string.
   * I had to implement my own because I need an Array, not an nsIDomNodeList
   * 
   * @param  rootElement      The DOM element at which to start searching (optional)
   * @param  className        A space separated list of classenames
   * @return microformatNodes An array of DOM Nodes, each representing a
                              microformat in the document.
   */
  getElementsByClassName: function getElementsByClassName(rootNode, className)
  {
    var returnElements = [];

    if ((rootNode.ownerDocument || rootNode).getElementsByClassName) {
    /* Firefox 3 - native getElementsByClassName */
      var col = rootNode.getElementsByClassName(className);
      for (let i = 0; i < col.length; i++) {
        returnElements[i] = col[i];
      }
    } else if ((rootNode.ownerDocument || rootNode).evaluate) {
    /* Firefox 2 and below - XPath */
      var xpathExpression;
      xpathExpression = ".//*[contains(concat(' ', @class, ' '), ' " + className + " ')]";
      var xpathResult = (rootNode.ownerDocument || rootNode).evaluate(xpathExpression, rootNode, null, 0, null);

      var node;
      while (node = xpathResult.iterateNext()) {
        returnElements.push(node);
      }
    } else {
    /* Slow fallback for testing */
      className = className.replace(/\-/g, "\\-");
      var elements = rootNode.getElementsByTagName("*");
      for (let i=0;i<elements.length;i++) {
        if (elements[i].className.match("(^|\\s)" + className + "(\\s|$)")) {
          returnElements.push(elements[i]);
        }
      }
    }
    return returnElements;
  },
  /**
   * Not intended for external consumption. Microformat implementations might use it.
   *
   * Retrieve elements matching an attribute and an attribute list in a space-separated string.
   * 
   * @param  rootElement      The DOM element at which to start searching (optional)
   * @param  atributeName     The attribute name to match against
   * @param  attributeValues  A space separated list of attribute values
   * @return microformatNodes An array of DOM Nodes, each representing a
                              microformat in the document.
   */
  getElementsByAttribute: function getElementsByAttribute(rootNode, attributeName, attributeValues)
  {
    var attributeList = attributeValues.split(" ");

    var returnElements = [];

    if ((rootNode.ownerDocument || rootNode).evaluate) {
    /* Firefox 3 and below - XPath */
      /* Create an XPath expression based on the attribute list */
      var xpathExpression = ".//*[";