dualfroz
93dd2b2f06
When maxDecimalPoints truncates a unit count to exactly 1, the rendered
number showed 1 while the word was pluralized from the raw count (for
example humanizeDuration(1500, { maxDecimalPoints: 0 }) returned
"1 seconds"). Pluralize from the normalized count that is actually
displayed so the number and word agree.
88 lines
2.4 KiB
Markdown
88 lines
2.4 KiB
Markdown
# Fix pluralization when `maxDecimalPoints` truncates the count to 1
|
|
|
|
## Problem
|
|
|
|
When the `maxDecimalPoints` option truncates a unit count down to exactly `1`,
|
|
the rendered number and the unit word disagree. The number is shown as `1` but
|
|
the word is pluralized:
|
|
|
|
```js
|
|
const humanizeDuration = require("humanize-duration");
|
|
|
|
humanizeDuration(1500, { maxDecimalPoints: 0 }); // "1 seconds" (expected "1 second")
|
|
humanizeDuration(1005, { maxDecimalPoints: 2 }); // "1 seconds" (expected "1 second")
|
|
```
|
|
|
|
The bug is language-independent; any language whose unit word is a function of
|
|
the count is affected (for example German returns `"1 Sekunden"` instead of
|
|
`"1 Sekunde"`).
|
|
|
|
## Root cause
|
|
|
|
`humanize-duration.js`, in `renderPiece`.
|
|
|
|
The displayed count is `normalizedUnitCount`, computed from `maxDecimalPoints`:
|
|
|
|
```js
|
|
var normalizedUnitCount =
|
|
maxDecimalPoints === void 0
|
|
? unitCount
|
|
: Math.floor(unitCount * Math.pow(10, maxDecimalPoints)) /
|
|
Math.pow(10, maxDecimalPoints);
|
|
var countStr = normalizedUnitCount.toString();
|
|
```
|
|
|
|
but the unit word was pluralized from the raw, un-truncated `unitCount`:
|
|
|
|
```js
|
|
if (typeof languageWord === "function") {
|
|
word = languageWord(unitCount);
|
|
}
|
|
```
|
|
|
|
So `1.5` seconds with `maxDecimalPoints: 0` renders the count as `"1"` (from
|
|
`normalizedUnitCount`) while `languageWord(1.5)` returns the plural form, giving
|
|
`"1 seconds"`.
|
|
|
|
## Fix
|
|
|
|
Pluralize based on the count that is actually rendered, `normalizedUnitCount`:
|
|
|
|
```js
|
|
if (typeof languageWord === "function") {
|
|
word = languageWord(normalizedUnitCount);
|
|
}
|
|
```
|
|
|
|
When `maxDecimalPoints` is unset, `normalizedUnitCount === unitCount`, so
|
|
existing behavior is unchanged. The fix only affects cases where the option
|
|
truncates the count.
|
|
|
|
## Test
|
|
|
|
Added a test to `test/humanizer.js` covering the truncated-to-one case, a
|
|
truncated count that stays plural, and the same mechanism in German:
|
|
|
|
```js
|
|
assert.strictEqual(humanizer({ maxDecimalPoints: 0 })(1500), "1 second");
|
|
assert.strictEqual(humanizer({ maxDecimalPoints: 2 })(1005), "1 second");
|
|
assert.strictEqual(humanizer({ maxDecimalPoints: 0 })(2500), "2 seconds");
|
|
assert.strictEqual(
|
|
humanizer({ language: "de", maxDecimalPoints: 0 })(1500),
|
|
"1 Sekunde"
|
|
);
|
|
```
|
|
|
|
Counterfactual (revert the one-line fix in `renderPiece`):
|
|
|
|
```
|
|
Expected values to be strictly equal:
|
|
expected: '1 second'
|
|
actual: '1 seconds'
|
|
```
|
|
|
|
## Gate
|
|
|
|
- `npm test` (runs `prettier --check` + `eslint` + `tsc --noEmit`, then
|
|
`node --test`): exit 0, all tests pass.
|