refactor: Simplify langium by abstracting common functions

packages/parser/src/language/common/index.ts

@@ -1,2 +1,2 @@
 export * from './commonLexer.js';
-export * from './commonValueConverters.js';
+export * from './valueConverter.js';

packages/parser/src/language/common/tokenBuilder.ts

@@ -0,0 +1,26 @@
+import type { GrammarAST, Stream, TokenBuilderOptions } from 'langium';
+import type { TokenType } from '../chevrotainWrapper.js';
+
+import { DefaultTokenBuilder } from 'langium';
+
+export class MermaidTokenBuilder extends DefaultTokenBuilder {
+  private keywords: Set<string>;
+  constructor(public _keywords: string[]) {
+    super();
+    this.keywords = new Set<string>(_keywords);
+  }
+
+  protected override buildKeywordTokens(
+    rules: Stream<GrammarAST.AbstractRule>,
+    terminalTokens: TokenType[],
+    options?: TokenBuilderOptions
+  ): TokenType[] {
+    const tokenTypes: TokenType[] = super.buildKeywordTokens(rules, terminalTokens, options);
+    tokenTypes.forEach((tokenType: TokenType): void => {
+      if (this.keywords.has(tokenType.name) && tokenType.PATTERN !== undefined) {
+        tokenType.PATTERN = new RegExp(tokenType.PATTERN.toString() + '(?!\\S)');
+      }
+    });
+    return tokenTypes;
+  }
+}

packages/parser/src/language/common/valueConverter.ts

@@ -1,38 +1,45 @@
+/* eslint-disable @typescript-eslint/no-unused-vars */
 import type { CstNode, GrammarAST, ValueType } from 'langium';
 import { DefaultValueConverter } from 'langium';
 
 import { accessibilityDescrRegex, accessibilityTitleRegex, titleRegex } from './commonMatcher.js';
 
-export class CommonValueConverter extends DefaultValueConverter {
+export abstract class MermaidValueConverter extends DefaultValueConverter {
+  /**
+   * A method contains convert logic to be used by class.
+   *
+   * @param rule - Parsed rule.
+   * @param input - Matched string.
+   * @param cstNode - Node in the Concrete Syntax Tree (CST).
+   * @returns converted the value if it's available or `undefined` if it's not.
+   */
+  protected abstract runCustomConverter(
+    rule: GrammarAST.AbstractRule,
+    input: string,
+    cstNode: CstNode
+  ): ValueType | undefined;
+
   protected override runConverter(
     rule: GrammarAST.AbstractRule,
     input: string,
     cstNode: CstNode
   ): ValueType {
-    const value: ValueType | undefined = CommonValueConverter.customRunConverter(
-      rule,
-      input,
-      cstNode
-    );
+    let value: ValueType | undefined = this.runCommonConverter(rule, input, cstNode);
+
+    if (value === undefined) {
+      value = this.runCustomConverter(rule, input, cstNode);
+    }
+
     if (value === undefined) {
       return super.runConverter(rule, input, cstNode);
-    } else {
-      return value;
     }
+
+    return value;
   }
 
-  /**
-   * A method contains convert logic to be used by class itself or `MermaidValueConverter`.
-   *
-   * @param rule - Parsed rule.
-   * @param input - Matched string.
-   * @param _cstNode - Node in the Concrete Syntax Tree (CST).
-   * @returns converted the value if it's common rule or `undefined` if it's not.
-   */
-  public static customRunConverter(
+  private runCommonConverter(
     rule: GrammarAST.AbstractRule,
     input: string,
-    // eslint-disable-next-line @typescript-eslint/no-unused-vars
     _cstNode: CstNode
   ): ValueType | undefined {
     let regex: RegExp | undefined;
@@ -72,3 +79,13 @@ export class CommonValueConverter extends DefaultValueConverter {
     return undefined;
   }
 }
+
+export class CommonValueConverter extends MermaidValueConverter {
+  protected runCustomConverter(
+    _rule: GrammarAST.AbstractRule,
+    _input: string,
+    _cstNode: CstNode
+  ): ValueType | undefined {
+    return undefined;
+  }
+}

packages/parser/src/language/info/infoModule.ts

@@ -9,7 +9,7 @@ import { EmptyFileSystem, createDefaultModule, createDefaultSharedModule, inject
 
 import { MermaidGeneratedSharedModule, InfoGeneratedModule } from '../generated/module.js';
 import { CommonLexer } from '../common/commonLexer.js';
-import { CommonValueConverter } from '../common/commonValueConverters.js';
+import { CommonValueConverter } from '../common/valueConverter.js';
 import { InfoTokenBuilder } from './infoTokenBuilder.js';
 
 /**

packages/parser/src/language/info/infoTokenBuilder.ts

@@ -1,24 +1,7 @@
-import type { GrammarAST, Stream, TokenBuilderOptions } from 'langium';
-import { DefaultTokenBuilder } from 'langium';
+import { MermaidTokenBuilder } from '../common/tokenBuilder.js';
 
-import type { TokenType } from '../chevrotainWrapper.js';
-
-export class InfoTokenBuilder extends DefaultTokenBuilder {
-  protected override buildKeywordTokens(
-    rules: Stream<GrammarAST.AbstractRule>,
-    terminalTokens: TokenType[],
-    options?: TokenBuilderOptions
-  ): TokenType[] {
-    const tokenTypes: TokenType[] = super.buildKeywordTokens(rules, terminalTokens, options);
-    // to restrict users, they mustn't have any non-whitespace characters after the keyword.
-    tokenTypes.forEach((tokenType: TokenType): void => {
-      if (
-        (tokenType.name === 'info' || tokenType.name === 'showInfo') &&
-        tokenType.PATTERN !== undefined
-      ) {
-        tokenType.PATTERN = new RegExp(tokenType.PATTERN.toString() + '(?!\\S)');
-      }
-    });
-    return tokenTypes;
+export class InfoTokenBuilder extends MermaidTokenBuilder {
+  constructor() {
+    super(['info', 'showInfo']);
   }
 }

packages/parser/src/language/pie/pieTokenBuilder.ts

@@ -1,23 +1,7 @@
-import type { GrammarAST, Stream, TokenBuilderOptions } from 'langium';
-import { DefaultTokenBuilder } from 'langium';
+import { MermaidTokenBuilder } from '../common/tokenBuilder.js';
 
-import type { TokenType } from '../chevrotainWrapper.js';
-
-export class PieTokenBuilder extends DefaultTokenBuilder {
-  protected override buildKeywordTokens(
-    rules: Stream<GrammarAST.AbstractRule>,
-    terminalTokens: TokenType[],
-    options?: TokenBuilderOptions
-  ): TokenType[] {
-    const tokenTypes: TokenType[] = super.buildKeywordTokens(rules, terminalTokens, options);
-    tokenTypes.forEach((tokenType: TokenType): void => {
-      if (
-        (tokenType.name === 'pie' || tokenType.name === 'showData') &&
-        tokenType.PATTERN !== undefined
-      ) {
-        tokenType.PATTERN = new RegExp(tokenType.PATTERN.toString() + '(?!\\S)');
-      }
-    });
-    return tokenTypes;
+export class PieTokenBuilder extends MermaidTokenBuilder {
+  constructor() {
+    super(['pie', 'showData']);
   }
 }

packages/parser/src/language/pie/pieValueConverter.ts

@@ -1,49 +1,17 @@
 import type { CstNode, GrammarAST, ValueType } from 'langium';
-import { DefaultValueConverter } from 'langium';
+import { MermaidValueConverter } from '../common/valueConverter.js';
 
-import { CommonValueConverter } from '../common/commonValueConverters.js';
-
-export class PieValueConverter extends DefaultValueConverter {
-  protected override runConverter(
-    rule: GrammarAST.AbstractRule,
-    input: string,
-    cstNode: CstNode
-  ): ValueType {
-    let value: ValueType | undefined = CommonValueConverter.customRunConverter(
-      rule,
-      input,
-      cstNode
-    );
-    if (value === undefined) {
-      value = PieValueConverter.customRunConverter(rule, input, cstNode);
-    }
-
-    if (value === undefined) {
-      return super.runConverter(rule, input, cstNode);
-    }
-    return value;
-  }
-
-  /**
-   * A method contains convert logic to be used by class itself or `MermaidValueConverter`.
-   *
-   * @param rule - Parsed rule.
-   * @param input - Matched string.
-   * @param _cstNode - Node in the Concrete Syntax Tree (CST).
-   * @returns converted the value if it's pie rule or `null` if it's not.
-   */
-  public static customRunConverter(
+export class PieValueConverter extends MermaidValueConverter {
+  override runCustomConverter(
     rule: GrammarAST.AbstractRule,
     input: string,
     // eslint-disable-next-line @typescript-eslint/no-unused-vars
     _cstNode: CstNode
   ): ValueType | undefined {
-    if (rule.name === 'PIE_SECTION_LABEL') {
-      return input
-        .replace(/"/g, '')
-        .trim()
-        .replaceAll(/[\t ]{2,}/gm, ' ');
+    if (rule.name !== 'PIE_SECTION_LABEL') {
+      return undefined;
     }
-    return undefined;
+
+    return input.replace(/"/g, '').trim();
   }
 }