sow-tools/internal/depot/integration_test.go

package depot

import (
	"bytes"
	"fmt"
	"net/http/httptest"
	"os"
	"path/filepath"
	"strings"
	"testing"
)

// TestIntegrationPushStatusPull drives status/push/pull end-to-end against a
// fake Bunny backend (real bunny HTTP path, in-memory blob store) since the
// command surface has no local->local push (push --target must be bunny).
func TestIntegrationPushStatusPull(t *testing.T) {
	manifestsDir := t.TempDir()
	sourceDir := t.TempDir()

	blobs := map[string]string{
		shaOf("blob-one"):   "blob-one",
		shaOf("blob-two"):   "blob-two",
		shaOf("blob-three"): "blob-three",
	}
	var manifest strings.Builder
	manifest.WriteString("assets:\n")
	for sha, content := range blobs {
		manifest.WriteString(fmt.Sprintf("  - path: %s\n    sha256: %s\n    size: %d\n", sha, sha, len(content)))
		blobPath := filepath.Join(sourceDir, BlobKey(sha))
		if err := os.MkdirAll(filepath.Dir(blobPath), 0755); err != nil {
			t.Fatal(err)
		}
		if err := os.WriteFile(blobPath, []byte(content), 0644); err != nil {
			t.Fatal(err)
		}
	}
	if err := os.WriteFile(filepath.Join(manifestsDir, "manifest.yml"), []byte(manifest.String()), 0644); err != nil {
		t.Fatal(err)
	}

	f := newFakeBunny()
	// Pre-seed one of the three blobs on the target so status starts with drift=2.
	var oneSHA string
	for sha := range blobs {
		oneSHA = sha
		break
	}
	f.blobs[oneSHA] = []byte(blobs[oneSHA])
	srv := httptest.NewServer(f.handler())
	defer srv.Close()

	env := map[string]string{
		"BUNNY_STORAGE_HOST":          hostPort(srv),
		"BUNNY_STORAGE_READ_PASSWORD": "readkey",
		"BUNNY_STORAGE_PASSWORD":      "writekey",
		"CDN_UNREACHABLE":             "1", // unused; CDNBase defaults to unreachable in test cfg path via storageURL fallback
	}
	getenv := testGetenv(env)

	// status: expect exit 1, missing=2
	var out, errb bytes.Buffer
	code := Run([]string{"status", "--manifests", manifestsDir, "--target", "bunny"}, &out, &errb, getenv)
	if code != 1 {
		t.Fatalf("status: expected 1, got %d (stdout=%s stderr=%s)", code, out.String(), errb.String())
	}
	if !bytesContains(out.String(), "missing=2") {
		t.Fatalf("status: expected missing=2, got %s", out.String())
	}

	// push: uploads the 2 missing blobs
	out.Reset()
	errb.Reset()
	code = Run([]string{"push", "--manifests", manifestsDir, "--source", sourceDir, "--target", "bunny"}, &out, &errb, getenv)
	if code != 0 {
		t.Fatalf("push: expected 0, got %d (stdout=%s stderr=%s)", code, out.String(), errb.String())
	}
	if !bytesContains(out.String(), "uploaded=2 failed=0") {
		t.Fatalf("push: expected uploaded=2 failed=0, got %s", out.String())
	}

	putCountAfterFirstPush := 0
	for _, r := range f.requestsSnapshot() {
		if r.Method == "PUT" {
			putCountAfterFirstPush++
		}
	}
	if putCountAfterFirstPush != 2 {
		t.Fatalf("expected 2 PUTs after first push, got %d", putCountAfterFirstPush)
	}

	// status: now clean
	out.Reset()
	errb.Reset()
	code = Run([]string{"status", "--manifests", manifestsDir, "--target", "bunny"}, &out, &errb, getenv)
	if code != 0 {
		t.Fatalf("status after push: expected 0, got %d (stdout=%s stderr=%s)", code, out.String(), errb.String())
	}

	// push again: idempotent, no new PUTs, uploaded=0
	out.Reset()
	errb.Reset()
	code = Run([]string{"push", "--manifests", manifestsDir, "--source", sourceDir, "--target", "bunny"}, &out, &errb, getenv)
	if code != 0 {
		t.Fatalf("second push: expected 0, got %d (stdout=%s stderr=%s)", code, out.String(), errb.String())
	}
	if !bytesContains(out.String(), "uploaded=0 failed=0") {
		t.Fatalf("second push: expected uploaded=0 failed=0, got %s", out.String())
	}
	putCountAfterSecondPush := 0
	for _, r := range f.requestsSnapshot() {
		if r.Method == "PUT" {
			putCountAfterSecondPush++
		}
	}
	if putCountAfterSecondPush != putCountAfterFirstPush {
		t.Fatalf("expected no additional PUTs on idempotent push, before=%d after=%d", putCountAfterFirstPush, putCountAfterSecondPush)
	}

	// pull into a dest dir: one pre-populated correct file (skipped, no GET),
	// one pre-populated corrupt file (re-downloaded), one absent (downloaded).
	destDir := t.TempDir()
	shas := make([]string, 0, len(blobs))
	for sha := range blobs {
		shas = append(shas, sha)
	}
	correctSHA := shas[0]
	corruptSHA := shas[1]
	// absentSHA := shas[2] // left absent on purpose

	correctPath := filepath.Join(destDir, BlobKey(correctSHA))
	if err := os.MkdirAll(filepath.Dir(correctPath), 0755); err != nil {
		t.Fatal(err)
	}
	if err := os.WriteFile(correctPath, []byte(blobs[correctSHA]), 0644); err != nil {
		t.Fatal(err)
	}

	corruptPath := filepath.Join(destDir, BlobKey(corruptSHA))
	if err := os.MkdirAll(filepath.Dir(corruptPath), 0755); err != nil {
		t.Fatal(err)
	}
	if err := os.WriteFile(corruptPath, []byte("corrupted-content"), 0644); err != nil {
		t.Fatal(err)
	}

	getCountBeforePull := 0
	for _, r := range f.requestsSnapshot() {
		if r.Method == "GET" && !strings.Contains(r.Range, "0-0") {
			getCountBeforePull++
		}
	}

	out.Reset()
	errb.Reset()
	code = Run([]string{"pull", "--manifests", manifestsDir, "--dest", destDir, "--target", "bunny"}, &out, &errb, getenv)
	if code != 0 {
		t.Fatalf("pull: expected 0, got %d (stdout=%s stderr=%s)", code, out.String(), errb.String())
	}
	if !bytesContains(out.String(), "pulled=2 present=1") {
		t.Fatalf("pull: expected pulled=2 present=1, got %s", out.String())
	}

	// verify the correct pre-existing file was never re-fetched with a full GET.
	fullGETsForCorrect := 0
	for _, r := range f.requestsSnapshot() {
		if r.Method == "GET" && strings.HasSuffix(r.Path, correctSHA) && r.Range != "bytes=0-0" {
			fullGETsForCorrect++
		}
	}
	if fullGETsForCorrect != 0 {
		t.Fatalf("expected no full GET for already-correct blob, got %d", fullGETsForCorrect)
	}

	// all three blobs should now be present and correct in destDir.
	for sha, content := range blobs {
		got, err := os.ReadFile(filepath.Join(destDir, BlobKey(sha)))
		if err != nil {
			t.Fatalf("dest blob %s: %v", sha, err)
		}
		if string(got) != content {
			t.Fatalf("dest blob %s: expected %q, got %q", sha, content, string(got))
		}
	}
}