@qqbrowser/openclaw-qbot 0.10.9 → 0.10.10

package/node_modules/fast-xml-parser/src/xmlparser/OrderedObjParser.js

@@ -7,6 +7,8 @@ import DocTypeReader from './DocTypeReader.js';
 import toNumber from "strnum";
 import getIgnoreAttributesFn from "../ignoreAttributes.js";
 import { Expression, Matcher } from 'path-expression-matcher';
+import { ExpressionSet } from 'path-expression-matcher';
+import { EntityDecoder, XML, CURRENCY, COMMON_HTML } from '@nodable/entities';
 
 // const regx =
 //   '<((!\\[CDATA\\[([\\s\\S]*?)(]]>))|((NAME:)?(NAME))([^>]*)>|((\\/)(NAME)\\s*>))([^<]*)'
@@ -67,36 +69,10 @@ function extractNamespace(rawTagName) {
 }
 
 export default class OrderedObjParser {
-  constructor(options) {
+  constructor(options, externalEntities) {
     this.options = options;
     this.currentNode = null;
     this.tagsNodeStack = [];
-    this.docTypeEntities = {};
-    this.lastEntities = {
-      "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.ampEntity = { regex: /&(amp|#38|#x26);/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: "₹" },
-      "num_dec": { regex: /&#([0-9]{1,7});/g, val: (_, str) => fromCodePoint(str, 10, "&#") },
-      "num_hex": { regex: /&#x([0-9a-fA-F]{1,6});/g, val: (_, str) => fromCodePoint(str, 16, "&#x") },
-    };
-    this.addExternalEntities = addExternalEntities;
     this.parseXml = parseXml;
     this.parseTextData = parseTextData;
     this.resolveNameSpace = resolveNameSpace;
@@ -109,6 +85,23 @@ export default class OrderedObjParser {
     this.ignoreAttributesFn = getIgnoreAttributesFn(this.options.ignoreAttributes)
     this.entityExpansionCount = 0;
     this.currentExpandedLength = 0;
+    let namedEntities = { ...XML };
+    if (this.options.entityDecoder) {
+      this.entityDecoder = this.options.entityDecoder
+    } else {
+      if (typeof this.options.htmlEntities === "object") namedEntities = this.options.htmlEntities;
+      else if (this.options.htmlEntities === true) namedEntities = { ...COMMON_HTML, ...CURRENCY };
+      this.entityDecoder = new EntityDecoder({
+        namedEntities: { ...namedEntities, ...externalEntities },
+        numericAllowed: this.options.htmlEntities,
+        limit: {
+          maxTotalExpansions: this.options.processEntities.maxTotalExpansions,
+          maxExpandedLength: this.options.processEntities.maxExpandedLength,
+          applyLimitsTo: this.options.processEntities.appliesTo,
+        }
+        //postCheck: resolved => resolved
+      });
+    }
 
     // Initialize path matcher for path-expression-matcher
     this.matcher = new Matcher();
@@ -121,34 +114,25 @@ export default class OrderedObjParser {
     this.isCurrentNodeStopNode = false;
 
     // Pre-compile stopNodes expressions
-    if (this.options.stopNodes && this.options.stopNodes.length > 0) {
-      this.stopNodeExpressions = [];
-      for (let i = 0; i < this.options.stopNodes.length; i++) {
-        const stopNodeExp = this.options.stopNodes[i];
+    this.stopNodeExpressionsSet = new ExpressionSet();
+    const stopNodesOpts = this.options.stopNodes;
+    if (stopNodesOpts && stopNodesOpts.length > 0) {
+      for (let i = 0; i < stopNodesOpts.length; i++) {
+        const stopNodeExp = stopNodesOpts[i];
         if (typeof stopNodeExp === 'string') {
           // Convert string to Expression object
-          this.stopNodeExpressions.push(new Expression(stopNodeExp));
+          this.stopNodeExpressionsSet.add(new Expression(stopNodeExp));
         } else if (stopNodeExp instanceof Expression) {
           // Already an Expression object
-          this.stopNodeExpressions.push(stopNodeExp);
+          this.stopNodeExpressionsSet.add(stopNodeExp);
         }
       }
+      this.stopNodeExpressionsSet.seal();
     }
   }
 
 }
 
-function addExternalEntities(externalEntities) {
-  const entKeys = Object.keys(externalEntities);
-  for (let i = 0; i < entKeys.length; i++) {
-    const ent = entKeys[i];
-    const escaped = ent.replace(/[.\-+*:]/g, '\\.');
-    this.lastEntities[ent] = {
-      regex: new RegExp("&" + escaped + ";", "g"),
-      val: externalEntities[ent]
-    }
-  }
-}
 
 /**
  * @param {string} val
@@ -160,28 +144,29 @@ function addExternalEntities(externalEntities) {
  * @param {boolean} escapeEntities
  */
 function parseTextData(val, tagName, jPath, dontTrim, hasAttributes, isLeafNode, escapeEntities) {
+  const options = this.options;
   if (val !== undefined) {
-    if (this.options.trimValues && !dontTrim) {
+    if (options.trimValues && !dontTrim) {
       val = val.trim();
     }
     if (val.length > 0) {
       if (!escapeEntities) val = this.replaceEntitiesValue(val, tagName, jPath);
 
       // Pass jPath string or matcher based on options.jPath setting
-      const jPathOrMatcher = this.options.jPath ? jPath.toString() : jPath;
-      const newval = this.options.tagValueProcessor(tagName, val, jPathOrMatcher, hasAttributes, isLeafNode);
+      const jPathOrMatcher = options.jPath ? jPath.toString() : jPath;
+      const newval = options.tagValueProcessor(tagName, val, jPathOrMatcher, 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 if (options.trimValues) {
+        return parseValue(val, options.parseTagValue, options.numberParseOptions);
       } else {
         const trimmedVal = val.trim();
         if (trimmedVal === val) {
-          return parseValue(val, this.options.parseTagValue, this.options.numberParseOptions);
+          return parseValue(val, options.parseTagValue, options.numberParseOptions);
         } else {
           return val;
         }
@@ -208,8 +193,9 @@ function resolveNameSpace(tagname) {
 //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, tagName) {
-  if (this.options.ignoreAttributes !== true && typeof attrStr === 'string') {
+function buildAttributesMap(attrStr, jPath, tagName, force = false) {
+  const options = this.options;
+  if (force === true || (options.ignoreAttributes !== true && typeof attrStr === 'string')) {
     // attrStr = attrStr.replace(/\r?\n/g, ' ');
     //attrStr = attrStr || attrStr.trim();
 
@@ -217,89 +203,80 @@ function buildAttributesMap(attrStr, jPath, tagName) {
     const len = matches.length; //don't make it inline
     const attrs = {};
 
-    // First pass: parse all attributes and update matcher with raw values
-    // This ensures the matcher has all attribute values when processors run
+    // Pre-process values once: trim + entity replacement
+    // Reused in both matcher update and second pass
+    const processedVals = new Array(len);
+    let hasRawAttrs = false;
     const rawAttrsForMatcher = {};
+
     for (let i = 0; i < len; i++) {
       const attrName = this.resolveNameSpace(matches[i][1]);
       const oldVal = matches[i][4];
 
       if (attrName.length && oldVal !== undefined) {
-        let parsedVal = oldVal;
-        if (this.options.trimValues) {
-          parsedVal = parsedVal.trim();
-        }
-        parsedVal = this.replaceEntitiesValue(parsedVal, tagName, this.readonlyMatcher);
-        rawAttrsForMatcher[attrName] = parsedVal;
+        let val = oldVal;
+        if (options.trimValues) val = val.trim();
+        val = this.replaceEntitiesValue(val, tagName, this.readonlyMatcher);
+        processedVals[i] = val;
+
+        rawAttrsForMatcher[attrName] = val;
+        hasRawAttrs = true;
       }
     }
 
-    // Update matcher with raw attribute values BEFORE running processors
-    if (Object.keys(rawAttrsForMatcher).length > 0 && typeof jPath === 'object' && jPath.updateCurrent) {
+    // Update matcher ONCE before second pass, if applicable
+    if (hasRawAttrs && typeof jPath === 'object' && jPath.updateCurrent) {
       jPath.updateCurrent(rawAttrsForMatcher);
     }
 
-    // Second pass: now process attributes with matcher having full attribute context
+    // Hoist toString() once — path doesn't change during attribute processing
+    const jPathStr = options.jPath ? jPath.toString() : this.readonlyMatcher;
+
+    // Second pass: apply processors, build final attrs
+    let hasAttrs = false;
     for (let i = 0; i < len; i++) {
       const attrName = this.resolveNameSpace(matches[i][1]);
 
-      // Convert jPath to string if needed for ignoreAttributesFn
-      const jPathStr = this.options.jPath ? jPath.toString() : this.readonlyMatcher;
-      if (this.ignoreAttributesFn(attrName, jPathStr)) {
-        continue
-      }
+      if (this.ignoreAttributesFn(attrName, jPathStr)) continue;
 
-      let oldVal = matches[i][4];
-      let aName = this.options.attributeNamePrefix + attrName;
+      let aName = options.attributeNamePrefix + attrName;
 
       if (attrName.length) {
-        if (this.options.transformAttributeName) {
-          aName = this.options.transformAttributeName(aName);
+        if (options.transformAttributeName) {
+          aName = options.transformAttributeName(aName);
         }
-        //if (aName === "__proto__") aName = "#__proto__";
-        aName = sanitizeName(aName, this.options);
+        aName = sanitizeName(aName, options);
 
-        if (oldVal !== undefined) {
-          if (this.options.trimValues) {
-            oldVal = oldVal.trim();
-          }
-          oldVal = this.replaceEntitiesValue(oldVal, tagName, this.readonlyMatcher);
+        if (matches[i][4] !== undefined) {
+          // Reuse already-processed value — no double entity replacement
+          const oldVal = processedVals[i];
 
-          // Pass jPath string or readonlyMatcher based on options.jPath setting
-          const jPathOrMatcher = this.options.jPath ? jPath.toString() : this.readonlyMatcher;
-          const newVal = this.options.attributeValueProcessor(attrName, oldVal, jPathOrMatcher);
+          const newVal = options.attributeValueProcessor(attrName, oldVal, jPathStr);
           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
-            );
+            attrs[aName] = parseValue(oldVal, options.parseAttributeValue, options.numberParseOptions);
           }
-        } else if (this.options.allowBooleanAttributes) {
+          hasAttrs = true;
+        } else if (options.allowBooleanAttributes) {
           attrs[aName] = true;
+          hasAttrs = true;
         }
       }
     }
 
-    if (!Object.keys(attrs).length) {
-      return;
-    }
-    if (this.options.attributesGroupName) {
+    if (!hasAttrs) return;
+
+    if (options.attributesGroupName && !options.preserveOrder) {
       const attrCollection = {};
-      attrCollection[this.options.attributesGroupName] = attrs;
+      attrCollection[options.attributesGroupName] = attrs;
       return attrCollection;
     }
-    return attrs
+    return attrs;
   }
 }
-
 const parseXml = function (xmlData) {
   xmlData = xmlData.replace(/\r\n?/g, "\n"); //TODO: remove this line
   const xmlObj = new xmlNode('!xml');
@@ -308,29 +285,32 @@ const parseXml = function (xmlData) {
 
   // Reset matcher for new document
   this.matcher.reset();
+  this.entityDecoder.reset();
 
   // Reset entity expansion counters for this document
   this.entityExpansionCount = 0;
   this.currentExpandedLength = 0;
-
-  const docTypeReader = new DocTypeReader(this.options.processEntities);
-  for (let i = 0; i < xmlData.length; i++) {//for each char in XML data
+  const options = this.options;
+  const docTypeReader = new DocTypeReader(options.processEntities);
+  const xmlLen = xmlData.length;
+  for (let i = 0; i < xmlLen; 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 c1 = xmlData.charCodeAt(i + 1);
+      if (c1 === 47) {//Closing Tag '/'
         const closeIndex = findClosingIndex(xmlData, ">", i, "Closing Tag is not closed.")
         let tagName = xmlData.substring(i + 2, closeIndex).trim();
 
-        if (this.options.removeNSPrefix) {
+        if (options.removeNSPrefix) {
           const colonIndex = tagName.indexOf(":");
           if (colonIndex !== -1) {
             tagName = tagName.substr(colonIndex + 1);
           }
         }
 
-        tagName = transformTagName(this.options.transformTagName, tagName, "", this.options).tagName;
+        tagName = transformTagName(options.transformTagName, tagName, "", options).tagName;
 
         if (currentNode) {
           textData = this.saveTextToParentTag(textData, currentNode, this.readonlyMatcher);
@@ -338,10 +318,10 @@ const parseXml = function (xmlData) {
 
         //check if last tag of nested tag was unpaired tag
         const lastTagName = this.matcher.getCurrentTag();
-        if (tagName && this.options.unpairedTags.indexOf(tagName) !== -1) {
+        if (tagName && options.unpairedTagsSet.has(tagName)) {
           throw new Error(`Unpaired tag can not be used as closing tag: </${tagName}>`);
         }
-        if (lastTagName && this.options.unpairedTags.indexOf(lastTagName) !== -1) {
+        if (lastTagName && options.unpairedTagsSet.has(lastTagName)) {
           // Pop the unpaired tag
           this.matcher.pop();
           this.tagsNodeStack.pop();
@@ -353,42 +333,51 @@ const parseXml = function (xmlData) {
         currentNode = this.tagsNodeStack.pop();//avoid recursion, set the parent tag scope
         textData = "";
         i = closeIndex;
-      } else if (xmlData[i + 1] === '?') {
+      } else if (c1 === 63) { //'?'
 
         let tagData = readTagExp(xmlData, i, false, "?>");
         if (!tagData) throw new Error("Pi Tag is not closed.");
 
         textData = this.saveTextToParentTag(textData, currentNode, this.readonlyMatcher);
-        if ((this.options.ignoreDeclaration && tagData.tagName === "?xml") || this.options.ignorePiTags) {
+        const attsMap = this.buildAttributesMap(tagData.tagExp, this.matcher, tagData.tagName, true);
+        if (attsMap) {
+          const ver = attsMap[this.options.attributeNamePrefix + "version"];
+          this.entityDecoder.setXmlVersion(Number(ver) || 1.0);
+        }
+        if ((options.ignoreDeclaration && tagData.tagName === "?xml") || options.ignorePiTags) {
           //do nothing
         } else {
 
           const childNode = new xmlNode(tagData.tagName);
-          childNode.add(this.options.textNodeName, "");
+          childNode.add(options.textNodeName, "");
 
-          if (tagData.tagName !== tagData.tagExp && tagData.attrExpPresent) {
-            childNode[":@"] = this.buildAttributesMap(tagData.tagExp, this.matcher, tagData.tagName);
+          if (tagData.tagName !== tagData.tagExp && tagData.attrExpPresent && options.ignoreAttributes !== true) {
+            childNode[":@"] = attsMap
           }
           this.addChild(currentNode, childNode, this.readonlyMatcher, i);
         }
 
 
         i = tagData.closeIndex + 1;
-      } else if (xmlData.substr(i + 1, 3) === '!--') {
+      } else if (c1 === 33
+        && xmlData.charCodeAt(i + 2) === 45
+        && xmlData.charCodeAt(i + 3) === 45) { //'!--'
         const endIndex = findClosingIndex(xmlData, "-->", i + 4, "Comment is not closed.")
-        if (this.options.commentPropName) {
+        if (options.commentPropName) {
           const comment = xmlData.substring(i + 4, endIndex - 2);
 
           textData = this.saveTextToParentTag(textData, currentNode, this.readonlyMatcher);
 
-          currentNode.add(this.options.commentPropName, [{ [this.options.textNodeName]: comment }]);
+          currentNode.add(options.commentPropName, [{ [options.textNodeName]: comment }]);
         }
         i = endIndex;
-      } else if (xmlData.substr(i + 1, 2) === '!D') {
+      } else if (c1 === 33
+        && xmlData.charCodeAt(i + 2) === 68) { //'!D'
         const result = docTypeReader.readDocType(xmlData, i);
-        this.docTypeEntities = result.entities;
+        this.entityDecoder.addInputEntities(result.entities);
         i = result.i;
-      } else if (xmlData.substr(i + 1, 2) === '![') {
+      } else if (c1 === 33
+        && xmlData.charCodeAt(i + 2) === 91) { // '!['
         const closeIndex = findClosingIndex(xmlData, "]]>", i, "CDATA is not closed.") - 2;
         const tagExp = xmlData.substring(i + 9, closeIndex);
 
@@ -398,20 +387,20 @@ const parseXml = function (xmlData) {
         if (val == undefined) val = "";
 
         //cdata should be set even if it is 0 length string
-        if (this.options.cdataPropName) {
-          currentNode.add(this.options.cdataPropName, [{ [this.options.textNodeName]: tagExp }]);
+        if (options.cdataPropName) {
+          currentNode.add(options.cdataPropName, [{ [options.textNodeName]: tagExp }]);
         } else {
-          currentNode.add(this.options.textNodeName, val);
+          currentNode.add(options.textNodeName, val);
         }
 
         i = closeIndex + 2;
       } else {//Opening tag
-        let result = readTagExp(xmlData, i, this.options.removeNSPrefix);
+        let result = readTagExp(xmlData, i, options.removeNSPrefix);
 
         // Safety check: readTagExp can return undefined
         if (!result) {
           // Log context for debugging
-          const context = xmlData.substring(Math.max(0, i - 50), Math.min(xmlData.length, i + 50));
+          const context = xmlData.substring(Math.max(0, i - 50), Math.min(xmlLen, i + 50));
           throw new Error(`readTagExp returned undefined at position ${i}. Context: "${context}"`);
         }
 
@@ -421,13 +410,13 @@ const parseXml = function (xmlData) {
         let attrExpPresent = result.attrExpPresent;
         let closeIndex = result.closeIndex;
 
-        ({ tagName, tagExp } = transformTagName(this.options.transformTagName, tagName, tagExp, this.options));
+        ({ tagName, tagExp } = transformTagName(options.transformTagName, tagName, tagExp, options));
 
-        if (this.options.strictReservedNames &&
-          (tagName === this.options.commentPropName
-            || tagName === this.options.cdataPropName
-            || tagName === this.options.textNodeName
-            || tagName === this.options.attributesGroupName
+        if (options.strictReservedNames &&
+          (tagName === options.commentPropName
+            || tagName === options.cdataPropName
+            || tagName === options.textNodeName
+            || tagName === options.attributesGroupName
           )) {
           throw new Error(`Invalid tag name: ${tagName}`);
         }
@@ -442,7 +431,7 @@ const parseXml = function (xmlData) {
 
         //check if last tag was unpaired tag
         const lastTag = currentNode;
-        if (lastTag && this.options.unpairedTags.indexOf(lastTag.tagname) !== -1) {
+        if (lastTag && options.unpairedTagsSet.has(lastTag.tagname)) {
           currentNode = this.tagsNodeStack.pop();
           this.matcher.pop();
         }
@@ -484,13 +473,14 @@ const parseXml = function (xmlData) {
 
           if (prefixedAttrs) {
             // Extract raw attributes (without prefix) for our use
-            rawAttrs = extractRawAttributes(prefixedAttrs, this.options);
+            //TODO: seems a performance overhead
+            rawAttrs = extractRawAttributes(prefixedAttrs, options);
           }
         }
 
         // Now check if this is a stop node (after attributes are set)
         if (tagName !== xmlObj.tagname) {
-          this.isCurrentNodeStopNode = this.isItStopNode(this.stopNodeExpressions, this.matcher);
+          this.isCurrentNodeStopNode = this.isItStopNode();
         }
 
         const startIndex = i;
@@ -502,7 +492,7 @@ const parseXml = function (xmlData) {
             i = result.closeIndex;
           }
           //unpaired tag
-          else if (this.options.unpairedTags.indexOf(tagName) !== -1) {
+          else if (options.unpairedTagsSet.has(tagName)) {
             i = result.closeIndex;
           }
           //normal tag
@@ -521,7 +511,7 @@ const parseXml = function (xmlData) {
           }
 
           // For stop nodes, store raw content as-is without any processing
-          childNode.add(this.options.textNodeName, tagContent);
+          childNode.add(options.textNodeName, tagContent);
 
           this.matcher.pop(); // Pop the stop node tag
           this.isCurrentNodeStopNode = false; // Reset flag
@@ -530,7 +520,7 @@ const parseXml = function (xmlData) {
         } else {
           //selfClosing tag
           if (isSelfClosing) {
-            ({ tagName, tagExp } = transformTagName(this.options.transformTagName, tagName, tagExp, this.options));
+            ({ tagName, tagExp } = transformTagName(options.transformTagName, tagName, tagExp, options));
 
             const childNode = new xmlNode(tagName);
             if (prefixedAttrs) {
@@ -540,7 +530,7 @@ const parseXml = function (xmlData) {
             this.matcher.pop(); // Pop self-closing tag
             this.isCurrentNodeStopNode = false; // Reset flag
           }
-          else if (this.options.unpairedTags.indexOf(tagName) !== -1) {//unpaired tag
+          else if (options.unpairedTagsSet.has(tagName)) {//unpaired tag
             const childNode = new xmlNode(tagName);
             if (prefixedAttrs) {
               childNode[":@"] = prefixedAttrs;
@@ -555,7 +545,7 @@ const parseXml = function (xmlData) {
           //opening tag
           else {
             const childNode = new xmlNode(tagName);
-            if (this.tagsNodeStack.length > this.options.maxNestedTags) {
+            if (this.tagsNodeStack.length > options.maxNestedTags) {
               throw new Error("Maximum nested tags exceeded");
             }
             this.tagsNodeStack.push(currentNode);
@@ -626,79 +616,7 @@ function replaceEntitiesValue(val, tagName, jPath) {
     }
   }
 
-  // Replace DOCTYPE entities
-  for (const entityName of Object.keys(this.docTypeEntities)) {
-    const entity = this.docTypeEntities[entityName];
-    const matches = val.match(entity.regx);
-
-    if (matches) {
-      // Track expansions
-      this.entityExpansionCount += matches.length;
-
-      // Check expansion limit
-      if (entityConfig.maxTotalExpansions &&
-        this.entityExpansionCount > entityConfig.maxTotalExpansions) {
-        throw new Error(
-          `Entity expansion limit exceeded: ${this.entityExpansionCount} > ${entityConfig.maxTotalExpansions}`
-        );
-      }
-
-      // Store length before replacement
-      const lengthBefore = val.length;
-      val = val.replace(entity.regx, entity.val);
-
-      // Check expanded length immediately after replacement
-      if (entityConfig.maxExpandedLength) {
-        this.currentExpandedLength += (val.length - lengthBefore);
-
-        if (this.currentExpandedLength > entityConfig.maxExpandedLength) {
-          throw new Error(
-            `Total expanded content size exceeded: ${this.currentExpandedLength} > ${entityConfig.maxExpandedLength}`
-          );
-        }
-      }
-    }
-  }
-  // Replace standard entities
-  for (const entityName of Object.keys(this.lastEntities)) {
-    const entity = this.lastEntities[entityName];
-    const matches = val.match(entity.regex);
-    if (matches) {
-      this.entityExpansionCount += matches.length;
-      if (entityConfig.maxTotalExpansions &&
-        this.entityExpansionCount > entityConfig.maxTotalExpansions) {
-        throw new Error(
-          `Entity expansion limit exceeded: ${this.entityExpansionCount} > ${entityConfig.maxTotalExpansions}`
-        );
-      }
-    }
-    val = val.replace(entity.regex, entity.val);
-  }
-  if (val.indexOf('&') === -1) return val;
-
-  // Replace HTML entities if enabled
-  if (this.options.htmlEntities) {
-    for (const entityName of Object.keys(this.htmlEntities)) {
-      const entity = this.htmlEntities[entityName];
-      const matches = val.match(entity.regex);
-      if (matches) {
-        //console.log(matches);
-        this.entityExpansionCount += matches.length;
-        if (entityConfig.maxTotalExpansions &&
-          this.entityExpansionCount > entityConfig.maxTotalExpansions) {
-          throw new Error(
-            `Entity expansion limit exceeded: ${this.entityExpansionCount} > ${entityConfig.maxTotalExpansions}`
-          );
-        }
-      }
-      val = val.replace(entity.regex, entity.val);
-    }
-  }
-
-  // Replace ampersand entity last
-  val = val.replace(this.ampEntity.regex, this.ampEntity.val);
-
-  return val;
+  return this.entityDecoder.decode(val);
 }
 
 
@@ -720,20 +638,14 @@ function saveTextToParentTag(textData, parentNode, matcher, isLeafNode) {
   return textData;
 }
 
-//TODO: use jPath to simplify the logic
 /**
  * @param {Array<Expression>} stopNodeExpressions - Array of compiled Expression objects
  * @param {Matcher} matcher - Current path matcher
  */
-function isItStopNode(stopNodeExpressions, matcher) {
-  if (!stopNodeExpressions || stopNodeExpressions.length === 0) return false;
+function isItStopNode() {
+  if (this.stopNodeExpressionsSet.size === 0) return false;
 
-  for (let i = 0; i < stopNodeExpressions.length; i++) {
-    if (matcher.matches(stopNodeExpressions[i])) {
-      return true;
-    }
-  }
-  return false;
+  return this.matcher.matchesAny(this.stopNodeExpressionsSet);
 }
 
 /**
@@ -743,32 +655,38 @@ function isItStopNode(stopNodeExpressions, matcher) {
  * @returns 
  */
 function tagExpWithClosingIndex(xmlData, i, closingChar = ">") {
-  let attrBoundary;
-  let tagExp = "";
-  for (let index = i; index < xmlData.length; index++) {
-    let ch = xmlData[index];
+  //TODO: ignore boolean attributes in tag expression
+  //TODO: if ignore attributes, dont read full attribute expression but the end. But read for xml declaration
+  let attrBoundary = 0;
+  const len = xmlData.length;
+  const closeCode0 = closingChar.charCodeAt(0);
+  const closeCode1 = closingChar.length > 1 ? closingChar.charCodeAt(1) : -1;
+
+  let result = '';
+  let segmentStart = i;
+
+  for (let index = i; index < len; index++) {
+    const code = xmlData.charCodeAt(index);
+
     if (attrBoundary) {
-      if (ch === attrBoundary) attrBoundary = "";//reset
-    } else if (ch === '"' || ch === "'") {
-      attrBoundary = ch;
-    } else if (ch === closingChar[0]) {
-      if (closingChar[1]) {
-        if (xmlData[index + 1] === closingChar[1]) {
-          return {
-            data: tagExp,
-            index: index
-          }
+      if (code === attrBoundary) attrBoundary = 0;
+    } else if (code === 34 || code === 39) { // " or '
+      attrBoundary = code;
+    } else if (code === closeCode0) {
+      if (closeCode1 !== -1) {
+        if (xmlData.charCodeAt(index + 1) === closeCode1) {
+          result += xmlData.substring(segmentStart, index);
+          return { data: result, index };
         }
       } else {
-        return {
-          data: tagExp,
-          index: index
-        }
+        result += xmlData.substring(segmentStart, index);
+        return { data: result, index };
       }
-    } else if (ch === '\t') {
-      ch = " "
+    } else if (code === 9 && !attrBoundary) { // \t - only replace with space outside attribute values
+      // Flush accumulated segment, add space, start new segment
+      result += xmlData.substring(segmentStart, index) + ' ';
+      segmentStart = index + 1;
     }
-    tagExp += ch;
   }
 }
 
@@ -781,6 +699,12 @@ function findClosingIndex(xmlData, str, i, errMsg) {
   }
 }
 
+function findClosingChar(xmlData, char, i, errMsg) {
+  const closingIndex = xmlData.indexOf(char, i);
+  if (closingIndex === -1) throw new Error(errMsg);
+  return closingIndex; // no offset needed
+}
+
 function readTagExp(xmlData, i, removeNSPrefix, closingChar = ">") {
   const result = tagExpWithClosingIndex(xmlData, i + 1, closingChar);
   if (!result) return;
@@ -822,10 +746,12 @@ function readStopNodeData(xmlData, tagName, i) {
   // Starting at 1 since we already have an open tag
   let openTagCount = 1;
 
-  for (; i < xmlData.length; i++) {
+  const xmllen = xmlData.length;
+  for (; i < xmllen; i++) {
     if (xmlData[i] === "<") {
-      if (xmlData[i + 1] === "/") {//close tag
-        const closeIndex = findClosingIndex(xmlData, ">", i, `${tagName} is not closed`);
+      const c1 = xmlData.charCodeAt(i + 1);
+      if (c1 === 47) {//close tag '/'
+        const closeIndex = findClosingChar(xmlData, ">", i, `${tagName} is not closed`);
         let closeTagName = xmlData.substring(i + 2, closeIndex).trim();
         if (closeTagName === tagName) {
           openTagCount--;
@@ -837,13 +763,16 @@ function readStopNodeData(xmlData, tagName, i) {
           }
         }
         i = closeIndex;
-      } else if (xmlData[i + 1] === '?') {
+      } else if (c1 === 63) { //?
         const closeIndex = findClosingIndex(xmlData, "?>", i + 1, "StopNode is not closed.")
         i = closeIndex;
-      } else if (xmlData.substr(i + 1, 3) === '!--') {
+      } else if (c1 === 33
+        && xmlData.charCodeAt(i + 2) === 45
+        && xmlData.charCodeAt(i + 3) === 45) { // '!--'
         const closeIndex = findClosingIndex(xmlData, "-->", i + 3, "StopNode is not closed.")
         i = closeIndex;
-      } else if (xmlData.substr(i + 1, 2) === '![') {
+      } else if (c1 === 33
+        && xmlData.charCodeAt(i + 2) === 91) { // '!['
         const closeIndex = findClosingIndex(xmlData, "]]>", i, "StopNode is not closed.") - 2;
         i = closeIndex;
       } else {