labmcp/internal/nixos/indexer_test.go

package nixos

import (
	"context"
	"os/exec"
	"testing"
	"time"

	"git.t-juice.club/torjus/labmcp/internal/database"
)

// TestNixpkgsRevision is the revision from flake.lock used for testing.
const TestNixpkgsRevision = "e6eae2ee2110f3d31110d5c222cd395303343b08"

// TestValidateRevision tests the revision validation function.
func TestValidateRevision(t *testing.T) {
	tests := []struct {
		name     string
		revision string
		wantErr  bool
	}{
		// Valid cases
		{"valid git hash", "e6eae2ee2110f3d31110d5c222cd395303343b08", false},
		{"valid short hash", "e6eae2e", false},
		{"valid channel name", "nixos-unstable", false},
		{"valid channel with version", "nixos-24.11", false},
		{"valid underscore", "some_branch", false},
		{"valid mixed", "release-24.05_beta", false},

		// Invalid cases - injection attempts
		{"injection semicolon", "foo; rm -rf /", true},
		{"injection quotes", `"; builtins.readFile /etc/passwd; "`, true},
		{"injection backticks", "foo`whoami`", true},
		{"injection dollar", "foo$(whoami)", true},
		{"injection newline", "foo\nbar", true},
		{"injection space", "foo bar", true},
		{"injection slash", "foo/bar", true},
		{"injection backslash", "foo\\bar", true},
		{"injection pipe", "foo|bar", true},
		{"injection ampersand", "foo&bar", true},
		{"injection redirect", "foo>bar", true},
		{"injection less than", "foo<bar", true},
		{"injection curly braces", "foo{bar}", true},
		{"injection parens", "foo(bar)", true},
		{"injection brackets", "foo[bar]", true},

		// Edge cases
		{"empty string", "", true},
		{"too long", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa", true},
		{"just dots", "...", false}, // dots are allowed, path traversal is handled elsewhere
		{"single char", "a", false},
		{"max length 64", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa", false},
		{"65 chars", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa", true},
	}

	for _, tt := range tests {
		t.Run(tt.name, func(t *testing.T) {
			err := ValidateRevision(tt.revision)
			if (err != nil) != tt.wantErr {
				t.Errorf("ValidateRevision(%q) error = %v, wantErr %v", tt.revision, err, tt.wantErr)
			}
		})
	}
}

// BenchmarkIndexRevision benchmarks indexing a full nixpkgs revision.
// This is a slow benchmark that requires nix to be installed.
// Run with: go test -bench=BenchmarkIndexRevision -benchtime=1x -timeout=30m ./internal/nixos/...
func BenchmarkIndexRevision(b *testing.B) {
	// Check if nix-build is available
	if _, err := exec.LookPath("nix-build"); err != nil {
		b.Skip("nix-build not found, skipping indexer benchmark")
	}

	// Use in-memory SQLite for the benchmark
	store, err := database.NewSQLiteStore(":memory:")
	if err != nil {
		b.Fatalf("Failed to create store: %v", err)
	}
	defer store.Close()

	ctx := context.Background()
	if err := store.Initialize(ctx); err != nil {
		b.Fatalf("Failed to initialize store: %v", err)
	}

	indexer := NewIndexer(store)

	b.ResetTimer()
	for i := 0; i < b.N; i++ {
		// Delete any existing revision first (for repeated runs)
		if rev, _ := store.GetRevision(ctx, TestNixpkgsRevision); rev != nil {
			store.DeleteRevision(ctx, rev.ID)
		}

		result, err := indexer.IndexRevision(ctx, TestNixpkgsRevision)
		if err != nil {
			b.Fatalf("IndexRevision failed: %v", err)
		}

		b.ReportMetric(float64(result.OptionCount), "options")
		if dur, ok := result.Duration.(time.Duration); ok {
			b.ReportMetric(float64(dur.Milliseconds()), "ms")
		}
	}
}

// BenchmarkIndexRevisionWithFiles benchmarks indexing with file content storage.
// This downloads the full nixpkgs tarball and stores allowed file types.
// Run with: go test -bench=BenchmarkIndexRevisionWithFiles -benchtime=1x -timeout=60m ./internal/nixos/...
func BenchmarkIndexRevisionWithFiles(b *testing.B) {
	// Check if nix-build is available
	if _, err := exec.LookPath("nix-build"); err != nil {
		b.Skip("nix-build not found, skipping indexer benchmark")
	}

	// Use in-memory SQLite for the benchmark
	store, err := database.NewSQLiteStore(":memory:")
	if err != nil {
		b.Fatalf("Failed to create store: %v", err)
	}
	defer store.Close()

	ctx := context.Background()
	if err := store.Initialize(ctx); err != nil {
		b.Fatalf("Failed to initialize store: %v", err)
	}

	indexer := NewIndexer(store)

	b.ResetTimer()
	for i := 0; i < b.N; i++ {
		// Delete any existing revision first
		if rev, _ := store.GetRevision(ctx, TestNixpkgsRevision); rev != nil {
			store.DeleteRevision(ctx, rev.ID)
		}

		result, err := indexer.IndexRevision(ctx, TestNixpkgsRevision)
		if err != nil {
			b.Fatalf("IndexRevision failed: %v", err)
		}

		fileStart := time.Now()
		fileCount, err := indexer.IndexFiles(ctx, result.Revision.ID, TestNixpkgsRevision)
		if err != nil {
			b.Fatalf("IndexFiles failed: %v", err)
		}
		fileDuration := time.Since(fileStart)

		b.ReportMetric(float64(result.OptionCount), "options")
		if dur, ok := result.Duration.(time.Duration); ok {
			b.ReportMetric(float64(dur.Milliseconds()), "options_ms")
		}
		b.ReportMetric(float64(fileCount), "files")
		b.ReportMetric(float64(fileDuration.Milliseconds()), "files_ms")
	}
}

// BenchmarkIndexFilesOnly benchmarks only the file indexing phase.
// Assumes options are already indexed. Useful for measuring file indexing in isolation.
// Run with: go test -bench=BenchmarkIndexFilesOnly -benchtime=1x -timeout=60m ./internal/nixos/...
func BenchmarkIndexFilesOnly(b *testing.B) {
	if _, err := exec.LookPath("nix-build"); err != nil {
		b.Skip("nix-build not found, skipping indexer benchmark")
	}

	store, err := database.NewSQLiteStore(":memory:")
	if err != nil {
		b.Fatalf("Failed to create store: %v", err)
	}
	defer store.Close()

	ctx := context.Background()
	if err := store.Initialize(ctx); err != nil {
		b.Fatalf("Failed to initialize store: %v", err)
	}

	indexer := NewIndexer(store)

	// Index options first (outside of benchmark timing)
	result, err := indexer.IndexRevision(ctx, TestNixpkgsRevision)
	if err != nil {
		b.Fatalf("IndexRevision failed: %v", err)
	}
	b.Logf("Pre-indexed %d options", result.OptionCount)

	b.ResetTimer()
	for i := 0; i < b.N; i++ {
		fileCount, err := indexer.IndexFiles(ctx, result.Revision.ID, TestNixpkgsRevision)
		if err != nil {
			b.Fatalf("IndexFiles failed: %v", err)
		}

		b.ReportMetric(float64(fileCount), "files")
	}
}

// TestIndexRevision is an integration test for the indexer.
// Run with: go test -run=TestIndexRevision -timeout=30m ./internal/nixos/...
func TestIndexRevision(t *testing.T) {
	if testing.Short() {
		t.Skip("Skipping integration test in short mode")
	}

	// Check if nix-build is available
	if _, err := exec.LookPath("nix-build"); err != nil {
		t.Skip("nix-build not found, skipping indexer test")
	}

	store, err := database.NewSQLiteStore(":memory:")
	if err != nil {
		t.Fatalf("Failed to create store: %v", err)
	}
	defer store.Close()

	ctx := context.Background()
	if err := store.Initialize(ctx); err != nil {
		t.Fatalf("Failed to initialize store: %v", err)
	}

	indexer := NewIndexer(store)

	t.Logf("Indexing nixpkgs revision %s...", TestNixpkgsRevision[:12])
	result, err := indexer.IndexRevision(ctx, TestNixpkgsRevision)
	if err != nil {
		t.Fatalf("IndexRevision failed: %v", err)
	}

	t.Logf("Indexed %d options in %s", result.OptionCount, result.Duration)

	// Verify we got a reasonable number of options (NixOS has thousands)
	if result.OptionCount < 1000 {
		t.Errorf("Expected at least 1000 options, got %d", result.OptionCount)
	}

	// Verify revision was stored
	rev, err := store.GetRevision(ctx, TestNixpkgsRevision)
	if err != nil {
		t.Fatalf("GetRevision failed: %v", err)
	}
	if rev == nil {
		t.Fatal("Revision not found after indexing")
	}
	if rev.OptionCount != result.OptionCount {
		t.Errorf("Stored option count %d != result count %d", rev.OptionCount, result.OptionCount)
	}

	// Test searching
	options, err := store.SearchOptions(ctx, rev.ID, "nginx", database.SearchFilters{Limit: 10})
	if err != nil {
		t.Fatalf("SearchOptions failed: %v", err)
	}
	if len(options) == 0 {
		t.Error("Expected to find nginx options")
	}
	t.Logf("Found %d nginx options", len(options))

	// Test getting a specific option
	opt, err := store.GetOption(ctx, rev.ID, "services.nginx.enable")
	if err != nil {
		t.Fatalf("GetOption failed: %v", err)
	}
	if opt == nil {
		t.Error("services.nginx.enable not found")
	} else {
		t.Logf("services.nginx.enable: type=%s", opt.Type)
		if opt.Type != "boolean" {
			t.Errorf("Expected type 'boolean', got %q", opt.Type)
		}
	}

	// Test getting children
	children, err := store.GetChildren(ctx, rev.ID, "services.nginx")
	if err != nil {
		t.Fatalf("GetChildren failed: %v", err)
	}
	if len(children) == 0 {
		t.Error("Expected services.nginx to have children")
	}
	t.Logf("services.nginx has %d direct children", len(children))
}