refactor(parser): Consume tokenizer events

packages/parse5/lib/common/token.ts

@@ -96,13 +96,9 @@ export interface EOFToken extends TokenBase {
     readonly type: TokenType.EOF;
 }
 
-interface HibernationToken extends TokenBase {
-    readonly type: TokenType.HIBERNATION;
-}
-
 export interface CharacterToken extends TokenBase {
     type: TokenType.CHARACTER | TokenType.NULL_CHARACTER | TokenType.WHITESPACE_CHARACTER;
     chars: string;
 }
 
-export type Token = DoctypeToken | TagToken | CommentToken | EOFToken | HibernationToken | CharacterToken;
+export type Token = DoctypeToken | TagToken | CommentToken | EOFToken | CharacterToken;

packages/parse5/lib/parser/index.ts

@@ -1,5 +1,4 @@
-import { TokenizerMode } from '../tokenizer/index.js';
-import { QueuedTokenizer } from '../tokenizer/queued.js';
+import { TokenHandler, Tokenizer, TokenizerMode } from '../tokenizer/index.js';
 import { OpenElementStack } from './open-element-stack.js';
 import { FormattingElementList, ElementEntry, EntryType } from './formatting-element-list.js';
 import * as defaultTreeAdapter from '../tree-adapters/default.js';
@@ -120,9 +119,9 @@ const defaultParserOptions = {
 };
 
 //Parser
-export class Parser<T extends TreeAdapterTypeMap> {
+export class Parser<T extends TreeAdapterTypeMap> implements TokenHandler {
     treeAdapter: TreeAdapter<T>;
-    private onParseError: ParserErrorHandler | null;
+    onParseError: ParserErrorHandler | null;
     private currentToken: Token | null = null;
     public options: Required<ParserOptions<T>>;
     public document: T['document'];
@@ -147,7 +146,7 @@ export class Parser<T extends TreeAdapterTypeMap> {
 
         this.document = document ?? this.treeAdapter.createDocument();
 
-        this.tokenizer = new QueuedTokenizer(this.options);
+        this.tokenizer = new Tokenizer(this.options, this);
         this.activeFormattingElements = new FormattingElementList(this.treeAdapter);
 
         this.fragmentContextID = fragmentContext ? getTagID(this.treeAdapter.getTagName(fragmentContext)) : $.UNKNOWN;
@@ -211,7 +210,8 @@ export class Parser<T extends TreeAdapterTypeMap> {
         return fragment;
     }
 
-    tokenizer: QueuedTokenizer;
+    tokenizer: Tokenizer;
+
     stopped = false;
     insertionMode = InsertionMode.INITIAL;
     originalInsertionMode = InsertionMode.INITIAL;
@@ -261,13 +261,7 @@ export class Parser<T extends TreeAdapterTypeMap> {
     //Parsing loop
     private _runParsingLoop(scriptHandler: null | ((scriptElement: T['element']) => void)): void {
         while (!this.stopped) {
-            const token = this.tokenizer.getNextToken();
-
-            this._processToken(token);
-
-            if (token.type === TokenType.START_TAG && token.selfClosing && !token.ackSelfClosing) {
-                this._err(token, ERR.nonVoidHtmlElementStartTagWithTrailingSolidus);
-            }
+            this.tokenizer.getNextToken();
 
             if (!this.tokenizer.active || (scriptHandler !== null && this.pendingScript)) {
                 break;
@@ -601,7 +595,7 @@ export class Parser<T extends TreeAdapterTypeMap> {
                 break;
             }
             case TokenType.START_TAG: {
-                this.onStartTag(token);
+                this._processStartTag(token);
                 break;
             }
             case TokenType.END_TAG: {
@@ -958,6 +952,23 @@ export class Parser<T extends TreeAdapterTypeMap> {
         this.skipNextNewLine = false;
         this.currentToken = token;
 
+        this._processStartTag(token);
+
+        if (token.selfClosing && !token.ackSelfClosing) {
+            this._err(token, ERR.nonVoidHtmlElementStartTagWithTrailingSolidus);
+        }
+    }
+    /**
+     * Processes a given start tag.
+     *
+     * `onStartTag` checks if a self-closing tag was recognized. When a token
+     * is moved inbetween multiple insertion modes, this check for self-closing
+     * could lead to false positives. To avoid this, `_processStartTag` is used
+     * for nested calls.
+     *
+     * @param token The token to process.
+     */
+    _processStartTag(token: TagToken): void {
         if (this.shouldProcessStartTagTokenInForeignContent(token)) {
             startTagInForeignContent(this, token);
         } else {
@@ -2627,7 +2638,7 @@ function tableStartTagInTable<T extends TreeAdapterTypeMap>(p: Parser<T>, token:
     if (p.openElements.hasInTableScope($.TABLE)) {
         p.openElements.popUntilTagNamePopped($.TABLE);
         p._resetInsertionMode();
-        p.onStartTag(token);
+        p._processStartTag(token);
     }
 }
 
@@ -3127,7 +3138,7 @@ function startTagInSelect<T extends TreeAdapterTypeMap>(p: Parser<T>, token: Tag
                 p._resetInsertionMode();
 
                 if (token.tagID !== $.SELECT) {
-                    p.onStartTag(token);
+                    p._processStartTag(token);
                 }
             }
             break;
@@ -3197,7 +3208,7 @@ function startTagInSelectInTable<T extends TreeAdapterTypeMap>(p: Parser<T>, tok
     ) {
         p.openElements.popUntilTagNamePopped($.SELECT);
         p._resetInsertionMode();
-        p.onStartTag(token);
+        p._processStartTag(token);
     } else {
         startTagInSelect(p, token);
     }

packages/parse5/lib/tokenizer/tokenizer-location-info.test.ts

@@ -1,7 +1,6 @@
 import * as assert from 'node:assert';
-import { Tokenizer, TokenizerMode } from './index.js';
-import { SinglePathHandler } from './queued.js';
-import { Location, EOFToken, Token } from '../common/token.js';
+import { Tokenizer, TokenizerMode, TokenHandler } from './index.js';
+import { Location, EOFToken, CharacterToken, DoctypeToken, TagToken, CommentToken } from '../common/token.js';
 import { getSubstringByLineCol, normalizeNewLine } from 'parse5-test-utils/utils/common.js';
 
 interface LocationInfoTestCase {
@@ -11,22 +10,17 @@ interface LocationInfoTestCase {
 }
 
 /** Receives events and immediately compares them against the expected values. */
-class LocationInfoHandler extends SinglePathHandler {
+class LocationInfoHandler implements TokenHandler {
     public sawEof = false;
     /** The index of the last html chunk. */
     private idx = 0;
     /** All of the lines in the input. */
     private lines: string[];
 
     constructor(private testCase: LocationInfoTestCase, private html: string) {
-        super();
         this.lines = html.split(/\r?\n/g);
     }
 
-    protected handleToken(token: Token): void {
-        this.validateLocation(token.location);
-    }
-
     private validateLocation(location: Location | null): void {
         assert.ok(location);
 
@@ -45,7 +39,28 @@ class LocationInfoHandler extends SinglePathHandler {
         this.idx += 1;
     }
 
-    override onEof({ location }: EOFToken): void {
+    onComment({ location }: CommentToken): void {
+        this.validateLocation(location);
+    }
+    onDoctype({ location }: DoctypeToken): void {
+        this.validateLocation(location);
+    }
+    onStartTag({ location }: TagToken): void {
+        this.validateLocation(location);
+    }
+    onEndTag({ location }: TagToken): void {
+        this.validateLocation(location);
+    }
+    onCharacter({ location }: CharacterToken): void {
+        this.validateLocation(location);
+    }
+    onNullCharacter({ location }: CharacterToken): void {
+        this.validateLocation(location);
+    }
+    onWhitespaceCharacter({ location }: CharacterToken): void {
+        this.validateLocation(location);
+    }
+    onEof({ location }: EOFToken): void {
         assert.ok(location);
         assert.strictEqual(location.endOffset, location.startOffset);
         assert.strictEqual(location.endOffset, this.html.length);