refactor: simplify encoders (#88)

Author: mdjastrzebski

src/builders.ts

@@ -1,6 +1,5 @@
 import type { RegexFlags, RegexSequence } from './types';
-import { encodeSequence } from './encoder/encoder';
-import { ensureArray } from './utils/elements';
+import { encode } from './encoder';
 
 /**
  * Generate RegExp object from elements with optional flags.
@@ -10,7 +9,7 @@ import { ensureArray } from './utils/elements';
  * @returns RegExp object
  */
 export function buildRegExp(sequence: RegexSequence, flags?: RegexFlags): RegExp {
-  const pattern = encodeSequence(ensureArray(sequence)).pattern;
+  const pattern = encode(sequence).pattern;
   const flagsString = encodeFlags(flags ?? {});
   return new RegExp(pattern, flagsString);
 }
@@ -21,7 +20,7 @@ export function buildRegExp(sequence: RegexSequence, flags?: RegexFlags): RegExp
  * @returns regex pattern string
  */
 export function buildPattern(sequence: RegexSequence): string {
-  return encodeSequence(ensureArray(sequence)).pattern;
+  return encode(sequence).pattern;
 }
 
 function encodeFlags(flags: RegexFlags): string {

src/constructs/__tests__/char-class.test.ts

@@ -37,6 +37,12 @@ test('`charClass` joins character escapes', () => {
   expect(charClass(word, nonDigit)).toEqualRegex(/[\w\D]/);
 });
 
+test('`charClass` throws on empty text', () => {
+  expect(() => charClass()).toThrowErrorMatchingInlineSnapshot(
+    `"\`charClass\` should receive at least one element"`,
+  );
+});
+
 test('`charRange` pattern', () => {
   expect(charRange('a', 'z')).toEqualRegex(/[a-z]/);
   expect(['x', charRange('0', '9')]).toEqualRegex(/x[0-9]/);
@@ -115,16 +121,7 @@ test('`negated` character class matching', () => {
 });
 
 test('`encodeCharacterClass` throws on empty text', () => {
-  expect(() =>
-    buildRegExp(
-      // @ts-expect-error
-      negated({
-        type: 'characterClass',
-        chars: [],
-        ranges: [],
-      }),
-    ),
-  ).toThrowErrorMatchingInlineSnapshot(
+  expect(() => buildRegExp(negated({ chars: [], ranges: [] }))).toThrowErrorMatchingInlineSnapshot(
     `"Character class should contain at least one character or character range"`,
   );
 });

src/encoder/__tests__/encoder.test.tsx renamed to src/constructs/__tests__/encoder.test.tsx

@@ -74,7 +74,11 @@ test('`buildRegExp` throws error on unknown element', () => {
   expect(() =>
     // @ts-expect-error intentionally passing incorrect object
     buildRegExp({ type: 'unknown' }),
-  ).toThrowErrorMatchingInlineSnapshot(`"\`encodeNode\`: unknown element type unknown"`);
+  ).toThrowErrorMatchingInlineSnapshot(`
+    "\`encodeElement\`: unknown element: {
+      "type": "unknown"
+    }"
+  `);
 });
 
 test('`buildPattern` throws on empty text', () => {

src/constructs/anchors.ts

@@ -1,43 +1,26 @@
-import type { EncodeResult } from '../encoder/types';
-import type { RegexConstruct } from '../types';
+import type { EncodedRegex } from '../types';
 
-export interface Anchor extends RegexConstruct {
-  type: 'anchor';
-  symbol: string;
-}
-
-export const startOfString: Anchor = {
-  type: 'anchor',
-  symbol: '^',
-  encode: encodeAnchor,
+export const startOfString: EncodedRegex = {
+  precedence: 'atom',
+  pattern: '^',
 };
 
-export const endOfString: Anchor = {
-  type: 'anchor',
-  symbol: '$',
-  encode: encodeAnchor,
+export const endOfString: EncodedRegex = {
+  precedence: 'atom',
+  pattern: '$',
 };
 
-export const wordBoundary: Anchor = {
-  type: 'anchor',
-  symbol: '\\b',
-  encode: encodeAnchor,
+export const wordBoundary: EncodedRegex = {
+  precedence: 'atom',
+  pattern: '\\b',
 };
 
-export const nonWordBoundary: Anchor = {
-  type: 'anchor',
-  symbol: '\\B',
-  encode: encodeAnchor,
+export const nonWordBoundary: EncodedRegex = {
+  precedence: 'atom',
+  pattern: '\\B',
 };
 
 /**
  * @deprecated Renamed to `nonWordBoundary`.
  */
 export const notWordBoundary = nonWordBoundary;
-
-function encodeAnchor(this: Anchor): EncodeResult {
-  return {
-    precedence: 'sequence',
-    pattern: this.symbol,
-  };
-}

src/constructs/capture.ts

@@ -1,13 +1,5 @@
-import { encodeSequence } from '../encoder/encoder';
-import type { EncodeResult } from '../encoder/types';
-import { ensureArray } from '../utils/elements';
-import type { RegexConstruct, RegexElement, RegexSequence } from '../types';
-
-export interface Capture extends RegexConstruct {
-  type: 'capture';
-  children: RegexElement[];
-  options?: CaptureOptions;
-}
+import { encode } from '../encoder';
+import type { EncodedRegex, RegexSequence } from '../types';
 
 export type CaptureOptions = {
   /**
@@ -16,8 +8,7 @@ export type CaptureOptions = {
   name?: string;
 };
 
-export interface Reference extends RegexConstruct {
-  type: 'reference';
+export interface Reference extends EncodedRegex {
   name: string;
 }
 
@@ -26,12 +17,18 @@ export interface Reference extends RegexConstruct {
  * - in the match results (`String.match`, `String.matchAll`, or `RegExp.exec`)
  * - in the regex itself, through {@link ref}
  */
-export function capture(sequence: RegexSequence, options?: CaptureOptions): Capture {
+export function capture(sequence: RegexSequence, options?: CaptureOptions): EncodedRegex {
+  const name = options?.name;
+  if (name) {
+    return {
+      precedence: 'atom',
+      pattern: `(?<${name}>${encode(sequence).pattern})`,
+    };
+  }
+
   return {
-    type: 'capture',
-    children: ensureArray(sequence),
-    options,
-    encode: encodeCapture,
+    precedence: 'atom',
+    pattern: `(${encode(sequence).pattern})`,
   };
 }
 
@@ -45,31 +42,9 @@ export function capture(sequence: RegexSequence, options?: CaptureOptions): Capt
  * @param name - Name of the capturing group to reference.
  */
 export function ref(name: string): Reference {
-  return {
-    type: 'reference',
-    name,
-    encode: encodeReference,
-  };
-}
-
-function encodeCapture(this: Capture): EncodeResult {
-  const name = this.options?.name;
-  if (name) {
-    return {
-      precedence: 'atom',
-      pattern: `(?<${name}>${encodeSequence(this.children).pattern})`,
-    };
-  }
-
   return {
     precedence: 'atom',
-    pattern: `(${encodeSequence(this.children).pattern})`,
-  };
-}
-
-function encodeReference(this: Reference): EncodeResult {
-  return {
-    precedence: 'atom',
-    pattern: `\\k<${this.name}>`,
+    pattern: `\\k<${name}>`,
+    name,
   };
 }

src/constructs/char-class.ts

@@ -1,27 +1,14 @@
-import type { EncodeResult } from '../encoder/types';
-import type { RegexConstruct } from '../types';
-import type { CharacterEscape } from './char-escape';
-
-/**
- * Character range from start to end (inclusive).
- */
-export interface CharacterRange {
-  start: string;
-  end: string;
-}
-
-export interface CharacterClass extends RegexConstruct {
-  type: 'characterClass';
-  chars: string[];
-  ranges?: CharacterRange[];
-}
+import { encodeCharClass } from '../encoder';
+import type { CharacterClass, CharacterEscape, EncodedRegex } from '../types';
 
 export function charClass(...elements: Array<CharacterClass | CharacterEscape>): CharacterClass {
+  if (!elements.length) {
+    throw new Error('`charClass` should receive at least one element');
+  }
+
   return {
-    type: 'characterClass',
     chars: elements.map((c) => c.chars).flat(),
     ranges: elements.map((c) => c.ranges ?? []).flat(),
-    encode: encodeCharacterClass,
   };
 }
 
@@ -39,10 +26,8 @@ export function charRange(start: string, end: string): CharacterClass {
   }
 
   return {
-    type: 'characterClass',
     chars: [],
     ranges: [{ start, end }],
-    encode: encodeCharacterClass,
   };
 }
 
@@ -54,47 +39,19 @@ export function anyOf(characters: string): CharacterClass {
   }
 
   return {
-    type: 'characterClass',
     chars,
-    encode: encodeCharacterClass,
   };
 }
 
-export function negated(element: CharacterClass | CharacterEscape): EncodeResult {
-  return encodeCharacterClass.call(element, true);
+export function negated(element: CharacterClass | CharacterEscape): EncodedRegex {
+  return encodeCharClass(element, true);
 }
 
 /**
  * @deprecated Renamed to `negated`.
  */
 export const inverted = negated;
 
-export function encodeCharacterClass(
-  this: CharacterClass | CharacterEscape,
-  isNegated?: boolean,
-): EncodeResult {
-  if (!this.chars.length && !this.ranges?.length) {
-    throw new Error('Character class should contain at least one character or character range');
-  }
-
-  // If passed characters includes hyphen (`-`) it need to be moved to
-  // first (or last) place in order to treat it as hyphen character and not a range.
-  // See: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Guide/Regular_expressions/Character_classes#types
-  const hyphen = this.chars.includes('-') ? '-' : '';
-  const caret = this.chars.includes('^') ? '^' : '';
-  const otherChars = this.chars.filter((c) => c !== '-' && c !== '^').join('');
-  const ranges = this.ranges?.map(({ start, end }) => `${start}-${end}`).join('') ?? '';
-  const negation = isNegated ? '^' : '';
-
-  let pattern = `[${negation}${ranges}${otherChars}${caret}${hyphen}]`;
-  if (pattern === '[^-]') pattern = '[\\^-]';
-
-  return {
-    precedence: 'atom',
-    pattern,
-  };
-}
-
 function escapeForCharacterClass(text: string): string {
   return text.replace(/[\]\\]/g, '\\$&'); // $& means the whole matched string
 }

src/constructs/char-escape.ts

@@ -1,18 +1,10 @@
-import type { EncodeResult } from '../encoder/types';
-
-export interface CharacterEscape extends EncodeResult {
-  kind: 'escape';
-
-  // `CharacterClass` compatibility
-  chars: string[];
-  ranges?: never;
-}
+import type { CharacterEscape, EncodedRegex } from '../types';
 
 /**
  * Matches any single character.
  * Specifically this one is NOT a character escape.
  */
-export const any: EncodeResult = {
+export const any: EncodedRegex = {
   precedence: 'atom',
   pattern: '.',
 };
@@ -21,42 +13,36 @@ export const digit: CharacterEscape = {
   precedence: 'atom',
   pattern: '\\d',
   chars: ['\\d'],
-  kind: 'escape',
 };
 
 export const nonDigit: CharacterEscape = {
   precedence: 'atom',
   pattern: '\\D',
   chars: ['\\D'],
-  kind: 'escape',
 };
 
 export const word: CharacterEscape = {
   precedence: 'atom',
   pattern: '\\w',
   chars: ['\\w'],
-  kind: 'escape',
 };
 
 export const nonWord: CharacterEscape = {
   precedence: 'atom',
   pattern: '\\W',
   chars: ['\\W'],
-  kind: 'escape',
 };
 
 export const whitespace: CharacterEscape = {
   precedence: 'atom',
   pattern: '\\s',
   chars: ['\\s'],
-  kind: 'escape',
 };
 
 export const nonWhitespace: CharacterEscape = {
   precedence: 'atom',
   pattern: '\\S',
   chars: ['\\S'],
-  kind: 'escape',
 };
 
 /**