refactor: Inline preprocessor & tokenizer error mixins

fb55 · fb55 · commit 54e1febc41bc · 2021-11-24T20:06:43.000Z
diff --git a/packages/parse5/lib/common/error-codes.ts b/packages/parse5/lib/common/error-codes.ts
@@ -1,3 +1,11 @@
+import { Location } from './token.js';
+
+export interface ParserError extends Location {
+    code: ERR;
+}
+
+export type ParserErrorHandler = (error: ParserError) => void;
+
 export enum ERR {
     controlCharacterInInputStream = 'control-character-in-input-stream',
     noncharacterInInputStream = 'noncharacter-in-input-stream',
diff --git a/packages/parse5/lib/extensions/error-reporting/mixin-base.ts b/packages/parse5/lib/extensions/error-reporting/mixin-base.ts
diff --git a/packages/parse5/lib/extensions/error-reporting/parser-mixin.ts b/packages/parse5/lib/extensions/error-reporting/parser-mixin.ts
@@ -1,23 +1,21 @@
-import { ErrorReportingMixinOptions, BASE_ERROR } from './mixin-base.js';
-import { ErrorReportingTokenizerMixin } from './tokenizer-mixin.js';
 import { Mixin } from '../../utils/mixin.js';
 import type { Location, Token } from '../../common/token.js';
 import type { Parser } from '../../parser/index.js';
 import type { TreeAdapterTypeMap } from '../../tree-adapters/interface.js';
-import type { ERR } from '../../common/error-codes.js';
+import type { ERR, ParserErrorHandler } from '../../common/error-codes.js';
 
 export class ErrorReportingParserMixin<T extends TreeAdapterTypeMap> extends Mixin<Parser<T>> {
-    private onParseError: ErrorReportingMixinOptions['onParseError'];
+    private onParseError: ParserErrorHandler;
     ctLoc: null | Location = null;
     locBeforeToken = false;
 
-    constructor(parser: Parser<T>, private opts: ErrorReportingMixinOptions) {
+    constructor(private parser: Parser<T>, opts: { onParseError: ParserErrorHandler }) {
         super(parser);
         this.onParseError = opts.onParseError;
     }
 
     _reportError(code: ERR) {
-        const err = { ...BASE_ERROR, code };
+        const err = this.parser.tokenizer!.preprocessor.getError(code);
 
         if (this.ctLoc) {
             err.startLine = this.ctLoc.startLine;
@@ -36,8 +34,6 @@ export class ErrorReportingParserMixin<T extends TreeAdapterTypeMap> extends Mix
         return {
             _bootstrap(this: Parser<T>, document: T['document'], fragmentContext: T['element'] | null) {
                 orig._bootstrap.call(this, document, fragmentContext);
-
-                Mixin.install(this.tokenizer, ErrorReportingTokenizerMixin, mxn.opts);
             },
 
             _processInputToken(token: Token) {
diff --git a/packages/parse5/lib/extensions/error-reporting/preprocessor-mixin.ts b/packages/parse5/lib/extensions/error-reporting/preprocessor-mixin.ts
diff --git a/packages/parse5/lib/extensions/error-reporting/tokenizer-mixin.ts b/packages/parse5/lib/extensions/error-reporting/tokenizer-mixin.ts
diff --git a/packages/parse5/lib/parser/index.ts b/packages/parse5/lib/parser/index.ts
@@ -17,9 +17,17 @@ import {
     DOCUMENT_MODE,
     isNumberedHeader,
 } from '../common/html.js';
-import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface';
-import type { ParserError } from '../extensions/error-reporting/mixin-base';
-import { TokenType, getTokenAttr, Token, CommentToken, CharacterToken, TagToken, DoctypeToken } from '../common/token';
+import type { TreeAdapter, TreeAdapterTypeMap } from '../tree-adapters/interface.js';
+import type { ParserError } from '../common/error-codes.js';
+import {
+    TokenType,
+    getTokenAttr,
+    Token,
+    CommentToken,
+    CharacterToken,
+    TagToken,
+    DoctypeToken,
+} from '../common/token.js';
 
 //Misc constants
 const HIDDEN_INPUT_TYPE = 'hidden';
diff --git a/packages/parse5/lib/tokenizer/index.test.ts b/packages/parse5/lib/tokenizer/index.test.ts
@@ -1,15 +1,12 @@
 import * as parse5 from '../index.js';
 import { Tokenizer } from './index.js';
-import { Mixin } from '../utils/mixin.js';
-import { ErrorReportingTokenizerMixin } from '../extensions/error-reporting/tokenizer-mixin.js';
 import { generateTokenizationTests } from '../../../../test/utils/generate-tokenization-tests.js';
 
 const dataPath = new URL('../../../../test/data/html5lib-tests/tokenizer', import.meta.url);
 
 generateTokenizationTests('tokenizer', 'Tokenizer', dataPath.pathname, ({ errors }) => {
-    const tokenizer = new Tokenizer({ sourceCodeLocationInfo: true });
-
-    Mixin.install(tokenizer, ErrorReportingTokenizerMixin, {
+    const tokenizer = new Tokenizer({
+        sourceCodeLocationInfo: true,
         onParseError(err) {
             errors.push({
                 code: err.code,
diff --git a/packages/parse5/lib/tokenizer/index.ts b/packages/parse5/lib/tokenizer/index.ts
@@ -19,7 +19,7 @@ import {
     Location,
 } from '../common/token.js';
 import { namedEntityData as neTree } from './named-entity-data.js';
-import { ERR } from '../common/error-codes.js';
+import { ERR, ParserErrorHandler } from '../common/error-codes.js';
 
 //C1 Unicode control character reference replacements
 const C1_CONTROLS_REFERENCE_REPLACEMENTS = new Map([
@@ -224,7 +224,7 @@ function findNamedEntityTreeBranch(nodeIx: number, cp: number): number {
 
 //Tokenizer
 export class Tokenizer {
-    preprocessor = new Preprocessor();
+    preprocessor: Preprocessor;
 
     tokenQueue: Token[] = [];
 
@@ -245,19 +245,22 @@ export class Tokenizer {
     currentAttr: Attribute = { name: '', value: '' };
 
     private addLocationInfo;
+    private onParseError;
 
-    constructor(options: { sourceCodeLocationInfo?: boolean }) {
+    constructor(options: { sourceCodeLocationInfo?: boolean; onParseError?: ParserErrorHandler | null }) {
         this.addLocationInfo = !!options.sourceCodeLocationInfo;
+        this.onParseError = options.onParseError ?? null;
+        this.preprocessor = new Preprocessor(options);
     }
 
     //Errors
-    _err(_err: string) {
-        // NOTE: err reporting is noop by default. Enabled by mixin.
+    private _err(code: ERR) {
+        this.onParseError?.(this.preprocessor.getError(code));
     }
 
-    private _errOnNextCodePoint(err: string) {
+    private _errOnNextCodePoint(code: ERR) {
         this._consume();
-        this._err(err);
+        this._err(code);
         this._unconsume();
     }
 
diff --git a/packages/parse5/lib/tokenizer/preprocessor.ts b/packages/parse5/lib/tokenizer/preprocessor.ts
@@ -1,5 +1,5 @@
 import * as unicode from '../common/unicode.js';
-import { ERR } from '../common/error-codes.js';
+import { ERR, ParserError, ParserErrorHandler } from '../common/error-codes.js';
 
 //Aliases
 const $ = unicode.CODE_POINTS;
@@ -25,6 +25,12 @@ export class Preprocessor {
     droppedBufferSize = 0;
     line = 1;
 
+    onParseError: ParserErrorHandler | null;
+
+    constructor(options: { onParseError?: ParserErrorHandler | null }) {
+        this.onParseError = options.onParseError ?? null;
+    }
+
     /** The column on the current line. If we just saw a gap (eg. a surrogate pair), return the index before. */
     get col(): number {
         return this.pos - this.lineStartPos + Number(this.lastGapPos !== this.pos);
@@ -34,8 +40,27 @@ export class Preprocessor {
         return this.droppedBufferSize + this.pos;
     }
 
-    _err(_err: string) {
-        // NOTE: err reporting is noop by default. Enabled by mixin.
+    public getError(code: ERR): ParserError {
+        const { line, col, offset } = this;
+
+        return {
+            code,
+            startLine: line,
+            endLine: line,
+            startCol: col,
+            endCol: col,
+            startOffset: offset,
+            endOffset: offset,
+        };
+    }
+
+    //NOTE: avoid reporting error twice on advance/retreat
+    private lastErrOffset = -1;
+    private _err(code: ERR) {
+        if (this.onParseError && this.lastErrOffset !== this.offset) {
+            this.lastErrOffset = this.offset;
+            this.onParseError(this.getError(code));
+        }
     }
 
     private _addGap() {
@@ -147,7 +172,8 @@ export class Preprocessor {
         //range (ASCII alphanumeric, whitespaces, big chunk of BMP)
         //before going into detailed performance cost validation.
         const isCommonValidRange =
-            (cp > 0x1f && cp < 0x7f) || cp === $.LINE_FEED || cp === $.CARRIAGE_RETURN || (cp > 0x9f && cp < 0xfdd0);
+            this.onParseError !== null &&
+            ((cp > 0x1f && cp < 0x7f) || cp === $.LINE_FEED || cp === $.CARRIAGE_RETURN || (cp > 0x9f && cp < 0xfdd0));
 
         if (!isCommonValidRange) {
             this._checkForProblematicCharacters(cp);
diff --git a/test/utils/generate-parsing-tests.ts b/test/utils/generate-parsing-tests.ts
@@ -1,5 +1,5 @@
-import { ParserOptions } from './../../packages/parse5/lib/parser/index';
-import { ParserError } from './../../packages/parse5/lib/extensions/error-reporting/mixin-base';
+import { ParserOptions } from './../../packages/parse5/lib/parser/index.js';
+import { ParserError } from './../../packages/parse5/lib/common/error-codes.js';
 import * as fs from 'fs';
 import * as path from 'path';
 import * as assert from 'assert';
