Merge pull request #41 from Delagen/master

parse html entities in text values, fixes #40

Author: amitguptagwl

lib/parser.js

@@ -5,11 +5,11 @@
   "main": "./src/parser.js",
   "scripts": {
     "test": "jasmine spec/*spec.js",
-    "bundle": "browserify src/parser.js --s parser > lib/parser.js",
+    "bundle": "browserify src/parser.js -s parser > lib/parser.js",
     "coverage": "node ./benchmark/perfTest.js; istanbul cover -x \"spec/*spec.js\" jasmine spec/*spec.js;",
     "coverage:check": "istanbul check-coverage --branch 90 --statement 90"
   },
-  "src": {
+  "bin": {
     "xml2js": "cli.js"
   },
   "repository": {
@@ -66,5 +66,8 @@
     "jasmine-core": "^2.8.0",
     "portfinder": "^1.0.13",
     "zombie": "^5.0.7"
+  },
+  "dependencies": {
+    "he": "~1.1.1"
   }
 }

package.json

@@ -86,7 +86,7 @@ describe("XMLParser", function () {
         var expected = {
                         "xml": {
                             "a": "text",
-                            "b": "        text    \n",
+                            "b": "\n       text    \n",
                             "c": "text",
                             "d": "text"
                         }
@@ -141,4 +141,4 @@ describe("XMLParser", function () {
         expect(result).toBe(true);
     });
 
-});
+});

spec/cdata_spec.js

@@ -33,13 +33,14 @@ describe("XMLParser", function () {
     });
 
     it("should parse all values of attributes as string", function () {
-        var xmlData = "<rootNode><tag int='045' float='65.34'>value</tag></rootNode>";
+        var xmlData = "<rootNode><tag int='045' float='65.34' text='foo&ampbar&apos;'>value&amp;\r\n&apos;</tag></rootNode>";
         var expected = {
             "rootNode": {
                 "tag": {
-                    "#text": "value",
+                    "#text": "value&\r\n'",
                     "@_int": "045",
-                    "@_float": "65.34"
+                    "@_float": "65.34",
+                    "@_text": "foo&ampbar'"
                 }
             }
         };

spec/xmlParser_spec.js

@@ -1,3 +1,4 @@
+var he = require("he");
 var getAllMatches = require("./util").getAllMatches;
 
 var xmlNode = function(tagname,parent,val){
@@ -73,7 +74,7 @@ var getTraversalObj =function (xmlData,options){
                 attrs[options.textNodeName] = val;
                 childNode.val = attrs;
             }else{
-                childNode.val = val;    
+                childNode.val = val;
             }
             currentNode.addChild(childNode);
             i++;
@@ -127,11 +128,13 @@ function resolveNameSpace(tagname,ignore){
     return tagname;
 }
 
-function parseValue(val,conversion){
+function parseValue(val,conversion,isAttribute){
     if(val){
         if(!conversion || isNaN(val)){
-            val = "" + val ;
-            val = val.replace("\n"," ");
+            val = "" + he.decode(val, {isAttributeValue:isAttribute, strict:true});
+            if(isAttribute) {
+                val = val.replace(/\r?\n/g, " ");
+            }
         }else{
             if(val.indexOf(".") !== -1){
                 if(parseFloat){
@@ -164,7 +167,7 @@ function buildAttributesArr(attrStr,ignore,prefix,attrNodeName,ignoreNS,conversi
         for (var i = 0; i < matches.length; i++) {
             var attrName = resolveNameSpace(matches[i][1],ignoreNS);
             if(attrName.length && attrName !== "xmlns") {
-                attrs[prefix + attrName] = parseValue(matches[i][3], conversion);
+                attrs[prefix + attrName] = parseValue(matches[i][3], conversion, true);
             }
         }
         if(!Object.keys(attrs).length){