Classify

Rasuvaeff\PropertyTesting\Classify

Class — Source

Records distribution labels for the current property run so the runner can report how often each case occurred. Use it to confirm a property is not passing vacuously — that the generators actually exercise the interesting inputs.

#[Property(runs: 500)]
public function holds(int $n): void
{
    Classify::when($n === 0, 'zero');
    Classify::when($n < 0, 'negative');
    Classify::label($n % 2 === 0 ? 'even' : 'odd');
    // ... assertions ...
}

After a fully passing property the runner prints the share of runs that hit each label. A label recorded several times within one run still counts once for that run.

State is per-run and process-local: the runner clears it before each run via beginRun() and drains it via flushRun(). Property runs are sequential, so the static buffer is never shared concurrently.

Methods

label()

static label(string $label): void

Record $label for the current run.

when()

static when(bool $condition, string $label): void

Record $label for the current run only when $condition holds.

cover()

static cover(
    bool $condition,
    string $label,
    float $minPercent,
): void

Like when(), but additionally REQUIRES the label to occur in at least $minPercent of the property's passing runs. When the requirement is not met the property fails with a CoverageViolationException even though every run passed — turning "the distribution looks wrong" from a printed hint into a CI failure.

Classify::cover($n % 2 === 0, 'even', 30.0);

beginRun()

static beginRun(): void

Clear the labels buffered for the current run.

flushRun()

static flushRun(): array

Return the labels recorded during the current run and clear the buffer.

flushRequirements()

static flushRequirements(): array

Return the coverage requirements registered during the property and clear them. The runner drains this once after the run loop (and defensively before it, in case a previous property aborted mid-flight).