Bugfix in the coalesce module

2026-04-27 20:29:55 -05:00
parent fb0519c561
commit 8a95dba276
4 changed files with 248 additions and 6 deletions
--- a/internal/coalesce/coalesce_test.go
+++ b/internal/coalesce/coalesce_test.go
@@ -165,6 +165,119 @@ func TestApplyDropsFillerCategoryFromMergedSameSpeakerRun(t *testing.T) {
 	}
 }

+func TestApplyMergesSameSpeakerBackchannelIntoDerivedRun(t *testing.T) {
+	first := segment("zach.json", 110, "Zach", 7811.778, 7812.478, "That makes sense.")
+	first.Categories = []string{"backchannel"}
+	second := model.Segment{
+		Source:      "zach.json",
+		SourceRef:   "coalesce:347",
+		DerivedFrom: []string{"zach.json#111", "zach.json#112"},
+		Speaker:     "Zach",
+		Start:       7812.498,
+		End:         7824.045,
+		Text:        "So, like, I'm above the silence field.",
+	}
+
+	got, summary := Apply(model.MergedTranscript{Segments: []model.Segment{first, second}}, 3)
+	if summary.OriginalSegmentsMerged != 2 || summary.CoalescedSegments != 1 {
+		t.Fatalf("summary = %#v", summary)
+	}
+	if len(got.Segments) != 1 {
+		t.Fatalf("segment count = %d, want 1", len(got.Segments))
+	}
+	if got.Segments[0].Text != "That makes sense. So, like, I'm above the silence field." {
+		t.Fatalf("text = %q", got.Segments[0].Text)
+	}
+	if got.Segments[0].SourceRef != "coalesce:1" {
+		t.Fatalf("source_ref = %q, want coalesce:1", got.Segments[0].SourceRef)
+	}
+	if !reflect.DeepEqual(got.Segments[0].DerivedFrom, []string{"zach.json#110", "coalesce:347"}) {
+		t.Fatalf("derived_from = %v", got.Segments[0].DerivedFrom)
+	}
+	if got.Segments[0].Categories != nil {
+		t.Fatalf("categories = %v, want nil", got.Segments[0].Categories)
+	}
+}
+
+func TestApplyMergesSameSpeakerFillerIntoDerivedRun(t *testing.T) {
+	first := segment("zach.json", 110, "Zach", 1, 1.5, "um")
+	first.Categories = []string{"filler"}
+	second := model.Segment{
+		Source:      "zach.json",
+		SourceRef:   "coalesce:12",
+		DerivedFrom: []string{"zach.json#111", "zach.json#112"},
+		Speaker:     "Zach",
+		Start:       1.6,
+		End:         4,
+		Text:        "next thought",
+	}
+
+	got, summary := Apply(model.MergedTranscript{Segments: []model.Segment{first, second}}, 3)
+	if summary.OriginalSegmentsMerged != 2 || summary.CoalescedSegments != 1 {
+		t.Fatalf("summary = %#v", summary)
+	}
+	if len(got.Segments) != 1 {
+		t.Fatalf("segment count = %d, want 1", len(got.Segments))
+	}
+	if got.Segments[0].Text != "um next thought" {
+		t.Fatalf("text = %q", got.Segments[0].Text)
+	}
+	if !reflect.DeepEqual(got.Segments[0].DerivedFrom, []string{"zach.json#110", "coalesce:12"}) {
+		t.Fatalf("derived_from = %v", got.Segments[0].DerivedFrom)
+	}
+	if got.Segments[0].Categories != nil {
+		t.Fatalf("categories = %v, want nil", got.Segments[0].Categories)
+	}
+}
+
+func TestApplyUsesSkippedBackchannelToSeedNextSameSpeakerRun(t *testing.T) {
+	mike := segment("mike.json", 367, "Mike", 7803.57, 7810.719, "It's very easy to notice.")
+	backchannel := segment("zach.json", 110, "Zach", 7811.778, 7812.478, "That makes sense.")
+	backchannel.Categories = []string{"backchannel"}
+	next := segment("zach.json", 111, "Zach", 7812.498, 7820, "So, like, next thought.")
+
+	got, summary := Apply(model.MergedTranscript{Segments: []model.Segment{mike, backchannel, next}}, 3)
+	if summary.OriginalSegmentsMerged != 2 || summary.CoalescedSegments != 1 {
+		t.Fatalf("summary = %#v", summary)
+	}
+	if len(got.Segments) != 2 {
+		t.Fatalf("segment count = %d, want 2", len(got.Segments))
+	}
+	if got.Segments[0].Text != "It's very easy to notice." {
+		t.Fatalf("first text = %q", got.Segments[0].Text)
+	}
+	if got.Segments[1].Speaker != "Zach" || got.Segments[1].Text != "That makes sense. So, like, next thought." {
+		t.Fatalf("second segment = %#v", got.Segments[1])
+	}
+	if !reflect.DeepEqual(got.Segments[1].DerivedFrom, []string{"zach.json#110", "zach.json#111"}) {
+		t.Fatalf("derived_from = %v", got.Segments[1].DerivedFrom)
+	}
+	if got.Segments[1].Categories != nil {
+		t.Fatalf("categories = %v, want nil", got.Segments[1].Categories)
+	}
+}
+
+func TestApplyUsesSkippedFillerToSeedNextSameSpeakerRun(t *testing.T) {
+	alice := segment("alice.json", 0, "Alice", 1, 2, "first")
+	filler := segment("bob.json", 0, "Bob", 2.1, 2.3, "um")
+	filler.Categories = []string{"filler"}
+	bob := segment("bob.json", 1, "Bob", 2.4, 4, "actual thought")
+
+	got, summary := Apply(model.MergedTranscript{Segments: []model.Segment{alice, filler, bob}}, 3)
+	if summary.OriginalSegmentsMerged != 2 || summary.CoalescedSegments != 1 {
+		t.Fatalf("summary = %#v", summary)
+	}
+	if len(got.Segments) != 2 {
+		t.Fatalf("segment count = %d, want 2", len(got.Segments))
+	}
+	if got.Segments[1].Text != "um actual thought" {
+		t.Fatalf("second text = %q", got.Segments[1].Text)
+	}
+	if got.Segments[1].Categories != nil {
+		t.Fatalf("categories = %v, want nil", got.Segments[1].Categories)
+	}
+}
+
 func TestApplySkipsDifferentSpeakerBackchannelAsMergeBlocker(t *testing.T) {
 	first := segment("a.json", 0, "Alice", 1, 2, "first")
 	backchannel := segment("b.json", 0, "Bob", 2.2, 2.5, "yeah")