package pipeline

import "testing"

func TestClusterRowsTwoBands(t *testing.T) {
	boxes := []indexedBox{
		{Text: "a", X: 0, Y: 100},
		{Text: "b", X: 10, Y: 102},
		{Text: "c", X: 0, Y: 300},
		{Text: "d", X: 10, Y: 301},
	}
	ClusterRows(boxes, 10, 1)
	if boxes[0].Cluster < 0 || boxes[1].Cluster < 0 {
		t.Fatal("first band should be clustered")
	}
	if boxes[0].Cluster != boxes[1].Cluster {
		t.Fatalf("same band: got %d and %d", boxes[0].Cluster, boxes[1].Cluster)
	}
	if boxes[2].Cluster != boxes[3].Cluster {
		t.Fatalf("second band: got %d and %d", boxes[2].Cluster, boxes[3].Cluster)
	}
	if boxes[0].Cluster == boxes[2].Cluster {
		t.Fatal("distant Y bands should differ")
	}
}

func TestClusterRowsSameBand(t *testing.T) {
	boxes := []indexedBox{
		{Text: "a", Y: 50},
		{Text: "b", Y: 52},
		{Text: "c", Y: 54},
	}
	ClusterRows(boxes, 10, 1)
	id := boxes[0].Cluster
	for i, b := range boxes {
		if b.Cluster != id {
			t.Fatalf("box %d cluster=%d, want %d", i, b.Cluster, id)
		}
	}
}

func TestClusterRowsEmpty(t *testing.T) {
	ClusterRows(nil, 10, 1) // no panic
	ClusterRows([]indexedBox{}, 10, 1)
}

func TestClusterRowsNoiseMinSamples(t *testing.T) {
	boxes := []indexedBox{
		{Text: "lonely", Y: 10},
		{Text: "pair-a", Y: 200},
		{Text: "pair-b", Y: 201},
	}
	// minSamples=2: lonely point stays noise (-1); pair forms a cluster
	ClusterRows(boxes, 5, 2)
	if boxes[0].Cluster != -1 {
		t.Fatalf("lonely cluster=%d, want -1 (noise)", boxes[0].Cluster)
	}
	if boxes[1].Cluster < 0 || boxes[1].Cluster != boxes[2].Cluster {
		t.Fatalf("pair clusters=%d,%d", boxes[1].Cluster, boxes[2].Cluster)
	}
}