+import { bind, call, identity, makeCallable } from "./function.js";
+import {
+ defineOwnProperties,
+ getOwnPropertyDescriptors,
+ getPrototype,
+ objectCreate,
+ setPrototype,
+} from "./object.js";
+import { type } from "./value.js";
+
+export const {
+ /**
+ * A RegExp·like object which only matches entire strings, and may
+ * have additional constraints specified.
+ *
+ * Matchers are callable objects and will return true if they are
+ * called with a string that they match, and false otherwise.
+ * Matchers will always return false if called with nonstrings,
+ * although other methods like `exec` coerce their arguments and may
+ * still return true.
+ */
+ Matcher,
+} = (() => {
+ const RE = RegExp;
+ const { prototype: rePrototype } = RE;
+ const { exec: reExec, toString: reToString } = rePrototype;
+ const getDotAll =
+ Object.getOwnPropertyDescriptor(rePrototype, "dotAll").get;
+ const getFlags =
+ Object.getOwnPropertyDescriptor(rePrototype, "flags").get;
+ const getGlobal =
+ Object.getOwnPropertyDescriptor(rePrototype, "global").get;
+ const getHasIndices =
+ Object.getOwnPropertyDescriptor(rePrototype, "hasIndices").get;
+ const getIgnoreCase =
+ Object.getOwnPropertyDescriptor(rePrototype, "ignoreCase").get;
+ const getMultiline =
+ Object.getOwnPropertyDescriptor(rePrototype, "multiline").get;
+ const getSource =
+ Object.getOwnPropertyDescriptor(rePrototype, "source").get;
+ const getSticky =
+ Object.getOwnPropertyDescriptor(rePrototype, "sticky").get;
+ const getUnicode =
+ Object.getOwnPropertyDescriptor(rePrototype, "unicode").get;
+
+ const Matcher = class extends identity {
+ #constraint;
+ #regExp;
+
+ /**
+ * Constructs a new Matcher from the provided source.
+ *
+ * If the provided source is a regular expression, then it must
+ * have the unicode flag set. Otherwise, it is interpreted as the
+ * string source of a regular expression with the unicode flag set.
+ *
+ * Other flags are taken from the provided regular expression
+ * object, if any are present.
+ *
+ * A name for the matcher may be provided as the second argument.
+ *
+ * A callable constraint on acceptable inputs may be provided as a
+ * third argument. If provided, it will be called with three
+ * arguments whenever a match appears successful: first, the string
+ * being matched, second, the match result, and third, the Matcher
+ * object itself. If the return value of this call is falsey, then
+ * the match will be considered a failure.
+ *
+ * ☡ If the provided source regular expression uses nongreedy
+ * quantifiers, it may not match the whole string even if a match
+ * with the whole string is possible. Surround the regular
+ * expression with `^(?:` and `)$` if you don’t want nongreedy
+ * regular expressions to fail when shorter matches are possible.
+ */
+ constructor(source, name = undefined, constraint = null) {
+ super(
+ ($) => {
+ if (typeof $ !== "string") {
+ // The provided value is not a string.
+ return false;
+ } else {
+ // The provided value is a string. Set the `lastIndex` of
+ // the regular expression to 0 and see if the first attempt
+ // at a match matches the whole string and passes the
+ // provided constraint (if present).
+ regExp.lastIndex = 0;
+ const result = call(reExec, regExp, [$]);
+ return result?.[0] === $ &&
+ (constraint === null || constraint($, result, this));
+ }
+ },
+ );
+ const regExp = this.#regExp = (() => {
+ try {
+ call(reExec, source, [""]); // throws if source not a RegExp
+ } catch {
+ return new RE(`${source}`, "u");
+ }
+ const unicode = call(getUnicode, source, []);
+ if (!unicode) {
+ // The provided regular expression does not have a unicode
+ // flag.
+ throw new TypeError(
+ `Piscēs: Cannot create Matcher from non‐Unicode RegExp: ${source}`,
+ );
+ } else {
+ // The provided regular expression has a unicode flag.
+ return new RE(source);
+ }
+ })();
+ if (constraint !== null && typeof constraint !== "function") {
+ throw new TypeError(
+ "Piscēs: Cannot construct Matcher: Constraint is not callable.",
+ );
+ } else {
+ this.#constraint = constraint;
+ return defineOwnProperties(
+ setPrototype(this, matcherPrototype),
+ {
+ lastIndex: {
+ configurable: false,
+ enumerable: false,
+ value: 0,
+ writable: false,
+ },
+ name: {
+ value: name != null
+ ? `${name}`
+ : `Matcher(${call(reToString, regExp, [])})`,
+ },
+ },
+ );
+ }
+ }
+
+ /** Gets whether the dotAll flag is present on this Matcher. */
+ get dotAll() {
+ return call(getDotAll, this.#regExp, []);
+ }
+
+ /**
+ * Executes this Matcher on the provided value and returns the
+ * result if there is a match, or null otherwise.
+ *
+ * Matchers only match if they can match the entire value on the
+ * first attempt.
+ *
+ * ☡ The match result returned by this method will be the same as
+ * that passed to the constraint function—and may have been
+ * modified by said function prior to being returned.
+ */
+ exec($) {
+ const regExp = this.#regExp;
+ const constraint = this.#constraint;
+ const string = `${$}`;
+ regExp.lastIndex = 0;
+ const result = call(reExec, regExp, [string]);
+ if (
+ result?.[0] === string &&
+ (constraint === null || constraint(string, result, this))
+ ) {
+ // The entire string was matched and the constraint, if
+ // present, returned a truthy value.
+ return result;
+ } else {
+ // The entire string was not matched or the constraint returned
+ // a falsey value.
+ return null;
+ }
+ }
+
+ /**
+ * Gets the flags present on this Matcher.
+ *
+ * ※ This needs to be defined because the internal RegExp object
+ * may have flags which are not yet recognized by ♓🌟 Piscēs.
+ */
+ get flags() {
+ return call(getFlags, this.#regExp, []);
+ }
+
+ /** Gets whether the global flag is present on this Matcher. */
+ get global() {
+ return call(getGlobal, this.#regExp, []);
+ }
+
+ /** Gets whether the hasIndices flag is present on this Matcher. */
+ get hasIndices() {
+ return call(getHasIndices, this.#regExp, []);
+ }
+
+ /** Gets whether the ignoreCase flag is present on this Matcher. */
+ get ignoreCase() {
+ return call(getIgnoreCase, this.#regExp, []);
+ }
+
+ /** Gets whether the multiline flag is present on this Matcher. */
+ get multiline() {
+ return call(getMultiline, this.#regExp, []);
+ }
+
+ /** Gets the regular expression source for this Matcher. */
+ get source() {
+ return call(getSource, this.#regExp, []);
+ }
+
+ /** Gets whether the sticky flag is present on this Matcher. */
+ get sticky() {
+ return call(getSticky, this.#regExp, []);
+ }
+
+ /**
+ * Gets whether the unicode flag is present on this Matcher.
+ *
+ * ※ This will always be true.
+ */
+ get unicode() {
+ return call(getUnicode, this.#regExp, []);
+ }
+ };
+
+ const matcherConstructor = defineOwnProperties(
+ class extends RegExp {
+ constructor(...args) {
+ return new Matcher(...args);
+ }
+ },
+ {
+ name: { value: "Matcher" },
+ length: { value: 1 },
+ },
+ );
+ const matcherPrototype = defineOwnProperties(
+ matcherConstructor.prototype,
+ getOwnPropertyDescriptors(Matcher.prototype),
+ { constructor: { value: matcherConstructor } },
+ );
+
+ return { Matcher: matcherConstructor };
+})();
+
+export const {
+ /**
+ * Returns the result of converting the provided value to A·S·C·I·I
+ * lowercase.
+ */
+ asciiLowercase,
+
+ /**
+ * Returns the result of converting the provided value to A·S·C·I·I
+ * uppercase.
+ */
+ asciiUppercase,
+} = (() => {
+ const {
+ toLowerCase: stringToLowercase,
+ toUpperCase: stringToUppercase,
+ } = String.prototype;
+ return {
+ asciiLowercase: ($) =>
+ stringReplaceAll(
+ `${$}`,
+ /[A-Z]/gu,
+ makeCallable(stringToLowercase),
+ ),
+ asciiUppercase: ($) =>
+ stringReplaceAll(
+ `${$}`,
+ /[a-z]/gu,
+ makeCallable(stringToUppercase),
+ ),
+ };
+})();
+
+export const {
+ /**
+ * Returns an iterator over the code units in the string
+ * representation of the provided value.
+ */
+ codeUnits,
+
+ /**
+ * Returns an iterator over the codepoints in the string
+ * representation of the provided value.
+ */
+ codepoints,
+
+ /**
+ * Returns an iterator over the scalar values in the string
+ * representation of the provided value.
+ *
+ * Codepoints which are not valid Unicode scalar values are replaced
+ * with U+FFFF.
+ */
+ scalarValues,
+
+ /**
+ * Returns the result of converting the provided value to a string of
+ * scalar values by replacing (unpaired) surrogate values with
+ * U+FFFD.
+ */
+ scalarValueString,
+} = (() => {
+ const {
+ iterator: iteratorSymbol,
+ toStringTag: toStringTagSymbol,
+ } = Symbol;
+ const { [iteratorSymbol]: arrayIterator } = Array.prototype;
+ const arrayIteratorPrototype = Object.getPrototypeOf(
+ [][iteratorSymbol](),
+ );
+ const { next: arrayIteratorNext } = arrayIteratorPrototype;
+ const iteratorPrototype = Object.getPrototypeOf(
+ arrayIteratorPrototype,
+ );
+ const { [iteratorSymbol]: stringIterator } = String.prototype;
+ const stringIteratorPrototype = Object.getPrototypeOf(
+ ""[iteratorSymbol](),
+ );
+ const { next: stringIteratorNext } = stringIteratorPrototype;
+
+ /**
+ * An iterator object for iterating over code values (either code
+ * units or codepoints) in a string.
+ *
+ * ※ This class is not exposed, although its methods are (through
+ * the prototypes of string code value iterator objects).
+ */
+ const StringCodeValueIterator = class extends identity {
+ #allowSurrogates;
+ #baseIterator;
+
+ /**
+ * Constructs a new string code value iterator from the provided
+ * base iterator.
+ *
+ * If the provided base iterator is an array iterator, this is a
+ * code unit iterator. If the provided iterator is a string
+ * iterator and surrogates are allowed, this is a codepoint
+ * iterator. If the provided iterator is a string iterator and
+ * surrogates are not allowed, this is a scalar value iterator.
+ */
+ constructor(baseIterator, allowSurrogates = true) {
+ super(objectCreate(stringCodeValueIteratorPrototype));
+ this.#allowSurrogates = !!allowSurrogates;
+ this.#baseIterator = baseIterator;
+ }
+
+ /** Provides the next code value in the iterator. */
+ next() {
+ const baseIterator = this.#baseIterator;
+ switch (getPrototype(baseIterator)) {
+ case arrayIteratorPrototype: {
+ // The base iterator is iterating over U·C·S characters.
+ const {
+ value: ucsCharacter,
+ done,
+ } = call(arrayIteratorNext, baseIterator, []);
+ return done
+ ? { value: undefined, done: true }
+ : { value: getCodeUnit(ucsCharacter, 0), done: false };
+ }
+ case stringIteratorPrototype: {
+ // The base iterator is iterating over Unicode characters.
+ const {
+ value: character,
+ done,
+ } = call(stringIteratorNext, baseIterator, []);
+ if (done) {
+ // The base iterator has been exhausted.
+ return { value: undefined, done: true };
+ } else {
+ // The base iterator provided a character; yield the
+ // codepoint.
+ const codepoint = getCodepoint(character, 0);
+ return {
+ value: this.#allowSurrogates || codepoint <= 0xD7FF ||
+ codepoint >= 0xE000
+ ? codepoint
+ : 0xFFFD,
+ done: false,
+ };
+ }
+ }
+ default: {
+ // Should not be possible!
+ throw new TypeError(
+ "Piscēs: Unrecognized base iterator type in %StringCodeValueIterator%.",
+ );
+ }
+ }
+ }
+ };
+
+ const {
+ next: stringCodeValueIteratorNext,
+ } = StringCodeValueIterator.prototype;
+ const stringCodeValueIteratorPrototype = objectCreate(
+ iteratorPrototype,
+ {
+ next: {
+ configurable: true,
+ enumerable: false,
+ value: stringCodeValueIteratorNext,
+ writable: true,
+ },
+ [toStringTagSymbol]: {
+ configurable: true,
+ enumerable: false,
+ value: "String Code Value Iterator",
+ writable: false,
+ },
+ },
+ );
+ const scalarValueIterablePrototype = {
+ [iteratorSymbol]() {
+ return {
+ next: bind(
+ stringCodeValueIteratorNext,
+ new StringCodeValueIterator(
+ call(stringIterator, this.source, []),
+ false,
+ ),
+ [],
+ ),
+ };
+ },
+ };
+
+ return {
+ codeUnits: ($) =>
+ new StringCodeValueIterator(call(arrayIterator, `${$}`, [])),
+ codepoints: ($) =>
+ new StringCodeValueIterator(
+ call(stringIterator, `${$}`, []),
+ true,
+ ),
+ scalarValues: ($) =>
+ new StringCodeValueIterator(
+ call(stringIterator, `${$}`, []),
+ false,
+ ),
+ scalarValueString: ($) =>
+ stringFromCodepoints(...objectCreate(
+ scalarValueIterablePrototype,
+ { source: { value: `${$}` } },
+ )),
+ };
+})();
+
Lady’s Gitweb / Pisces / blobdiff