htmlparser2 - npm Package Compare versions

Comparing version 9.0.0 to 9.1.0

lib/esm/index.d.ts

@@ -1,4 +0,5 @@
-import { Parser, ParserOptions } from "./Parser.js";
-export { Parser, type ParserOptions } from "./Parser.js";
-import { DomHandlerOptions, ChildNode, Element, Document } from "domhandler";
+import { Parser, type ParserOptions } from "./Parser.js";
+export type { Handler, ParserOptions } from "./Parser.js";
+export { Parser } from "./Parser.js";
+import { type DomHandlerOptions, type ChildNode, type Element, type Document } from "domhandler";
 export { DomHandler, DomHandler as DefaultHandler, type DomHandlerOptions, } from "domhandler";
@@ -41,3 +42,3 @@ export type Options = ParserOptions & DomHandlerOptions;
 export declare function createDomStream(callback: (error: Error | null, dom: ChildNode[]) => void, options?: Options, elementCallback?: (element: Element) => void): Parser;
-export { default as Tokenizer, type Callbacks as TokenizerCallbacks, } from "./Tokenizer.js";
+export { default as Tokenizer, type Callbacks as TokenizerCallbacks, QuoteType, } from "./Tokenizer.js";
 export * as ElementType from "domelementtype";
@@ -44,0 +45,0 @@ import { type Feed } from "domutils";

lib/esm/index.js

@@ -55,3 +55,3 @@ import { Parser } from "./Parser.js";
 }
-export { default as Tokenizer, } from "./Tokenizer.js";
+export { default as Tokenizer, QuoteType, } from "./Tokenizer.js";
 /*
@@ -58,0 +58,0 @@ * All of the following exports exist for backwards-compatibility.

lib/esm/Parser.d.ts

@@ -1,2 +0,2 @@
-import Tokenizer, { Callbacks, QuoteType } from "./Tokenizer.js";
+import Tokenizer, { type Callbacks, QuoteType } from "./Tokenizer.js";
 export interface ParserOptions {
@@ -100,2 +100,3 @@ /**
     private readonly lowerCaseAttributeNames;
+    private readonly recognizeSelfClosing;
     /** We are parsing HTML. Inverse of the `xmlMode` option. */
@@ -102,0 +103,0 @@ private readonly htmlMode;

lib/esm/Parser.js

@@ -101,3 +101,3 @@ import Tokenizer, { QuoteType } from "./Tokenizer.js";
     constructor(cbs, options = {}) {
-        var _a, _b, _c, _d, _e;
+        var _a, _b, _c, _d, _e, _f;
         this.options = options;
@@ -129,5 +129,7 @@ /** The start index of the last event. */
             (_b = options.lowerCaseAttributeNames) !== null && _b !== void 0 ? _b : this.htmlMode;
-        this.tokenizer = new ((_c = options.Tokenizer) !== null && _c !== void 0 ? _c : Tokenizer)(this.options, this);
+        this.recognizeSelfClosing =
+            (_c = options.recognizeSelfClosing) !== null && _c !== void 0 ? _c : !this.htmlMode;
+        this.tokenizer = new ((_d = options.Tokenizer) !== null && _d !== void 0 ? _d : Tokenizer)(this.options, this);
         this.foreignContext = [!this.htmlMode];
-        (_e = (_d = this.cbs).onparserinit) === null || _e === void 0 ? void 0 : _e.call(_d, this);
+        (_f = (_e = this.cbs).onparserinit) === null || _f === void 0 ? void 0 : _f.call(_e, this);
     }
@@ -251,3 +253,3 @@ // Tokenizer event handlers
         this.endIndex = endIndex;
-        if (this.options.recognizeSelfClosing || this.foreignContext[0]) {
+        if (this.recognizeSelfClosing || this.foreignContext[0]) {
             this.closeCurrentTag(false);
@@ -254,0 +256,0 @@ // Set `startIndex` for next node

lib/esm/Tokenizer.d.ts

@@ -107,2 +107,3 @@ export declare enum QuoteType {
     private stateBeforeSpecialS;
+    private stateBeforeSpecialT;
     private startEntity;
@@ -109,0 +110,0 @@ private stateInEntity;

lib/esm/Tokenizer.js

@@ -62,5 +62,6 @@ import { EntityDecoder, DecodingMode, htmlDecodeTree, xmlDecodeTree, } from "entities/lib/decode.js";
     State[State["BeforeSpecialS"] = 22] = "BeforeSpecialS";
-    State[State["SpecialStartSequence"] = 23] = "SpecialStartSequence";
-    State[State["InSpecialTag"] = 24] = "InSpecialTag";
-    State[State["InEntity"] = 25] = "InEntity";
+    State[State["BeforeSpecialT"] = 23] = "BeforeSpecialT";
+    State[State["SpecialStartSequence"] = 24] = "SpecialStartSequence";
+    State[State["InSpecialTag"] = 25] = "InSpecialTag";
+    State[State["InEntity"] = 26] = "InEntity";
 })(State || (State = {}));
@@ -95,8 +96,11 @@ function isWhitespace(c) {
 const Sequences = {
-    Cdata: new Uint8Array([0x43, 0x44, 0x41, 0x54, 0x41, 0x5b]),
-    CdataEnd: new Uint8Array([0x5d, 0x5d, 0x3e]),
-    CommentEnd: new Uint8Array([0x2d, 0x2d, 0x3e]),
-    ScriptEnd: new Uint8Array([0x3c, 0x2f, 0x73, 0x63, 0x72, 0x69, 0x70, 0x74]),
-    StyleEnd: new Uint8Array([0x3c, 0x2f, 0x73, 0x74, 0x79, 0x6c, 0x65]),
+    Cdata: new Uint8Array([0x43, 0x44, 0x41, 0x54, 0x41, 0x5b]), // CDATA[
+    CdataEnd: new Uint8Array([0x5d, 0x5d, 0x3e]), // ]]>
+    CommentEnd: new Uint8Array([0x2d, 0x2d, 0x3e]), // `-->`
+    ScriptEnd: new Uint8Array([0x3c, 0x2f, 0x73, 0x63, 0x72, 0x69, 0x70, 0x74]), // `</script`
+    StyleEnd: new Uint8Array([0x3c, 0x2f, 0x73, 0x74, 0x79, 0x6c, 0x65]), // `</style`
     TitleEnd: new Uint8Array([0x3c, 0x2f, 0x74, 0x69, 0x74, 0x6c, 0x65]), // `</title`
+    TextareaEnd: new Uint8Array([
+        0x3c, 0x2f, 0x74, 0x65, 0x78, 0x74, 0x61, 0x72, 0x65, 0x61,
+    ]), // `</textarea`
 };
@@ -324,10 +328,13 @@ export default class Tokenizer {
             this.sectionStart = this.index;
-            if (!this.xmlMode && lower === Sequences.TitleEnd[2]) {
-                this.startSpecial(Sequences.TitleEnd, 3);
+            if (this.xmlMode) {
+                this.state = State.InTagName;
             }
+            else if (lower === Sequences.ScriptEnd[2]) {
+                this.state = State.BeforeSpecialS;
+            }
+            else if (lower === Sequences.TitleEnd[2]) {
+                this.state = State.BeforeSpecialT;
+            }
             else {
-                this.state =
-                    !this.xmlMode && lower === Sequences.ScriptEnd[2]
-                        ? State.BeforeSpecialS
-                        : State.InTagName;
+                this.state = State.InTagName;
             }
@@ -415,3 +422,3 @@ }
             this.cbs.onattribname(this.sectionStart, this.index);
-            this.sectionStart = -1;
+            this.sectionStart = this.index;
             this.state = State.AfterAttributeName;
@@ -426,3 +433,4 @@ this.stateAfterAttributeName(c);
         else if (c === CharCodes.Slash || c === CharCodes.Gt) {
-            this.cbs.onattribend(QuoteType.NoValue, this.index);
+            this.cbs.onattribend(QuoteType.NoValue, this.sectionStart);
+            this.sectionStart = -1;
             this.state = State.BeforeAttributeName;
@@ -432,3 +440,3 @@ this.stateBeforeAttributeName(c);
         else if (!isWhitespace(c)) {
-            this.cbs.onattribend(QuoteType.NoValue, this.index);
+            this.cbs.onattribend(QuoteType.NoValue, this.sectionStart);
             this.state = State.InAttributeName;
@@ -460,3 +468,3 @@ this.sectionStart = this.index;
                 ? QuoteType.Double
-                : QuoteType.Single, this.index);
+                : QuoteType.Single, this.index + 1);
             this.state = State.BeforeAttributeName;
@@ -544,2 +552,15 @@ }
     }
+    stateBeforeSpecialT(c) {
+        const lower = c | 0x20;
+        if (lower === Sequences.TitleEnd[3]) {
+            this.startSpecial(Sequences.TitleEnd, 4);
+        }
+        else if (lower === Sequences.TextareaEnd[3]) {
+            this.startSpecial(Sequences.TextareaEnd, 4);
+        }
+        else {
+            this.state = State.InTagName;
+            this.stateInTagName(c); // Consume the token again
+        }
+    }
     startEntity() {
@@ -673,2 +694,6 @@ this.baseState = this.state;
                 }
+                case State.BeforeSpecialT: {
+                    this.stateBeforeSpecialT(c);
+                    break;
+                }
                 case State.InAttributeValueNq: {
@@ -675,0 +700,0 @@ this.stateInAttributeValueNoQuotes(c);

lib/esm/WritableStream.d.ts

 /// <reference types="node" />
 /// <reference types="node" />
-import { Handler, ParserOptions } from "./Parser.js";
+import { type Handler, type ParserOptions } from "./Parser.js";
 import { Writable } from "node:stream";
@@ -5,0 +5,0 @@ /**

lib/index.d.ts

@@ -1,4 +0,5 @@
-import { Parser, ParserOptions } from "./Parser.js";
-export { Parser, type ParserOptions } from "./Parser.js";
-import { DomHandlerOptions, ChildNode, Element, Document } from "domhandler";
+import { Parser, type ParserOptions } from "./Parser.js";
+export type { Handler, ParserOptions } from "./Parser.js";
+export { Parser } from "./Parser.js";
+import { type DomHandlerOptions, type ChildNode, type Element, type Document } from "domhandler";
 export { DomHandler, DomHandler as DefaultHandler, type DomHandlerOptions, } from "domhandler";
@@ -41,3 +42,3 @@ export type Options = ParserOptions & DomHandlerOptions;
 export declare function createDomStream(callback: (error: Error | null, dom: ChildNode[]) => void, options?: Options, elementCallback?: (element: Element) => void): Parser;
-export { default as Tokenizer, type Callbacks as TokenizerCallbacks, } from "./Tokenizer.js";
+export { default as Tokenizer, type Callbacks as TokenizerCallbacks, QuoteType, } from "./Tokenizer.js";
 export * as ElementType from "domelementtype";
@@ -44,0 +45,0 @@ import { type Feed } from "domutils";

lib/index.js

@@ -29,3 +29,3 @@ "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.DomUtils = exports.parseFeed = exports.getFeed = exports.ElementType = exports.Tokenizer = exports.createDomStream = exports.createDocumentStream = exports.parseDOM = exports.parseDocument = exports.DefaultHandler = exports.DomHandler = exports.Parser = void 0;
+exports.DomUtils = exports.parseFeed = exports.getFeed = exports.ElementType = exports.QuoteType = exports.Tokenizer = exports.createDomStream = exports.createDocumentStream = exports.parseDOM = exports.parseDocument = exports.DefaultHandler = exports.DomHandler = exports.Parser = void 0;
 var Parser_js_1 = require("./Parser.js");
@@ -93,2 +93,3 @@ var Parser_js_2 = require("./Parser.js");
 Object.defineProperty(exports, "Tokenizer", { enumerable: true, get: function () { return __importDefault(Tokenizer_js_1).default; } });
+Object.defineProperty(exports, "QuoteType", { enumerable: true, get: function () { return Tokenizer_js_1.QuoteType; } });
 /*
@@ -95,0 +96,0 @@ * All of the following exports exist for backwards-compatibility.

lib/Parser.d.ts

@@ -1,2 +0,2 @@
-import Tokenizer, { Callbacks, QuoteType } from "./Tokenizer.js";
+import Tokenizer, { type Callbacks, QuoteType } from "./Tokenizer.js";
 export interface ParserOptions {
@@ -100,2 +100,3 @@ /**
     private readonly lowerCaseAttributeNames;
+    private readonly recognizeSelfClosing;
     /** We are parsing HTML. Inverse of the `xmlMode` option. */
@@ -102,0 +103,0 @@ private readonly htmlMode;

lib/Parser.js

@@ -128,3 +128,3 @@ "use strict";
         if (options === void 0) { options = {}; }
-        var _a, _b, _c, _d, _e;
+        var _a, _b, _c, _d, _e, _f;
         this.options = options;
@@ -156,5 +156,7 @@ /** The start index of the last event. */
             (_b = options.lowerCaseAttributeNames) !== null && _b !== void 0 ? _b : this.htmlMode;
-        this.tokenizer = new ((_c = options.Tokenizer) !== null && _c !== void 0 ? _c : Tokenizer_js_1.default)(this.options, this);
+        this.recognizeSelfClosing =
+            (_c = options.recognizeSelfClosing) !== null && _c !== void 0 ? _c : !this.htmlMode;
+        this.tokenizer = new ((_d = options.Tokenizer) !== null && _d !== void 0 ? _d : Tokenizer_js_1.default)(this.options, this);
         this.foreignContext = [!this.htmlMode];
-        (_e = (_d = this.cbs).onparserinit) === null || _e === void 0 ? void 0 : _e.call(_d, this);
+        (_f = (_e = this.cbs).onparserinit) === null || _f === void 0 ? void 0 : _f.call(_e, this);
     }
@@ -278,3 +280,3 @@ // Tokenizer event handlers
         this.endIndex = endIndex;
-        if (this.options.recognizeSelfClosing || this.foreignContext[0]) {
+        if (this.recognizeSelfClosing || this.foreignContext[0]) {
             this.closeCurrentTag(false);
@@ -281,0 +283,0 @@ // Set `startIndex` for next node

lib/Tokenizer.d.ts

@@ -107,2 +107,3 @@ export declare enum QuoteType {
     private stateBeforeSpecialS;
+    private stateBeforeSpecialT;
     private startEntity;
@@ -109,0 +110,0 @@ private stateInEntity;

lib/Tokenizer.js

@@ -65,5 +65,6 @@ "use strict";
     State[State["BeforeSpecialS"] = 22] = "BeforeSpecialS";
-    State[State["SpecialStartSequence"] = 23] = "SpecialStartSequence";
-    State[State["InSpecialTag"] = 24] = "InSpecialTag";
-    State[State["InEntity"] = 25] = "InEntity";
+    State[State["BeforeSpecialT"] = 23] = "BeforeSpecialT";
+    State[State["SpecialStartSequence"] = 24] = "SpecialStartSequence";
+    State[State["InSpecialTag"] = 25] = "InSpecialTag";
+    State[State["InEntity"] = 26] = "InEntity";
 })(State || (State = {}));
@@ -90,3 +91,3 @@ function isWhitespace(c) {
     QuoteType[QuoteType["Double"] = 3] = "Double";
-})(QuoteType = exports.QuoteType || (exports.QuoteType = {}));
+})(QuoteType || (exports.QuoteType = QuoteType = {}));
 /**
@@ -99,8 +100,11 @@ * Sequences used to match longer strings.
 var Sequences = {
-    Cdata: new Uint8Array([0x43, 0x44, 0x41, 0x54, 0x41, 0x5b]),
-    CdataEnd: new Uint8Array([0x5d, 0x5d, 0x3e]),
-    CommentEnd: new Uint8Array([0x2d, 0x2d, 0x3e]),
-    ScriptEnd: new Uint8Array([0x3c, 0x2f, 0x73, 0x63, 0x72, 0x69, 0x70, 0x74]),
-    StyleEnd: new Uint8Array([0x3c, 0x2f, 0x73, 0x74, 0x79, 0x6c, 0x65]),
+    Cdata: new Uint8Array([0x43, 0x44, 0x41, 0x54, 0x41, 0x5b]), // CDATA[
+    CdataEnd: new Uint8Array([0x5d, 0x5d, 0x3e]), // ]]>
+    CommentEnd: new Uint8Array([0x2d, 0x2d, 0x3e]), // `-->`
+    ScriptEnd: new Uint8Array([0x3c, 0x2f, 0x73, 0x63, 0x72, 0x69, 0x70, 0x74]), // `</script`
+    StyleEnd: new Uint8Array([0x3c, 0x2f, 0x73, 0x74, 0x79, 0x6c, 0x65]), // `</style`
     TitleEnd: new Uint8Array([0x3c, 0x2f, 0x74, 0x69, 0x74, 0x6c, 0x65]), // `</title`
+    TextareaEnd: new Uint8Array([
+        0x3c, 0x2f, 0x74, 0x65, 0x78, 0x74, 0x61, 0x72, 0x65, 0x61,
+    ]), // `</textarea`
 };
@@ -330,10 +334,13 @@ var Tokenizer = /** @class */ (function () {
             this.sectionStart = this.index;
-            if (!this.xmlMode && lower === Sequences.TitleEnd[2]) {
-                this.startSpecial(Sequences.TitleEnd, 3);
+            if (this.xmlMode) {
+                this.state = State.InTagName;
             }
+            else if (lower === Sequences.ScriptEnd[2]) {
+                this.state = State.BeforeSpecialS;
+            }
+            else if (lower === Sequences.TitleEnd[2]) {
+                this.state = State.BeforeSpecialT;
+            }
             else {
-                this.state =
-                    !this.xmlMode && lower === Sequences.ScriptEnd[2]
-                        ? State.BeforeSpecialS
-                        : State.InTagName;
+                this.state = State.InTagName;
             }
@@ -421,3 +428,3 @@ }
             this.cbs.onattribname(this.sectionStart, this.index);
-            this.sectionStart = -1;
+            this.sectionStart = this.index;
             this.state = State.AfterAttributeName;
@@ -432,3 +439,4 @@ this.stateAfterAttributeName(c);
         else if (c === CharCodes.Slash || c === CharCodes.Gt) {
-            this.cbs.onattribend(QuoteType.NoValue, this.index);
+            this.cbs.onattribend(QuoteType.NoValue, this.sectionStart);
+            this.sectionStart = -1;
             this.state = State.BeforeAttributeName;
@@ -438,3 +446,3 @@ this.stateBeforeAttributeName(c);
         else if (!isWhitespace(c)) {
-            this.cbs.onattribend(QuoteType.NoValue, this.index);
+            this.cbs.onattribend(QuoteType.NoValue, this.sectionStart);
             this.state = State.InAttributeName;
@@ -466,3 +474,3 @@ this.sectionStart = this.index;
                 ? QuoteType.Double
-                : QuoteType.Single, this.index);
+                : QuoteType.Single, this.index + 1);
             this.state = State.BeforeAttributeName;
@@ -550,2 +558,15 @@ }
     };
+    Tokenizer.prototype.stateBeforeSpecialT = function (c) {
+        var lower = c | 0x20;
+        if (lower === Sequences.TitleEnd[3]) {
+            this.startSpecial(Sequences.TitleEnd, 4);
+        }
+        else if (lower === Sequences.TextareaEnd[3]) {
+            this.startSpecial(Sequences.TextareaEnd, 4);
+        }
+        else {
+            this.state = State.InTagName;
+            this.stateInTagName(c); // Consume the token again
+        }
+    };
     Tokenizer.prototype.startEntity = function () {
@@ -679,2 +700,6 @@ this.baseState = this.state;
                 }
+                case State.BeforeSpecialT: {
+                    this.stateBeforeSpecialT(c);
+                    break;
+                }
                 case State.InAttributeValueNq: {
@@ -681,0 +706,0 @@ this.stateInAttributeValueNoQuotes(c);

lib/WritableStream.d.ts

 /// <reference types="node" />
 /// <reference types="node" />
-import { Handler, ParserOptions } from "./Parser.js";
+import { type Handler, type ParserOptions } from "./Parser.js";
 import { Writable } from "node:stream";
@@ -5,0 +5,0 @@ /**

package.json

 {
     "name": "htmlparser2",
     "description": "Fast & forgiving HTML/XML parser",
-    "version": "9.0.0",
+    "version": "9.1.0",
     "author": "Felix Boehm <[e-mail]>",
@@ -70,14 +70,14 @@ "funding": [
     "devDependencies": {
-        "@types/jest": "^29.5.1",
-        "@types/node": "^20.1.1",
-        "@typescript-eslint/eslint-plugin": "^5.59.5",
-        "@typescript-eslint/parser": "^5.59.5",
-        "eslint": "^8.40.0",
-        "eslint-config-prettier": "^8.8.0",
-        "eslint-plugin-n": "^15.7.0",
-        "eslint-plugin-unicorn": "^47.0.0",
-        "jest": "^29.5.0",
-        "prettier": "^2.8.8",
-        "ts-jest": "^29.1.0",
-        "typescript": "^5.0.4"
+        "@types/jest": "^29.5.11",
+        "@types/node": "^20.10.6",
+        "@typescript-eslint/eslint-plugin": "^6.17.0",
+        "@typescript-eslint/parser": "^6.17.0",
+        "eslint": "^8.56.0",
+        "eslint-config-prettier": "^9.1.0",
+        "eslint-plugin-n": "^16.6.1",
+        "eslint-plugin-unicorn": "^50.0.1",
+        "jest": "^29.7.0",
+        "prettier": "^3.1.1",
+        "ts-jest": "^29.1.1",
+        "typescript": "^5.3.3"
     },
@@ -84,0 +84,0 @@ "jest": {

README.md

@@ -72,3 +72,3 @@ # htmlparser2
 parser.write(
-    "Xyz <script type='text/javascript'>const foo = '<<bar>>';</script>"
+    "Xyz <script type='text/javascript'>const foo = '<<bar>>';</script>",
 );
@@ -75,0 +75,0 @@ parser.end();

No alert changes

Improved metrics

Total package byte prevSize

253719

increased by3.17%

Lines of code

3633

increased by1.74%

No dependency changes