Add support for ES2025 RegExp Duplicate named capturing groups

acornjs · Apr 14, 2024 · f03e191 · f03e191
1 parent ed4a7a1
commit f03e191
Show file tree

Hide file tree

Showing 5 changed files with 55 additions and 3 deletions.
diff --git a/acorn/src/acorn.d.ts b/acorn/src/acorn.d.ts
@@ -573,7 +573,7 @@ export function tokenizer(input: string, options: Options): {
   [Symbol.iterator](): Iterator<Token>
 }
 
-export type ecmaVersion = 3 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 2015 | 2016 | 2017 | 2018 | 2019 | 2020 | 2021 | 2022 | 2023 | 2024 | "latest"
+export type ecmaVersion = 3 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 2015 | 2016 | 2017 | 2018 | 2019 | 2020 | 2021 | 2022 | 2023 | 2024 | 2025 | "latest"
 
 export interface Options {
   /**

diff --git a/acorn/src/regexp.js b/acorn/src/regexp.js
@@ -23,6 +23,8 @@ export class RegExpValidationState {
     this.numCapturingParens = 0
     this.maxBackReference = 0
     this.groupNames = []
+    this.groupNamesInDisjunction = []
+    this.groupNamesInAlternative = []
     this.backReferenceNames = []
   }
 
@@ -169,6 +171,8 @@ pp.regexp_pattern = function(state) {
   state.numCapturingParens = 0
   state.maxBackReference = 0
   state.groupNames.length = 0
+  state.groupNamesInDisjunction.length = 0
+  state.groupNamesInAlternative.length = 0
   state.backReferenceNames.length = 0
 
   this.regexp_disjunction(state)
@@ -194,11 +198,27 @@ pp.regexp_pattern = function(state) {
 
 // https://www.ecma-international.org/ecma-262/8.0/#prod-Disjunction
 pp.regexp_disjunction = function(state) {
+  let upperGroupNamesInDisjunction
+  if (this.options.ecmaVersion >= 16) {
+    upperGroupNamesInDisjunction = state.groupNamesInDisjunction
+    state.groupNamesInDisjunction = []
+  }
+
   this.regexp_alternative(state)
   while (state.eat(0x7C /* | */)) {
     this.regexp_alternative(state)
   }
 
+  if (this.options.ecmaVersion >= 16) {
+    // Adds the group name that appears in current Disjunction
+    // as the group name of the current Alternative and upper Disjunction.
+    for (const groupName of state.groupNamesInDisjunction) {
+      upperGroupNamesInDisjunction.push(groupName)
+      state.groupNamesInAlternative.push(groupName)
+    }
+    state.groupNamesInDisjunction = upperGroupNamesInDisjunction
+  }
+
   // Make the same message as V8.
   if (this.regexp_eatQuantifier(state, true)) {
     state.raise("Nothing to repeat")
@@ -210,8 +230,18 @@ pp.regexp_disjunction = function(state) {
 
 // https://www.ecma-international.org/ecma-262/8.0/#prod-Alternative
 pp.regexp_alternative = function(state) {
+  let upperGroupNamesInAlternative
+  if (this.options.ecmaVersion >= 16) {
+    upperGroupNamesInAlternative = state.groupNamesInAlternative
+    state.groupNamesInAlternative = [...state.groupNamesInAlternative]
+  }
+
   while (state.pos < state.source.length && this.regexp_eatTerm(state))
     ;
+
+  if (this.options.ecmaVersion >= 16) {
+    state.groupNamesInAlternative = upperGroupNamesInAlternative
+  }
 }
 
 // https://www.ecma-international.org/ecma-262/8.0/#prod-annexB-Term
@@ -448,10 +478,15 @@ pp.regexp_eatExtendedPatternCharacter = function(state) {
 pp.regexp_groupSpecifier = function(state) {
   if (state.eat(0x3F /* ? */)) {
     if (this.regexp_eatGroupName(state)) {
-      if (state.groupNames.indexOf(state.lastStringValue) !== -1) {
+      const groupNames = this.options.ecmaVersion >= 16 ? state.groupNamesInAlternative : state.groupNames
+      if (groupNames.indexOf(state.lastStringValue) !== -1) {
         state.raise("Duplicate capture group name")
       }
       state.groupNames.push(state.lastStringValue)
+      if (this.options.ecmaVersion >= 16) {
+        state.groupNamesInAlternative.push(state.lastStringValue)
+        state.groupNamesInDisjunction.push(state.lastStringValue)
+      }
       return
     }
     state.raise("Invalid group")

diff --git a/bin/test262.unsupported-features b/bin/test262.unsupported-features
@@ -1,3 +1,2 @@
 decorators
 import-assertions
-regexp-duplicate-named-groups
diff --git a/test/run.js b/test/run.js
@@ -15,6 +15,7 @@
   require("./tests-regexp-2020.js");
   require("./tests-regexp-2022.js");
   require("./tests-regexp-2024.js");
+  require("./tests-regexp-2025.js");
   require("./tests-json-superset.js");
   require("./tests-optional-catch-binding.js");
   require("./tests-bigint.js");

diff --git a/test/tests-regexp-2025.js b/test/tests-regexp-2025.js
@@ -0,0 +1,17 @@
+if (typeof exports !== "undefined") {
+  var test = require("./driver.js").test
+  var testFail = require("./driver.js").testFail
+}
+
+test("/(?<x>a)|(?<x>b)/", {}, { ecmaVersion: 2025 })
+testFail("/(?<x>a)|(?<x>b)/", "Invalid regular expression: /(?<x>a)|(?<x>b)/: Duplicate capture group name (1:1)", { ecmaVersion: 2024 })
+testFail("/(?<x>a)(?<x>b)/", "Invalid regular expression: /(?<x>a)(?<x>b)/: Duplicate capture group name (1:1)", { ecmaVersion: 2025 })
+test("/(?:(?<x>a)|(?<x>b))\\k<x>/", {}, { ecmaVersion: 2025 })
+testFail("/(?:(?<x>a)|(?<x>b))\\k<x>/", "Invalid regular expression: /(?:(?<x>a)|(?<x>b))\\k<x>/: Duplicate capture group name (1:1)", { ecmaVersion: 2024 })
+testFail("/(?:(?<x>a)(?<x>b))\\k<x>/", "Invalid regular expression: /(?:(?<x>a)(?<x>b))\\k<x>/: Duplicate capture group name (1:1)", { ecmaVersion: 2025 })
+test("/(?<y>a)(?<x>a)|(?<x>b)(?<y>b)/", {}, { ecmaVersion: 2025 })
+test("/(?<x>a)|(?<x>b)|(?<x>c)/", {}, { ecmaVersion: 2025 })
+test("/(?<x>a)|\\k<x>/", {}, { ecmaVersion: 2025 })
+testFail("/(?<x>a)|(?<x>b)(?<x>c)/", "Invalid regular expression: /(?<x>a)|(?<x>b)(?<x>c)/: Duplicate capture group name (1:1)", { ecmaVersion: 2025 })
+testFail("/(?:(?<x>a)|(?<x>b))(?<x>c)/", "Invalid regular expression: /(?:(?<x>a)|(?<x>b))(?<x>c)/: Duplicate capture group name (1:1)", { ecmaVersion: 2025 })
+testFail("/(?:(?:(?<x>a)|(?<x>b))|(?:))(?<x>c)/", "Invalid regular expression: /(?:(?:(?<x>a)|(?<x>b))|(?:))(?<x>c)/: Duplicate capture group name (1:1)", { ecmaVersion: 2025 })