fix(parser): Fix indices for end, CDATA, add indices to tests

src/Parser.ts

@@ -403,7 +403,7 @@ export class Parser {
     }
 
     oncdata(value: string): void {
-        this.updatePosition(1);
+        this.updatePosition(9);
         if (this.options.xmlMode || this.options.recognizeCDATA) {
             this.cbs.oncdatastart?.();
             this.cbs.ontext?.(value);
@@ -419,6 +419,8 @@ export class Parser {
 
     onend(): void {
         if (this.cbs.onclosetag) {
+            // Set start- and end indices for remaining tags
+            this.startIndex = this.endIndex = this.tokenizer.getAbsoluteIndex();
             for (
                 let i = this.stack.length;
                 i > 0;

src/Tokenizer.ts

@@ -736,6 +736,7 @@ export default class Tokenizer {
 
         if (this.trieIndex < 0) {
             this.emitNamedEntity();
+            this._index--;
             return;
         }
 
@@ -770,7 +771,6 @@ export default class Tokenizer {
 
         this.sectionStart = this._index - this.trieExcess + 1;
         this._state = this.baseState;
-        this._index--;
     }
 
     private decodeNumericEntity(base: 10 | 16, strict: boolean) {

src/WritableStream.spec.ts

@@ -34,10 +34,9 @@ helper.createSuite("Stream", (test, cb) => {
                     const handler = helper.getEventCollector(cb);
                     const stream = new WritableStream(handler, test.options);
 
-                    fs.readFile(filePath, (err, data) => {
-                        if (err) throw err;
-                        stream.end(data);
-                    });
+                    fs.readFile(filePath, (err, data) =>
+                        err ? cb(err) : stream.end(data)
+                    );
                 }),
                 test.options
             )

src/__fixtures__/Events/01-simple.json

@@ -1,9 +1,11 @@
 {
     "name": "simple",
-    "html": "<h1 class=test>adsf</h1>",
+    "input": "<h1 class=test>adsf</h1>",
     "expected": [
         {
             "event": "opentagname",
+            "startIndex": 0,
+            "endIndex": 3,
             "data": ["h1"]
         },
         {
@@ -12,6 +14,8 @@
         },
         {
             "event": "opentag",
+            "startIndex": 0,
+            "endIndex": 14,
             "data": [
                 "h1",
                 {
@@ -25,6 +29,8 @@
         },
         {
             "event": "closetag",
+            "startIndex": 19,
+            "endIndex": 23,
             "data": ["h1"]
         }
     ]

src/__fixtures__/Events/02-template.json

@@ -1,17 +1,23 @@
 {
     "name": "Template script tags",
-    "html": "<p><script type=\"text/template\"><h1>Heading1</h1></script></p>",
+    "input": "<p><script type=\"text/template\"><h1>Heading1</h1></script></p>",
     "expected": [
         {
             "event": "opentagname",
+            "startIndex": 0,
+            "endIndex": 2,
             "data": ["p"]
         },
         {
             "event": "opentag",
+            "startIndex": 0,
+            "endIndex": 2,
             "data": ["p", {}]
         },
         {
             "event": "opentagname",
+            "startIndex": 3,
+            "endIndex": 10,
             "data": ["script"]
         },
         {
@@ -20,6 +26,8 @@
         },
         {
             "event": "opentag",
+            "startIndex": 3,
+            "endIndex": 31,
             "data": [
                 "script",
                 {
@@ -33,10 +41,14 @@
         },
         {
             "event": "closetag",
+            "startIndex": 49,
+            "endIndex": 57,
             "data": ["script"]
         },
         {
             "event": "closetag",
+            "startIndex": 58,
+            "endIndex": 61,
             "data": ["p"]
         }
     ]

src/__fixtures__/Events/03-lowercase_tags.json

@@ -1,14 +1,16 @@
 {
     "name": "Lowercase tags",
+    "input": "<H1 class=test>adsf</H1>",
     "options": {
         "parser": {
             "lowerCaseTags": true
         }
     },
-    "html": "<H1 class=test>adsf</H1>",
     "expected": [
         {
             "event": "opentagname",
+            "startIndex": 0,
+            "endIndex": 3,
             "data": ["h1"]
         },
         {
@@ -17,6 +19,8 @@
         },
         {
             "event": "opentag",
+            "startIndex": 0,
+            "endIndex": 14,
             "data": [
                 "h1",
                 {
@@ -30,6 +34,8 @@
         },
         {
             "event": "closetag",
+            "startIndex": 19,
+            "endIndex": 23,
             "data": ["h1"]
         }
     ]

src/__fixtures__/Events/04-cdata.json

@@ -1,20 +1,28 @@
 {
     "name": "CDATA",
+    "input": "<tag><![CDATA[ asdf ><asdf></adsf><> fo]]></tag><![CD>",
     "options": {
-        "parser": { "xmlMode": true }
+        "parser": {
+            "xmlMode": true
+        }
     },
-    "html": "<tag><![CDATA[ asdf ><asdf></adsf><> fo]]></tag><![CD>",
     "expected": [
         {
             "event": "opentagname",
+            "startIndex": 0,
+            "endIndex": 4,
             "data": ["tag"]
         },
         {
             "event": "opentag",
+            "startIndex": 0,
+            "endIndex": 4,
             "data": ["tag", {}]
         },
         {
             "event": "cdatastart",
+            "startIndex": 5,
+            "endIndex": 41,
             "data": []
         },
         {
@@ -23,14 +31,20 @@
         },
         {
             "event": "cdataend",
+            "startIndex": 5,
+            "endIndex": 41,
             "data": []
         },
         {
             "event": "closetag",
+            "startIndex": 42,
+            "endIndex": 47,
             "data": ["tag"]
         },
         {
             "event": "processinginstruction",
+            "startIndex": 48,
+            "endIndex": 53,
             "data": ["![CD", "![CD"]
         }
     ]

src/__fixtures__/Events/05-cdata-special.json

@@ -1,13 +1,17 @@
 {
     "name": "CDATA (inside special)",
-    "html": "<script>/*<![CDATA[*/ asdf ><asdf></adsf><> fo/*]]>*/</script>",
+    "input": "<script>/*<![CDATA[*/ asdf ><asdf></adsf><> fo/*]]>*/</script>",
     "expected": [
         {
             "event": "opentagname",
+            "startIndex": 0,
+            "endIndex": 7,
             "data": ["script"]
         },
         {
             "event": "opentag",
+            "startIndex": 0,
+            "endIndex": 7,
             "data": ["script", {}]
         },
         {
@@ -16,6 +20,8 @@
         },
         {
             "event": "closetag",
+            "startIndex": 53,
+            "endIndex": 61,
             "data": ["script"]
         }
     ]

src/__fixtures__/Events/06-leading-lt.json

@@ -1,6 +1,6 @@
 {
     "name": "leading lt",
-    "html": ">a>",
+    "input": ">a>",
     "expected": [
         {
             "event": "text",

src/__fixtures__/Events/07-self-closing.json

@@ -1,9 +1,11 @@
 {
     "name": "Self-closing tags",
-    "html": "<a href=http://test.com/>Foo</a><hr / >",
+    "input": "<a href=http://test.com/>Foo</a><hr / >",
     "expected": [
         {
             "event": "opentagname",
+            "startIndex": 0,
+            "endIndex": 2,
             "data": ["a"]
         },
         {
@@ -12,6 +14,8 @@
         },
         {
             "event": "opentag",
+            "startIndex": 0,
+            "endIndex": 24,
             "data": [
                 "a",
                 {
@@ -25,18 +29,26 @@
         },
         {
             "event": "closetag",
+            "startIndex": 28,
+            "endIndex": 31,
             "data": ["a"]
         },
         {
             "event": "opentagname",
+            "startIndex": 32,
+            "endIndex": 35,
             "data": ["hr"]
         },
         {
             "event": "opentag",
+            "startIndex": 32,
+            "endIndex": 38,
             "data": ["hr", {}]
         },
         {
             "event": "closetag",
+            "startIndex": 32,
+            "endIndex": 38,
             "data": ["hr"]
         }
     ]