gristlabs_grist-core/test/common/NumberParse.ts
Jarosław Sadziński a52d56f613 (core) Moving client and common tests to core
Summary:
- Moved /test/client and /test/common to core.
- Moved two files (CircularArray and RecentItems) from app/common to core/app/common.
- Moved resetOrg test to gen-server.
- `testrun.sh` is now invoking common and client test from core.
- Added missing packages to core's package.json (and revealed underscore as it is used in the main app).
- Removed Coord.js as it is not used anywhere.

Test Plan: Existing tests

Reviewers: paulfitz

Reviewed By: paulfitz

Subscribers: paulfitz

Differential Revision: https://phab.getgrist.com/D3590
2022-08-23 19:20:10 +02:00

393 lines
15 KiB
TypeScript
Raw Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

import {getCurrency, locales} from 'app/common/Locales';
import {NumMode, parseNumMode} from 'app/common/NumberFormat';
import NumberParse from 'app/common/NumberParse';
import {assert} from 'chai';
import * as _ from 'lodash';
describe("NumberParse", function() {
let parser = new NumberParse("en", "USD");
function check(str: string, expected: number | null) {
const parsed = parser.parse(str);
assert.equal(parsed?.result ?? null, expected);
}
it("can do basic parsing", function() {
check("123", 123);
check("-123", -123);
check("-123.456", -123.456);
check("-1.234e56", -1.234e56);
check("1.234e-56", 1.234e-56);
check("(1.234e56)", -1.234e56);
check("($1.23)", -1.23);
check("($ 1.23)", -1.23);
check("$ 1.23", 1.23);
check("$1.23", 1.23);
check("12.34%", 0.1234);
check("1,234,567.89", 1234567.89);
check(".89", .89);
check(".89000", .89);
check("0089", 89);
// The digit separator is ',' but spaces are always removed anyway
check("1 234 567.89", 1234567.89);
assert.equal(parser.parse(""), null);
check(" ", null);
check("()", null);
check(" ( ) ", null);
check(" (,) ", null);
check(" (.) ", null);
check(",", null);
check(",.", null);
check(".,", null);
check(",,,", null);
check("...", null);
check(".", null);
check("%", null);
check("$", null);
check("(ABC)", null);
check("ABC", null);
check("USD", null);
check("NaN", null);
check("NAN", null);
check("nan", null);
// Currency symbol can only appear once
check("$$1.23", null);
// Other currency symbols not allowed
check("USD 1.23", null);
check("€ 1.23", null);
check("£ 1.23", null);
check("$ 1.23", 1.23);
// Parentheses represent negative numbers,
// so the number inside can't also be negative or 0
check("(0)", null);
check("(-1.23)", null);
check("(1.23)", -1.23);
check("-1.23", -1.23);
// Only one % allowed
check("12.34%%", null);
check("12.34%", 0.1234);
});
it("can handle different minus sign positions", function() {
parser = new NumberParse("fy", "EUR");
let formatter = Intl.NumberFormat("fy", {style: "currency", currency: "EUR"});
assert.isTrue(parser.currencyEndsInMinusSign);
// Note the '-' is at the end
assert.equal(formatter.format(-1), "€ 1,00-");
// The parser can handle this, it also allows the '-' in the beginning as usual
check("€ 1,00-", -1);
check("€ -1,00", -1);
check("-€ 1,00", -1);
// But it's only allowed at the end for currency amounts, to match the formatter
check("1,00-", null);
check("-1,00", -1);
// By contrast, this locale doesn't put '-' at the end so the parser doesn't allow that
parser = new NumberParse("en", "USD");
formatter = Intl.NumberFormat("en", {style: "currency", currency: "USD"});
assert.isFalse(parser.currencyEndsInMinusSign);
assert.equal(formatter.format(-1), "-$1.00");
check("-$1.00", -1);
check("$-1.00", -1);
check("$1.00-", null);
check("-1.00", -1);
check("1.00-", null);
});
it("can handle different separators", function() {
let formatter = Intl.NumberFormat("en", {useGrouping: true});
assert.equal(formatter.format(123456789.123), "123,456,789.123");
parser = new NumberParse("en", "USD");
assert.equal(parser.digitGroupSeparator, ",");
assert.equal(parser.digitGroupSeparatorCurrency, ",");
assert.equal(parser.decimalSeparator, ".");
check("123,456,789.123", 123456789.123);
// The typical separator is ',' but spaces are always removed anyway
check("123 456 789.123", 123456789.123);
// There must be at least two digits after the separator
check("123,456", 123456);
check("12,34,56", 123456);
check("1,2,3,4,5,6", null);
check("123,,456", null);
check("1,234", 1234);
check("123,4", null);
// This locale uses 'opposite' separators to the above, i.e. ',' and '.' have swapped roles
formatter = Intl.NumberFormat("de-AT", {useGrouping: true, currency: "EUR", style: "currency"});
assert.equal(formatter.format(123456789.123), '€ 123.456.789,12');
// But only for currency amounts! Non-currency amounts use NBSP (non-breaking space) for the digit separator
formatter = Intl.NumberFormat("de-AT", {useGrouping: true});
assert.equal(formatter.format(123456789.123), '123 456 789,123');
parser = new NumberParse("de-AT", "EUR");
assert.equal(parser.digitGroupSeparator, " ");
assert.equal(parser.digitGroupSeparatorCurrency, ".");
assert.equal(parser.decimalSeparator, ",");
check("€ 123.456.789,123", 123456789.123);
check("€ 123 456 789,123", 123456789.123);
// The parser allows the currency separator for non-currency amounts
check(" 123.456.789,123", 123456789.123);
check(" 123 456 789,123", 123456789.123); // normal space
check(" 123 456 789,123", 123456789.123); // NBSP
formatter = Intl.NumberFormat("en-ZA", {useGrouping: true});
assert.equal(formatter.format(123456789.123), '123 456 789,123');
parser = new NumberParse("en-ZA", "ZAR");
assert.equal(parser.digitGroupSeparator, " ");
assert.equal(parser.digitGroupSeparatorCurrency, " ");
assert.equal(parser.decimalSeparator, ",");
// ',' is the official decimal separator of this locale,
// but in general '.' will also work as long as it's not the digit separator.
check("123 456 789,123", 123456789.123);
check("123 456 789.123", 123456789.123);
});
it("returns basic info about formatting options for a single string", function() {
parser = new NumberParse("en", "USD");
assert.isNull(parser.parse(""));
assert.isNull(parser.parse("a b"));
const defaultOptions = {
isCurrency: false,
isParenthesised: false,
hasDigitGroupSeparator: false,
isScientific: false,
isPercent: false,
};
assert.deepEqual(parser.parse("1"),
{result: 1, cleaned: "1", options: defaultOptions});
assert.deepEqual(parser.parse("$1"),
{result: 1, cleaned: "1", options: {...defaultOptions, isCurrency: true}});
assert.deepEqual(parser.parse("100%"),
{result: 1, cleaned: "100", options: {...defaultOptions, isPercent: true}});
assert.deepEqual(parser.parse("1,000"),
{result: 1000, cleaned: "1000", options: {...defaultOptions, hasDigitGroupSeparator: true}});
assert.deepEqual(parser.parse("1E2"),
{result: 100, cleaned: "1e2", options: {...defaultOptions, isScientific: true}});
assert.deepEqual(parser.parse("$1,000"),
{result: 1000, cleaned: "1000", options: {...defaultOptions, isCurrency: true, hasDigitGroupSeparator: true}});
});
it("guesses formatting options", function() {
parser = new NumberParse("en", "USD");
assert.deepEqual(parser.guessOptions([]), {});
assert.deepEqual(parser.guessOptions([""]), {});
assert.deepEqual(parser.guessOptions([null]), {});
assert.deepEqual(parser.guessOptions(["", null]), {});
assert.deepEqual(parser.guessOptions(["abc"]), {});
assert.deepEqual(parser.guessOptions(["1"]), {});
assert.deepEqual(parser.guessOptions(["1", "", null, "abc"]), {});
assert.deepEqual(parser.guessOptions(["$1,000"]), {numMode: "currency", decimals: 0});
assert.deepEqual(parser.guessOptions(["1,000%"]), {numMode: "percent"});
assert.deepEqual(parser.guessOptions(["1,000"]), {numMode: "decimal"});
assert.deepEqual(parser.guessOptions(["1E2"]), {numMode: "scientific"});
// Choose the most common mode when there are several candidates
assert.deepEqual(parser.guessOptions(["$1", "$2", "3%"]), {numMode: "currency", decimals: 0});
assert.deepEqual(parser.guessOptions(["$1", "2%", "3%"]), {numMode: "percent"});
assert.deepEqual(parser.guessOptions(["(2)"]), {numSign: 'parens'});
assert.deepEqual(parser.guessOptions(["(2)", "3"]), {numSign: 'parens'});
// If we see a negative number not surrounded by parens, assume that other parens mean something else
assert.deepEqual(parser.guessOptions(["(2)", "-3"]), {});
assert.deepEqual(parser.guessOptions(["($2)"]), {numSign: 'parens', numMode: "currency", decimals: 0});
// Guess 'decimal' (i.e. with thousands separators) even if most numbers don't have separators
assert.deepEqual(parser.guessOptions(["1", "10", "100", "1,000"]), {numMode: "decimal"});
// For USD, currencies are formatted with minimum 2 decimal places by default,
// so if the data doesn't have that many decimals we have to explicitly specify the number of decimals, default 0.
// The number of digits for other currencies is defaultNumDecimalsCurrency, tested a bit further down.
assert.deepEqual(parser.guessOptions(["$1"]), {numMode: "currency", decimals: 0});
assert.deepEqual(parser.guessOptions(["$1.2"]), {numMode: "currency", decimals: 0});
assert.deepEqual(parser.guessOptions(["$1.23"]), {numMode: "currency"});
assert.deepEqual(parser.guessOptions(["$1.234"]), {numMode: "currency"});
// Otherwise decimal places are guessed based on trailing zeroes
assert.deepEqual(parser.guessOptions(["$1.0"]), {numMode: "currency", decimals: 1});
assert.deepEqual(parser.guessOptions(["$1.00"]), {numMode: "currency", decimals: 2});
assert.deepEqual(parser.guessOptions(["$1.000"]), {numMode: "currency", decimals: 3});
assert.deepEqual(parser.guessOptions(["1E2"]), {numMode: "scientific"});
assert.deepEqual(parser.guessOptions(["1.3E2"]), {numMode: "scientific"});
assert.deepEqual(parser.guessOptions(["1.34E2"]), {numMode: "scientific"});
assert.deepEqual(parser.guessOptions(["1.0E2"]), {numMode: "scientific", decimals: 1});
assert.deepEqual(parser.guessOptions(["1.30E2"]), {numMode: "scientific", decimals: 2});
assert.equal(parser.defaultNumDecimalsCurrency, 2);
parser = new NumberParse("en", "TND");
assert.equal(parser.defaultNumDecimalsCurrency, 3);
parser = new NumberParse("en", "ZMK");
assert.equal(parser.defaultNumDecimalsCurrency, 0);
});
// Nice mixture of numbers of different sizes and containing all digits
const numbers = [
..._.range(1, 12),
..._.range(3, 20).map(n => Math.pow(3, n)),
..._.range(10).map(n => Math.pow(10, -n) * 1234560798),
];
numbers.push(...numbers.map(n => -n));
numbers.push(...numbers.map(n => 1 / n));
numbers.push(0); // added at the end because of the division just before
// Formatter to compare numbers that only differ because of floating point precision errors
const basicFormatter = Intl.NumberFormat("en", {
maximumSignificantDigits: 15,
useGrouping: false,
});
// All values supported by parseNumMode
const numModes: Array<NumMode | undefined> = ['currency', 'decimal', 'percent', 'scientific', undefined];
// Generate a test suite for every supported locale
for (const locale of locales) {
describe(`with ${locale.code} locale (${locale.name})`, function() {
const currency = getCurrency(locale.code);
beforeEach(() => {
parser = new NumberParse(locale.code, currency);
});
it("has sensible parser attributes", function() {
// These don't strictly need to have length 1, but it's nice to know
assert.lengthOf(parser.percentageSymbol, 1);
assert.lengthOf(parser.minusSign, 1);
assert.lengthOf(parser.decimalSeparator, 1);
// These *do* need to be a single character since the regex uses `[]`.
assert.lengthOf(parser.digitGroupSeparator, 1);
// This is the only symbol that's allowed to be empty
assert.include([0, 1], parser.digitGroupSeparatorCurrency.length);
assert.isNotEmpty(parser.exponentSeparator);
assert.isNotEmpty(parser.currencySymbol);
const symbols = [
parser.percentageSymbol,
parser.minusSign,
parser.decimalSeparator,
parser.digitGroupSeparator,
parser.exponentSeparator,
parser.currencySymbol,
...parser.digitsMap.keys(),
];
// All the symbols must be distinct
assert.equal(symbols.length, new Set(symbols).size);
// The symbols mustn't contain characters that the parser removes (e.g. spaces)
// or they won't be replaced correctly.
// The digit group separators are OK because they're removed anyway, and often the separator is a space.
// Currency is OK because it gets removed before these characters.
for (const symbol of symbols) {
if (![
parser.digitGroupSeparator,
parser.digitGroupSeparatorCurrency,
parser.currencySymbol,
].includes(symbol)) {
assert.equal(symbol, symbol.replace(NumberParse.removeCharsRegex, "REMOVED"));
}
}
// Decimal and digit separators have to be different.
// We checked digitGroupSeparator already with the Set above,
// but not digitGroupSeparatorCurrency because it can equal digitGroupSeparator.
assert.notEqual(parser.decimalSeparator, parser.digitGroupSeparator);
assert.notEqual(parser.decimalSeparator, parser.digitGroupSeparatorCurrency);
for (const key of parser.digitsMap.keys()) {
assert.lengthOf(key, 1);
assert.lengthOf(parser.digitsMap.get(key)!, 1);
}
});
it("can parse formatted numbers", function() {
for (const numMode of numModes) {
const formatter = Intl.NumberFormat(locale.code, {
...parseNumMode(numMode, currency),
maximumFractionDigits: 15,
maximumSignificantDigits: 15,
});
for (const num of numbers) {
const fnum = formatter.format(num);
const formattedNumbers = [fnum];
if (num > 0 && fnum[0] === "0") {
// E.g. test that '.5' is parsed as '0.5'
formattedNumbers.push(fnum.substring(1));
}
if (num < 0) {
formattedNumbers.push(`(${formatter.format(-num)})`);
}
for (const formatted of formattedNumbers) {
const parsed = parser.parse(formatted)?.result;
// Fast check, particularly to avoid formatting the numbers
// Makes the tests about 1.5s/30% faster.
if (parsed === num) {
continue;
}
try {
assert.exists(parsed);
assert.equal(
basicFormatter.format(parsed!),
basicFormatter.format(num),
);
} catch (e) {
// Handy information for understanding failures
// tslint:disable-next-line:no-console
console.log({
num, formatted, parsed, numMode, parser,
parts: formatter.formatToParts(num),
formattedChars: [...formatted].map(char => ({
char,
// To see invisible characters, e.g. RTL/LTR marks
codePoint: char.codePointAt(0),
codePointHex: char.codePointAt(0)!.toString(16),
})),
formatterOptions: formatter.resolvedOptions(),
});
throw e;
}
}
}
}
});
});
}
});