refactor: Simplify TokenBuilder and ValueConverter

packages/parser/src/language/common/common.langium

@@ -1,14 +1,19 @@
+interface Common {
+  accDescr?: string;
+  accTitle?: string;
+  title?: string;
+}
+
 fragment TitleAndAccessibilities:
   ((accDescr=ACC_DESCR | accTitle=ACC_TITLE | title=TITLE) NEWLINE+)+
 ;
 
 terminal NEWLINE: /\r?\n/;
-terminal ACC_DESCR: /accDescr(?:[\t ]*:[\t ]*[^\n\r]*?(?=%%)|\s*{[^}]*})|accDescr(?:[\t ]*:[\t ]*[^\n\r]*|\s*{[^}]*})/;
-terminal ACC_TITLE: /accTitle[\t ]*:[\t ]*[^\n\r]*?(?=%%)|accTitle[\t ]*:[\t ]*[^\n\r]*/;
-terminal TITLE: /title(?:[\t ]+[^\n\r]*?|)(?=%%)|title(?:[\t ]+[^\n\r]*|)/;
+terminal ACC_DESCR: /accDescr([\t ]*:[^\n\r]*(?=%%)|\s*{[^}]*})|accDescr([\t ]*:[^\n\r]*|\s*{[^}]*})/;
+terminal ACC_TITLE: /accTitle[\t ]*:[^\n\r]*(?=%%)|accTitle[\t ]*:[^\n\r]*/;
+terminal TITLE: /title([\t ][^\n\r]*|)(?=%%)|title([\t ][^\n\r]*|)/;
 
 hidden terminal WHITESPACE: /[\t ]+/;
-// TODO: add YAML_COMMENT hidden rule without interfere actual grammar
 hidden terminal YAML: /---[\t ]*\r?\n[\S\s]*?---[\t ]*(?!.)/;
 hidden terminal DIRECTIVE: /[\t ]*%%{[\S\s]*?}%%\s*/;
 hidden terminal SINGLE_LINE_COMMENT: /[\t ]*%%[^\n\r]*/;

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

@@ -1,2 +1,3 @@
-export * from './commonLexer.js';
-export * from './commonValueConverters.js';
+export * from './lexer.js';
+export * from './tokenBuilder.js';
+export { MermaidValueConverter } from './valueConverter.js';

packages/parser/src/language/info/infoTokenBuilder.ts → packages/parser/src/language/common/tokenBuilder.ts

@@ -1,9 +1,16 @@
 import type { GrammarAST, Stream, TokenBuilderOptions } from 'langium';
+import type { TokenType } from '../chevrotainWrapper.js';
+
 import { DefaultTokenBuilder } from 'langium';
 
-import type { TokenType } from '../chevrotainWrapper.js';
+export abstract class MermaidTokenBuilder extends DefaultTokenBuilder {
+  private keywords: Set<string>;
+
+  public constructor(keywords: string[]) {
+    super();
+    this.keywords = new Set<string>(keywords);
+  }
 
-export class InfoTokenBuilder extends DefaultTokenBuilder {
   protected override buildKeywordTokens(
     rules: Stream<GrammarAST.AbstractRule>,
     terminalTokens: TokenType[],
@@ -12,10 +19,7 @@ export class InfoTokenBuilder extends DefaultTokenBuilder {
     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
-      ) {
+      if (this.keywords.has(tokenType.name) && tokenType.PATTERN !== undefined) {
         tokenType.PATTERN = new RegExp(tokenType.PATTERN.toString() + '(?!\\S)');
       }
     });

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

@@ -0,0 +1,82 @@
+/* eslint-disable @typescript-eslint/no-unused-vars */
+import type { CstNode, GrammarAST, ValueType } from 'langium';
+import { DefaultValueConverter } from 'langium';
+
+import { accessibilityDescrRegex, accessibilityTitleRegex, titleRegex } from './matcher.js';
+
+const rulesRegexes: Record<string, RegExp> = {
+  ACC_DESCR: accessibilityDescrRegex,
+  ACC_TITLE: accessibilityTitleRegex,
+  TITLE: titleRegex,
+};
+
+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 {
+    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);
+    }
+
+    return value;
+  }
+
+  private runCommonConverter(
+    rule: GrammarAST.AbstractRule,
+    input: string,
+    _cstNode: CstNode
+  ): ValueType | undefined {
+    const regex: RegExp | undefined = rulesRegexes[rule.name];
+    if (regex === undefined) {
+      return undefined;
+    }
+    const match = regex.exec(input);
+    if (match === null) {
+      return undefined;
+    }
+    // single line title, accTitle, accDescr
+    if (match[1] !== undefined) {
+      return match[1].trim().replace(/[\t ]{2,}/gm, ' ');
+    }
+    // multi line accDescr
+    if (match[2] !== undefined) {
+      return match[2]
+        .replace(/^\s*/gm, '')
+        .replace(/\s+$/gm, '')
+        .replace(/[\t ]{2,}/gm, ' ')
+        .replace(/[\n\r]{2,}/gm, '\n');
+    }
+    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/index.ts

@@ -1 +1 @@
-export * from './infoModule.js';
+export * from './module.js';

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

@@ -7,10 +7,10 @@ import type {
 } from 'langium';
 import { EmptyFileSystem, createDefaultModule, createDefaultSharedModule, inject } from 'langium';
 
-import { MermaidGeneratedSharedModule, InfoGeneratedModule } from '../generated/module.js';
-import { CommonLexer } from '../common/commonLexer.js';
-import { CommonValueConverter } from '../common/commonValueConverters.js';
-import { InfoTokenBuilder } from './infoTokenBuilder.js';
+import { CommonLexer } from '../common/lexer.js';
+import { CommonValueConverter } from '../common/valueConverter.js';
+import { InfoGeneratedModule, MermaidGeneratedSharedModule } from '../generated/module.js';
+import { InfoTokenBuilder } from './tokenBuilder.js';
 
 /**
  * Declaration of `Info` services.
@@ -34,7 +34,7 @@ export type InfoServices = LangiumServices & InfoAddedServices;
  */
 export const InfoModule: Module<InfoServices, PartialLangiumServices & InfoAddedServices> = {
   parser: {
-    Lexer: (services) => new CommonLexer(services),
+    Lexer: (services: InfoServices) => new CommonLexer(services),
     TokenBuilder: () => new InfoTokenBuilder(),
     ValueConverter: () => new CommonValueConverter(),
   },

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

@@ -0,0 +1,7 @@
+import { MermaidTokenBuilder } from '../common/index.js';
+
+export class InfoTokenBuilder extends MermaidTokenBuilder {
+  public constructor() {
+    super(['info', 'showInfo']);
+  }
+}