microformats.js

this is Microformats.js for linux

var EXPORTED_SYMBOLS = ["Microformats", "adr", "tag", "hCard", "hCalendar", "geo"];

var Microformats = {
  /* When a microformat is added, the name is placed in this list */
  list: [],
  /* Custom iterator so that microformats can be enumerated as */
  /* for (i in Microformats) */
  __iterator__: function () {
    for (let i=0; i < this.list.length; i++) {
      yield this.list[i];
    }
  },
  /**
   * Retrieves microformats objects of the given type from a document
   * 
   * @param  name          The name of the microformat (required)
   * @param  rootElement   The DOM element at which to start searching (required)
   * @param  options       Literal object with the following options:
   *                       recurseExternalFrames - Whether or not to search child frames
   *                       that reference external pages (with a src attribute)
   *                       for microformats (optional - defaults to true)
   *                       showHidden -  Whether or not to add hidden microformat
   *                       (optional - defaults to false)
   *                       debug - Whether or not we are in debug mode (optional
   *                       - defaults to false)
   * @param  targetArray  An array of microformat objects to which is added the results (optional)
   * @return A new array of microformat objects or the passed in microformat 
   *         object array with the new objects added
   */
  get: function(name, rootElement, options, targetArray) {
    function isAncestor(haystack, needle) {
      var parent = needle;
      while (parent = parent.parentNode) {
        /* We need to check parentNode because defaultView.frames[i].frameElement */
        /* isn't a real DOM node */
        if (parent == needle.parentNode) {
          return true;
        }
      }
      return false;
    }
    if (!Microformats[name] || !rootElement) {
      return;
    }
    targetArray = targetArray || [];

    /* Root element might not be the document - we need the document's default view */
    /* to get frames and to check their ancestry */
    var defaultView = rootElement.defaultView || rootElement.ownerDocument.defaultView;
    var rootDocument = rootElement.ownerDocument || rootElement;

    /* If recurseExternalFrames is undefined or true, look through all child frames for microformats */
    if (!options || !options.hasOwnProperty("recurseExternalFrames") || options.recurseExternalFrames) {
      if (defaultView && defaultView.frames.length > 0) {
        for (let i=0; i < defaultView.frames.length; i++) {
          if (isAncestor(rootDocument, defaultView.frames[i].frameElement)) {
            Microformats.get(name, defaultView.frames[i].document, options, targetArray);
          }
        }
      }
    }

    /* Get the microformat nodes for the document */
    var microformatNodes = [];
    if (Microformats[name].className) {
      microformatNodes = Microformats.getElementsByClassName(rootElement,
                                        Microformats[name].className);
      /* alternateClassName is for cases where a parent microformat is inferred by the children */
      /* If we find alternateClassName, the entire document becomes the microformat */
      if ((microformatNodes.length == 0) && Microformats[name].alternateClassName) {
        var altClass = Microformats.getElementsByClassName(rootElement, Microformats[name].alternateClassName);
        if (altClass.length > 0) {
          microformatNodes.push(rootElement); 
        }
      }
    } else if (Microformats[name].attributeValues) {
      microformatNodes =
        Microformats.getElementsByAttribute(rootElement,
                                            Microformats[name].attributeName,
                                            Microformats[name].attributeValues);
      
    }
    /* Create objects for the microformat nodes and put them into the microformats */
    /* array */
    for (let i = 0; i < microformatNodes.length; i++) {
      /* If showHidden undefined or false, don't add microformats to the list that aren't visible */
      if (!options || !options.hasOwnProperty("showHidden") || !options.showHidden) {
        if (microformatNodes[i].ownerDocument) {
          if (microformatNodes[i].getBoundingClientRect) {
            var box = microformatNodes[i].getBoundingClientRect();
            box.width = box.right - box.left;
            box.height = box.bottom - box.top;
          } else {
            var box = microformatNodes[i].ownerDocument.getBoxObjectFor(microformatNodes[i]);
          }
          if ((box.height == 0) || (box.width == 0)) {
            continue;
          }
        }
      }
      try {
        if (options && options.debug) {
          /* Don't validate in the debug case so that we don't get errors thrown */
          /* in the debug case, we want all microformats, even if they are invalid */
          targetArray.push(new Microformats[name].mfObject(microformatNodes[i], false));
        } else {
          targetArray.push(new Microformats[name].mfObject(microformatNodes[i], true));
        }
      } catch (ex) {
        /* Creation of individual object probably failed because it is invalid. */
        /* This isn't a problem, because the page might have invalid microformats */
      }
    }
    return targetArray;
  },
  /**
   * Counts microformats objects of the given type from a document
   * 
   * @param  name          The name of the microformat (required)
   * @param  rootElement   The DOM element at which to start searching (required)
   * @param  options       Literal object with the following options:
   *                       recurseExternalFrames - Whether or not to search child frames
   *                       that reference external pages (with a src attribute)
   *                       for microformats (optional - defaults to true)
   *                       showHidden -  Whether or not to add hidden microformat
   *                       (optional - defaults to false)
   *                       debug - Whether or not we are in debug mode (optional
   *                       - defaults to false)
   * @return The new count
   */
  count: function(name, rootElement, options) {
    var mfArray = Microformats.get(name, rootElement, options);
    if (mfArray) {
      return mfArray.length;
    }
    return 0;
  },
  /**
   * Returns true if the passed in node is a microformat. Does NOT return true
   * if the passed in node is a child of a microformat.
   *
   * @param  node          DOM node to check
   * @return true if the node is a microformat, false if it is not
   */
  isMicroformat: function(node) {
    for (let i in Microformats)
    {
      if (Microformats[i].className) {
        if (Microformats.matchClass(node, Microformats[i].className)) {
            return true;
        }
      } else {
        var attribute;
        if (attribute = node.getAttribute(Microformats[i].attributeName)) {
          var attributeList = Microformats[i].attributeValues.split(" ");
          for (let j=0; j < attributeList.length; j++) {
            if (attribute.match("(^|\\s)" + attributeList[j] + "(\\s|$)")) {
              return true;
            }
          }
        }
      }
    }
    return false;
  },
  /**
   * This function searches a given nodes ancestors looking for a microformat
   * and if it finds it, returns it. It does NOT include self, so if the passed
   * in node is a microformat, it will still search ancestors for a microformat.
   *
   * @param  node          DOM node to check
   * @return If the node is contained in a microformat, it returns the parent
   *         DOM node, otherwise returns null
   */
  getParent: function(node) {
    var xpathExpression;
    var xpathResult;

    xpathExpression = "ancestor::*[";
    for (let i=0; i < Microformats.list.length; i++) {
      var mfname = Microformats.list[i];
      if (i != 0) {
        xpathExpression += " or ";
      }
      if (Microformats[mfname].className) {
        xpathExpression += "contains(concat(' ', @class, ' '), ' " + Microformats[mfname].className + " ')";
      } else {
        var attributeList = Microformats[mfname].attributeValues.split(" ");
        for (let j=0; j < attributeList.length; j++) {
          if (j != 0) {
            xpathExpression += " or ";
          }
          xpathExpression += "contains(concat(' ', @" + Microformats[mfname].attributeName + ", ' '), ' " + attributeList[j] + " ')";
        }
      }
    }
    xpathExpression += "][1]";
    xpathResult = (node.ownerDocument || node).evaluate(xpathExpression, node, null,  Components.interfaces.nsIDOMXPathResult.FIRST_ORDERED_NODE_TYPE, null);
    if (xpathResult.singleNodeValue) {
      xpathResult.singleNodeValue.microformat = mfname;
      return xpathResult.singleNodeValue;
    }
    return null;
  },
  /**
   * If the passed in node is a microformat, this function returns a space 
   * separated list of the microformat names that correspond to this node
   *
   * @param  node          DOM node to check
   * @return If the node is a microformat, a space separated list of microformat
   *         names, otherwise returns nothing
   */
  getNamesFromNode: function(node) {
    var microformatNames = [];
    var xpathExpression;
    var xpathResult;
    for (let i in Microformats)
    {
      if (Microformats[i]) {
        if (Microformats[i].className) {
          if (Microformats.matchClass(node, Microformats[i].className)) {
            microformatNames.push(i);
            continue;
          }
        } else if (Microformats[i].attributeValues) {
          var attribute;
          if (attribute = node.getAttribute(Microformats[i].attributeName)) {
            var attributeList = Microformats[i].attributeValues.split(" ");
            for (let j=0; j < attributeList.length; j++) {
              /* If we match any attribute, we've got a microformat */
              if (attribute.match("(^|\\s)" + attributeList[j] + "(\\s|$)")) {
                microformatNames.push(i);
                break;
              }
            }
          }
        }
      }
    }
    return microformatNames.join(" ");
  },
  /**
   * Outputs the contents of a microformat object for debug purposes.
   *
   * @param  microformatObject JavaScript object that represents a microformat
   * @return string containing a visual representation of the contents of the microformat
   */
  debug: function debug(microformatObject) {
    function dumpObject(item, indent)
    {
      if (!indent) {
        indent = "";
      }
      var toreturn = "";
      var testArray = [];
      
      for (let i in item)
      {
        if (testArray[i]) {
          continue;
        }
        if (typeof item[i] == "object") {
          if ((i != "node") && (i != "resolvedNode")) {
            if (item[i] && item[i].semanticType) {
              toreturn += indent + item[i].semanticType + " [" + i + "] { \n";
            } else {
              toreturn += indent + "object " + i + " { \n";
            }
            toreturn += dumpObject(item[i], indent + "\t");
            toreturn += indent + "}\n";
          }
        } else if ((typeof item[i] != "function") && (i != "semanticType")) {
          if (item[i]) {
            toreturn += indent + i + "=" + item[i] + "\n";
          }
        }
      }
      if (!toreturn && item) {
        toreturn = item.toString();
      }
      return toreturn;
    }
    return dumpObject(microformatObject);
  },
  add: function add(microformat, microformatDefinition) {
    /* We always replace an existing definition with the new one */
    if (!Microformats[microformat]) {
      Microformats.list.push(microformat);
    }
    Microformats[microformat] = microformatDefinition;
    microformatDefinition.mfObject.prototype.debug =
      function(microformatObject) {
        return Microformats.debug(microformatObject)
      };
  },
  /* All parser specific functions are contained in this object */
  parser: {
    /**
     * Uses the microformat patterns to decide what the correct text for a
     * given microformat property is. This includes looking at things like
     * abbr, img/alt, area/alt and value excerpting.
     *
     * @param  propnode   The DOMNode to check
     * @param  parentnode The parent node of the property. If it is a subproperty,
     *                    this is the parent property node. If it is not, this is the
     *                    microformat node.
     & @param  datatype   HTML/text - whether to use innerHTML or innerText - defaults to text
     * @return A string with the value of the property
     */
    defaultGetter: function(propnode, parentnode, datatype) {
      if (((((propnode.localName.toLowerCase() == "abbr") || (propnode.localName.toLowerCase() == "html:abbr")) && !propnode.namespaceURI) || 
         ((propnode.localName.toLowerCase() == "abbr") && (propnode.namespaceURI == "http://www.w3.org/1999/xhtml"))) && (propnode.getAttribute("title"))) {
        return propnode.getAttribute("title");
      } else if ((propnode.nodeName.toLowerCase() == "img") && (propnode.getAttribute("alt"))) {
        return propnode.getAttribute("alt");
      } else if ((propnode.nodeName.toLowerCase() == "area") && (propnode.getAttribute("alt"))) {
        return propnode.getAttribute("alt");
      } else if ((propnode.nodeName.toLowerCase() == "textarea") ||
                 (propnode.nodeName.toLowerCase() == "select") ||
                 (propnode.nodeName.toLowerCase() == "input")) {
        return propnode.value;
      } else {
        var values = Microformats.getElementsByClassName(propnode, "value");
        /* Verify that values are children of the propnode */
        for (let i = values.length-1; i >= 0; i--) {
          if (values[i].parentNode != propnode) {
            values.splice(i,1);
          }
        }
        if (values.length > 0) {
          var value = "";
          for (let j=0;j<values.length;j++) {
            value += Microformats.parser.defaultGetter(values[j], propnode, datatype);
          }
          return value;
        }
        var s;
        if (datatype == "HTML") {
          s = propnode.innerHTML;
        } else {
          if (propnode.innerText) {
            s = propnode.innerText;
          } else {
            s = propnode.textContent;
          }
        }
        /* If we are processing a value node, don't remove whitespace */
        if (!Microformats.matchClass(propnode, "value")) {
          /* Remove new lines, carriage returns and tabs */
          s	= s.replace(/[\n\r\t]/gi, ' ');
          /* Replace any double spaces with single spaces */
          s	= s.replace(/\s{2,}/gi, ' ');
          /* Remove any double spaces that are left */
          s	= s.replace(/\s{2,}/gi, '');
          /* Remove any spaces at the beginning */
          s	= s.replace(/^\s+/, '');
          /* Remove any spaces at the end */
          s	= s.replace(/\s+$/, '');
        }