From ac6d4529bd50c4bec8ff18b3b250e0695b91ba55 Mon Sep 17 00:00:00 2001
From: Jeff Schiller <codedread@gmail.com>
Date: Thu, 29 Dec 2011 22:31:48 -0800
Subject: [PATCH] Remove namespaced attributes.  Fix DOM element tagName
 parsing.  Update HTML so that each scoured file links persist.

---
 lite/index.html     | 48 +++++++++++++++++++++++-------------------
 lite/pdom.js        | 51 ++++++++++++++++++++++++++++++++++-----------
 lite/pdom_test.html | 25 +++++++++++++++++++++-
 lite/scour.js       | 39 +++++++++++++++++++++++++++++++++-
 4 files changed, 127 insertions(+), 36 deletions(-)

diff --git a/lite/index.html b/lite/index.html
index 834708b..e230dd9 100644
--- a/lite/index.html
+++ b/lite/index.html
@@ -22,16 +22,16 @@
     <p id='progressPara' class='hidden'>
       <span id='stage'>Progress</span>: <progress id='progress' max='100'></progress>
     </p>
-    <div>
-      <div id='originalSvg' class='hidden'>
+    <div id='links' class='hidden'>
+      <div id='originalSvg'>
         Original:
-          <a id='originalLinkText' href='#'>[text]</a>
-          <a id='originalLinkSvg' href='#'>[svg]</a>
+          <a id='originalLinkText' target='_blank'>[text]</a>
+          <a id='originalLinkSvg' target='_blank'>[svg]</a>
       </div>
-      <div id='scouredSvg' class='hidden'>
+      <div id='scouredSvg'>
         Scoured:
-          <a id='scouredLinkText' href='#'>[text]</a>
-          <a id='scouredLinkSvg' href='#'>[svg]</a>
+          <a id='scouredLinkText' target='_blank'>[text]</a>
+          <a id='scouredLinkSvg' target='_blank'>[svg]</a>
       </div>
     </div>
     <div id='status' class='hidden'></div>
@@ -44,19 +44,6 @@
     var originalSvg;
     var scouredSvg;
 
-    document.getElementById('originalLinkText').onclick = function(evt) {
-      window.open('data:text/plain;base64,' + window.btoa(originalSvg));
-    };
-    document.getElementById('originalLinkSvg').onclick = function(evt) {
-      window.open('data:image/svg+xml;base64,' + window.btoa(originalSvg));
-    };
-    document.getElementById('scouredLinkText').onclick = function(evt) {
-      window.open('data:text/plain;base64,' + window.btoa(scouredSvg));
-    };
-    document.getElementById('scouredLinkSvg').onclick = function(evt) {
-      window.open('data:image/svg+xml;base64,' + window.btoa(scouredSvg));
-    };
-
     var handleMessage = function(evt) {
       if (typeof evt.data == 'string') {
         var p = document.createElement('p');
@@ -73,13 +60,28 @@
         if (evt.data.progress) {
           handleProgress(evt.data.progress);
         }
+        if (evt.data.update) {
+          var status = document.getElementById('status').firstChild;
+          status.innerHTML = status.innerHTML + evt.data.update;
+        }
         if (evt.data.scouredSvg) {
-          document.getElementById('scouredSvg').className = '';
           scouredSvg = evt.data.scouredSvg;
           handleMessage({
             data: 'Scoured SVG came out to be ' +
               evt.data.scouredSvg.length + ' bytes'
           });
+
+          var status = document.getElementById('status');
+          var links = document.getElementById('links');
+          document.getElementById('originalLinkText').setAttribute('href',
+              'data:text/plain;base64,' + window.btoa(originalSvg));
+          document.getElementById('originalLinkSvg').setAttribute('href',
+              'data:image/svg+xml;base64,' + window.btoa(originalSvg));
+          document.getElementById('scouredLinkText').setAttribute('href',
+              'data:text/plain;base64,' + window.btoa(scouredSvg));
+          document.getElementById('scouredLinkSvg').setAttribute('href',
+              'data:image/svg+xml;base64,' + window.btoa(scouredSvg));
+          status.innerHTML = links.innerHTML + status.innerHTML;
         }
       }
     };
@@ -89,7 +91,7 @@
     };
     var getFile = function(evt) {
       stageSpan.innerHTML = 'Loading';
-      var showElems = ['progressPara', 'status', 'originalSvg'];
+      var showElems = ['progressPara', 'status'];
       for (var i in showElems) {
         document.getElementById(showElems[i]).className = '';
       }
@@ -102,6 +104,8 @@
         // TODO: Use addEventListener when WebKit supports it
         // https://bugs.webkit.org/show_bug.cgi?id=42723
         fr.onload = function(evt) {
+          handleMessage({data:
+              '-------------------------------------------------------------------------------'});
           handleMessage({data:'Loaded \'' + theFile.name + '\'' + ' (' + theFile.size + ' bytes)'});
           var worker = new Worker('scour.js');
           worker.addEventListener('message', handleMessage);
diff --git a/lite/pdom.js b/lite/pdom.js
index c4b9d76..5d6a947 100644
--- a/lite/pdom.js
+++ b/lite/pdom.js
@@ -109,10 +109,11 @@ pdom.NodeList.prototype.item = function(index) {
 /**
  * @param {Object.<string, pdom.Node>} nodeMap An object containing the
  *     attribute name-Node pairs.
+ * @param {pdom.Node} opt_refNode An optional reference node.
  * @constructor
  */
-pdom.NamedNodeMap = function(nodeMap) {
-  this.setNodeMapInternal(nodeMap);
+pdom.NamedNodeMap = function(nodeMap, opt_refNode) {
+  this.setNodeMapInternal(nodeMap, opt_refNode);
   this.__defineGetter__('length', function() { return this.attrs_.length });
 };
 
@@ -136,12 +137,33 @@ pdom.NamedNodeMap.prototype.nodeMap_ = {};
 /**
  * Sets the internal node map (and updates the array).
  * @param {Object.<string, pdom.Node>} The node map.
+ * @param {pdom.Node} opt_refNode The optional reference node for namespace prefix checking.
  */
-pdom.NamedNodeMap.prototype.setNodeMapInternal = function(nodeMap) {
+pdom.NamedNodeMap.prototype.setNodeMapInternal = function(nodeMap, opt_refNode) {
+  var nsPrefixMap = {};
+  if (opt_refNode) {
+    var node = opt_refNode;
+    while (node) {
+      for (var prefix in node.nsPrefixMapInternal) {
+        nsPrefixMap[prefix] = node.nsPrefixMapInternal[prefix];
+      }
+      node = node.parentNode;
+    }
+  }
   this.nodeMap_ = {};
   this.attrs_ = [];
   for (var name in nodeMap) {
-    var attr = new pdom.Attr(name, nodeMap[name]);
+    // If the attribute name includes a colon, resolve the namespace prefix.
+    var colonIndex = name.indexOf(':');
+    var namespaceURI = null;
+    if (colonIndex != -1) {
+      var prefix = name.substring(0, colonIndex);
+      var uri = nsPrefixMap[prefix];
+      if (uri) {
+        namespaceURI = uri;
+      }
+    }
+    var attr = new pdom.Attr(name, nodeMap[name], namespaceURI);
     this.attrs_.push(attr);
     this.nodeMap_[name] = attr;
   }
@@ -334,16 +356,18 @@ pdom.inherits(pdom.DocumentType, pdom.Node);
  *
  * @param {string} name The name of the attribute.
  * @param {string} value The value of the attribute.
+ * @param {stirng} opt_namespaceURI Optional namespace URI.
  * @constructor
  * @extends {pdom.Attr}
  */
-pdom.Attr = function(name, value) {
+pdom.Attr = function(name, value, opt_namespaceURI) {
   pdom.base(this, null);
   
   this.__defineGetter__('nodeType', function() {
     return pdom.Node.ATTRIBUTE_NODE;
   });
   this.__defineGetter__('name', function() { return name });
+  this.__defineGetter__('namespaceURI', function() { return opt_namespaceURI });
 
   /**  
    * @type {string}
@@ -375,7 +399,7 @@ pdom.Element = function(tagName, opt_parentNode, opt_attrs) {
 
   this.__defineGetter__('attributes', function() {
     if (!this.attributeMap_) {
-      this.attributeMap_ = new pdom.NamedNodeMap(this.attributes_);
+      this.attributeMap_ = new pdom.NamedNodeMap(this.attributes_, this);
     }
     return this.attributeMap_;
   });
@@ -453,7 +477,7 @@ pdom.Element.prototype.getAttribute = function(attrName) {
 pdom.Element.prototype.setAttribute = function(name, value) {
   this.attributes_[name] = value;
   if (this.attributeMap_) {
-    this.attributeMap_.setNodeMapInternal(this.attributes_);
+    this.attributeMap_.setNodeMapInternal(this.attributes_, this);
   }
 };
 
@@ -464,7 +488,7 @@ pdom.Element.prototype.setAttribute = function(name, value) {
 pdom.Element.prototype.removeAttribute = function(name) {
   delete this.attributes_[name];
   if (this.attributeMap_) {
-    this.attributeMap_.setNodeMapInternal(this.attributes_);
+    this.attributeMap_.setNodeMapInternal(this.attributes_, this);
   }
 };
 
@@ -658,7 +682,7 @@ pdom.parse.parseOneNode_ = function(parsingContext) {
     if (endDocType == -1) {
       throw 'Could not find the end of the DOCTYPE (>)';
     }
-    var newDocType = new pdom.DocType();
+    var newDocType = new pdom.DocumentType();
     parsingContext.currentNode.childNodes_.push(newDocType);
     parsingContext.offset = endDocType + 1;
     return newDocType;
@@ -706,9 +730,12 @@ pdom.parse.parseOneNode_ = function(parsingContext) {
 
     var attrs = {};
 
-    // TODO: This should be whitespace, not space.
-    var tagNameIndex = xmlText.indexOf(' ', i + 1);
-    if (tagNameIndex == -1 || tagNameIndex > endStartTagIndex) {
+    // Find if whitespace occurs before the end of the start tag.
+    var wsMatches = xmlText.substring(i + 1).match(/\s+/);
+    var tagNameIndex = wsMatches && wsMatches.length > 0 ?
+        wsMatches.index + i + 1 :
+        endStartTagIndex;
+    if (tagNameIndex > endStartTagIndex) {
       tagNameIndex = endStartTagIndex;
     } else {
       // Find all attributes and record them.
diff --git a/lite/pdom_test.html b/lite/pdom_test.html
index 45ca74b..8104955 100644
--- a/lite/pdom_test.html
+++ b/lite/pdom_test.html
@@ -31,7 +31,7 @@
 
     function testParseSimpleGetAttribute() {
       var doc = new pdom.DOMParser().parseFromString(
-          '<?xml version="1.0"?><simple bar="foo" foo=\'bar\'' +
+          '<simple bar="foo" foo=\'bar\'' +
           '  baz = " blah blah "' +
           '/>');
 
@@ -43,6 +43,15 @@
       muther.assert(elem.getAttribute('foo') == 'bar', 'foo attribute not correct');
       muther.assert(elem.getAttribute('baz') == ' blah blah ', 'baz attribute not correct');
     },
+
+    function testParseTagNameWithEOL() {
+      var doc = new pdom.DOMParser().parseFromString('<parent\n/>');
+
+      var parent = doc.documentElement;
+      muther.assert(parent, 'Could not parse documentElement');
+      muther.assert(parent.tagName == 'parent', 'Did not parse the parent tagName: \'' +
+          parent.tagName + '\'');
+    },
     
     function testNodeTypes() {
       var doc = new pdom.DOMParser().parseFromString('<simple foo="bar"><!-- Comment -->Text</simple>');
@@ -111,6 +120,20 @@
       muther.assert(grandChild.namespaceURI === "http://bar/",
           'prefixed namespaceURI did not work');
     },
+
+    function testNamespacedAttributes() {
+      var NS_BAR = 'http://bar/';
+      var doc = new pdom.DOMParser().parseFromString(
+          '<parent xmlns="http://foo/" xmlns:bar="http://bar/" bar:foo="baz">' +
+          '<child bar:baz="blah"/></parent>');
+      var parent = doc.documentElement;
+      var child = parent.firstChild;
+
+      muther.assert(parent.attributes.getNamedItem('bar:foo').namespaceURI == NS_BAR,
+          'bar: attribute on parent does not have the proper namespaceURI');
+      muther.assert(child.attributes.item(0).namespaceURI == NS_BAR,
+          'bar: attribute on child does not have the proper namespaceURI');
+    },
     
     function testTagName() {
       var doc = new pdom.DOMParser().parseFromString('<simple/>');
diff --git a/lite/scour.js b/lite/scour.js
index 48c13d1..f540270 100644
--- a/lite/scour.js
+++ b/lite/scour.js
@@ -48,6 +48,7 @@ var removeNamespacedElements = function(node, namespaces) {
       var child = node.childNodes.item(i);
       if (namespaces.indexOf(child.namespaceURI) != -1) {
         childrenToRemove.push(child);
+        postMessage({update: '.'});
       }
     }
     
@@ -63,6 +64,35 @@ var removeNamespacedElements = function(node, namespaces) {
 };
 
 
+/**
+ * @param {pdom.Node|Node} node The parent node.
+ * @param {Array.<string>} namespaces An array of namespace URIs.
+ */
+var removeNamespacedAttributes = function(node, namespaces) {
+  if (node.nodeType == 1) {
+    // Remove all namespace'd attributes from this element.
+    var attrsToRemove = [];
+    for (var i = 0; i < node.attributes.length; ++i) {
+      var attr = node.attributes.item(i);
+      if (namespaces.indexOf(attr.namespaceURI) != -1 ||
+          (namespaces.indexOf(attr.value) != -1 && attr.name.indexOf('xmlns:') == 0)) {
+        attrsToRemove.push(attr);
+        postMessage({update: '.'});
+      }
+    }
+    
+    for (var i = 0; i < attrsToRemove.length; ++i) {
+      node.removeAttribute(attrsToRemove[i].name);
+    }
+
+    // Now recurse for children.
+    for (var i = 0; i < node.childNodes.length; ++i) {
+      removeNamespacedAttributes(node.childNodes.item(i), namespaces);
+    }
+  }
+};
+
+
 /**
  * @param {string} in_string The SVG document as a string.
  * @param {object} opt_options An optional set of options.
@@ -76,8 +106,15 @@ var scourString = function(in_string, opt_options) {
 
   // Remove editor stuff.
   if (!options.keep_editor_data) {
+    postMessage({message: 'Removing namespaced elements '});
     removeNamespacedElements(doc.documentElement, unwanted_ns);
-    postMessage({message: 'Removed namespaced elements'});
+    postMessage({update: ' done!'});
+    postMessage({progress: {loaded: 45, total: 100}});
+
+    postMessage({message: 'Removing namespaced attributes '});
+    removeNamespacedAttributes(doc.documentElement, unwanted_ns);
+    postMessage({update: ' done!'});
+    postMessage({progress: {loaded: 90, total: 100}});
   }
 
   return new pdom.XMLSerializer().serializeToString(doc);