opusdev/vector-similarity-api

1'use strict';2 3/** Highest positive signed 32-bit float value */4const maxInt = 2147483647; // aka. 0x7FFFFFFF or 2^31-15 6/** Bootstring parameters */7const base = 36;8const tMin = 1;9const tMax = 26;10const skew = 38;11const damp = 700;12const initialBias = 72;13const initialN = 128; // 0x8014const delimiter = '-'; // '\x2D'15 16/** Regular expressions */17const regexPunycode = /^xn--/;18const regexNonASCII = /[^\0-\x7F]/; // Note: U+007F DEL is excluded too.19const regexSeparators = /[\x2E\u3002\uFF0E\uFF61]/g; // RFC 3490 separators20 21/** Error messages */22const errors = {23	'overflow': 'Overflow: input needs wider integers to process',24	'not-basic': 'Illegal input >= 0x80 (not a basic code point)',25	'invalid-input': 'Invalid input'26};27 28/** Convenience shortcuts */29const baseMinusTMin = base - tMin;30const floor = Math.floor;31const stringFromCharCode = String.fromCharCode;32 33/*--------------------------------------------------------------------------*/34 35/**36 * A generic error utility function.37 * @private38 * @param {String} type The error type.39 * @returns {Error} Throws a `RangeError` with the applicable error message.40 */41function error(type) {42	throw new RangeError(errors[type]);43}44 45/**46 * A generic `Array#map` utility function.47 * @private48 * @param {Array} array The array to iterate over.49 * @param {Function} callback The function that gets called for every array50 * item.51 * @returns {Array} A new array of values returned by the callback function.52 */53function map(array, callback) {54	const result = [];55	let length = array.length;56	while (length--) {57		result[length] = callback(array[length]);58	}59	return result;60}61 62/**63 * A simple `Array#map`-like wrapper to work with domain name strings or email64 * addresses.65 * @private66 * @param {String} domain The domain name or email address.67 * @param {Function} callback The function that gets called for every68 * character.69 * @returns {String} A new string of characters returned by the callback70 * function.71 */72function mapDomain(domain, callback) {73	const parts = domain.split('@');74	let result = '';75	if (parts.length > 1) {76		// In email addresses, only the domain name should be punycoded. Leave77		// the local part (i.e. everything up to `@`) intact.78		result = parts[0] + '@';79		domain = parts[1];80	}81	// Avoid `split(regex)` for IE8 compatibility. See #17.82	domain = domain.replace(regexSeparators, '\x2E');83	const labels = domain.split('.');84	const encoded = map(labels, callback).join('.');85	return result + encoded;86}87 88/**89 * Creates an array containing the numeric code points of each Unicode90 * character in the string. While JavaScript uses UCS-2 internally,91 * this function will convert a pair of surrogate halves (each of which92 * UCS-2 exposes as separate characters) into a single code point,93 * matching UTF-16.94 * @see `punycode.ucs2.encode`95 * @see <https://mathiasbynens.be/notes/javascript-encoding>96 * @memberOf punycode.ucs297 * @name decode98 * @param {String} string The Unicode input string (UCS-2).99 * @returns {Array} The new array of code points.100 */101function ucs2decode(string) {102	const output = [];103	let counter = 0;104	const length = string.length;105	while (counter < length) {106		const value = string.charCodeAt(counter++);107		if (value >= 0xD800 && value <= 0xDBFF && counter < length) {108			// It's a high surrogate, and there is a next character.109			const extra = string.charCodeAt(counter++);110			if ((extra & 0xFC00) == 0xDC00) { // Low surrogate.111				output.push(((value & 0x3FF) << 10) + (extra & 0x3FF) + 0x10000);112			} else {113				// It's an unmatched surrogate; only append this code unit, in case the114				// next code unit is the high surrogate of a surrogate pair.115				output.push(value);116				counter--;117			}118		} else {119			output.push(value);120		}121	}122	return output;123}124 125/**126 * Creates a string based on an array of numeric code points.127 * @see `punycode.ucs2.decode`128 * @memberOf punycode.ucs2129 * @name encode130 * @param {Array} codePoints The array of numeric code points.131 * @returns {String} The new Unicode string (UCS-2).132 */133const ucs2encode = codePoints => String.fromCodePoint(...codePoints);134 135/**136 * Converts a basic code point into a digit/integer.137 * @see `digitToBasic()`138 * @private139 * @param {Number} codePoint The basic numeric code point value.140 * @returns {Number} The numeric value of a basic code point (for use in141 * representing integers) in the range `0` to `base - 1`, or `base` if142 * the code point does not represent a value.143 */144const basicToDigit = function(codePoint) {145	if (codePoint >= 0x30 && codePoint < 0x3A) {146		return 26 + (codePoint - 0x30);147	}148	if (codePoint >= 0x41 && codePoint < 0x5B) {149		return codePoint - 0x41;150	}151	if (codePoint >= 0x61 && codePoint < 0x7B) {152		return codePoint - 0x61;153	}154	return base;155};156 157/**158 * Converts a digit/integer into a basic code point.159 * @see `basicToDigit()`160 * @private161 * @param {Number} digit The numeric value of a basic code point.162 * @returns {Number} The basic code point whose value (when used for163 * representing integers) is `digit`, which needs to be in the range164 * `0` to `base - 1`. If `flag` is non-zero, the uppercase form is165 * used; else, the lowercase form is used. The behavior is undefined166 * if `flag` is non-zero and `digit` has no uppercase form.167 */168const digitToBasic = function(digit, flag) {169	//  0..25 map to ASCII a..z or A..Z170	// 26..35 map to ASCII 0..9171	return digit + 22 + 75 * (digit < 26) - ((flag != 0) << 5);172};173 174/**175 * Bias adaptation function as per section 3.4 of RFC 3492.176 * https://tools.ietf.org/html/rfc3492#section-3.4177 * @private178 */179const adapt = function(delta, numPoints, firstTime) {180	let k = 0;181	delta = firstTime ? floor(delta / damp) : delta >> 1;182	delta += floor(delta / numPoints);183	for (/* no initialization */; delta > baseMinusTMin * tMax >> 1; k += base) {184		delta = floor(delta / baseMinusTMin);185	}186	return floor(k + (baseMinusTMin + 1) * delta / (delta + skew));187};188 189/**190 * Converts a Punycode string of ASCII-only symbols to a string of Unicode191 * symbols.192 * @memberOf punycode193 * @param {String} input The Punycode string of ASCII-only symbols.194 * @returns {String} The resulting string of Unicode symbols.195 */196const decode = function(input) {197	// Don't use UCS-2.198	const output = [];199	const inputLength = input.length;200	let i = 0;201	let n = initialN;202	let bias = initialBias;203 204	// Handle the basic code points: let `basic` be the number of input code205	// points before the last delimiter, or `0` if there is none, then copy206	// the first basic code points to the output.207 208	let basic = input.lastIndexOf(delimiter);209	if (basic < 0) {210		basic = 0;211	}212 213	for (let j = 0; j < basic; ++j) {214		// if it's not a basic code point215		if (input.charCodeAt(j) >= 0x80) {216			error('not-basic');217		}218		output.push(input.charCodeAt(j));219	}220 221	// Main decoding loop: start just after the last delimiter if any basic code222	// points were copied; start at the beginning otherwise.223 224	for (let index = basic > 0 ? basic + 1 : 0; index < inputLength; /* no final expression */) {225 226		// `index` is the index of the next character to be consumed.227		// Decode a generalized variable-length integer into `delta`,228		// which gets added to `i`. The overflow checking is easier229		// if we increase `i` as we go, then subtract off its starting230		// value at the end to obtain `delta`.231		const oldi = i;232		for (let w = 1, k = base; /* no condition */; k += base) {233 234			if (index >= inputLength) {235				error('invalid-input');236			}237 238			const digit = basicToDigit(input.charCodeAt(index++));239 240			if (digit >= base) {241				error('invalid-input');242			}243			if (digit > floor((maxInt - i) / w)) {244				error('overflow');245			}246 247			i += digit * w;248			const t = k <= bias ? tMin : (k >= bias + tMax ? tMax : k - bias);249 250			if (digit < t) {251				break;252			}253 254			const baseMinusT = base - t;255			if (w > floor(maxInt / baseMinusT)) {256				error('overflow');257			}258 259			w *= baseMinusT;260 261		}262 263		const out = output.length + 1;264		bias = adapt(i - oldi, out, oldi == 0);265 266		// `i` was supposed to wrap around from `out` to `0`,267		// incrementing `n` each time, so we'll fix that now:268		if (floor(i / out) > maxInt - n) {269			error('overflow');270		}271 272		n += floor(i / out);273		i %= out;274 275		// Insert `n` at position `i` of the output.276		output.splice(i++, 0, n);277 278	}279 280	return String.fromCodePoint(...output);281};282 283/**284 * Converts a string of Unicode symbols (e.g. a domain name label) to a285 * Punycode string of ASCII-only symbols.286 * @memberOf punycode287 * @param {String} input The string of Unicode symbols.288 * @returns {String} The resulting Punycode string of ASCII-only symbols.289 */290const encode = function(input) {291	const output = [];292 293	// Convert the input in UCS-2 to an array of Unicode code points.294	input = ucs2decode(input);295 296	// Cache the length.297	const inputLength = input.length;298 299	// Initialize the state.300	let n = initialN;301	let delta = 0;302	let bias = initialBias;303 304	// Handle the basic code points.305	for (const currentValue of input) {306		if (currentValue < 0x80) {307			output.push(stringFromCharCode(currentValue));308		}309	}310 311	const basicLength = output.length;312	let handledCPCount = basicLength;313 314	// `handledCPCount` is the number of code points that have been handled;315	// `basicLength` is the number of basic code points.316 317	// Finish the basic string with a delimiter unless it's empty.318	if (basicLength) {319		output.push(delimiter);320	}321 322	// Main encoding loop:323	while (handledCPCount < inputLength) {324 325		// All non-basic code points < n have been handled already. Find the next326		// larger one:327		let m = maxInt;328		for (const currentValue of input) {329			if (currentValue >= n && currentValue < m) {330				m = currentValue;331			}332		}333 334		// Increase `delta` enough to advance the decoder's <n,i> state to <m,0>,335		// but guard against overflow.336		const handledCPCountPlusOne = handledCPCount + 1;337		if (m - n > floor((maxInt - delta) / handledCPCountPlusOne)) {338			error('overflow');339		}340 341		delta += (m - n) * handledCPCountPlusOne;342		n = m;343 344		for (const currentValue of input) {345			if (currentValue < n && ++delta > maxInt) {346				error('overflow');347			}348			if (currentValue === n) {349				// Represent delta as a generalized variable-length integer.350				let q = delta;351				for (let k = base; /* no condition */; k += base) {352					const t = k <= bias ? tMin : (k >= bias + tMax ? tMax : k - bias);353					if (q < t) {354						break;355					}356					const qMinusT = q - t;357					const baseMinusT = base - t;358					output.push(359						stringFromCharCode(digitToBasic(t + qMinusT % baseMinusT, 0))360					);361					q = floor(qMinusT / baseMinusT);362				}363 364				output.push(stringFromCharCode(digitToBasic(q, 0)));365				bias = adapt(delta, handledCPCountPlusOne, handledCPCount === basicLength);366				delta = 0;367				++handledCPCount;368			}369		}370 371		++delta;372		++n;373 374	}375	return output.join('');376};377 378/**379 * Converts a Punycode string representing a domain name or an email address380 * to Unicode. Only the Punycoded parts of the input will be converted, i.e.381 * it doesn't matter if you call it on a string that has already been382 * converted to Unicode.383 * @memberOf punycode384 * @param {String} input The Punycoded domain name or email address to385 * convert to Unicode.386 * @returns {String} The Unicode representation of the given Punycode387 * string.388 */389const toUnicode = function(input) {390	return mapDomain(input, function(string) {391		return regexPunycode.test(string)392			? decode(string.slice(4).toLowerCase())393			: string;394	});395};396 397/**398 * Converts a Unicode string representing a domain name or an email address to399 * Punycode. Only the non-ASCII parts of the domain name will be converted,400 * i.e. it doesn't matter if you call it with a domain that's already in401 * ASCII.402 * @memberOf punycode403 * @param {String} input The domain name or email address to convert, as a404 * Unicode string.405 * @returns {String} The Punycode representation of the given domain name or406 * email address.407 */408const toASCII = function(input) {409	return mapDomain(input, function(string) {410		return regexNonASCII.test(string)411			? 'xn--' + encode(string)412			: string;413	});414};415 416/*--------------------------------------------------------------------------*/417 418/** Define the public API */419const punycode = {420	/**421	 * A string representing the current Punycode.js version number.422	 * @memberOf punycode423	 * @type String424	 */425	'version': '2.3.1',426	/**427	 * An object of methods to convert from JavaScript's internal character428	 * representation (UCS-2) to Unicode code points, and back.429	 * @see <https://mathiasbynens.be/notes/javascript-encoding>430	 * @memberOf punycode431	 * @type Object432	 */433	'ucs2': {434		'decode': ucs2decode,435		'encode': ucs2encode436	},437	'decode': decode,438	'encode': encode,439	'toASCII': toASCII,440	'toUnicode': toUnicode441};442 443module.exports = punycode;444