More UnicodeSet and CharMap method (#76)

Author: RunDevelopment

src/char-map.ts

@@ -3,7 +3,27 @@ import { CharRange, CharSet } from "./char-set";
 import { filterMut } from "./util";
 
 export interface ReadonlyCharMap<T> extends Iterable<[CharRange, T]> {
+	/**
+	 * Returns whether this map is empty.
+	 *
+	 * This is equivalent to `this.size === 0` and `this.entryCount === 0`.
+	 */
 	readonly isEmpty: boolean;
+	/**
+	 * The number of characters in this map. This is different from {@link entryCount}.
+	 *
+	 * This is equivalent to `[...this.keys()].reduce((count, range) => count + range.max - range.min + 1, 0)`.
+	 */
+	readonly size: number;
+	/**
+	 * The number of entires in this map.
+	 *
+	 * This is different from {@link size}. In general, you should use {@link size}, because it has the same semantics
+	 * as `Set#size` and `Map#size`.
+	 *
+	 * This is equivalent to `[...this.entries()].length`.
+	 */
+	readonly entryCount: number;
 
 	/**
 	 * Returns whether the given character is a key in the map.
@@ -44,14 +64,37 @@ export interface ReadonlyCharMap<T> extends Iterable<[CharRange, T]> {
 	 * @param callback
 	 */
 	forEach(callback: (value: T, chars: CharRange, map: ReadonlyCharMap<T>) => void): void;
+	/**
+	 * Returns all ranges of characters that are keys in the map.
+	 *
+	 * Keys will be returned in the same order as `this.entries()`.
+	 */
 	keys(): Iterable<CharRange>;
+	/**
+	 * Returns all values in the map. Values might not be unique if more than one range maps to the same value.
+	 *
+	 * Values will be returned in the same order as `this.entries()`.
+	 */
 	values(): Iterable<T>;
-	entries(range?: CharRange): Iterable<[CharRange, T]>;
+	/**
+	 * Returns all key-value pairs in the map.
+	 *
+	 * Entries will be returned in the order of ascending ranges.
+	 */
+	entries(): Iterable<[CharRange, T]>;
 
 	/**
 	 * Returns a mapping from the values of this map to its keys.
 	 */
 	invert(maxCharacter: Char): Map<T, CharSet>;
+
+	/**
+	 * Returns a new map with all values mapped by the given function.
+	 *
+	 * If no function is given, the identity function is used.
+	 */
+	copy(): CharMap<T>;
+	copy<U>(mapFn: (value: T) => U): CharMap<U>;
 }
 
 interface Item<T> {
@@ -76,6 +119,16 @@ export class CharMap<T> implements ReadonlyCharMap<T> {
 	get isEmpty(): boolean {
 		return this._array.length === 0;
 	}
+	get size(): number {
+		let count = 0;
+		for (const { range } of this._array) {
+			count += range.max - range.min + 1;
+		}
+		return count;
+	}
+	get entryCount(): number {
+		return this._array.length;
+	}
 
 	private _indexOf(char: Char): number | undefined {
 		let l = 0;
@@ -402,6 +455,34 @@ export class CharMap<T> implements ReadonlyCharMap<T> {
 		this._array = [];
 	}
 
+	copy(): CharMap<T>;
+	copy<U>(mapFn: (value: T) => U): CharMap<U>;
+	copy<U>(mapFn?: (value: T) => U): CharMap<U> {
+		if (!mapFn) {
+			const map = new CharMap<T>();
+			map._array = this._array.map(item => {
+				return { range: item.range, value: item.value };
+			});
+			return map as unknown as CharMap<U>;
+		}
+
+		const map = new CharMap<U>();
+		map._array = this._array.map(item => {
+			return { range: item.range, value: mapFn(item.value) };
+		});
+
+		// merge adjacent
+		filterMut(map._array, (item, prev) => {
+			if (prev && prev.range.max + 1 === item.range.min && prev.value === item.value) {
+				prev.range = { min: prev.range.min, max: item.range.max };
+				return false;
+			}
+			return true;
+		});
+
+		return map;
+	}
+
 	map(mapFn: (value: T, chars: CharRange, map: ReadonlyCharMap<T>) => T): void {
 		for (const item of this._array) {
 			item.value = mapFn(item.value, item.range, this);

src/js/char-case-folding.ts

@@ -5,7 +5,23 @@ import { Maximum } from "./maximum";
 import { UnicodeCaseFolding } from "./unicode";
 import { UTF16CaseFolding } from "./utf16-case-folding";
 
+/**
+ * A set of functions that can be used to perform case-insensitive matching.
+ *
+ * It must fulfill the following conditions:
+ *
+ * 1. `canonicalize` must be idempotent, i.e. `canonicalize(canonicalize(char)) === canonicalize(char)`.
+ * 2. `toCharSet(canonicalize(a))` is the set of all characters `c` such that `canonicalize(a) === canonicalize(c)`.
+ */
 export interface CharCaseFolding {
+	/**
+	 * The canonicalization function. This typically maps characters to their lowercase form.
+	 *
+	 * If no function is given, then the identity function is used. This also implies that `toCharSet` must return a
+	 * set containing only the given character.
+	 *
+	 * @default char => char
+	 */
 	readonly canonicalize?: (char: Char) => Char;
 	readonly toCharSet: (char: Char) => CharSet;
 }

src/js/parser.ts

@@ -730,27 +730,8 @@ export class Parser {
 			return context.nc.newCharClass(element, chars.chars);
 		}
 
-		// ECMAScript spec says that alternatives are sorted by descending length.
-		// This isn't enough for uniqueness though, so we also sort by code point.
-		const words = [...chars.accept.wordSets];
-		if (!chars.chars.isEmpty) {
-			words.push([chars.chars]);
-		}
-		words.sort((a, b) => {
-			if (a.length !== b.length) {
-				return b.length - a.length;
-			}
-			for (let i = 0; i < a.length; i++) {
-				const diff = a[i].compare(b[i]);
-				if (diff !== 0) {
-					return diff;
-				}
-			}
-			return 0;
-		});
-
 		const alternation = context.nc.newAlt(element);
-		for (const word of words) {
+		for (const word of chars.wordSets) {
 			const alternative = context.nc.newConcat(element);
 			alternation.alternatives.push(alternative);