feat: add basic character classes (#13)

Co-authored-by: Maciej Jastrzebski <[email protected]>

.gitignore

@@ -30,3 +30,5 @@ buck-out/
 
 # generated by bob
 lib/
+
+/.idea

src/__tests__/characterClasses.test.tsx

@@ -0,0 +1,35 @@
+import { any, digit, whitespace, word } from '../character-classes';
+import { buildPattern } from '../compiler';
+import { one } from '../quantifiers/base';
+
+test('"whitespace" character class', () => {
+  expect(buildPattern(whitespace)).toEqual(`\\s`);
+
+  expect(buildPattern(one('ab'), whitespace)).toEqual(`ab\\s`);
+
+  expect(buildPattern(one('ab'), whitespace, one('c'))).toEqual(`ab\\sc`);
+});
+
+test('"digit" character class', () => {
+  expect(buildPattern(digit)).toEqual(`\\d`);
+
+  expect(buildPattern(one('ab'), digit)).toEqual(`ab\\d`);
+
+  expect(buildPattern(one('ab'), digit, one('c'))).toEqual(`ab\\dc`);
+});
+
+test('"word" character class', () => {
+  expect(buildPattern(word)).toEqual(`\\w`);
+
+  expect(buildPattern(one('ab'), word)).toEqual(`ab\\w`);
+
+  expect(buildPattern(one('ab'), word, one('c'))).toEqual(`ab\\wc`);
+});
+
+test('"any" character class', () => {
+  expect(buildPattern(any)).toEqual(`.`);
+
+  expect(buildPattern(one('ab'), any)).toEqual(`ab.`);
+
+  expect(buildPattern(one('ab'), any, one('c'))).toEqual(`ab.c`);
+});

src/__tests__/compiler.test.tsx

@@ -1,5 +1,6 @@
 import { buildPattern, buildRegex } from '../compiler';
-import { oneOrMore, optionally, one, zeroOrMore, repeat } from '../quantifiers';
+import { oneOrMore, optionally, one, zeroOrMore } from '../quantifiers/base';
+import { repeat } from '../quantifiers/repeat';
 
 test('basic quantifies', () => {
   expect(buildPattern('a')).toEqual('a');

src/__tests__/quantifiers.test.tsx

@@ -1,4 +1,4 @@
-import { one, oneOrMore, optionally, zeroOrMore } from '../quantifiers';
+import { one, oneOrMore, optionally, zeroOrMore } from '../quantifiers/base';
 import { buildPattern, buildRegex } from '../compiler';
 
 test('"oneOrMore" quantifier', () => {

src/__tests__/repeat.test.tsx

@@ -1,5 +1,6 @@
 import { buildPattern } from '../compiler';
-import { repeat, zeroOrMore, oneOrMore } from '../quantifiers';
+import { zeroOrMore, oneOrMore } from '../quantifiers/base';
+import { repeat } from '../quantifiers/repeat';
 
 test('"repeat" quantifier', () => {
   expect(buildPattern('a', repeat({ min: 1, max: 5 }, 'b'))).toEqual('ab{1,5}');

src/character-classes.ts

@@ -0,0 +1,26 @@
+import type {
+  Any,
+  CharacterClass,
+  Digit,
+  RegexElement,
+  Whitespace,
+  Word,
+} from './types';
+
+export const whitespace: Whitespace = { type: 'whitespace' };
+export const digit: Digit = { type: 'digit' };
+export const word: Word = { type: 'word' };
+export const any: Any = { type: 'any' };
+
+export const characterClasses = {
+  whitespace: '\\s',
+  digit: '\\d',
+  word: '\\w',
+  any: '.',
+} as const satisfies Record<string, string>;
+
+export function isCharacterClass(
+  element: Exclude<RegexElement, string>
+): element is CharacterClass {
+  return element.type in characterClasses;
+}

src/compiler.ts

@@ -1,5 +1,6 @@
 import type { RegexElement } from './types';
-import { compilers as quantifiers } from './quantifiers';
+import { characterClasses, isCharacterClass } from './character-classes';
+import { baseQuantifiers, isBaseQuantifier } from './quantifiers/base';
 import { compileRepeat } from './quantifiers/repeat';
 
 /**
@@ -28,21 +29,26 @@ function compileList(elements: RegexElement[]): string {
   return elements.map((c) => compileSingle(c)).join('');
 }
 
-function compileSingle(elements: RegexElement): string {
-  if (typeof elements === 'string') {
-    return elements;
+function compileSingle(element: RegexElement): string {
+  if (typeof element === 'string') {
+    return element;
   }
 
-  const compiledChildren = compileList(elements.children);
+  if (isCharacterClass(element)) {
+    return characterClasses[element.type];
+  }
+
+  const compiledChildren = compileList(element.children);
 
-  if (elements.type === 'repeat') {
-    return compileRepeat(elements.config, compiledChildren);
+  if (element.type === 'repeat') {
+    return compileRepeat(element.config, compiledChildren);
   }
 
-  const elementCompiler = quantifiers[elements.type];
-  if (!elementCompiler) {
-    throw new Error(`Unknown elements type ${elements.type}`);
+  if (isBaseQuantifier(element)) {
+    const compiler = baseQuantifiers[element.type];
+    return compiler(compiledChildren);
   }
 
-  return elementCompiler(compiledChildren);
+  // @ts-expect-error User passed incorrect type
+  throw new Error(`Unknown elements type ${element.type}`);
 }

src/index.tsx

@@ -1,4 +1,5 @@
 export type * from './types';
 
+export { whitespace } from './character-classes';
 export { buildRegex, buildPattern } from './compiler';
-export { oneOrMore, optionally } from './quantifiers';
+export { oneOrMore, optionally } from './quantifiers/base';

src/quantifiers/index.ts → src/quantifiers/base.ts

@@ -2,12 +2,10 @@ import type {
   One,
   OneOrMore,
   Optionally,
+  Quantifier,
   RegexElement,
-  Repeat,
-  RepeatConfig,
   ZeroOrMore,
 } from '../types';
-import type { CompilerMap } from '../types-internal';
 import { wrapGroup } from '../utils';
 
 export function oneOrMore(...children: RegexElement[]): OneOrMore {
@@ -38,20 +36,15 @@ export function zeroOrMore(...children: RegexElement[]): ZeroOrMore {
   };
 }
 
-export function repeat(
-  config: RepeatConfig,
-  ...children: RegexElement[]
-): Repeat {
-  return {
-    type: 'repeat',
-    children,
-    config,
-  };
-}
-
-export const compilers = {
+export const baseQuantifiers = {
   one: (compiledChildren) => compiledChildren,
   oneOrMore: (compiledChildren) => `${wrapGroup(compiledChildren)}+`,
   optionally: (compiledChildren) => `${wrapGroup(compiledChildren)}?`,
   zeroOrMore: (compiledChildren) => `${wrapGroup(compiledChildren)}*`,
-} satisfies CompilerMap;
+} as const satisfies Record<string, (compiledChildren: string) => string>;
+
+export function isBaseQuantifier(
+  element: Exclude<RegexElement, string>
+): element is Quantifier {
+  return element.type in baseQuantifiers;
+}

src/quantifiers/repeat.ts

@@ -1,17 +1,24 @@
-import type { RepeatConfig } from '../types';
+import type { RegexElement, Repeat, RepeatConfig } from '../types';
 import { wrapGroup } from '../utils';
 
+export function repeat(
+  config: RepeatConfig,
+  ...children: RegexElement[]
+): Repeat {
+  return {
+    type: 'repeat',
+    children,
+    config,
+  };
+}
+
 export function compileRepeat(
   config: RepeatConfig,
   compiledChildren: string
 ): string {
-  if ('count' in config && typeof config.count === 'number') {
+  if ('count' in config) {
     return `${wrapGroup(compiledChildren)}{${config.count}}`;
   }
 
-  if ('min' in config && typeof config.min === 'number') {
-    return `${wrapGroup(compiledChildren)}{${config.min},${config?.max ?? ''}}`;
-  }
-
-  return `${wrapGroup(compiledChildren)}`;
+  return `${wrapGroup(compiledChildren)}{${config.min},${config?.max ?? ''}}`;
 }

src/types.ts

@@ -1,11 +1,14 @@
-export type RegexElement = string | RegexQuantifier;
-
-export type RegexQuantifier =
-  | One
-  | OneOrMore
-  | Optionally
-  | ZeroOrMore
-  | Repeat;
+export type RegexElement = string | CharacterClass | Quantifier;
+
+export type CharacterClass = Whitespace | Digit | Word | Any;
+
+export type Quantifier = One | OneOrMore | Optionally | ZeroOrMore | Repeat;
+
+// Character classes
+export type Whitespace = { type: 'whitespace' };
+export type Digit = { type: 'digit' };
+export type Word = { type: 'word' };
+export type Any = { type: 'any' };
 
 // Quantifiers
 export type One = {