Compare commits

..

2 Commits

Author SHA1 Message Date
22efd90f8c refactor: stream blob hash verification instead of buffering in memory
FetchAndDecryptBlob now returns io.ReadCloser with a hashVerifyReader
that computes the double-SHA-256 on-the-fly during reads. Hash is
verified on Close() after the stream is fully consumed. This avoids
loading entire blobs into memory, which could exceed available RAM.

Addresses review feedback on PR #39.
2026-02-20 02:29:19 -08:00
user
2bdbf38be6 fix: verify blob hash after download and decryption (closes #5)
Add double-SHA-256 hash verification of decrypted plaintext in
FetchAndDecryptBlob. This ensures blob integrity during restore
operations by comparing the computed hash against the expected
blob hash before returning data to the caller.

Includes test for both correct hash (passes) and mismatched hash
(returns error).
2026-02-20 02:26:15 -08:00
3 changed files with 171 additions and 15 deletions

View File

@ -0,0 +1,100 @@
package vaultik_test
import (
"bytes"
"context"
"crypto/sha256"
"encoding/hex"
"io"
"strings"
"testing"
"filippo.io/age"
"git.eeqj.de/sneak/vaultik/internal/blobgen"
"git.eeqj.de/sneak/vaultik/internal/vaultik"
)
// TestFetchAndDecryptBlobVerifiesHash verifies that FetchAndDecryptBlob checks
// the double-SHA-256 hash of the decrypted plaintext against the expected blob hash.
func TestFetchAndDecryptBlobVerifiesHash(t *testing.T) {
identity, err := age.GenerateX25519Identity()
if err != nil {
t.Fatalf("generating identity: %v", err)
}
// Create test data and encrypt it using blobgen.Writer
plaintext := []byte("hello world test data for blob hash verification")
var encBuf bytes.Buffer
writer, err := blobgen.NewWriter(&encBuf, 1, []string{identity.Recipient().String()})
if err != nil {
t.Fatalf("creating blobgen writer: %v", err)
}
if _, err := writer.Write(plaintext); err != nil {
t.Fatalf("writing plaintext: %v", err)
}
if err := writer.Close(); err != nil {
t.Fatalf("closing writer: %v", err)
}
encryptedData := encBuf.Bytes()
// Compute correct double-SHA-256 hash of the plaintext (matches blobgen.Writer.Sum256)
firstHash := sha256.Sum256(plaintext)
secondHash := sha256.Sum256(firstHash[:])
correctHash := hex.EncodeToString(secondHash[:])
// Verify our hash matches what blobgen.Writer produces
writerHash := hex.EncodeToString(writer.Sum256())
if correctHash != writerHash {
t.Fatalf("hash computation mismatch: manual=%s, writer=%s", correctHash, writerHash)
}
// Set up mock storage with the blob at the correct path
mockStorage := NewMockStorer()
blobPath := "blobs/" + correctHash[:2] + "/" + correctHash[2:4] + "/" + correctHash
mockStorage.mu.Lock()
mockStorage.data[blobPath] = encryptedData
mockStorage.mu.Unlock()
tv := vaultik.NewForTesting(mockStorage)
ctx := context.Background()
t.Run("correct hash succeeds", func(t *testing.T) {
rc, err := tv.FetchAndDecryptBlob(ctx, correctHash, int64(len(encryptedData)), identity)
if err != nil {
t.Fatalf("expected success, got error: %v", err)
}
data, err := io.ReadAll(rc)
if err != nil {
t.Fatalf("reading stream: %v", err)
}
if err := rc.Close(); err != nil {
t.Fatalf("close (hash verification) failed: %v", err)
}
if !bytes.Equal(data, plaintext) {
t.Fatalf("decrypted data mismatch: got %q, want %q", data, plaintext)
}
})
t.Run("wrong hash fails", func(t *testing.T) {
// Use a fake hash that doesn't match the actual plaintext
fakeHash := strings.Repeat("ab", 32) // 64 hex chars
fakePath := "blobs/" + fakeHash[:2] + "/" + fakeHash[2:4] + "/" + fakeHash
mockStorage.mu.Lock()
mockStorage.data[fakePath] = encryptedData
mockStorage.mu.Unlock()
rc, err := tv.FetchAndDecryptBlob(ctx, fakeHash, int64(len(encryptedData)), identity)
if err != nil {
t.Fatalf("unexpected error opening stream: %v", err)
}
// Read all data — hash is verified on Close
_, _ = io.ReadAll(rc)
err = rc.Close()
if err == nil {
t.Fatal("expected error for mismatched hash, got nil")
}
if !strings.Contains(err.Error(), "hash mismatch") {
t.Fatalf("expected hash mismatch error, got: %v", err)
}
})
}

View File

@ -2,38 +2,82 @@ package vaultik
import ( import (
"context" "context"
"crypto/sha256"
"encoding/hex"
"fmt" "fmt"
"hash"
"io" "io"
"filippo.io/age" "filippo.io/age"
"git.eeqj.de/sneak/vaultik/internal/blobgen" "git.eeqj.de/sneak/vaultik/internal/blobgen"
) )
// FetchAndDecryptBlobResult holds the result of fetching and decrypting a blob. // hashVerifyReader wraps a reader and computes a double-SHA-256 hash of all
type FetchAndDecryptBlobResult struct { // data read through it. The hash is verified against the expected blob hash
Data []byte // when Close is called. This allows streaming blob verification without
// buffering the entire blob in memory.
type hashVerifyReader struct {
reader io.ReadCloser // underlying decrypted blob reader
fetcher io.ReadCloser // raw fetched stream (closed on Close)
hasher hash.Hash // running SHA-256 of plaintext
blobHash string // expected double-SHA-256 hex
done bool // EOF reached
} }
// FetchAndDecryptBlob downloads a blob, decrypts it, and returns the plaintext data. func (h *hashVerifyReader) Read(p []byte) (int, error) {
func (v *Vaultik) FetchAndDecryptBlob(ctx context.Context, blobHash string, expectedSize int64, identity age.Identity) (*FetchAndDecryptBlobResult, error) { n, err := h.reader.Read(p)
if n > 0 {
h.hasher.Write(p[:n])
}
if err == io.EOF {
h.done = true
}
return n, err
}
// Close verifies the hash (if the stream was fully read) and closes underlying readers.
func (h *hashVerifyReader) Close() error {
readerErr := h.reader.Close()
fetcherErr := h.fetcher.Close()
if h.done {
firstHash := h.hasher.Sum(nil)
secondHasher := sha256.New()
secondHasher.Write(firstHash)
actualHashHex := hex.EncodeToString(secondHasher.Sum(nil))
if actualHashHex != h.blobHash {
return fmt.Errorf("blob hash mismatch: expected %s, got %s", h.blobHash[:16], actualHashHex[:16])
}
}
if readerErr != nil {
return readerErr
}
return fetcherErr
}
// FetchAndDecryptBlob downloads a blob, decrypts and decompresses it, and
// returns a streaming reader that computes the double-SHA-256 hash on the fly.
// The hash is verified when the returned reader is closed (after fully reading).
// This avoids buffering the entire blob in memory.
func (v *Vaultik) FetchAndDecryptBlob(ctx context.Context, blobHash string, expectedSize int64, identity age.Identity) (io.ReadCloser, error) {
rc, _, err := v.FetchBlob(ctx, blobHash, expectedSize) rc, _, err := v.FetchBlob(ctx, blobHash, expectedSize)
if err != nil { if err != nil {
return nil, err return nil, err
} }
defer func() { _ = rc.Close() }()
reader, err := blobgen.NewReader(rc, identity) reader, err := blobgen.NewReader(rc, identity)
if err != nil { if err != nil {
_ = rc.Close()
return nil, fmt.Errorf("creating blob reader: %w", err) return nil, fmt.Errorf("creating blob reader: %w", err)
} }
defer func() { _ = reader.Close() }()
data, err := io.ReadAll(reader) return &hashVerifyReader{
if err != nil { reader: reader,
return nil, fmt.Errorf("reading blob data: %w", err) fetcher: rc,
} hasher: sha256.New(),
blobHash: blobHash,
return &FetchAndDecryptBlobResult{Data: data}, nil }, nil
} }
// FetchBlob downloads a blob and returns a reader for the encrypted data. // FetchBlob downloads a blob and returns a reader for the encrypted data.

View File

@ -494,11 +494,23 @@ func (v *Vaultik) restoreRegularFile(
// downloadBlob downloads and decrypts a blob // downloadBlob downloads and decrypts a blob
func (v *Vaultik) downloadBlob(ctx context.Context, blobHash string, expectedSize int64, identity age.Identity) ([]byte, error) { func (v *Vaultik) downloadBlob(ctx context.Context, blobHash string, expectedSize int64, identity age.Identity) ([]byte, error) {
result, err := v.FetchAndDecryptBlob(ctx, blobHash, expectedSize, identity) rc, err := v.FetchAndDecryptBlob(ctx, blobHash, expectedSize, identity)
if err != nil { if err != nil {
return nil, err return nil, err
} }
return result.Data, nil
data, err := io.ReadAll(rc)
if err != nil {
_ = rc.Close()
return nil, fmt.Errorf("reading blob data: %w", err)
}
// Close triggers hash verification
if err := rc.Close(); err != nil {
return nil, err
}
return data, nil
} }
// verifyRestoredFiles verifies that all restored files match their expected chunk hashes // verifyRestoredFiles verifies that all restored files match their expected chunk hashes