VAULT-20405 chunk decompression to prevent loading full decompressed …

…data into memory at once (#26464)

* VAULT-20405 chunk decompression to prevent loading full decompressed data into memory at once

* Add changelog

changelog/26464.txt

@@ -0,0 +1,3 @@
+```release-note:improvement
+sdk/decompression: DecompressWithCanary will now chunk the decompression in memory to prevent loading it all at once.
+```

sdk/helper/compressutil/compress.go

@@ -11,7 +11,6 @@ import (
 	"io"
 
 	"github.com/golang/snappy"
-	"github.com/hashicorp/errwrap"
 	"github.com/pierrec/lz4"
 )
 
@@ -34,7 +33,7 @@ const (
 	CompressionCanaryLZ4 byte = '4'
 )
 
-// SnappyReadCloser embeds the snappy reader which implements the io.Reader
+// CompressUtilReadCloser embeds the snappy reader which implements the io.Reader
 // interface. The decompress procedure in this utility expects an
 // io.ReadCloser. This type implements the io.Closer interface to retain the
 // generic way of decompression.
@@ -98,7 +97,7 @@ func Compress(data []byte, config *CompressionConfig) ([]byte, error) {
 			// These are valid compression levels
 		default:
 			// If compression level is set to NoCompression or to
-			// any invalid value, fallback to Defaultcompression
+			// any invalid value, fallback to DefaultCompression
 			config.GzipCompressionLevel = gzip.DefaultCompression
 		}
 		writer, err = gzip.NewWriterLevel(&buf, config.GzipCompressionLevel)
@@ -116,7 +115,7 @@ func Compress(data []byte, config *CompressionConfig) ([]byte, error) {
 	}
 
 	if err != nil {
-		return nil, errwrap.Wrapf("failed to create a compression writer: {{err}}", err)
+		return nil, fmt.Errorf("failed to create a compression writer: %w", err)
 	}
 
 	if writer == nil {
@@ -126,7 +125,7 @@ func Compress(data []byte, config *CompressionConfig) ([]byte, error) {
 	// Compress the input and place it in the same buffer containing the
 	// canary byte.
 	if _, err = writer.Write(data); err != nil {
-		return nil, errwrap.Wrapf("failed to compress input data: err: {{err}}", err)
+		return nil, fmt.Errorf("failed to compress input data: err: %w", err)
 	}
 
 	// Close the io.WriteCloser
@@ -206,7 +205,7 @@ func DecompressWithCanary(data []byte) ([]byte, string, bool, error) {
 		return nil, "", true, nil
 	}
 	if err != nil {
-		return nil, "", false, errwrap.Wrapf("failed to create a compression reader: {{err}}", err)
+		return nil, "", false, fmt.Errorf("failed to create a compression reader: %w", err)
 	}
 	if reader == nil {
 		return nil, "", false, fmt.Errorf("failed to create a compression reader")
@@ -217,8 +216,18 @@ func DecompressWithCanary(data []byte) ([]byte, string, bool, error) {
 
 	// Read all the compressed data into a buffer
 	var buf bytes.Buffer
-	if _, err = io.Copy(&buf, reader); err != nil {
-		return nil, "", false, err
+
+	// Read the compressed data into a buffer, but do so
+	// slowly to prevent reading all the data into memory
+	// at once (protecting against e.g. zip bombs).
+	for {
+		_, err := io.CopyN(&buf, reader, 1024)
+		if err != nil {
+			if err == io.EOF {
+				break
+			}
+			return nil, "", false, err
+		}
 	}
 
 	return buf.Bytes(), compressionType, false, nil

sdk/helper/compressutil/compress_test.go

@@ -116,3 +116,40 @@ func TestCompressUtil_InvalidConfigurations(t *testing.T) {
 		t.Fatal("expected an error")
 	}
 }
+
+// TestDecompressWithCanaryLargeInput tests that DecompressWithCanary works
+// as expected even with large values.
+func TestDecompressWithCanaryLargeInput(t *testing.T) {
+	t.Parallel()
+
+	inputJSON := `{"sample":"data`
+	for i := 0; i < 100000; i++ {
+		inputJSON += " and data"
+	}
+	inputJSON += `"}`
+	inputJSONBytes := []byte(inputJSON)
+
+	compressedJSONBytes, err := Compress(inputJSONBytes, &CompressionConfig{Type: CompressionTypeGzip, GzipCompressionLevel: gzip.BestCompression})
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	decompressedJSONBytes, wasNotCompressed, err := Decompress(compressedJSONBytes)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	// Check if the input for decompress was not compressed in the first place
+	if wasNotCompressed {
+		t.Fatalf("bytes were not compressed as expected")
+	}
+
+	if len(decompressedJSONBytes) == 0 {
+		t.Fatalf("bytes were not compressed as expected")
+	}
+
+	// Compare the value after decompression
+	if !bytes.Equal(inputJSONBytes, decompressedJSONBytes) {
+		t.Fatalf("decompressed value differs: decompressed value;\nexpected: %q\nactual: %q", string(inputJSONBytes), string(decompressedJSONBytes))
+	}
+}