replace CRLF with single space for attribute value only

Author: amitguptagwl

package.json

@@ -1,6 +1,6 @@
 {
   "name": "fast-xml-parser",
-  "version": "3.0.0",
+  "version": "3.0.1",
   "description": "Validate XML or Parse XML to JS/JSON very fast without C/C++ based libraries",
   "main": "./src/parser.js",
   "scripts": {

spec/assets/mixed.xml

@@ -0,0 +1,36 @@
+<?xml version="1.0" standalone="yes" ?>
+<!--open the DOCTYPE declaration -
+  the open square bracket indicates an internal DTD-->
+<!DOCTYPE foo [
+
+
+<!--define the internal DTD-->
+<!ELEMENT foo (#PCDATA)>
+<!--close the DOCTYPE declaration-->
+]>
+<ns:root xmlns:soap="http://schemas.xmlsoap.org/soap/envelope/">
+    <ptag attr="val" boolean>some data
+        <nestedtag>
+            nesteddata<!--single line comment-->
+        </nestedtag>
+        after
+    </ptag>
+    <ptag>before text
+    <![CDATA[
+        <nestedtag>
+            nested cdata 1<!--single line comment-->
+        </nestedtag>
+    ]]>middle
+    <![CDATA[
+        <nestedtag>
+            nested cdata 2<!--multi line
+             comment-->
+        </nestedtag>
+    ]]>after
+    <![CDATA[
+        <nestedtag>
+            nested cdata 3
+        </nestedtag>
+    ]]>end
+    </ptag>
+</ns:root>

spec/attr_spec.js

@@ -73,7 +73,7 @@ describe("XMLParser", function () {
     });
 
     it("should decode HTML entities / char", function () {
-        var xmlData = '<element id="7" data="foo\nbar" bug="foo&ampbar&apos;"/>';
+        var xmlData = '<element id="7" data="foo\r\nbar" bug="foo&ampbar&apos;"/>';
         var expected = {
             "element": {
                 "id"    :     7,

spec/cdata_spec.js

@@ -189,4 +189,34 @@ describe("XMLParser", function () {
         expect(result).toEqual(expected);
     });
 
+    it("should validate XML with repeated multiline CDATA and comments", function () {
+        var fs = require("fs");
+        var path = require("path");
+        var fileNamePath = path.join(__dirname, "assets/mixed.xml");
+        var xmlData = fs.readFileSync(fileNamePath).toString();
+
+
+        var expected = {
+            "ns:root": {
+                "ptag": [
+                    {
+                        "nestedtag": "nesteddata",
+                        "@_attr": "val",
+                        "@_boolean": true,
+                        "#text": "some dataafter"
+                    },
+                    "before text\n        <nestedtag>\n            nested cdata 1\n        </nestedtag>\n    middle\n        <nestedtag>\n            nested cdata 2\n        </nestedtag>\n    after\n        <nestedtag>\n            nested cdata 3\n        </nestedtag>\n    end"
+                ],
+                "@_xmlns:soap": "http://schemas.xmlsoap.org/soap/envelope/"
+            }
+        };
+
+            var result = parser.parse(xmlData,{
+                ignoreAttributes:false,
+                allowBooleanAttributes:true
+            });
+        //console.log(JSON.stringify(result,null,4));
+        expect(result).toEqual(expected);
+    });
+
 });

src/parser.js

@@ -48,13 +48,13 @@ var buildOptions = function (options){
 
 var getTraversalObj =function (xmlData,options){
     options = buildOptions(options);
-    xmlData = xmlData.replace(/\r?\n/g, " ");//make it single line
-    xmlData = xmlData.replace(/<!--.*?-->/g, "");//Remove  comments
+    //xmlData = xmlData.replace(/\r?\n/g, " ");//make it single line
+    xmlData = xmlData.replace(/<!--[\s\S]*?-->/g, "");//Remove  comments
 
     var xmlObj = new xmlNode('!xml');
     var currentNode = xmlObj;
 
-    var tagsRegx = new RegExp("<((!\\[CDATA\\[(.*?)(\\]\\]>))|((\\w*:)?([\\w:\\-\\._]+))([^>]*)>|((\\/)((\\w*:)?([\\w:\\-\\._]+))>))([^<]*)","g");
+    var tagsRegx = new RegExp("<((!\\[CDATA\\[([\\s\\S]*?)(\\]\\]>))|((\\w*:)?([\\w:\\-\\._]+))([^>]*)>|((\\/)((\\w*:)?([\\w:\\-\\._]+))>))([^<]*)","g");
     var tag = tagsRegx.exec(xmlData);
     var nextTag = tagsRegx.exec(xmlData);
     var previousMatch,nextMatch;
@@ -167,6 +167,7 @@ function parseValue(val,shouldParse){
 var attrsRegx = new RegExp("([^\\s=]+)\\s*(=\\s*(['\"])(.*?)\\3)?","g");
 function buildAttributesMap(attrStr,options){
     if( !options.ignoreAttributes && typeof attrStr === "string" ){
+        attrStr = attrStr.replace(/\r?\n/g, " ");
         //attrStr = attrStr || attrStr.trim();
 
         var matches = util.getAllMatches(attrStr,attrsRegx);

src/validator.js

@@ -213,7 +213,7 @@ function readAttributeStr(xmlData,i){
 /**
  * Select all the attributes whether valid or invalid.
  */
-var validAttrStrRegxp = new RegExp("(\\s*)([^\\s=]+)(\\s*=)?(\\s*(['\"])((.|\\n)*?)\\5)?", "g");
+var validAttrStrRegxp = new RegExp("(\\s*)([^\\s=]+)(\\s*=)?(\\s*(['\"])(([\\s\\S])*?)\\5)?", "g");
 
 //attr, ="sd", a="amit's", a="sd"b="saf", ab  cd=""