Reuse Char constants in meta transformer

Author: wimpyprogrammer

src/constants/index.ts

@@ -3,9 +3,12 @@ const whitespace = ` \t\r\n${nbsp}`.split('');
 
 const digits = '0123456789'.split('');
 
+const underscore = '_';
 const basicLowercase = 'abcdefghijklmnopqrstuvwxyz'.split('');
 const basicUppercase = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ'.split('');
-const basicSpecial = '~`!@#$%^&*()-_=+<,>.?/[]{}|\\:;"\''.split('');
+const basicAlpha = [...basicLowercase, ...basicUppercase, underscore];
+
+const basicSpecial = '~`!@#$%^&*()-=+<,>.?/[]{}|\\:;"\''.split('');
 
 const extendedLowercase = 'àáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ'.split('');
 const extendedUppercase = 'ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞß'.split('');
@@ -16,16 +19,26 @@ const extendedSpecial = `¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼
 // https://www.ascii-code.com/#extendedASCIIDescription
 const windows1252Special = '€‚ƒ„…†‡ˆ‰Š‹ŒŽ‘’“”•–—˜™š›œžŸ'.split('');
 
-const all = ([] as string[]).concat(
-	whitespace,
-	digits,
-	basicLowercase,
-	basicUppercase,
-	basicSpecial,
-	extendedLowercase,
-	extendedUppercase,
-	extendedSpecial,
-	windows1252Special
-);
+const extended = [
+	...extendedLowercase,
+	...extendedUppercase,
+	...extendedSpecial,
+	...windows1252Special,
+];
 
-export const Chars = { all };
+const all = [
+	...whitespace,
+	...digits,
+	...basicAlpha,
+	...basicSpecial,
+	...extended,
+];
+
+export const Chars = {
+	all,
+	basicAlpha,
+	basicSpecial,
+	digits,
+	extended,
+	whitespace,
+};

src/transforms/meta-to-char-class-transform.ts

@@ -7,13 +7,9 @@ import {
 	CharacterClass,
 	SpecialChar,
 } from 'regexp-tree/ast';
+import { Chars } from '../constants';
 import * as Guards from '../types/regexp-tree-guards';
-import {
-	createClassRange,
-	createEscapedSimpleChar,
-	createSimpleChar,
-	createSimpleChars,
-} from './utils';
+import { createEscapedSimpleChar, createSimpleChar } from './utils';
 
 type Replace<ParentType extends AstClass> = (
 	parentNode: AsExpression<ParentType>,
@@ -59,73 +55,63 @@ const replacer: NodeReplacer = {
 	},
 };
 
-const optionsAlpha = [createClassRange('a', 'z'), createClassRange('A', 'Z')];
-const optionsDigit = createClassRange('0', '9');
-const optionUnderscore = createEscapedSimpleChar('_');
-const optionsWhitespaceNoBreak = [
-	...createSimpleChars(' \t'),
-	createSimpleChar(String.fromCharCode(160)), // &nbsp;
-];
-const optionsWhitespace = [
-	...optionsWhitespaceNoBreak,
-	...createSimpleChars('\r\n'),
-];
+const optionsAlpha = Chars.basicAlpha.map(createSimpleChar);
+const optionsDigit = Chars.digits.map(createSimpleChar);
+
+const optionsWhitespace = Chars.whitespace.map(createSimpleChar);
+
+const needEscape = [']', '-', '\\'];
+const noEscape = Chars.basicSpecial.filter(c => !needEscape.includes(c));
 const optionsOther = [
-	...createSimpleChars('~`!@#$%^&*()=+<,>.?/[{}|:;"\''),
-	createEscapedSimpleChar(']'),
-	createEscapedSimpleChar('-'),
-	createEscapedSimpleChar('\\'),
-];
-const optionsNewLine = createSimpleChar('\n');
-const optionsExtendedAscii = [
-	...createSimpleChars('àáâãäåæçèéêëìíîïðñòóôõöøùúûüýþÿ'),
-	...createSimpleChars('ÀÁÂÃÄÅÆÇÈÉÊËÌÍÎÏÐÑÒÓÔÕÖØÙÚÛÜÝÞß'),
-	...createSimpleChars('¡¢£¤¥¦§¨©ª«¬®¯°±²³´µ¶·¸¹º»¼½¾¿'),
-	...createSimpleChars('€‚ƒ„…†‡ˆ‰Š‹ŒŽ‘’“”•–—˜™š›œžŸ×÷'),
-	createSimpleChar(String.fromCharCode(173)), // &shy;
+	...noEscape.map(createSimpleChar),
+	...needEscape.map(createEscapedSimpleChar),
 ];
 
+const optionsExtended = Chars.extended.map(createSimpleChar);
+
 function getMetaCharExpressions(
 	metaChar: SpecialChar,
 	regExpFlags: string
 ): CharacterClass['expressions'] {
 	switch (metaChar.value) {
 		case '.': {
-			const dotAllOptions = regExpFlags.includes('s') ? [optionsNewLine] : [];
+			const optionsNewLine = createSimpleChar('\n');
+			const optionsDotAll = regExpFlags.includes('s') ? [optionsNewLine] : [];
+			const whitespaceNoBreaks = Chars.whitespace.filter(
+				c => !'\r\n'.includes(c)
+			);
+			const optionsWhitespaceNoBreak = whitespaceNoBreaks.map(createSimpleChar);
 
 			return [
 				...optionsAlpha,
-				optionsDigit,
+				...optionsDigit,
 				...optionsWhitespaceNoBreak,
 				...optionsOther,
-				optionUnderscore,
-				...optionsExtendedAscii,
-				...dotAllOptions,
+				...optionsExtended,
+				...optionsDotAll,
 			];
 		}
 		case '\\w':
-			return [...optionsAlpha, optionsDigit, optionUnderscore];
+			return [...optionsAlpha, ...optionsDigit];
 		case '\\W':
-			return [...optionsWhitespace, ...optionsOther, ...optionsExtendedAscii];
+			return [...optionsWhitespace, ...optionsOther, ...optionsExtended];
 		case '\\d':
-			return [optionsDigit];
+			return optionsDigit;
 		case '\\D':
 			return [
 				...optionsAlpha,
 				...optionsWhitespace,
 				...optionsOther,
-				optionUnderscore,
-				...optionsExtendedAscii,
+				...optionsExtended,
 			];
 		case '\\s':
 			return optionsWhitespace;
 		case '\\S':
 			return [
 				...optionsAlpha,
-				optionsDigit,
+				...optionsDigit,
 				...optionsOther,
-				optionUnderscore,
-				...optionsExtendedAscii,
+				...optionsExtended,
 			];
 		default:
 			return [];

src/transforms/utils.ts

@@ -1,4 +1,4 @@
-import { ClassRange, SimpleChar } from 'regexp-tree/ast';
+import { SimpleChar } from 'regexp-tree/ast';
 
 export function createSimpleChar(value: string): SimpleChar {
 	if (value.length !== 1) throw new Error('value must be a char');
@@ -18,15 +18,3 @@ export function createEscapedSimpleChar(value: string): SimpleChar {
 		escaped: true,
 	};
 }
-
-export function createSimpleChars(values: string): SimpleChar[] {
-	return values.split('').map(createSimpleChar);
-}
-
-export function createClassRange(from: string, to: string): ClassRange {
-	return {
-		from: createSimpleChar(from),
-		to: createSimpleChar(to),
-		type: 'ClassRange',
-	};
-}