fix: use correct hashes for empty files (#4620)

Signed-off-by: Paweł Pałucha <pawel.palucha@chainguard.dev>

internal/file/digest.go

@@ -24,6 +24,16 @@ func supportedHashAlgorithms() []crypto.Hash {
 	}
 }
 
+// Pre-computed hash values for empty files
+const (
+	emptyMD5    = "d41d8cd98f00b204e9800998ecf8427e"
+	emptySHA1   = "da39a3ee5e6b4b0d3255bfef95601890afd80709"
+	emptySHA224 = "d14a028c2a3a2bc9476102bb288234c415a2b01f828ea62ac5b3e42f"
+	emptySHA256 = "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
+	emptySHA384 = "38b060a751ac96384cd9327eb1b1e36a21fdb71114be07434c0cc7bf63f6e1da274edebfe76f65fbd51ad2f14898b95b"
+	emptySHA512 = "cf83e1357eefb8bdf1542850d66d8007d620e4050b5715dc83f4a921d36ce9ce47d0d13c5d85f2b0ff8318d2877eec2f63b931bd47417a81a538327af927da3e"
+)
+
 func NewDigestsFromFile(ctx context.Context, closer io.ReadCloser, hashes []crypto.Hash) ([]file.Digest, error) {
 	hashes = NormalizeHashes(hashes)
 	// create a set of hasher objects tied together with a single writer to feed content into
@@ -39,14 +49,40 @@ func NewDigestsFromFile(ctx context.Context, closer io.ReadCloser, hashes []cryp
 		return nil, err
 	}
 
+	// For empty files, use pre-computed constants for better performance
 	if size == 0 {
-		return make([]file.Digest, 0), nil
+		result := make([]file.Digest, len(hashes))
+		for idx, hashObj := range hashes {
+			var value string
+			switch hashObj {
+			case crypto.MD5:
+				value = emptyMD5
+			case crypto.SHA1:
+				value = emptySHA1
+			case crypto.SHA224:
+				value = emptySHA224
+			case crypto.SHA256:
+				value = emptySHA256
+			case crypto.SHA384:
+				value = emptySHA384
+			case crypto.SHA512:
+				value = emptySHA512
+			default:
+				// Fallback to calculated hash for unsupported algorithms
+				value = fmt.Sprintf("%+x", hashers[idx].Sum(nil))
+			}
+			result[idx] = file.Digest{
+				Algorithm: CleanDigestAlgorithmName(hashObj.String()),
+				Value:     value,
+			}
+		}
+		return result, nil
 	}
 
 	result := make([]file.Digest, len(hashes))
-	// only capture digests when there is content. It is important to do this based on SIZE and not
+	// Capture digests for all files with content. It is important to base this on actual
-	// FILE TYPE. The reasoning is that it is possible for a tar to be crafted with a header-only
+	// content SIZE rather than FILE TYPE, as it is possible for a tar to be crafted with
-	// file type but a body is still allowed.
+	// a header-only file type but a body is still allowed.
 	for idx, hasher := range hashers {
 		result[idx] = file.Digest{
 			Algorithm: CleanDigestAlgorithmName(hashes[idx].String()),

internal/file/digest_test.go

@@ -72,6 +72,37 @@ func TestNewDigestsFromFile(t *testing.T) {
 				},
 			},
 		},
+		{
+			name:    "empty file has valid checksums",
+			fixture: "test-fixtures/empty.txt",
+			hashes:  supportedHashAlgorithms(),
+			want: []file.Digest{
+				{
+					Algorithm: "md5",
+					Value:     "d41d8cd98f00b204e9800998ecf8427e",
+				},
+				{
+					Algorithm: "sha1",
+					Value:     "da39a3ee5e6b4b0d3255bfef95601890afd80709",
+				},
+				{
+					Algorithm: "sha224",
+					Value:     "d14a028c2a3a2bc9476102bb288234c415a2b01f828ea62ac5b3e42f",
+				},
+				{
+					Algorithm: "sha256",
+					Value:     "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855",
+				},
+				{
+					Algorithm: "sha384",
+					Value:     "38b060a751ac96384cd9327eb1b1e36a21fdb71114be07434c0cc7bf63f6e1da274edebfe76f65fbd51ad2f14898b95b",
+				},
+				{
+					Algorithm: "sha512",
+					Value:     "cf83e1357eefb8bdf1542850d66d8007d620e4050b5715dc83f4a921d36ce9ce47d0d13c5d85f2b0ff8318d2877eec2f63b931bd47417a81a538327af927da3e",
+				},
+			},
+		},
 	}
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {

syft/file/cataloger/filedigest/cataloger_test.go

@@ -34,12 +34,7 @@ func testDigests(t testing.TB, root string, files []string, hashes ...crypto.Has
 			t.Fatalf("could not read %q : %+v", f, err)
 		}
 
-		if len(b) == 0 {
+		// Calculate digests for all files, including empty files
-			// we don't keep digests for empty files
-			digests[file.NewLocation(f).Coordinates] = []file.Digest{}
-			continue
-		}
-
 		for _, hash := range hashes {
 			h := hash.New()
 			h.Write(b)
@@ -65,13 +60,13 @@ func TestDigestsCataloger(t *testing.T) {
 			name:     "md5",
 			digests:  []crypto.Hash{crypto.MD5},
 			files:    []string{"test-fixtures/last/empty/empty", "test-fixtures/last/path.txt"},
-			expected: testDigests(t, "test-fixtures/last", []string{"path.txt"}, crypto.MD5),
+			expected: testDigests(t, "test-fixtures/last", []string{"empty/empty", "path.txt"}, crypto.MD5),
 		},
 		{
 			name:     "md5-sha1-sha256",
 			digests:  []crypto.Hash{crypto.MD5, crypto.SHA1, crypto.SHA256},
 			files:    []string{"test-fixtures/last/empty/empty", "test-fixtures/last/path.txt"},
-			expected: testDigests(t, "test-fixtures/last", []string{"path.txt"}, crypto.MD5, crypto.SHA1, crypto.SHA256),
+			expected: testDigests(t, "test-fixtures/last", []string{"empty/empty", "path.txt"}, crypto.MD5, crypto.SHA1, crypto.SHA256),
 		},
 	}