feat: inverted character class (#27)

Author: mdjastrzebski

src/components/__tests__/character-class.test.ts

@@ -5,9 +5,11 @@ import {
   anyOf,
   digit,
   encodeCharacterClass,
+  inverted,
   whitespace,
   word,
 } from '../character-class';
+import { execRegex } from '../../test-utils';
 
 test('"whitespace" character class', () => {
   expect(buildPattern(whitespace)).toEqual(`\\s`);
@@ -60,17 +62,33 @@ test('"anyOf" moves hyphen to the first position', () => {
   expect(buildPattern(anyOf('a-bc'))).toBe('[-abc]');
 });
 
-test('`anyOf` throws on empty text', () => {
+test('"anyOf" throws on empty text', () => {
   expect(() => anyOf('')).toThrowErrorMatchingInlineSnapshot(
     `"\`anyOf\` should received at least one character"`
   );
 });
 
+test('"inverted" character class', () => {
+  expect(buildPattern(inverted(anyOf('a')))).toBe('[^a]');
+  expect(buildPattern(inverted(anyOf('abc')))).toBe('[^abc]');
+});
+
+test('"inverted" character class double inversion', () => {
+  expect(buildPattern(inverted(inverted(anyOf('a'))))).toBe('a');
+  expect(buildPattern(inverted(inverted(anyOf('abc'))))).toBe('[abc]');
+});
+
+test('"inverted" character class execution', () => {
+  expect(execRegex('aa', [inverted(anyOf('a'))])).toBeNull();
+  expect(execRegex('aba', [inverted(anyOf('a'))])).toEqual(['b']);
+});
+
 test('buildPattern throws on empty text', () => {
   expect(() =>
     encodeCharacterClass({
       type: 'characterClass',
       characters: [],
+      inverted: false,
     })
   ).toThrowErrorMatchingInlineSnapshot(
     `"Character class should contain at least one character"`

src/components/character-class.ts

@@ -5,21 +5,25 @@ import type { CharacterClass } from './types';
 export const any: CharacterClass = {
   type: 'characterClass',
   characters: ['.'],
+  inverted: false,
 };
 
 export const whitespace: CharacterClass = {
   type: 'characterClass',
   characters: ['\\s'],
+  inverted: false,
 };
 
 export const digit: CharacterClass = {
   type: 'characterClass',
   characters: ['\\d'],
+  inverted: false,
 };
 
 export const word: CharacterClass = {
   type: 'characterClass',
   characters: ['\\w'],
+  inverted: false,
 };
 
 export function anyOf(characters: string): CharacterClass {
@@ -31,26 +35,36 @@ export function anyOf(characters: string): CharacterClass {
   return {
     type: 'characterClass',
     characters: charactersArray,
+    inverted: false,
   };
 }
 
-export function encodeCharacterClass({
-  characters,
-}: CharacterClass): EncoderNode {
-  if (characters.length === 0) {
+export function inverted(characterClass: CharacterClass): CharacterClass {
+  return {
+    type: 'characterClass',
+    characters: characterClass.characters,
+    inverted: !characterClass.inverted,
+  };
+}
+
+export function encodeCharacterClass(
+  characterClass: CharacterClass
+): EncoderNode {
+  if (characterClass.characters.length === 0) {
     throw new Error('Character class should contain at least one character');
   }
 
-  if (characters.length === 1) {
+  if (characterClass.characters.length === 1 && !characterClass.inverted) {
     return {
       precedence: EncoderPrecedence.Atom,
-      pattern: characters[0]!,
+      pattern: characterClass.characters[0]!,
     };
   }
 
+  const characterString = reorderHyphen(characterClass.characters).join('');
   return {
     precedence: EncoderPrecedence.Atom,
-    pattern: `[${reorderHyphen(characters).join('')}]`,
+    pattern: `[${characterClass.inverted ? '^' : ''}${characterString}]`,
   };
 }
 

src/components/types.ts

@@ -5,6 +5,7 @@ export type Quantifier = One | OneOrMore | Optionally | ZeroOrMore | Repeat;
 export type CharacterClass = {
   type: 'characterClass';
   characters: string[];
+  inverted: boolean;
 };
 
 // Components

src/test-utils.ts

@@ -6,5 +6,6 @@ export function execRegex(
   elements: Array<RegexElement | string>
 ) {
   const regex = buildRegex(...elements);
-  return [...regex.exec(text)!];
+  const result = regex.exec(text);
+  return result ? [...result] : null;
 }