hashing: move to repository package

internal/repository/hashing/reader.go

@@ -0,0 +1,28 @@
+package hashing
+
+import (
+	"hash"
+	"io"
+)
+
+// A Reader hashes all data read from the underlying reader.
+type Reader struct {
+	r io.Reader
+	h hash.Hash
+}
+
+// NewReader returns a new Reader that uses the hash h.
+func NewReader(r io.Reader, h hash.Hash) *Reader {
+	return &Reader{r: r, h: h}
+}
+
+func (h *Reader) Read(p []byte) (int, error) {
+	n, err := h.r.Read(p)
+	_, _ = h.h.Write(p[:n]) // Never returns an error.
+	return n, err
+}
+
+// Sum returns the hash of the data read so far.
+func (h *Reader) Sum(d []byte) []byte {
+	return h.h.Sum(d)
+}

internal/repository/hashing/reader_test.go

@@ -0,0 +1,72 @@
+package hashing
+
+import (
+	"bytes"
+	"crypto/rand"
+	"crypto/sha256"
+	"io"
+	"testing"
+)
+
+func TestReader(t *testing.T) {
+	tests := []int{5, 23, 2<<18 + 23, 1 << 20}
+
+	for _, size := range tests {
+		data := make([]byte, size)
+		_, err := io.ReadFull(rand.Reader, data)
+		if err != nil {
+			t.Fatalf("ReadFull: %v", err)
+		}
+
+		expectedHash := sha256.Sum256(data)
+
+		rd := NewReader(bytes.NewReader(data), sha256.New())
+		n, err := io.Copy(io.Discard, rd)
+		if err != nil {
+			t.Fatal(err)
+		}
+
+		if n != int64(size) {
+			t.Errorf("Reader: invalid number of bytes written: got %d, expected %d",
+				n, size)
+		}
+
+		resultingHash := rd.Sum(nil)
+
+		if !bytes.Equal(expectedHash[:], resultingHash) {
+			t.Errorf("Reader: hashes do not match: expected %02x, got %02x",
+				expectedHash, resultingHash)
+		}
+	}
+}
+
+func BenchmarkReader(b *testing.B) {
+	buf := make([]byte, 1<<22)
+	_, err := io.ReadFull(rand.Reader, buf)
+	if err != nil {
+		b.Fatal(err)
+	}
+
+	expectedHash := sha256.Sum256(buf)
+
+	b.SetBytes(int64(len(buf)))
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		rd := NewReader(bytes.NewReader(buf), sha256.New())
+		n, err := io.Copy(io.Discard, rd)
+		if err != nil {
+			b.Fatal(err)
+		}
+
+		if n != int64(len(buf)) {
+			b.Errorf("Reader: invalid number of bytes written: got %d, expected %d",
+				n, len(buf))
+		}
+
+		resultingHash := rd.Sum(nil)
+		if !bytes.Equal(expectedHash[:], resultingHash) {
+			b.Errorf("Reader: hashes do not match: expected %02x, got %02x",
+				expectedHash, resultingHash)
+		}
+	}
+}

internal/repository/hashing/writer.go

@@ -0,0 +1,40 @@
+package hashing
+
+import (
+	"hash"
+	"io"
+)
+
+// Writer transparently hashes all data while writing it to the underlying writer.
+type Writer struct {
+	w io.Writer
+	h hash.Hash
+}
+
+// NewWriter wraps the writer w and feeds all data written to the hash h.
+func NewWriter(w io.Writer, h hash.Hash) *Writer {
+	return &Writer{
+		h: h,
+		w: w,
+	}
+}
+
+// Write wraps the write method of the underlying writer and also hashes all data.
+func (h *Writer) Write(p []byte) (int, error) {
+	// write the data to the underlying writing
+	n, err := h.w.Write(p)
+
+	// according to the interface documentation, Write() on a hash.Hash never
+	// returns an error.
+	_, hashErr := h.h.Write(p[:n])
+	if hashErr != nil {
+		panic(hashErr)
+	}
+
+	return n, err
+}
+
+// Sum returns the hash of all data written so far.
+func (h *Writer) Sum(d []byte) []byte {
+	return h.h.Sum(d)
+}

internal/repository/hashing/writer_test.go

@@ -0,0 +1,73 @@
+package hashing
+
+import (
+	"bytes"
+	"crypto/rand"
+	"crypto/sha256"
+	"io"
+	"testing"
+)
+
+func TestWriter(t *testing.T) {
+	tests := []int{5, 23, 2<<18 + 23, 1 << 20}
+
+	for _, size := range tests {
+		data := make([]byte, size)
+		_, err := io.ReadFull(rand.Reader, data)
+		if err != nil {
+			t.Fatalf("ReadFull: %v", err)
+		}
+
+		expectedHash := sha256.Sum256(data)
+
+		wr := NewWriter(io.Discard, sha256.New())
+
+		n, err := io.Copy(wr, bytes.NewReader(data))
+		if err != nil {
+			t.Fatal(err)
+		}
+
+		if n != int64(size) {
+			t.Errorf("Writer: invalid number of bytes written: got %d, expected %d",
+				n, size)
+		}
+
+		resultingHash := wr.Sum(nil)
+
+		if !bytes.Equal(expectedHash[:], resultingHash) {
+			t.Errorf("Writer: hashes do not match: expected %02x, got %02x",
+				expectedHash, resultingHash)
+		}
+	}
+}
+
+func BenchmarkWriter(b *testing.B) {
+	buf := make([]byte, 1<<22)
+	_, err := io.ReadFull(rand.Reader, buf)
+	if err != nil {
+		b.Fatal(err)
+	}
+
+	expectedHash := sha256.Sum256(buf)
+
+	b.SetBytes(int64(len(buf)))
+	b.ResetTimer()
+	for i := 0; i < b.N; i++ {
+		wr := NewWriter(io.Discard, sha256.New())
+		n, err := io.Copy(wr, bytes.NewReader(buf))
+		if err != nil {
+			b.Fatal(err)
+		}
+
+		if n != int64(len(buf)) {
+			b.Errorf("Writer: invalid number of bytes written: got %d, expected %d",
+				n, len(buf))
+		}
+
+		resultingHash := wr.Sum(nil)
+		if !bytes.Equal(expectedHash[:], resultingHash) {
+			b.Errorf("Writer: hashes do not match: expected %02x, got %02x",
+				expectedHash, resultingHash)
+		}
+	}
+}