@eslint-community/regexpp 4.5.1 → 4.6.1

package/README.md

@@ -79,7 +79,7 @@ Parse a regular expression literal.
 - **Return:**
     - The AST of the regular expression.
 
-#### parser.parsePattern(source, start?, end?, uFlag?)
+#### parser.parsePattern(source, start?, end?, flags?)
 
 Parse a regular expression pattern.
 
@@ -87,7 +87,7 @@ Parse a regular expression pattern.
     - `source` (`string`) The source code to parse. E.g. `"abc"`.
     - `start?` (`number`) The start index in the source code. Default is `0`.
     - `end?` (`number`) The end index in the source code. Default is `source.length`.
-    - `uFlag?` (`boolean`) The flag to enable Unicode mode.
+    - `flags?` (`{ unicode?: boolean, unicodeSets?: boolean }`) The flags to enable Unicode mode, and Unicode Set mode.
 - **Return:**
     - The AST of the regular expression pattern.
 
@@ -118,7 +118,7 @@ Validate a regular expression literal.
     - `start?` (`number`) The start index in the source code. Default is `0`.
     - `end?` (`number`) The end index in the source code. Default is `source.length`.
 
-#### validator.validatePattern(source, start, end, uFlag)
+#### validator.validatePattern(source, start, end, flags)
 
 Validate a regular expression pattern.
 
@@ -126,7 +126,7 @@ Validate a regular expression pattern.
     - `source` (`string`) The source code to validate.
     - `start?` (`number`) The start index in the source code. Default is `0`.
     - `end?` (`number`) The end index in the source code. Default is `source.length`.
-    - `uFlag?` (`boolean`) The flag to enable Unicode mode.
+    - `flags?` (`{ unicode?: boolean, unicodeSets?: boolean }`) The flags to enable Unicode mode, and Unicode Set mode.
 
 #### validator.validateFlags(source, start, end)
 
@@ -172,6 +172,6 @@ Please use GitHub's Issues/PRs.
 - `npm run watch` runs tests with `--watch` option.
 
 [`AST.Node`]: src/ast.ts#L4
-[`RegExpParser.Options`]: src/parser.ts#L539
-[`RegExpValidator.Options`]: src/validator.ts#L127
-[`RegExpVisitor.Handlers`]: src/visitor.ts#L204
+[`RegExpParser.Options`]: src/parser.ts#L743
+[`RegExpValidator.Options`]: src/validator.ts#L220
+[`RegExpVisitor.Handlers`]: src/visitor.ts#L291

package/index.d.ts

@@ -44,11 +44,16 @@ declare module "@eslint-community/regexpp/ast" {
     | CapturingGroup
     | CharacterClass
     | CharacterClassRange
+    | ClassIntersection
+    | ClassStringDisjunction
+    | ClassSubtraction
+    | ExpressionCharacterClass
     | Group
     | LookaroundAssertion
     | Pattern
     | Quantifier
-    | RegExpLiteral;
+    | RegExpLiteral
+    | StringAlternative;
   /**
    * The type which includes all leaf nodes.
    */
@@ -71,16 +76,28 @@ declare module "@eslint-community/regexpp/ast" {
     | Character
     | CharacterClass
     | CharacterSet
+    | ExpressionCharacterClass
     | Group
     | LookaheadAssertion;
   /**
    * The type which includes all character class atom nodes.
    */
   export type CharacterClassElement =
+    | ClassRangesCharacterClassElement
+    | UnicodeSetsCharacterClassElement;
+  export type ClassRangesCharacterClassElement =
     | Character
     | CharacterClassRange
     | EscapeCharacterSet
     | UnicodePropertyCharacterSet;
+  export type UnicodeSetsCharacterClassElement =
+    | Character
+    | CharacterClassRange
+    | ClassStringDisjunction
+    | EscapeCharacterSet
+    | ExpressionCharacterClass
+    | UnicodePropertyCharacterSet
+    | UnicodeSetsCharacterClass;
   /**
    * The type which defines common properties for all node types.
    */
@@ -184,12 +201,35 @@ declare module "@eslint-community/regexpp/ast" {
    * The character class.
    * E.g. `[ab]`, `[^ab]`
    */
-  export interface CharacterClass extends NodeBase {
+  export type CharacterClass =
+    | ClassRangesCharacterClass
+    | UnicodeSetsCharacterClass;
+  interface BaseCharacterClass extends NodeBase {
     type: "CharacterClass";
-    parent: Alternative | Quantifier;
+    parent:
+      | Alternative
+      | ExpressionCharacterClass
+      | Quantifier
+      | UnicodeSetsCharacterClass;
+    unicodeSets: boolean;
     negate: boolean;
     elements: CharacterClassElement[];
   }
+  export interface ClassRangesCharacterClass extends BaseCharacterClass {
+    parent: Alternative | Quantifier;
+    unicodeSets: false;
+    elements: ClassRangesCharacterClassElement[];
+  }
+  /** UnicodeSetsCharacterClass is the CharacterClass when in Unicode sets mode. So it may contain strings. */
+  export interface UnicodeSetsCharacterClass extends BaseCharacterClass {
+    parent:
+      | Alternative
+      | ExpressionCharacterClass
+      | Quantifier
+      | UnicodeSetsCharacterClass;
+    unicodeSets: true;
+    elements: UnicodeSetsCharacterClassElement[];
+  }
   /**
    * The character class.
    * E.g. `[a-b]`
@@ -249,7 +289,12 @@ declare module "@eslint-community/regexpp/ast" {
    */
   export interface EscapeCharacterSet extends NodeBase {
     type: "CharacterSet";
-    parent: Alternative | CharacterClass | Quantifier;
+    parent:
+      | Alternative
+      | CharacterClass
+      | ClassIntersection
+      | ClassSubtraction
+      | Quantifier;
     kind: "digit" | "space" | "word";
     negate: boolean;
   }
@@ -257,14 +302,92 @@ declare module "@eslint-community/regexpp/ast" {
    * The unicode property escape.
    * E.g. `\p{ASCII}`, `\P{ASCII}`, `\p{Script=Hiragana}`
    */
-  export interface UnicodePropertyCharacterSet extends NodeBase {
+  export type UnicodePropertyCharacterSet =
+    | CharacterUnicodePropertyCharacterSet
+    | StringsUnicodePropertyCharacterSet;
+  interface BaseUnicodePropertyCharacterSet extends NodeBase {
     type: "CharacterSet";
-    parent: Alternative | CharacterClass | Quantifier;
+    parent:
+      | Alternative
+      | CharacterClass
+      | ClassIntersection
+      | ClassSubtraction
+      | Quantifier;
     kind: "property";
+    strings: boolean;
     key: string;
     value: string | null;
     negate: boolean;
   }
+  export interface CharacterUnicodePropertyCharacterSet
+    extends BaseUnicodePropertyCharacterSet {
+    strings: false;
+    value: string | null;
+    negate: boolean;
+  }
+  /** StringsUnicodePropertyCharacterSet is Unicode property escape with property of strings. */
+  export interface StringsUnicodePropertyCharacterSet
+    extends BaseUnicodePropertyCharacterSet {
+    strings: true;
+    value: null;
+    negate: false;
+  }
+  /**
+   * The expression character class.
+   * E.g. `[a--b]`, `[a&&b]`,`[^a--b]`, `[^a&&b]`
+   */
+  export interface ExpressionCharacterClass extends NodeBase {
+    type: "ExpressionCharacterClass";
+    parent:
+      | Alternative
+      | ExpressionCharacterClass
+      | Quantifier
+      | UnicodeSetsCharacterClass;
+    negate: boolean;
+    expression: ClassIntersection | ClassSubtraction;
+  }
+  export type ClassSetOperand =
+    | Character
+    | ClassStringDisjunction
+    | EscapeCharacterSet
+    | ExpressionCharacterClass
+    | UnicodePropertyCharacterSet
+    | UnicodeSetsCharacterClass;
+  /**
+   * The character class intersection.
+   * E.g. `a&&b`
+   */
+  export interface ClassIntersection extends NodeBase {
+    type: "ClassIntersection";
+    parent: ClassIntersection | ExpressionCharacterClass;
+    left: ClassIntersection | ClassSetOperand;
+    right: ClassSetOperand;
+  }
+  /**
+   * The character class subtraction.
+   * E.g. `a--b`
+   */
+  export interface ClassSubtraction extends NodeBase {
+    type: "ClassSubtraction";
+    parent: ClassSubtraction | ExpressionCharacterClass;
+    left: ClassSetOperand | ClassSubtraction;
+    right: ClassSetOperand;
+  }
+  /**
+   * The character class string disjunction.
+   * E.g. `\q{a|b}`
+   */
+  export interface ClassStringDisjunction extends NodeBase {
+    type: "ClassStringDisjunction";
+    parent: ClassIntersection | ClassSubtraction | UnicodeSetsCharacterClass;
+    alternatives: StringAlternative[];
+  }
+  /** StringAlternative is only used for `\q{alt}`({@link ClassStringDisjunction}). */
+  export interface StringAlternative extends NodeBase {
+    type: "StringAlternative";
+    parent: ClassStringDisjunction;
+    elements: Character[];
+  }
   /**
    * The character.
    * This includes escape sequences which mean a character.
@@ -272,7 +395,14 @@ declare module "@eslint-community/regexpp/ast" {
    */
   export interface Character extends NodeBase {
     type: "Character";
-    parent: Alternative | CharacterClass | CharacterClassRange | Quantifier;
+    parent:
+      | Alternative
+      | CharacterClass
+      | CharacterClassRange
+      | ClassIntersection
+      | ClassSubtraction
+      | Quantifier
+      | StringAlternative;
     value: number;
   }
   /**
@@ -298,7 +428,9 @@ declare module "@eslint-community/regexpp/ast" {
     multiline: boolean;
     sticky: boolean;
     unicode: boolean;
+    unicodeSets: boolean;
   }
+  export {};
 }
 
 declare module "@eslint-community/regexpp/parser" {
@@ -318,13 +450,14 @@ declare module "@eslint-community/regexpp/parser" {
        */
       strict?: boolean;
       /**
-       * ECMAScript version. Default is `2023`.
+       * ECMAScript version. Default is `2024`.
        * - `2015` added `u` and `y` flags.
        * - `2018` added `s` flag, Named Capturing Group, Lookbehind Assertion,
        *   and Unicode Property Escape.
        * - `2019`, `2020`, and `2021` added more valid Unicode Property Escapes.
        * - `2022` added `d` flag.
        * - `2023` added more valid Unicode Property Escapes.
+       * - `2024` added `v` flag.
        */
       ecmaVersion?: EcmaVersion;
     }
@@ -356,6 +489,25 @@ declare module "@eslint-community/regexpp/parser" {
      * @param source The source code to parse.
      * @param start The start index in the source code.
      * @param end The end index in the source code.
+     * @param flags The flags.
+     * @returns The AST of the given pattern.
+     */
+    parsePattern(
+      source: string,
+      start?: number,
+      end?: number,
+      flags?: {
+        unicode?: boolean;
+        unicodeSets?: boolean;
+      }
+    ): Pattern;
+    /**
+     * @deprecated Backward compatibility
+     * Use object `flags` instead of boolean `uFlag`.
+     *
+     * @param source The source code to parse.
+     * @param start The start index in the source code.
+     * @param end The end index in the source code.
      * @param uFlag The flag to set unicode mode.
      * @returns The AST of the given pattern.
      */
@@ -380,13 +532,14 @@ declare module "@eslint-community/regexpp/validator" {
        */
       strict?: boolean;
       /**
-       * ECMAScript version. Default is `2023`.
+       * ECMAScript version. Default is `2024`.
        * - `2015` added `u` and `y` flags.
        * - `2018` added `s` flag, Named Capturing Group, Lookbehind Assertion,
        *   and Unicode Property Escape.
        * - `2019`, `2020`, and `2021` added more valid Unicode Property Escapes.
        * - `2022` added `d` flag.
        * - `2023` added more valid Unicode Property Escapes.
+       * - `2024` added `v` flag.
        */
       ecmaVersion?: EcmaVersion;
       /**
@@ -411,6 +564,7 @@ declare module "@eslint-community/regexpp/validator" {
        * @param flags.sticky `y` flag.
        * @param flags.dotAll `s` flag.
        * @param flags.hasIndices `d` flag.
+       * @param flags.unicodeSets `v` flag.
        */
       onRegExpFlags?: (
         start: number,
@@ -423,6 +577,7 @@ declare module "@eslint-community/regexpp/validator" {
           sticky: boolean;
           dotAll: boolean;
           hasIndices: boolean;
+          unicodeSets: boolean;
         }
       ) => void;
       /**
@@ -604,6 +759,7 @@ declare module "@eslint-community/regexpp/validator" {
        * @param key The property name.
        * @param value The property value.
        * @param negate The flag which represents that the character set is negative.
+       * @param strings If true, the given property is property of strings.
        */
       onUnicodePropertyCharacterSet?: (
         start: number,
@@ -611,7 +767,8 @@ declare module "@eslint-community/regexpp/validator" {
         kind: "property",
         key: string,
         value: string | null,
-        negate: boolean
+        negate: boolean,
+        strings: boolean
       ) => void;
       /**
        * A function that is called when the validator found a character.
@@ -635,8 +792,13 @@ declare module "@eslint-community/regexpp/validator" {
        * A function that is called when the validator entered a character class.
        * @param start The 0-based index of the first character.
        * @param negate The flag which represents that the character class is negative.
+       * @param unicodeSets `true` if unicodeSets mode.
        */
-      onCharacterClassEnter?: (start: number, negate: boolean) => void;
+      onCharacterClassEnter?: (
+        start: number,
+        negate: boolean,
+        unicodeSets: boolean
+      ) => void;
       /**
        * A function that is called when the validator left a character class.
        * @param start The 0-based index of the first character.
@@ -661,6 +823,46 @@ declare module "@eslint-community/regexpp/validator" {
         min: number,
         max: number
       ) => void;
+      /**
+       * A function that is called when the validator found a class intersection.
+       * @param start The 0-based index of the first character.
+       * @param end The next 0-based index of the last character.
+       */
+      onClassIntersection?: (start: number, end: number) => void;
+      /**
+       * A function that is called when the validator found a class subtraction.
+       * @param start The 0-based index of the first character.
+       * @param end The next 0-based index of the last character.
+       */
+      onClassSubtraction?: (start: number, end: number) => void;
+      /**
+       * A function that is called when the validator entered a class string disjunction.
+       * @param start The 0-based index of the first character.
+       */
+      onClassStringDisjunctionEnter?: (start: number) => void;
+      /**
+       * A function that is called when the validator left a class string disjunction.
+       * @param start The 0-based index of the first character.
+       * @param end The next 0-based index of the last character.
+       */
+      onClassStringDisjunctionLeave?: (start: number, end: number) => void;
+      /**
+       * A function that is called when the validator entered a string alternative.
+       * @param start The 0-based index of the first character.
+       * @param index The 0-based index of alternatives in a disjunction.
+       */
+      onStringAlternativeEnter?: (start: number, index: number) => void;
+      /**
+       * A function that is called when the validator left a string alternative.
+       * @param start The 0-based index of the first character.
+       * @param end The next 0-based index of the last character.
+       * @param index The 0-based index of alternatives in a disjunction.
+       */
+      onStringAlternativeLeave?: (
+        start: number,
+        end: number,
+        index: number
+      ) => void;
     }
   }
   /**
@@ -691,6 +893,23 @@ declare module "@eslint-community/regexpp/validator" {
      * @param source The source code to validate.
      * @param start The start index in the source code.
      * @param end The end index in the source code.
+     * @param flags The flags.
+     */
+    validatePattern(
+      source: string,
+      start?: number,
+      end?: number,
+      flags?: {
+        unicode?: boolean;
+        unicodeSets?: boolean;
+      }
+    ): void;
+    /**
+     * @deprecated Backward compatibility
+     * Use object `flags` instead of boolean `uFlag`.
+     * @param source The source code to validate.
+     * @param start The start index in the source code.
+     * @param end The end index in the source code.
      * @param uFlag The flag to set unicode mode.
      */
     validatePattern(
@@ -712,12 +931,17 @@ declare module "@eslint-community/regexpp/visitor" {
     CharacterClass,
     CharacterClassRange,
     CharacterSet,
+    ClassIntersection,
+    ClassStringDisjunction,
+    ClassSubtraction,
+    ExpressionCharacterClass,
     Flags,
     Group,
     Node,
     Pattern,
     Quantifier,
     RegExpLiteral,
+    StringAlternative,
   } from "@eslint-community/regexpp/ast";
   /**
    * The visitor to walk on AST.
@@ -752,6 +976,18 @@ declare module "@eslint-community/regexpp/visitor" {
       onCharacterClassRangeLeave?: (node: CharacterClassRange) => void;
       onCharacterSetEnter?: (node: CharacterSet) => void;
       onCharacterSetLeave?: (node: CharacterSet) => void;
+      onClassIntersectionEnter?: (node: ClassIntersection) => void;
+      onClassIntersectionLeave?: (node: ClassIntersection) => void;
+      onClassStringDisjunctionEnter?: (node: ClassStringDisjunction) => void;
+      onClassStringDisjunctionLeave?: (node: ClassStringDisjunction) => void;
+      onClassSubtractionEnter?: (node: ClassSubtraction) => void;
+      onClassSubtractionLeave?: (node: ClassSubtraction) => void;
+      onExpressionCharacterClassEnter?: (
+        node: ExpressionCharacterClass
+      ) => void;
+      onExpressionCharacterClassLeave?: (
+        node: ExpressionCharacterClass
+      ) => void;
       onFlagsEnter?: (node: Flags) => void;
       onFlagsLeave?: (node: Flags) => void;
       onGroupEnter?: (node: Group) => void;
@@ -762,6 +998,8 @@ declare module "@eslint-community/regexpp/visitor" {
       onQuantifierLeave?: (node: Quantifier) => void;
       onRegExpLiteralEnter?: (node: RegExpLiteral) => void;
       onRegExpLiteralLeave?: (node: RegExpLiteral) => void;
+      onStringAlternativeEnter?: (node: StringAlternative) => void;
+      onStringAlternativeLeave?: (node: StringAlternative) => void;
     }
   }
 }
@@ -777,5 +1015,7 @@ declare module "@eslint-community/regexpp/ecma-versions" {
     | 2020
     | 2021
     | 2022
-    | 2023;
+    | 2023
+    | 2024;
+  export const latestEcmaVersion = 2024;
 }