fix #501: parse for entities only once

NaturalIntelligence · Oct 5, 2022 · 8cfd915 · 8cfd915
1 parent 443e8ed
commit 8cfd915
Show file tree

Hide file tree

Showing 3 changed files with 29 additions and 1 deletion.
diff --git a/spec/entities_spec.js b/spec/entities_spec.js
@@ -501,4 +501,28 @@ describe("XMLParser External Entites", function() {
         // console.log(output);
         expect(output.replace(/\s+/g, "")).toEqual(expected.replace(/\s+/g, ""));
     });
+
+    it("should replace '&amp;lt;' with '&lt;'", function() {
+        const xmlData = `<SimpleScalarPropertiesInputOutput>
+        <stringValue>&amp;lt;</stringValue>
+      </SimpleScalarPropertiesInputOutput>`;
+
+        const expected = {
+            "SimpleScalarPropertiesInputOutput": {
+              "stringValue": "&lt;"
+            }
+          };
+
+        const options = {
+            attributeNamePrefix: "",
+            ignoreAttributes:    false,
+            processEntities: true,
+            // preserveOrder: true
+        };
+        const parser = new XMLParser(options);
+        let result = parser.parse(xmlData);
+        //console.log(JSON.stringify(result,null,4));
+
+        expect(result).toEqual(expected);
+    });
 });
diff --git a/src/xmlparser/OrderedObjParser.js b/src/xmlparser/OrderedObjParser.js
@@ -20,12 +20,12 @@ class OrderedObjParser{
     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.ampEntity = { regex: /&(amp|#38|#x26);/g, val : "&"};
     this.htmlEntities = {
       "space": { regex: /&(nbsp|#160);/g, val: " " },
       // "lt" : { regex: /&(lt|#60);/g, val: "<" },
@@ -364,6 +364,7 @@ const parseXml = function(xmlData) {
 }
 
 const replaceEntitiesValue = function(val){
+
   if(this.options.processEntities){
     for(let entityName in this.docTypeEntities){
       const entity = this.docTypeEntities[entityName];
@@ -379,6 +380,7 @@ const replaceEntitiesValue = function(val){
         val = val.replace( entity.regex, entity.val);
       }
     }
+    val = val.replace( this.ampEntity.regex, this.ampEntity.val);
   }
   return val;
 }

diff --git a/src/xmlparser/XMLParser.js b/src/xmlparser/XMLParser.js
@@ -47,6 +47,8 @@ class XMLParser{
             throw new Error("Entity value can't have '&'")
         }else if(key.indexOf("&") !== -1 || key.indexOf(";") !== -1){
             throw new Error("An entity must be set without '&' and ';'. Eg. use '#xD' for '&#xD;'")
+        }else if(value === "&"){
+            throw new Error("An entity with value '&' is not permitted");
         }else{
             this.externalEntities[key] = value;
         }