@fable-org/fable-library-js 2.6.0 → 2.7.1

package/TimeOnlyTemporal.js

@@ -0,0 +1,142 @@
+import { toInt64, fromFloat64 } from "./BigInt.js";
+import { fromTicks as TimeSpan_fromTicks, totalNanoseconds as TimeSpan_totalNanoseconds } from "./TimeSpanTemporal.js";
+import { Exception } from "./Util.js";
+export const PlainTime = Temporal.PlainTime;
+// The generic equality/comparison/hashing helpers in Util.ts dispatch at runtime
+// on .NET-style Equals/CompareTo/GetHashCode methods. Attach them so TimeOnly
+// also works in generic contexts: records, tuples, erased generics, etc.
+const proto = Temporal.PlainTime.prototype;
+proto.Equals = function (other) { return this.equals(other); };
+proto.CompareTo = function (other) { return Temporal.PlainTime.compare(this, other); };
+proto.GetHashCode = function () { return hash(this); };
+// Lets String.Format format this value without String.ts importing this module (see String.ts)
+proto[Symbol.for("Fable.DateTimeFormattable")] = function (format) { return toString(this, format); };
+const nanosecondsPerDay = 86400000000000;
+const nanosecondsPerDayBig = 86400000000000n;
+function totalNanoseconds(t) {
+    return ((t.hour * 60 + t.minute) * 60 + t.second) * 1000000000
+        + t.millisecond * 1000000 + t.microsecond * 1000 + t.nanosecond;
+}
+function fromNanoseconds(n) {
+    return new Temporal.PlainTime(Math.floor(n / 3600000000000), Math.floor(n / 60000000000) % 60, Math.floor(n / 1000000000) % 60, Math.floor(n / 1000000) % 1000, Math.floor(n / 1000) % 1000, n % 1000);
+}
+export function create(h = 0, m = 0, s = 0, ms = 0, us = 0) {
+    return new Temporal.PlainTime(h, m, s, ms, us);
+}
+export function fromTicks(ticks) {
+    return fromNanoseconds(Number(BigInt(ticks) * 100n));
+}
+export function fromTimeSpan(timeSpan) {
+    const ns = Number(TimeSpan_totalNanoseconds(timeSpan));
+    if (ns < 0 || ns >= nanosecondsPerDay)
+        throw new Exception("The TimeSpan describes an unrepresentable TimeOnly.");
+    return fromNanoseconds(ns);
+}
+export function fromDateTime(d) {
+    // Under the Temporal representation a DateTime is a PlainDateTime (kind-agnostic wall-clock)
+    return new Temporal.PlainTime(d.hour, d.minute, d.second, d.millisecond, d.microsecond, d.nanosecond);
+}
+export function minValue() {
+    return new Temporal.PlainTime();
+}
+export function maxValue() {
+    // This is "23:59:59.9999999" (.NET tick precision)
+    return new Temporal.PlainTime(23, 59, 59, 999, 999, 900);
+}
+export function ticks(t) {
+    return toInt64(fromFloat64(totalNanoseconds(t) / 100));
+}
+export function toTimeSpan(t) {
+    // TimeOnly is tick-precise, so the nanosecond count is always a multiple of 100.
+    return TimeSpan_fromTicks(Math.round(totalNanoseconds(t) / 100));
+}
+// Kept in bigint: a TimeOnly itself always fits a float64 exactly (under 2^47 ns),
+// but the amount added does not — a TimeSpan of more than ~104 days already
+// exceeds Number.MAX_SAFE_INTEGER in nanoseconds, and .NET wraps it modulo a day
+// rather than saturating, so the low ticks are the part that must survive.
+function addNanoseconds(t, deltaNs, wrappedDays) {
+    const totalNs = BigInt(totalNanoseconds(t)) + deltaNs;
+    let days = totalNs / nanosecondsPerDayBig;
+    let remainder = totalNs % nanosecondsPerDayBig;
+    // BigInt division truncates toward zero; .NET wraps toward the previous day.
+    if (remainder < 0n) {
+        remainder += nanosecondsPerDayBig;
+        days -= 1n;
+    }
+    if (wrappedDays !== undefined) {
+        wrappedDays.contents = Number(days);
+    }
+    return fromNanoseconds(Number(remainder));
+}
+export function add(t, ts, wrappedDays) {
+    return addNanoseconds(t, TimeSpan_totalNanoseconds(ts), wrappedDays);
+}
+export function addHours(t, h) {
+    // Scale to ticks first, as .NET does, then widen — `h` hours in nanoseconds
+    // overflows the exact float64 range two orders of magnitude sooner.
+    return addNanoseconds(t, BigInt(Math.round(h * 36000000000)) * 100n);
+}
+export function addMinutes(t, m) {
+    return addNanoseconds(t, BigInt(Math.round(m * 600000000)) * 100n);
+}
+export function isBetween(t, start, end) {
+    return Temporal.PlainTime.compare(start, end) <= 0
+        ? (Temporal.PlainTime.compare(start, t) <= 0 && Temporal.PlainTime.compare(end, t) > 0)
+        : (Temporal.PlainTime.compare(start, t) <= 0 || Temporal.PlainTime.compare(end, t) > 0);
+}
+export function equals(x, y) {
+    return x.equals(y);
+}
+export function compare(x, y) {
+    return Temporal.PlainTime.compare(x, y);
+}
+export function hash(t) {
+    return totalNanoseconds(t) % 2147483647;
+}
+export function op_Subtraction(left, right) {
+    // Returns the elapsed TimeSpan, wrapping around midnight
+    const ns = (totalNanoseconds(left) - totalNanoseconds(right) + nanosecondsPerDay) % nanosecondsPerDay;
+    return TimeSpan_fromTicks(Math.round(ns / 100));
+}
+export function toString(t, format = "t", _provider) {
+    switch (format) {
+        case "t":
+            return t.toString({ smallestUnit: "minute" });
+        case "r":
+        case "R":
+        case "T":
+            return t.toString({ smallestUnit: "second" });
+        case "o":
+        case "O":
+            // .NET tick precision is 7 fractional digits
+            return t.toString({ fractionalSecondDigits: 7 });
+        default:
+            throw new Exception("Custom formats are not supported");
+    }
+}
+export function parse(str) {
+    // Allowed format types:
+    // hh:mm
+    // hh:mm:ss
+    // hh:mm:ss.fffffff
+    const r = /^\s*([0-1]?\d|2[0-3])\s*:\s*([0-5]?\d)(\s*:\s*([0-5]?\d)(\.(\d+))?)?\s*$/.exec(str);
+    if (r != null && r[1] != null && r[2] != null) {
+        const h = +r[1];
+        const m = +r[2];
+        const s = r[4] != null ? +r[4] : 0;
+        // .NET resolves the fraction down to a single tick (7 digits) and truncates
+        // anything finer, so the digits are read as ticks rather than milliseconds.
+        const subSecondTicks = r[6] != null ? +r[6].padEnd(7, "0").substring(0, 7) : 0;
+        return fromNanoseconds(((h * 60 + m) * 60 + s) * 1000000000 + subSecondTicks * 100);
+    }
+    throw new Exception(`String '${str}' was not recognized as a valid TimeOnly.`);
+}
+export function tryParse(v, defValue) {
+    try {
+        defValue.contents = parse(v);
+        return true;
+    }
+    catch {
+        return false;
+    }
+}

package/TimeSpanTemporal.js

@@ -0,0 +1,259 @@
+import { toInt64 } from "./BigInt.js";
+import { Exception, padWithZeros } from "./Util.js";
+export const Duration = Temporal.Duration;
+// The generic equality/comparison/hashing helpers in Util.ts dispatch at runtime
+// on .NET-style Equals/CompareTo/GetHashCode methods. Attach them so TimeSpan
+// also works in generic contexts: records, tuples, erased generics, etc.
+const proto = Temporal.Duration.prototype;
+proto.Equals = function (other) { return Temporal.Duration.compare(this, other) === 0; };
+proto.CompareTo = function (other) { return Temporal.Duration.compare(this, other); };
+proto.GetHashCode = function () { return hash(this); };
+// Lets String.Format format this value without String.ts importing this module (see String.ts)
+proto[Symbol.for("Fable.DateTimeFormattable")] = function (format) { return toString(this, format); };
+const nsPerMillisecond = 1000000n;
+const nsPerSecond = 1000000000n;
+const nsPerMinute = 60000000000n;
+const nsPerHour = 3600000000000n;
+const nsPerDay = 86400000000000n;
+export function totalNanoseconds(ts) {
+    return BigInt(ts.days) * nsPerDay
+        + BigInt(ts.hours) * nsPerHour
+        + BigInt(ts.minutes) * nsPerMinute
+        + BigInt(ts.seconds) * nsPerSecond
+        + BigInt(ts.milliseconds) * nsPerMillisecond
+        + BigInt(ts.microseconds) * 1000n
+        + BigInt(ts.nanoseconds);
+}
+function fromNanoseconds(totalNs) {
+    const negative = totalNs < 0n;
+    // A .NET TimeSpan is a whole number of 100ns ticks, so anything finer than a
+    // tick is dropped (toward zero, as .NET does) rather than carried in the
+    // Duration's nanosecond field — otherwise values that cannot exist in .NET
+    // would leak out of Parse/FromMilliseconds/arithmetic.
+    let n = (negative ? -totalNs : totalNs) / 100n * 100n;
+    const days = Number(n / nsPerDay);
+    n %= nsPerDay;
+    const hours = Number(n / nsPerHour);
+    n %= nsPerHour;
+    const minutes = Number(n / nsPerMinute);
+    n %= nsPerMinute;
+    const seconds = Number(n / nsPerSecond);
+    n %= nsPerSecond;
+    const milliseconds = Number(n / nsPerMillisecond);
+    n %= nsPerMillisecond;
+    const microseconds = Number(n / 1000n);
+    n %= 1000n;
+    const d = new Temporal.Duration(0, 0, 0, days, hours, minutes, seconds, milliseconds, microseconds, Number(n));
+    return negative ? d.negated() : d;
+}
+// Converts a possibly fractional unit count into an exact nanosecond amount,
+// keeping the integer part exact for the full Int64 tick range.
+function unitToNanoseconds(value, nsPerUnit) {
+    const whole = Math.trunc(value);
+    const frac = value - whole;
+    return BigInt(whole) * nsPerUnit + BigInt(Math.round(frac * Number(nsPerUnit)));
+}
+export function create(d = 0, h = 0, m = 0, s = 0, ms = 0, us = 0) {
+    switch (arguments.length) {
+        case 1:
+            // ticks
+            return fromTicks(arguments[0]);
+        case 3:
+            // h,m,s
+            d = 0, h = arguments[0], m = arguments[1], s = arguments[2], ms = 0;
+            break;
+        default:
+            // d,h,m,s,ms,us
+            break;
+    }
+    return fromNanoseconds(unitToNanoseconds(d, nsPerDay) + unitToNanoseconds(h, nsPerHour) + unitToNanoseconds(m, nsPerMinute)
+        + unitToNanoseconds(s, nsPerSecond) + unitToNanoseconds(ms, nsPerMillisecond) + unitToNanoseconds(us, 1000n));
+}
+export function fromTicks(ticks) {
+    return fromNanoseconds(BigInt(ticks) * 100n);
+}
+export function zero() {
+    return new Temporal.Duration();
+}
+const ticksPerUnit = {
+    days: 864000000000,
+    hours: 36000000000,
+    minutes: 600000000,
+    seconds: 10000000,
+    milliseconds: 10000,
+};
+// The DateTime/DateTimeOffset/TimeOnly Add* members take a double and are
+// tick-precise in .NET: the value is scaled to ticks and rounded, so
+// AddSeconds(0.0000001) moves exactly one tick and AddDays(1.5) exactly 36 hours.
+// Temporal's own `add({ days: 1.5 })` rejects non-integers, so the scaling has to
+// happen here.
+export function fromUnits(value, unit) {
+    return fromTicks(BigInt(Math.round(value * ticksPerUnit[unit])));
+}
+export function fromDays(d, h = 0, m = 0n, s = 0n, ms = 0n) {
+    return create(d, h, Number(m), Number(s), Number(ms));
+}
+export function fromHours(h, m = 0n, s = 0n, ms = 0n) {
+    return create(0, h, Number(m), Number(s), Number(ms));
+}
+export function fromMinutes(m, s = 0n, ms = 0n) {
+    return create(0, 0, Number(m), Number(s), Number(ms));
+}
+export function fromSeconds(s, ms = 0n) {
+    return create(0, 0, 0, Number(s), Number(ms));
+}
+export function fromMilliseconds(ms) {
+    return fromNanoseconds(unitToNanoseconds(Number(ms), nsPerMillisecond));
+}
+export function ticks(ts) {
+    return toInt64(totalNanoseconds(ts) / 100n);
+}
+// .NET computes each total as (double)Ticks / TicksPerUnit; match that exactly
+// (Temporal's own .total() balances differently and can be 1 ULP off).
+function totalTicks(ts) {
+    return Number(totalNanoseconds(ts) / 100n);
+}
+export function totalDays(ts) {
+    return totalTicks(ts) / 864000000000;
+}
+export function totalHours(ts) {
+    return totalTicks(ts) / 36000000000;
+}
+export function totalMinutes(ts) {
+    return totalTicks(ts) / 600000000;
+}
+export function totalSeconds(ts) {
+    return totalTicks(ts) / 10000000;
+}
+export function totalMilliseconds(ts) {
+    return totalTicks(ts) / 10000;
+}
+export function negate(ts) {
+    return ts.negated();
+}
+export function add(ts1, ts2) {
+    return fromNanoseconds(totalNanoseconds(ts1) + totalNanoseconds(ts2));
+}
+export function subtract(ts1, ts2) {
+    return fromNanoseconds(totalNanoseconds(ts1) - totalNanoseconds(ts2));
+}
+// .NET scales the tick count as a double and rounds: `new TimeSpan((long)Math
+// .Round(Ticks * factor))`. Scaling ticks rather than nanoseconds keeps the
+// magnitude — and so the rounding — identical to .NET's.
+export function multiply(ts, factor) {
+    return fromTicks(BigInt(Math.round(Number(ticks(ts)) * factor)));
+}
+export function divide(ts, factor) {
+    return fromTicks(BigInt(Math.round(Number(ticks(ts)) / factor)));
+}
+export function divideByTimeSpan(ts, other) {
+    return Number(ticks(ts)) / Number(ticks(other));
+}
+export const op_Addition = add;
+export const op_Subtraction = subtract;
+export const op_Multiply = multiply;
+export const op_UnaryNegation = negate;
+export function compare(x, y) {
+    return Temporal.Duration.compare(x, y);
+}
+export const compareTo = compare;
+export function equals(x, y) {
+    return Temporal.Duration.compare(x, y) === 0;
+}
+export function hash(ts) {
+    return Number(totalNanoseconds(ts) % 2147483647n);
+}
+export function duration(ts) {
+    return ts.abs();
+}
+export function toString(ts, format = "c", _provider) {
+    if (["c", "g", "G"].indexOf(format) === -1) {
+        throw new Exception("Custom formats are not supported");
+    }
+    const d = Math.abs(ts.days);
+    const h = Math.abs(ts.hours);
+    const m = Math.abs(ts.minutes);
+    const s = Math.abs(ts.seconds);
+    const ticks = Math.abs(ts.milliseconds) * 10000
+        + Math.abs(ts.microseconds) * 10
+        + Math.trunc(Math.abs(ts.nanoseconds) / 100);
+    const sign = ts.sign < 0 ? "-" : "";
+    const days = d === 0 && format !== "G" ? "" : format === "c" ? d + "." : d + ":";
+    const time = `${format === "g" ? h : padWithZeros(h, 2)}:${padWithZeros(m, 2)}:${padWithZeros(s, 2)}`;
+    // .NET prints seven tick digits for "c" and "G", trims trailing zeros for "g", and omits a zero fraction for "c" and "g".
+    const fraction = format === "g"
+        ? (ticks === 0 ? "" : "." + padWithZeros(ticks, 7).replace(/0+$/, ""))
+        : (ticks === 0 && format === "c" ? "" : "." + padWithZeros(ticks, 7));
+    return `${sign}${days}${time}${fraction}`;
+}
+export function parse(str) {
+    const firstDot = str.search("\\.");
+    const firstColon = str.search("\\:");
+    if (firstDot === -1 && firstColon === -1) { // There is only a day ex: 4
+        const d = parseInt(str, 0);
+        if (isNaN(d)) {
+            throw new Exception(`String '${str}' was not recognized as a valid TimeSpan.`);
+        }
+        else {
+            return create(d, 0, 0, 0, 0);
+        }
+    }
+    if (firstColon > 0) { // process time part
+        // WIP: (-?)(((\d+)\.)?([0-9]|0[0-9]|1[0-9]|2[0-3]):(\d+)(:\d+(\.\d{1,7})?)?|\d+(?:(?!\.)))
+        const r = /^(-?)((\d+)\.)?(?:0*)([0-9]|0[0-9]|1[0-9]|2[0-3]):(?:0*)([0-5][0-9]|[0-9])(:(?:0*)([0-5][0-9]|[0-9]))?\.?(\d+)?$/.exec(str);
+        if (r != null && r[4] != null && r[5] != null) {
+            let d = 0;
+            let ms = 0;
+            let s = 0;
+            const sign = r[1] != null && r[1] === "-" ? -1 : 1;
+            const h = +r[4];
+            const m = +r[5];
+            if (r[3] != null) {
+                d = +r[3];
+            }
+            if (r[7] != null) {
+                s = +r[7];
+            }
+            if (r[8] != null) {
+                // Depending on the number of decimals passed, we need to adapt the numbers
+                switch (r[8].length) {
+                    case 1:
+                        ms = +r[8] * 100;
+                        break;
+                    case 2:
+                        ms = +r[8] * 10;
+                        break;
+                    case 3:
+                        ms = +r[8];
+                        break;
+                    case 4:
+                        ms = +r[8] / 10;
+                        break;
+                    case 5:
+                        ms = +r[8] / 100;
+                        break;
+                    case 6:
+                        ms = +r[8] / 1000;
+                        break;
+                    case 7:
+                        ms = +r[8] / 10000;
+                        break;
+                    default:
+                        throw new Exception(`String '${str}' was not recognized as a valid TimeSpan.`);
+                }
+            }
+            const ts = create(d, h, m, s, ms);
+            return sign < 0 ? negate(ts) : ts;
+        }
+    }
+    throw new Exception(`String '${str}' was not recognized as a valid TimeSpan.`);
+}
+export function tryParse(v, defValue) {
+    try {
+        defValue.contents = parse(v);
+        return true;
+    }
+    catch {
+        return false;
+    }
+}

package/lib/big.js

@@ -210,6 +210,9 @@ function round(x, sd, rm, more) {
     }
     return x;
 }
+function hasNonzeroDigit(x) {
+    return x.c.some(d => d !== 0);
+}
 /*
  * Return a string representing the value of Big x in normal or exponential notation.
  * Handles P.toExponential, P.toFixed, P.toJSON, P.toPrecision, P.toString and P.valueOf.
@@ -728,7 +731,7 @@ P.times = P.mul = function (y) {
  * rm? {number} Rounding mode: 0 (down), 1 (half-up), 2 (half-even) or 3 (up).
  */
 P.toExponential = function (dp, rm) {
-    var x = this, n = x.c[0];
+    var x = this, n = hasNonzeroDigit(x);
     if (dp !== UNDEFINED) {
         if (dp !== ~~dp || dp < 0 || dp > MAX_DP) {
             throw Error(INVALID_DP);
@@ -737,7 +740,7 @@ P.toExponential = function (dp, rm) {
         for (; x.c.length < dp;)
             x.c.push(0);
     }
-    return stringify(x, true, !!n);
+    return stringify(x, true, n);
 };
 /*
  * Return a string representing the value of this Big in normal notation rounded to dp fixed
@@ -746,21 +749,22 @@ P.toExponential = function (dp, rm) {
  * dp? {number} Decimal places: integer, 0 to MAX_DP inclusive.
  * rm? {number} Rounding mode: 0 (down), 1 (half-up), 2 (half-even) or 3 (up).
  *
- * (-0).toFixed(0) is '0', but (-0.1).toFixed(0) is '-0'.
- * (-0).toFixed(1) is '0.0', but (-0.01).toFixed(1) is '-0.0'.
+ * Unlike JavaScript, the sign is omitted when the rounded result is zero:
+ * (-0.1).toFixed(0) is '0' and (-0.01).toFixed(1) is '0.0', matching .NET.
  */
 P.toFixed = function (dp, rm) {
-    var x = this, n = x.c[0];
+    var x = this;
     if (dp !== UNDEFINED) {
         if (dp !== ~~dp || dp < 0 || dp > MAX_DP) {
             throw Error(INVALID_DP);
         }
-        x = round(new x.constructor(x), dp + x.e + 1, rm);
+        x = new x.constructor(x);
+        x = round(x, dp + x.e + 1, rm);
         // x.e may have changed if the value is rounded up.
         for (dp = dp + x.e + 1; x.c.length < dp;)
             x.c.push(0);
     }
-    return stringify(x, false, !!n);
+    return stringify(x, false, hasNonzeroDigit(x));
 };
 /*
  * Return a string representing the value of this Big.
@@ -770,7 +774,7 @@ P.toFixed = function (dp, rm) {
  */
 P.toJSON = P.toString = function () {
     var x = this, Big = x.constructor;
-    return stringify(x, x.e <= Big.NE || x.e >= Big.PE, !!x.c[0]);
+    return stringify(x, x.e <= Big.NE || x.e >= Big.PE, hasNonzeroDigit(x));
 };
 /*
  * Return the value of this Big as a primitve number.
@@ -792,7 +796,7 @@ P.toNumber = function () {
  * rm? {number} Rounding mode: 0 (down), 1 (half-up), 2 (half-even) or 3 (up).
  */
 P.toPrecision = function (sd, rm) {
-    var x = this, Big = x.constructor, n = x.c[0];
+    var x = this, Big = x.constructor, n = hasNonzeroDigit(x);
     if (sd !== UNDEFINED) {
         if (sd !== ~~sd || sd < 1 || sd > MAX_DP) {
             throw Error(INVALID + 'precision');
@@ -801,7 +805,7 @@ P.toPrecision = function (sd, rm) {
         for (; x.c.length < sd;)
             x.c.push(0);
     }
-    return stringify(x, sd <= x.e || x.e <= Big.NE || x.e >= Big.PE, !!n);
+    return stringify(x, sd <= x.e || x.e <= Big.NE || x.e >= Big.PE, n);
 };
 /*
  * Return a string representing the value of this Big.

package/package.json

@@ -3,7 +3,7 @@
     "private": false,
     "type": "module",
     "name": "@fable-org/fable-library-js",
-    "version": "2.6.0",
+    "version": "2.7.1",
     "description": "Core library used by F# projects compiled with fable.io",
     "author": "Fable Contributors",
     "license": "MIT",