feat: base (non-named) capture (#24)

Author: mdjastrzebski

src/__tests__/capture.test.tsx

@@ -0,0 +1,21 @@
+import { buildPattern } from '..';
+import { capture } from '../capture';
+import { oneOrMore } from '../quantifiers/base';
+import { execRegex } from '../test-utils';
+
+test('"capture" base cases', () => {
+  expect(buildPattern(capture('a'))).toBe('(a)');
+  expect(buildPattern(capture('abc'))).toBe('(abc)');
+  expect(buildPattern(capture(oneOrMore('abc')))).toBe('((?:abc)+)');
+  expect(buildPattern(oneOrMore(capture('abc')))).toBe('(abc)+');
+});
+
+test('"capture" captures group', () => {
+  expect(execRegex('ab', [capture('b')])).toEqual(['b', 'b']);
+  expect(execRegex('ab', ['a', capture('b')])).toEqual(['ab', 'b']);
+  expect(execRegex('abc', ['a', capture('b'), capture('c')])).toEqual([
+    'abc',
+    'b',
+    'c',
+  ]);
+});

src/capture.ts

@@ -0,0 +1,16 @@
+import type { Capture, RegexElement } from './types';
+import { EncoderPriority, type EncoderNode } from './types-internal';
+
+export function capture(...children: RegexElement[]): Capture {
+  return {
+    type: 'capture',
+    children,
+  };
+}
+
+export function encodeCapture(node: EncoderNode): EncoderNode {
+  return {
+    pattern: `(${node.pattern})`,
+    priority: EncoderPriority.Atom,
+  };
+}

src/encoder.ts

@@ -10,6 +10,7 @@ import {
 } from './quantifiers/base';
 import { encodeRepeat } from './quantifiers/repeat';
 import { concatNodes, escapeText } from './utils';
+import { encodeCapture } from './capture';
 
 export function encodeSequence(elements: RegexElement[]): EncoderNode {
   return concatNodes(elements.map((c) => encodeElement(c)));
@@ -48,6 +49,10 @@ export function encodeElement(element: RegexElement): EncoderNode {
     return encodeZeroOrMore(encodeSequence(element.children));
   }
 
+  if (element.type === 'capture') {
+    return encodeCapture(encodeSequence(element.children));
+  }
+
   // @ts-expect-error User passed incorrect type
   throw new Error(`Unknown elements type ${element.type}`);
 }

src/test-utils.ts

@@ -0,0 +1,12 @@
+import { buildRegex } from '.';
+import type { RegexElement } from './types';
+
+export function execRegex(text: string, elements: RegexElement[]) {
+  const regex = buildRegex(...elements);
+  return [...regex.exec(text)!];
+}
+
+export function execRegexFull(text: string, elements: RegexElement[]) {
+  const regex = buildRegex(...elements);
+  return regex.exec(text)!;
+}

src/types.ts

@@ -1,4 +1,9 @@
-export type RegexElement = string | ChoiceOf | CharacterClass | Quantifier;
+export type RegexElement =
+  | string
+  | CharacterClass
+  | ChoiceOf
+  | Quantifier
+  | Capture;
 
 export type Quantifier = One | OneOrMore | Optionally | ZeroOrMore | Repeat;
 
@@ -41,3 +46,9 @@ export type Repeat = {
 };
 
 export type RepeatConfig = { count: number } | { min: number; max?: number };
+
+// Captures
+export type Capture = {
+  type: 'capture';
+  children: RegexElement[];
+};