tapir/internal/reconcile/reconcile_spots_test.go

package reconcile_test

import (
	"testing"

	"github.com/stretchr/testify/assert"
	"github.com/stretchr/testify/require"

	"git.d-ma.be/mathias/tapir/internal/domain"
	"git.d-ma.be/mathias/tapir/internal/reconcile"
)

// Every judged spot is stored with a few words of left context, and the saved segments
// carry the merged words while keeping the base's timing (ADR-035 decisions 6 and 7).

func TestReconcileSpots_AKeptSpotAndAChangedSpotAreBothStoredWithTheirVerdict(t *testing.T) {
	adj := &scripted{fn: func(i int, _ reconcile.Request) (reconcile.Verdict, error) {
		if i == 0 {
			return reconcile.VerdictBase, nil
		}
		return reconcile.VerdictAlt, nil
	}}

	out := merge(t, tr(progBase), tr(progAlt), adj)

	require.Len(t, out.Reconciliation.Contested, 2)
	assert.Equal(t, "base", out.Reconciliation.Contested[0].Used)
	assert.Equal(t, "1015", out.Reconciliation.Contested[0].Base)
	assert.Equal(t, "alt", out.Reconciliation.Contested[1].Used)
	assert.Equal(t, 1, out.Reconciliation.Changed, "the exact counts are unchanged by storing more")
}

func TestReconcileSpots_AutoResolvedDisputesAreNotStored(t *testing.T) {
	out := merge(t, tr("i takrenns delen ska vi ju betala"), tr("i takrennsdelen ska vi betala"), always(reconcile.VerdictAlt))

	assert.Empty(t, out.Reconciliation.Contested, "compounds and fillers were never judged; they are counted, not listed")
}

func TestReconcileSpots_ASpotCarriesTheWordsJustBeforeIt(t *testing.T) {
	out := merge(t, tr("grannen målar ett vanligt staket mot gatan"), tr("grannen målar ett KORT staket mot gatan"), always(reconcile.VerdictAlt))

	require.Len(t, out.Reconciliation.Contested, 1)
	assert.Equal(t, "grannen målar ett", out.Reconciliation.Contested[0].Before)
}

func TestReconcileSpots_TheContextIsAFewWordsNotTheWholeMeeting(t *testing.T) {
	long := "ett två tre fyra fem sex sju åtta nio tio elva tolv"
	out := merge(t, tr(long+" X slut"), tr(long+" Y slut"), always(reconcile.VerdictAlt))

	require.Len(t, out.Reconciliation.Contested, 1)
	assert.Equal(t, "fem sex sju åtta nio tio elva tolv", out.Reconciliation.Contested[0].Before,
		"eight words: enough to find the place, small enough to stay out of the way")
}

func segmented(text string, segs ...domain.Segment) domain.Transcription {
	t := tr(text)
	t.Segments = segs
	return t
}

func TestReconcileSegments_AChangedWordReplacesTheBaseWordInItsOwnSegmentOnly(t *testing.T) {
	base := segmented("ett två tre. fyra fem sex. sju åtta nio.",
		domain.Segment{Start: 0, End: 10, Text: "ett två tre."},
		domain.Segment{Start: 10, End: 20, Text: "fyra fem sex."},
		domain.Segment{Start: 20, End: 30, Text: "sju åtta nio."})

	out := merge(t, base, tr("ett två tre. fyra femm sex. sju åtta nio."), always(reconcile.VerdictAlt))

	require.Len(t, out.Segments, 3)
	assert.Equal(t, "ett två tre.", out.Segments[0].Text, "an untouched segment keeps its exact text")
	assert.Equal(t, "fyra femm sex.", out.Segments[1].Text)
	assert.Equal(t, "sju åtta nio.", out.Segments[2].Text)
	assert.InDelta(t, 10.0, out.Segments[1].Start, 1e-9, "timing stays the base's")
	assert.InDelta(t, 20.0, out.Segments[1].End, 1e-9)
}

func TestReconcileSegments_AWordOnlyTheAlternativeHeardLandsInTheSegmentItBelongsTo(t *testing.T) {
	base := segmented("a b c. d e f.", domain.Segment{Start: 0, End: 5, Text: "a b c."}, domain.Segment{Start: 5, End: 10, Text: "d e f."})

	out := merge(t, base, tr("a b c. d NY e f."), always(reconcile.VerdictAlt))

	assert.Equal(t, "a b c.", out.Segments[0].Text)
	assert.Equal(t, "d NY e f.", out.Segments[1].Text)
}

func TestReconcileSegments_AWordOnlyTheBaseHeardIsRemovedFromItsSegment(t *testing.T) {
	base := segmented("a b c. d X e f.", domain.Segment{Start: 0, End: 5, Text: "a b c."}, domain.Segment{Start: 5, End: 10, Text: "d X e f."})

	out := merge(t, base, tr("a b c. d e f."), always(reconcile.VerdictAlt))

	assert.Equal(t, "d e f.", out.Segments[1].Text)
}

func TestReconcileSegments_AKeptBaseLeavesEverySegmentAsItWas(t *testing.T) {
	base := segmented("a x b", domain.Segment{Start: 0, End: 4, Text: "a x b"})

	out := merge(t, base, tr("a y b"), always(reconcile.VerdictBase))

	assert.Equal(t, base.Segments, out.Segments)
}

func TestReconcileSegments_TheCallersSegmentsAreNeverChangedInPlace(t *testing.T) {
	segs := []domain.Segment{{Start: 0, End: 4, Text: "a x b"}}
	base := segmented("a x b", segs...)

	_ = merge(t, base, tr("a y b"), always(reconcile.VerdictAlt))

	assert.Equal(t, "a x b", segs[0].Text, "Reconcile builds new segments; it does not edit the base transcription's")
}

func TestReconcileSegments_SegmentsThatDoNotLineUpWithTheTextAreLeftAsTheBases(t *testing.T) {
	// A word wrapped across a segment edge makes the segments' word count differ from
	// the joined text's. Guessing would put words in the wrong place, so leave them.
	base := segmented("a x b c", domain.Segment{Start: 0, End: 4, Text: "a x"}, domain.Segment{Start: 4, End: 8, Text: "b c d"})

	out := merge(t, base, tr("a y b c"), always(reconcile.VerdictAlt))

	assert.Equal(t, base.Segments, out.Segments)
	assert.Equal(t, "a y b c", out.Text, "the merged text is unaffected")
}

func TestReconcileSegments_ASpanThatCrossesASegmentEdgeIsRemovedFromBothSegments(t *testing.T) {
	base := segmented("a b c d", domain.Segment{Start: 0, End: 5, Text: "a b"}, domain.Segment{Start: 5, End: 10, Text: "c d"})

	out := merge(t, base, tr("a X d"), always(reconcile.VerdictAlt))

	require.Equal(t, "a X d", out.Text)
	assert.Equal(t, "a X", out.Segments[0].Text)
	assert.Equal(t, "d", out.Segments[1].Text, "'c' was replaced too: it must not linger in the second segment")
}