javascript: Prefer the longest keywords (#407)

Author: davidjgoss

CHANGELOG.md

@@ -14,6 +14,7 @@ This document is formatted according to the principles of [Keep A CHANGELOG](htt
 - [.NET] Prefer the longest step keyword ([#405](https://github.com/cucumber/gherkin/pull/405))
 - [PHP] Prefer the longest step keyword ([#403](https://github.com/cucumber/gherkin/pull/403))
 - [Go] Prefer the longest step keyword ([#403](https://github.com/cucumber/gherkin/pull/404))
+- [JavaScript] Prefer the longest step keyword ([#407](https://github.com/cucumber/gherkin/pull/407))
 
 ## [32.1.1] - 2025-04-11
 ### Fixed

javascript/src/GherkinClassicTokenMatcher.ts

@@ -6,6 +6,7 @@ import * as messages from '@cucumber/messages'
 import { TokenType } from './Parser'
 import ITokenMatcher from './ITokenMatcher'
 import countSymbols from './countSymbols'
+import { compareStepKeywords } from './compareStepKeywords'
 
 const DIALECT_DICT: { [key: string]: Dialect } = DIALECTS
 const LANGUAGE_PATTERN = /^\s*#\s*language\s*:\s*([a-zA-Z\-_]+)\s*$/
@@ -25,6 +26,7 @@ export default class GherkinClassicTokenMatcher implements ITokenMatcher<TokenTy
   private activeDocStringSeparator: string
   private indentToRemove: number
   private keywordTypesMap: { [key: string]: messages.StepKeywordType[] }
+  private sortedStepKeywords: readonly string[]
 
   constructor(private readonly defaultDialectName: string = 'en') {
     this.reset()
@@ -39,6 +41,7 @@ export default class GherkinClassicTokenMatcher implements ITokenMatcher<TokenTy
     this.dialectName = newDialectName
     this.dialect = newDialect
     this.initializeKeywordTypes()
+    this.initializeSortedStepKeywords()
   }
 
   reset() {
@@ -59,6 +62,16 @@ export default class GherkinClassicTokenMatcher implements ITokenMatcher<TokenTy
                            messages.StepKeywordType.CONJUNCTION)
   }
 
+  initializeSortedStepKeywords() {
+    this.sortedStepKeywords = []
+      .concat(this.dialect.given)
+      .concat(this.dialect.when)
+      .concat(this.dialect.then)
+      .concat(this.dialect.and)
+      .concat(this.dialect.but)
+      .sort(compareStepKeywords)
+  }
+
   match_TagLine(token: IToken<TokenType>) {
     if (token.line.startsWith('@')) {
       this.setTokenMatched(token, TokenType.TagLine, null, null, null, null, this.getTags(token.line))
@@ -164,13 +177,7 @@ export default class GherkinClassicTokenMatcher implements ITokenMatcher<TokenTy
   }
 
   match_StepLine(token: IToken<TokenType>) {
-    const keywords = []
-      .concat(this.dialect.given)
-      .concat(this.dialect.when)
-      .concat(this.dialect.then)
-      .concat(this.dialect.and)
-      .concat(this.dialect.but)
-    for (const keyword of keywords) {
+    for (const keyword of this.sortedStepKeywords) {
       if (token.line.startsWith(keyword)) {
         const title = token.line.getRestTrimmed(keyword.length)
         const keywordTypes = this.keywordTypesMap[keyword]

javascript/src/GherkinInMarkdownTokenMatcher.ts

@@ -5,6 +5,7 @@ import DIALECTS from './gherkin-languages.json'
 import { Item } from './IToken'
 import * as messages from '@cucumber/messages'
 import { NoSuchLanguageException } from './Errors'
+import { compareStepKeywords } from './compareStepKeywords'
 
 const DIALECT_DICT: { [key: string]: Dialect } = DIALECTS
 const DEFAULT_DOC_STRING_SEPARATOR = /^(```[`]*)(.*)/
@@ -38,6 +39,7 @@ export default class GherkinInMarkdownTokenMatcher implements ITokenMatcher<Toke
       .concat(this.dialect.and)
       .concat(this.dialect.but)
       .filter((value, index, self) => value !== '* ' && self.indexOf(value) === index)
+      .sort(compareStepKeywords)
     this.initializeKeywordTypes()
 
     this.stepRegexp = new RegExp(

javascript/src/compareStepKeywords.ts

@@ -0,0 +1,10 @@
+/**
+ * Compares two step keywords based on length. Can be used with Array.sort to
+ * sort keywords by length, descending.
+ *
+ * @param a - the first step keyword
+ * @param b - the second step keyword
+ */
+export function compareStepKeywords(a: string, b: string): number {
+    return b.length - a.length
+}