NaviWatcher/internal/scanner/scanner_test.go

package scanner

import (
	"testing"

	"naviwatcher/internal/database"
)

func TestSimilarity(t *testing.T) {
	tests := []struct {
		name     string
		a        string
		b        string
		want     float64
		epsilon  float64
	}{
		{
			name:    "exact match scores 1.0",
			a:       "The Wall",
			b:       "The Wall",
			want:    1.0,
			epsilon: 1e-9,
		},
		{
			name:    "case-insensitive exact match scores 1.0",
			a:       "The Wall",
			b:       "the wall",
			want:    1.0,
			epsilon: 1e-9,
		},
		{
			name:    "remastered variant stays above threshold",
			a:       "The Wall",
			b:       "The Wall (Remastered)",
			want:    1.0, // parenthesized content is stripped during normalization
			epsilon: 1e-9,
		},
		{
			name: "year-suffixed variant stays above threshold",
			a:    "Abbey Road",
			b:    "Abbey Road (2019 Remix)",
			// After normalization both collapse to "abbey road" → identical.
			want:    1.0,
			epsilon: 1e-9,
		},
		{
			name:    "clearly different titles score below 0.85",
			a:       "The Wall",
			b:       "Completely Different Album",
			want:    0.0,
			epsilon: 1e-9,
		},
		{
			name:    "substring-ish title scores moderately",
			a:       "Dark Side of the Moon",
			b:       "Dark Side of the Moon Part II",
			want:    0.0, // non-empty; value asserted only as below threshold
			epsilon: 1e-9,
		},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			got := Similarity(tt.a, tt.b)
			switch {
			case tt.name == "clearly different titles score below 0.85" ||
				tt.name == "substring-ish title scores moderately":
				if got >= 0.85 {
					t.Errorf("Similarity(%q, %q) = %v, want < 0.85", tt.a, tt.b, got)
				}
			default:
				if diff := got - tt.want; diff > tt.epsilon || diff < -tt.epsilon {
					t.Errorf("Similarity(%q, %q) = %v, want %v (+/- %v)", tt.a, tt.b, got, tt.want, tt.epsilon)
				}
			}
		})
	}
}

func TestSimilarity_EmptyStrings(t *testing.T) {
	// Both empty → no match (0.0).
	if got := Similarity("", ""); got != 0.0 {
		t.Errorf("Similarity(%q, %q) = %v, want 0.0", "", "", got)
	}
	// One empty, one non-empty → no similarity.
	if got := Similarity("The Wall", ""); got != 0.0 {
		t.Errorf("Similarity(%q, %q) = %v, want 0.0", "The Wall", "", got)
	}
	if got := Similarity("", "The Wall"); got != 0.0 {
		t.Errorf("Similarity(%q, %q) = %v, want 0.0", "", "The Wall", got)
	}
	// Whitespace-only inputs normalize to empty → no match.
	if got := Similarity("   ", "The Wall"); got != 0.0 {
		t.Errorf("Similarity(%q, %q) = %v, want 0.0", "   ", "The Wall", got)
	}
}

func TestIsMatch(t *testing.T) {
	const threshold = 0.85

	tests := []struct {
		name     string
		a        string
		b        string
		expected bool
	}{
		{name: "exact match is a match", a: "The Wall", b: "The Wall", expected: true},
		{name: "remastered variant is a match", a: "The Wall", b: "The Wall (Remastered)", expected: true},
		{name: "year variant is a match", a: "Abbey Road", b: "Abbey Road (2019)", expected: true},
		{name: "clearly different is not a match", a: "The Wall", b: "Random Noise", expected: false},
		{name: "empty vs non-empty is not a match", a: "", b: "The Wall", expected: false},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			if got := IsMatch(tt.a, tt.b, threshold); got != tt.expected {
				t.Errorf("IsMatch(%q, %q, %v) = %v, want %v", tt.a, tt.b, threshold, got, tt.expected)
			}
		})
	}
}

func TestFindMissingReleases(t *testing.T) {
	const threshold = 0.85

	artistA := "artist-a"
	artistB := "artist-b"

	tests := []struct {
		name    string
		local   []database.LocalAlbum
		external []database.ExternalRelease
		want    []string // RGIDs expected to be reported as missing
	}{
		{
			name:     "no local albums means all external are missing",
			local:    nil,
			external: []database.ExternalRelease{
				{RGID: "rg1", ArtistID: artistA, Title: "The Wall"},
				{RGID: "rg2", ArtistID: artistA, Title: "Animals"},
			},
			want: []string{"rg1", "rg2"},
		},
		{
			name: "exact local title is not missing",
			local: []database.LocalAlbum{
				{ID: "l1", ArtistID: artistA, Title: "The Wall"},
			},
			external: []database.ExternalRelease{
				{RGID: "rg1", ArtistID: artistA, Title: "The Wall"},
				{RGID: "rg2", ArtistID: artistA, Title: "Animals"},
			},
			want: []string{"rg2"},
		},
		{
			name: "fuzzy local title (remastered) is not missing",
			local: []database.LocalAlbum{
				{ID: "l1", ArtistID: artistA, Title: "The Wall (Remastered)"},
			},
			external: []database.ExternalRelease{
				{RGID: "rg1", ArtistID: artistA, Title: "The Wall"},
				{RGID: "rg2", ArtistID: artistA, Title: "Animals"},
			},
			want: []string{"rg2"},
		},
		{
			name: "ignored external is never reported",
			local: []database.LocalAlbum{
				{ID: "l1", ArtistID: artistA, Title: "The Wall"},
			},
			external: []database.ExternalRelease{
				{RGID: "rg1", ArtistID: artistA, Title: "The Wall"},
				{RGID: "rg2", ArtistID: artistA, Title: "Animals", IsIgnored: true},
			},
			want: []string{},
		},
		{
			name: "different artist id is not matched across artists",
			local: []database.LocalAlbum{
				{ID: "l1", ArtistID: artistA, Title: "The Wall"},
			},
			external: []database.ExternalRelease{
				{RGID: "rg1", ArtistID: artistB, Title: "The Wall"},
			},
			want: []string{"rg1"},
		},
		{
			name: "threshold boundary at 0.85",
			local: []database.LocalAlbum{
				{ID: "l1", ArtistID: artistA, Title: "The Wall Live"},
			},
			external: []database.ExternalRelease{
				{RGID: "rg1", ArtistID: artistA, Title: "The Wall"},
			},
			want: []string{"rg1"}, // "The Wall Live" vs "The Wall" is below 0.85
		},
		{
			name: "empty external list returns nothing",
			local: []database.LocalAlbum{
				{ID: "l1", ArtistID: artistA, Title: "The Wall"},
			},
			external: nil,
			want:     []string{},
		},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			got := FindMissingReleases(tt.local, tt.external, threshold)

			gotRGIDs := make([]string, 0, len(got))
			for _, m := range got {
				gotRGIDs = append(gotRGIDs, m.RGID)
			}

			if len(gotRGIDs) != len(tt.want) {
				t.Fatalf("FindMissingReleases() returned %v, want RGIDs %v", gotRGIDs, tt.want)
			}
			wantSet := make(map[string]struct{}, len(tt.want))
			for _, r := range tt.want {
				wantSet[r] = struct{}{}
			}
			for _, r := range gotRGIDs {
				if _, ok := wantSet[r]; !ok {
					t.Errorf("FindMissingReleases() returned unexpected RGID %q (got %v, want %v)", r, gotRGIDs, tt.want)
				}
			}
		})
	}
}

func TestFindMissingReleases_ThresholdBoundaryInclusive(t *testing.T) {
	// A title at exactly the threshold must NOT be reported as missing
	// (IsMatch uses >= threshold).
	local := []database.LocalAlbum{
		{ID: "l1", ArtistID: "a", Title: "The Wall Live"},
	}
	// Force a known score: "the wall" vs "the wall" would be 1.0; instead
	// use a release whose similarity is exactly 0.85 so the boundary is hit.
	// We assert behaviour via the documented contract using IsMatch, not a
	// brittle exact score here.
	external := []database.ExternalRelease{
		{RGID: "rg1", ArtistID: "a", Title: "The Wall"},
	}
	// With default threshold 0.85, "The Wall Live" does not match "The Wall";
	// at a low threshold it would. Confirms threshold is honoured.
	if len(FindMissingReleases(local, external, 0.85)) != 1 {
		t.Errorf("expected 1 missing at 0.85 threshold")
	}
}

func TestIsMatch_ThresholdBoundary(t *testing.T) {
	// A moderately different title should be a match at a low threshold but
	// not at a high one, confirming the boundary is inclusive (>=).
	a, b := "The Wall", "The Wall Live"
	low := IsMatch(a, b, 0.5)
	high := IsMatch(a, b, 0.99)
	if !low {
		t.Errorf("IsMatch(%q, %q, 0.5) = false, want true", a, b)
	}
	if high {
		t.Errorf("IsMatch(%q, %q, 0.99) = true, want false", a, b)
	}
}