Add support for parsing HTML numeric entities (#645)

NaturalIntelligence · Mar 16, 2024 · 391f24f · 391f24f
1 parent 072b2b0
commit 391f24f
Show file tree

Hide file tree

Showing 5 changed files with 43 additions and 0 deletions.
diff --git a/docs/v4/5.Entities.md b/docs/v4/5.Entities.md
@@ -132,6 +132,8 @@ Following HTML entities are supported by the parser by default when `htmlEntitie
 | ₹      | Indian Rupee               | `&inr;`       | `&#8377;`        |
 ---
 
+In addition, [numeric character references](https://html.spec.whatwg.org/multipage/syntax.html#syntax-charref) are also supported. Both decimal (`num_dec`) and hexadecimal(`num_hex`).
+
 In future version of FXP, we'll be supporting more features of DOCTYPE such as `ELEMENT`, reading content for an entity from a file etc.
 
 ## External Entities

diff --git a/spec/entities_spec.js b/spec/entities_spec.js
@@ -377,6 +377,41 @@ describe("XMLParser Entities", function() {
         expect(result).toEqual(expected);
     });
 
+
+    it("should parse HTML numeric entities when htmlEntities:true", function() {
+        const xmlData = `
+        <?xml version="1.0" encoding="UTF-8"?>
+        <note>
+            <heading>Bear</heading>
+            <body face="&#x295;&#x2022;&#x1D25;&#x2022;&#x294;">Bears are called B&#228;ren in German!</body>
+        </note> `;
+
+        const expected = {
+            "?xml": {
+                "version": "1.0",
+                "encoding": "UTF-8"
+            },
+            "note": {
+                "heading": "Bear",
+                "body": {
+                    "#text": "Bears are called Bären in German!",
+                    "face": "ʕ•ᴥ•ʔ"
+                }
+            }
+        };
+
+        const options = {
+            attributeNamePrefix: "",
+            ignoreAttributes:    false,
+            processEntities: true,
+            htmlEntities: true,
+        };
+        const parser = new XMLParser(options);
+        let result = parser.parse(xmlData);
+
+        expect(result).toEqual(expected);
+    });
+
     it("should throw error if an entity name contains special char", function() {
         const xmlData = `
         <?xml version="1.0" encoding="UTF-8"?>

diff --git a/src/v5/EntitiesParser.js b/src/v5/EntitiesParser.js
@@ -13,6 +13,8 @@ const htmlEntities = {
     "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) => String.fromCharCode(Number.parseInt(str, 10)) },
+    "num_hex": { regex: /&#x([0-9a-fA-F]{1,6});/g, val : (_, str) => String.fromCharCode(Number.parseInt(str, 16)) },
 };
 
 class EntitiesParser{

diff --git a/src/v5/valueParsers/EntitiesParser.js b/src/v5/valueParsers/EntitiesParser.js
@@ -13,6 +13,8 @@ const htmlEntities = {
     "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) => String.fromCharCode(Number.parseInt(str, 10)) },
+    "num_hex": { regex: /&#x([0-9a-fA-F]{1,6});/g, val : (_, str) => String.fromCharCode(Number.parseInt(str, 16)) },
 };
 
 class EntitiesParser{

diff --git a/src/xmlparser/OrderedObjParser.js b/src/xmlparser/OrderedObjParser.js
@@ -40,6 +40,8 @@ class OrderedObjParser{
       "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) => String.fromCharCode(Number.parseInt(str, 10)) },
+      "num_hex": { regex: /&#x([0-9a-fA-F]{1,6});/g, val : (_, str) => String.fromCharCode(Number.parseInt(str, 16)) },
     };
     this.addExternalEntities = addExternalEntities;
     this.parseXml = parseXml;