fast-xml-parser 3.21.0 → 4.0.0-beta.3

package/src/xmlparser/OrderedObjParser.js

@@ -0,0 +1,486 @@
+'use strict';
+
+const util = require('../util');
+const xmlNode = require('./xmlNode');
+const readDocType = require("./DocTypeReader");
+const toNumber = require("strnum");
+
+const regx =
+  '<((!\\[CDATA\\[([\\s\\S]*?)(]]>))|((NAME:)?(NAME))([^>]*)>|((\\/)(NAME)\\s*>))([^<]*)'
+  .replace(/NAME/g, util.nameRegexp);
+
+//const tagsRegx = new RegExp("<(\\/?[\\w:\\-\._]+)([^>]*)>(\\s*"+cdataRegx+")*([^<]+)?","g");
+//const tagsRegx = new RegExp("<(\\/?)((\\w*:)?([\\w:\\-\._]+))([^>]*)>([^<]*)("+cdataRegx+"([^<]*))*([^<]+)?","g");
+
+class OrderedObjParser{
+  constructor(options){
+    this.options = options;
+    this.currentNode = null;
+    this.tagsNodeStack = [];
+    this.docTypeEntities = {};
+    this.lastEntities = {
+      "amp" : { regex: /&(amp|#38|#x26);/g, val : "&"},
+      "apos" : { regex: /&(apos|#39|#x27);/g, val : "'"},
+      "gt" : { regex: /&(gt|#62|#x3E);/g, val : ">"},
+      "lt" : { regex: /&(lt|#60|#x3C);/g, val : "<"},
+      "quot" : { regex: /&(quot|#34|#x22);/g, val : "\""},
+    };
+    this.htmlEntities = {
+      "space": { regex: /&(nbsp|#160);/g, val: " " },
+      // "lt" : { regex: /&(lt|#60);/g, val: "<" },
+      // "gt" : { regex: /&(gt|#62);/g, val: ">" },
+      // "amp" : { regex: /&(amp|#38);/g, val: "&" },
+      // "quot" : { regex: /&(quot|#34);/g, val: "\"" },
+      // "apos" : { regex: /&(apos|#39);/g, val: "'" },
+      "cent" : { regex: /&(cent|#162);/g, val: "¢" },
+      "pound" : { regex: /&(pound|#163);/g, val: "£" },
+      "yen" : { regex: /&(yen|#165);/g, val: "¥" },
+      "euro" : { regex: /&(euro|#8364);/g, val: "€" },
+      "copyright" : { regex: /&(copy|#169);/g, val: "©" },
+      "reg" : { regex: /&(reg|#174);/g, val: "®" },
+      "inr" : { regex: /&(inr|#8377);/g, val: "₹" },
+    };
+    this.parseXml = parseXml;
+    this.parseTextData = parseTextData;
+    this.resolveNameSpace = resolveNameSpace;
+    this.buildAttributesMap = buildAttributesMap;
+    this.isItStopNode = isItStopNode;
+    this.replaceEntitiesValue = replaceEntitiesValue;
+    this.readTagExp = readTagExp;
+    this.readStopNodeData = readStopNodeData;
+  }
+
+}
+
+/**
+ * @param {string} val
+ * @param {string} tagName
+ * @param {string} jPath
+ * @param {boolean} dontTrim
+ * @param {boolean} hasAttributes
+ * @param {boolean} isLeafNode
+ */
+function parseTextData(val, tagName, jPath, dontTrim, hasAttributes, isLeafNode) {
+  if (val !== undefined) {
+    if (this.options.trimValues && !dontTrim) {
+      val = val.trim();
+    }
+    if(val.length > 0){
+      val = this.replaceEntitiesValue(val);
+      
+      const newval = this.options.tagValueProcessor(tagName, val, jPath, hasAttributes, isLeafNode);
+      if(newval === null || newval === undefined){
+        //don't parse
+        return val;
+      }else if(typeof newval !== typeof val || newval !== val){
+        //overwrite
+        return newval;
+      }else if(this.options.trimValues){
+        return parseValue(val, this.options.parseTagValue, this.options.numberParseOptions);
+      }else{
+        const trimmedVal = val.trim();
+        if(trimmedVal === val){
+          return parseValue(val, this.options.parseTagValue, this.options.numberParseOptions);
+        }else{
+          return val;
+        }
+      }
+    }
+  }
+}
+
+function resolveNameSpace(tagname) {
+  if (this.options.removeNSPrefix) {
+    const tags = tagname.split(':');
+    const prefix = tagname.charAt(0) === '/' ? '/' : '';
+    if (tags[0] === 'xmlns') {
+      return '';
+    }
+    if (tags.length === 2) {
+      tagname = prefix + tags[1];
+    }
+  }
+  return tagname;
+}
+
+//TODO: change regex to capture NS
+//const attrsRegx = new RegExp("([\\w\\-\\.\\:]+)\\s*=\\s*(['\"])((.|\n)*?)\\2","gm");
+const attrsRegx = new RegExp('([^\\s=]+)\\s*(=\\s*([\'"])([\\s\\S]*?)\\3)?', 'gm');
+
+function buildAttributesMap(attrStr, jPath) {
+  if (!this.options.ignoreAttributes && typeof attrStr === 'string') {
+    // attrStr = attrStr.replace(/\r?\n/g, ' ');
+    //attrStr = attrStr || attrStr.trim();
+
+    const matches = util.getAllMatches(attrStr, attrsRegx);
+    const len = matches.length; //don't make it inline
+    const attrs = {};
+    for (let i = 0; i < len; i++) {
+      const attrName = this.resolveNameSpace(matches[i][1]);
+      let oldVal = matches[i][4];
+      const aName = this.options.attributeNamePrefix + attrName;
+      if (attrName.length) {
+        if (oldVal !== undefined) {
+          if (this.options.trimValues) {
+            oldVal = oldVal.trim();
+          }
+          oldVal = this.replaceEntitiesValue(oldVal);
+          const newVal = this.options.attributeValueProcessor(attrName, oldVal, jPath);
+          if(newVal === null || newVal === undefined){
+            //don't parse
+            attrs[aName] = oldVal;
+          }else if(typeof newVal !== typeof oldVal || newVal !== oldVal){
+            //overwrite
+            attrs[aName] = newVal;
+          }else{
+            //parse
+            attrs[aName] = parseValue(
+              oldVal,
+              this.options.parseAttributeValue,
+              this.options.numberParseOptions
+            );
+          }
+        } else if (this.options.allowBooleanAttributes) {
+          attrs[aName] = true;
+        }
+      }
+    }
+    if (!Object.keys(attrs).length) {
+      return;
+    }
+    if (this.options.attributesGroupName) {
+      const attrCollection = {};
+      attrCollection[this.options.attributesGroupName] = attrs;
+      return attrCollection;
+    }
+    return attrs;
+  }
+}
+
+const parseXml = function(xmlData) {
+  xmlData = xmlData.replace(/\r\n?/g, "\n"); //TODO: remove this line
+  const xmlObj = new xmlNode('!xml');
+  let currentNode = xmlObj;
+  let textData = "";
+  let jPath = "";
+  for(let i=0; i< xmlData.length; i++){//for each char in XML data
+    const ch = xmlData[i];
+    if(ch === '<'){
+      // const nextIndex = i+1;
+      // const _2ndChar = xmlData[nextIndex];
+      if( xmlData[i+1] === '/') {//Closing Tag
+        const closeIndex = findClosingIndex(xmlData, ">", i, "Closing Tag is not closed.")
+        let tagName = xmlData.substring(i+2,closeIndex).trim();
+
+        if(this.options.removeNSPrefix){
+          const colonIndex = tagName.indexOf(":");
+          if(colonIndex !== -1){
+            tagName = tagName.substr(colonIndex+1);
+          }
+        }
+
+        if(currentNode){
+          textData = this.parseTextData(textData
+            , currentNode.tagname
+            , jPath
+            ,false
+            , currentNode.attributes ? Object.keys(currentNode.attributes).length !== 0 : false
+            , Object.keys(currentNode.child).length === 0);
+          if(textData !== undefined &&  textData !== "") currentNode.add(this.options.textNodeName, textData);
+          textData = "";
+        }
+
+        jPath = jPath.substr(0, jPath.lastIndexOf("."));
+        
+        currentNode = this.tagsNodeStack.pop();//avoid recurssion, set the parent tag scope
+        textData = "";
+        i = closeIndex;
+      } else if( xmlData[i+1] === '?') {
+        i = findClosingIndex(xmlData, "?>", i, "Pi Tag is not closed.")
+      } else if(xmlData.substr(i + 1, 3) === '!--') {
+        const endIndex = findClosingIndex(xmlData, "-->", i, "Comment is not closed.")
+        if(this.options.commentPropName){
+          const comment = xmlData.substring(i + 4, endIndex - 2);
+
+          //TODO: remove repeated code
+          if(textData){ //store previously collected data as textNode
+            textData = this.parseTextData(textData
+              , currentNode.tagname
+              , jPath
+              ,false
+              , currentNode.attributes ? Object.keys(currentNode.attributes).length !== 0 : false
+              , Object.keys(currentNode.child).length === 0);
+  
+            if(textData !== undefined &&  textData !== "") currentNode.add(this.options.textNodeName, textData);
+            textData = "";
+          }
+          currentNode.add(this.options.commentPropName, [ { [this.options.textNodeName] : comment } ]);
+        }
+        i = endIndex;
+      } else if( xmlData.substr(i + 1, 2) === '!D') {
+        const result = readDocType(xmlData, i);
+        this.docTypeEntities = result.entities;
+        i = result.i;
+      }else if(xmlData.substr(i + 1, 2) === '![') {
+        const closeIndex = findClosingIndex(xmlData, "]]>", i, "CDATA is not closed.") - 2;
+        const tagExp = xmlData.substring(i + 9,closeIndex);
+
+        if(textData){ //store previously collected data as textNode
+          textData = this.parseTextData(textData
+            , currentNode.tagname
+            , jPath
+            ,false
+            , currentNode.attributes ? Object.keys(currentNode.attributes).length !== 0 : false
+            , Object.keys(currentNode.child).length === 0);
+
+          if(textData !== undefined &&  textData !== "") currentNode.add(this.options.textNodeName, textData);
+          textData = "";
+        }
+
+        //cdata should be set even if it is 0 length string
+        if(this.options.cdataPropName){
+          // let val = this.parseTextData(tagExp, this.options.cdataPropName, jPath + "." + this.options.cdataPropName, true, false, true);
+          // if(!val) val = "";
+          currentNode.add(this.options.cdataPropName, [ { [this.options.textNodeName] : tagExp } ]);
+        }else{
+          let val = this.parseTextData(tagExp, currentNode.tagname, jPath, true, false, true);
+          if(!val) val = "";
+          currentNode.add(this.options.textNodeName, val);
+        }
+        
+        i = closeIndex + 2;
+      }else {//Opening tag
+       
+        let result = this.readTagExp(xmlData,i);
+        let tagName= result.tagName;
+        let tagExp = result.tagExp;
+        let attrExpPresent = result.attrExpPresent;
+        let closeIndex = result.closeIndex;
+        
+        //save text as child node
+        if (currentNode && textData) {
+          if(currentNode.tagname !== '!xml'){
+            //when nested tag is found
+            textData = this.parseTextData(textData
+              , currentNode.tagname
+              , jPath
+              , false
+              , currentNode.attributes ? Object.keys(currentNode.attributes).length !== 0 : false
+              , false);
+            if(textData !== undefined &&  textData !== "") currentNode.add(this.options.textNodeName, textData);
+            textData = "";
+          }
+        }
+
+        if(tagName !== xmlObj.tagname){
+          jPath += jPath ? "." + tagName : tagName;
+        }
+
+        //check if last tag was unpaired tag
+        const lastTag = currentNode;
+        if(lastTag && this.options.unpairedTags.indexOf(lastTag.tagname) !== -1 ){
+          currentNode = this.tagsNodeStack.pop();
+        }
+
+        if (this.isItStopNode(this.options.stopNodes, jPath, tagName)) { //TODO: namespace
+          let tagContent = "";
+          //self-closing tag
+          if(tagExp.length > 0 && tagExp.lastIndexOf("/") === tagExp.length - 1){}
+          //boolean tag
+          else if(this.options.unpairedTags.indexOf(tagName) !== -1){}
+          //normal tag
+          else{
+            //read until closing tag is found
+            const result = this.readStopNodeData(xmlData, tagName, closeIndex + 1);
+            if(!result) throw new Error(`Unexpected end of ${tagName}`);
+            i = result.i;
+            tagContent = result.tagContent;
+          }
+
+          const childNode = new xmlNode(tagName);
+          if(tagName !== tagExp && attrExpPresent){
+            childNode.attributes = this.buildAttributesMap(tagExp, jPath);
+          }
+          jPath = jPath.substr(0, jPath.lastIndexOf("."));
+          childNode.add(this.options.textNodeName, tagContent);
+          
+          currentNode.addChild(childNode);
+        }else{
+  //selfClosing tag
+          if(tagExp.length > 0 && tagExp.lastIndexOf("/") === tagExp.length - 1){
+            
+            if(tagName[tagName.length - 1] === "/"){ //remove trailing '/'
+              tagName = tagName.substr(0, tagName.length - 1);
+              tagExp = tagName;
+            }else{
+              tagExp = tagExp.substr(0, tagExp.length - 1);
+            }
+
+            const childNode = new xmlNode(tagName);
+            if(tagName !== tagExp && attrExpPresent){
+              childNode.attributes = this.buildAttributesMap(tagExp, jPath);
+            }
+            jPath = jPath.substr(0, jPath.lastIndexOf("."));
+            currentNode.addChild(childNode);
+          }
+    //opening tag
+          else{
+            const childNode = new xmlNode( tagName);
+            this.tagsNodeStack.push(currentNode);
+            
+            if(tagName !== tagExp && attrExpPresent){
+              childNode.attributes = this.buildAttributesMap(tagExp, jPath);
+            }
+            currentNode.addChild(childNode);
+            currentNode = childNode;
+          }
+          textData = "";
+          i = closeIndex;
+        }
+      }
+    }else{
+      textData += xmlData[i];
+    }
+  }
+  return xmlObj.child;
+}
+
+const replaceEntitiesValue = function(val){
+  if(this.options.processEntities){
+    for(let entityName in this.docTypeEntities){
+      const entity = this.docTypeEntities[entityName];
+      val = val.replace( entity.regx, entity.val);
+    }
+    for(let entityName in this.lastEntities){
+      const entity = this.lastEntities[entityName];
+      val = val.replace( entity.regex, entity.val);
+    }
+    if(this.options.htmlEntities){
+      for(let entityName in this.htmlEntities){
+        const entity = this.htmlEntities[entityName];
+        val = val.replace( entity.regex, entity.val);
+      }
+    }
+  }
+  return val;
+}
+//TODO: use jPath to simplify the logic
+/**
+ * 
+ * @param {string[]} stopNodes 
+ * @param {string} jPath
+ * @param {string} currentTagName 
+ */
+function isItStopNode(stopNodes, jPath, currentTagName){
+  const allNodesExp = "*." + currentTagName;
+  for (const stopNodePath in stopNodes) {
+    const stopNodeExp = stopNodes[stopNodePath];
+    if( allNodesExp === stopNodeExp || jPath === stopNodeExp  ) return true;
+  }
+  return false;
+}
+
+/**
+ * Returns the tag Expression and where it is ending handling single-dobule quotes situation
+ * @param {string} xmlData 
+ * @param {number} i starting index
+ * @returns 
+ */
+function tagExpWithClosingIndex(xmlData, i){
+  let attrBoundary;
+  let tagExp = "";
+  for (let index = i; index < xmlData.length; index++) {
+    let ch = xmlData[index];
+    if (attrBoundary) {
+        if (ch === attrBoundary) attrBoundary = "";//reset
+    } else if (ch === '"' || ch === "'") {
+        attrBoundary = ch;
+    } else if (ch === '>') {
+        return {
+          data: tagExp,
+          index: index
+        }
+    } else if (ch === '\t') {
+      ch = " "
+    }
+    tagExp += ch;
+  }
+}
+
+function findClosingIndex(xmlData, str, i, errMsg){
+  const closingIndex = xmlData.indexOf(str, i);
+  if(closingIndex === -1){
+    throw new Error(errMsg)
+  }else{
+    return closingIndex + str.length - 1;
+  }
+}
+
+function readTagExp(xmlData,i){
+  const result = tagExpWithClosingIndex(xmlData, i+1);
+  let tagExp = result.data;
+  const closeIndex = result.index;
+  const separatorIndex = tagExp.search(/\s/);
+  let tagName = tagExp;
+  let attrExpPresent = true;
+  if(separatorIndex !== -1){//separate tag name and attributes expression
+    tagName = tagExp.substr(0, separatorIndex).replace(/\s\s*$/, '');
+    tagExp = tagExp.substr(separatorIndex + 1);
+  }
+
+  if(this. options.removeNSPrefix){
+    const colonIndex = tagName.indexOf(":");
+    if(colonIndex !== -1){
+      tagName = tagName.substr(colonIndex+1);
+      attrExpPresent = tagName !== result.data.substr(colonIndex + 1);
+    }
+  }
+
+  return {
+    tagName: tagName,
+    tagExp: tagExp,
+    closeIndex: closeIndex,
+    attrExpPresent: attrExpPresent,
+  }
+}
+/**
+ * find paired tag for a stop node
+ * @param {string} xmlData 
+ * @param {string} tagName 
+ * @param {number} i 
+ */
+function readStopNodeData(xmlData, tagName, i){
+  const startIndex = i;
+  for (; i < xmlData.length; i++) {
+    if( xmlData[i] === "<" && xmlData[i+1] === "/"){ 
+        const closeIndex = findClosingIndex(xmlData, ">", i, `${tagName} is not closed`);
+        let closeTagName = xmlData.substring(i+2,closeIndex).trim();
+        if(closeTagName === tagName){
+          return {
+            tagContent: xmlData.substring(startIndex, i),
+            i : closeIndex
+          }
+        }
+        i=closeIndex;
+      }
+  }//end for loop
+}
+
+function parseValue(val, shouldParse, options) {
+  if (shouldParse && typeof val === 'string') {
+    //console.log(options)
+    const newval = val.trim();
+    if(newval === 'true' ) return true;
+    else if(newval === 'false' ) return false;
+    else return toNumber(val, options);
+  } else {
+    if (util.isExist(val)) {
+      return val;
+    } else {
+      return '';
+    }
+  }
+}
+
+
+module.exports = OrderedObjParser;

package/src/xmlparser/XMLParser.js

@@ -0,0 +1,37 @@
+const { buildOptions} = require("./OptionsBuilder");
+const OrderedObjParser = require("./OrderedObjParser");
+const { prettify} = require("./node2json");
+const validator = require('../validator');
+
+class XMLParser{
+    constructor(options){
+        this.options = buildOptions(options);
+    }
+    /**
+     * Parse XML dats to JS object 
+     * @param {string|Buffer} xmlData 
+     * @param {boolean|Object} validationOption 
+     */
+    parse(xmlData,validationOption){
+        if(typeof xmlData === "string"){
+        }else if( xmlData.toString){
+            xmlData = xmlData.toString();
+        }else{
+            throw new Error("XML data is accepted in String or Bytes[] form.")
+        }
+        if( validationOption){
+            if(validationOption === true) validationOption = {}; //validate with default options
+            
+            const result = validator.validate(xmlData, validationOption);
+            if (result !== true) {
+              throw Error( `${result.err.msg}:${result.err.line}:${result.err.col}` )
+            }
+          }
+        const orderedObjParser = new OrderedObjParser(this.options);
+        const orderedResult = orderedObjParser.parseXml(xmlData);
+        if(this.options.preserveOrder || orderedResult === undefined) return orderedResult;
+        else return prettify(orderedResult, this.options);
+    }
+}
+
+module.exports = XMLParser;

package/src/xmlparser/node2json.js

@@ -0,0 +1,101 @@
+'use strict';
+
+/**
+ * 
+ * @param {array} node 
+ * @param {any} options 
+ * @returns 
+ */
+function prettify(node, options){
+  return compress( node, options);
+}
+
+/**
+ * 
+ * @param {array} arr 
+ * @param {object} options 
+ * @param {string} jPath 
+ * @returns object
+ */
+function compress(arr, options, jPath){
+  let text;
+  const compressedObj = {};
+  for (let i = 0; i < arr.length; i++) {
+    const tagObj = arr[i];
+    const property = propName(tagObj);
+    let newJpath = "";
+    if(jPath === undefined) newJpath = property;
+    else newJpath = jPath + "." + property;
+
+    if(property === options.textNodeName){
+      if(text === undefined) text = tagObj[property];
+      else text += "" + tagObj[property];
+    }else if(property === undefined){
+      continue;
+    }else if(tagObj[property]){
+      
+      let val = compress(tagObj[property], options, newJpath);
+      const isLeaf = isLeafTag(val, options);
+
+      if(tagObj.attributes){
+        assignAttributes( val, tagObj.attributes, newJpath, options);
+      }else if(Object.keys(val).length === 1 && val[options.textNodeName] !== undefined && !options.alwaysCreateTextNode){
+        val = val[options.textNodeName];
+      }else if(Object.keys(val).length === 0){
+        if(options.alwaysCreateTextNode) val[options.textNodeName] = "";
+        else val = "";
+      }
+
+      if(compressedObj[property] !== undefined) {
+        if(!Array.isArray(compressedObj[property])) {
+          compressedObj[property] = [ compressedObj[property] ];
+        }
+        compressedObj[property].push(val);
+      }else{
+        //TODO: if a node is not an array, then check if it should be an array
+        //also determine if it is a leaf node
+        if (options.isArray(property, newJpath, isLeaf )) {
+          compressedObj[property] = [val];
+        }else{
+          compressedObj[property] = val;
+        }
+      }
+    }
+    
+  }
+  // if(text && text.length > 0) compressedObj[options.textNodeName] = text;
+  if(typeof text === "string"){
+    if(text.length > 0) compressedObj[options.textNodeName] = text;
+  }else if(text !== undefined) compressedObj[options.textNodeName] = text;
+  return compressedObj;
+}
+
+function propName(obj){
+  const keys = Object.keys(obj);
+  for (let i = 0; i < keys.length; i++) {
+    const key = keys[i];
+    if(key !== "attributes") return key;
+  }
+}
+
+function assignAttributes(obj, attrMap, jpath, options){
+  if (attrMap) {
+    const keys = Object.keys(attrMap);
+    const len = keys.length; //don't make it inline
+    for (let i = 0; i < len; i++) {
+      const atrrName = keys[i];
+      if (options.isArray(atrrName, jpath + "." + atrrName, true, true)) {
+        obj[atrrName] = [ attrMap[atrrName] ];
+      } else {
+        obj[atrrName] = attrMap[atrrName];
+      }
+    }
+  }
+}
+
+function isLeafTag(obj, options){
+  const propCount = Object.keys(obj).length;
+  if( propCount === 0 || (propCount === 1 && obj[options.textNodeName]) ) return true;
+  return false;
+}
+exports.prettify = prettify;

package/src/xmlparser/xmlNode.js

@@ -0,0 +1,23 @@
+'use strict';
+
+class XmlNode{
+  constructor(tagname) {
+    this.tagname = tagname;
+    this.child = []; //nested tags, text, cdata, comments in order
+    this.attributes = {}; //attributes map
+  }
+  add(key,val){
+    // this.child.push( {name : key, val: val, isCdata: isCdata });
+    this.child.push( {[key]: val });
+  }
+  addChild(node) {
+    if(node.attributes && Object.keys(node.attributes).length > 0){
+      this.child.push( { [node.tagname]: node.child, attributes: node.attributes });
+    }else{
+      this.child.push( { [node.tagname]: node.child });
+    }
+  };
+};
+
+
+module.exports = XmlNode;