feat(kzip): change Unit.Digest() to return hex string representation (#…

…4044)

kythe/go/platform/kcd/kcd.go

@@ -241,9 +241,9 @@ type Unit interface {
 	// canonicalization is unit-dependent, and may safely be a no-op.
 	Canonicalize()
 
-	// Digest writes a unique representation of the unit to w sufficient to
-	// generate a content-addressable digest.
-	Digest(w io.Writer)
+	// Digest produces a unique string representation of a unit sufficient to
+	// serve as a content-addressable digest.
+	Digest() string
 }
 
 // Index represents the indexable terms of a compilation.
@@ -261,14 +261,6 @@ func HexDigest(data []byte) string {
 	return hex.EncodeToString(sum[:])
 }
 
-// UnitDigest computes a hex-encoded SHA256 digest of the data in a compilation
-// unit.
-func UnitDigest(u Unit) string {
-	sha := sha256.New()
-	u.Digest(sha)
-	return hex.EncodeToString(sha.Sum(nil)[:])
-}
-
 // IsValidDigest reports whether s is valid as a digest computed by the
 // HexDigest function. It does not check whether s could have actually been
 // generated by the hash function, only the structure of the value.

kythe/go/platform/kcd/kcd_test.go

@@ -17,7 +17,6 @@
 package kcd
 
 import (
-	"io"
 	"regexp"
 	"testing"
 	"time"
@@ -35,21 +34,9 @@ func TestHexDigest(t *testing.T) {
 		if got := HexDigest([]byte(input)); got != want {
 			t.Errorf("HexDigest(%q): got %q, want %q", input, got, want)
 		}
-
-		u := stubUnit{bits: input}
-		if got := UnitDigest(u); got != want {
-			t.Errorf("UnitDigest(%q): got %q, want %q", input, got, want)
-		}
 	}
 }
 
-type stubUnit struct {
-	bits string
-	Unit
-}
-
-func (s stubUnit) Digest(w io.Writer) { w.Write([]byte(s.bits)) }
-
 func regexps(exprs ...string) (res []*regexp.Regexp) {
 	for _, expr := range exprs {
 		res = append(res, regexp.MustCompile(expr))

kythe/go/platform/kcd/kythe/units.go

@@ -18,8 +18,9 @@
 package kythe // import "kythe.io/kythe/go/platform/kcd/kythe"
 
 import (
+	"crypto/sha256"
+	"encoding/hex"
 	"fmt"
-	"io"
 	"sort"
 	"strings"
 
@@ -94,15 +95,16 @@ func (u Unit) Canonicalize() {
 }
 
 // Digest satisfies part of the kcd.Unit interface.
-func (u Unit) Digest(w io.Writer) {
+func (u Unit) Digest() string {
+	sha := sha256.New()
 	pb := u.Proto
 	if pb == nil {
 		pb = new(apb.CompilationUnit)
 	}
 	put := func(tag string, ss ...string) {
-		fmt.Fprintln(w, tag)
+		fmt.Fprintln(sha, tag)
 		for _, s := range ss {
-			fmt.Fprint(w, s, "\x00")
+			fmt.Fprint(sha, s, "\x00")
 		}
 	}
 	putv := func(tag string, v *spb.VName) {
@@ -124,6 +126,7 @@ func (u Unit) Digest(w io.Writer) {
 	for _, d := range pb.Details {
 		put("DET", d.TypeUrl, string(d.Value))
 	}
+	return hex.EncodeToString(sha.Sum(nil)[:])
 }
 
 // ConvertUnit reports whether v can be converted to a Kythe kcd.Unit, and if

kythe/go/platform/kcd/kythe/units_test.go

@@ -17,7 +17,6 @@
 package kythe
 
 import (
-	"bytes"
 	"reflect"
 	"sort"
 	"testing"
@@ -129,7 +128,7 @@ func keys(v interface{}) (keys []string) {
 }
 
 func TestDigest(t *testing.T) {
-	const empty = "CU\n\x00\x00\x00\x00\x00ARG\nOUT\n\x00SRC\nCWD\n\x00CTX\n\x00"
+	const empty = "56bf5044e1b5c4c1cc7c4b131ac2fb979d288460e63352b10eef80ca35bd0a7b"
 	tests := []struct {
 		unit *apb.CompilationUnit
 		want string
@@ -144,12 +143,7 @@ func TestDigest(t *testing.T) {
 				Language:  "L",
 			},
 			Argument: []string{"a1", "a2"},
-		}, "CU\nS\x00C\x00\x00P\x00L\x00" +
-			"ARG\na1\x00a2\x00" +
-			"OUT\n\x00" +
-			"SRC\n" +
-			"CWD\n\x00" +
-			"CTX\n\x00",
+		}, "e9e170dcfca53c8126755bbc8b703994dedd3af32584291e01fba164ab5d3f32",
 		},
 		{&apb.CompilationUnit{
 			RequiredInput: []*apb.CompilationUnit_FileInput{{
@@ -165,20 +159,11 @@ func TestDigest(t *testing.T) {
 				TypeUrl: "type",
 				Value:   []byte("nasaldemons"),
 			}},
-		}, "CU\n\x00\x00\x00\x00\x00" +
-			"RI\nRIS\x00\x00\x00\x00\x00" +
-			"IN\npath\x00digest\x00" +
-			"ARG\n" +
-			"OUT\nblah\x00" +
-			"SRC\nCWD\n\x00CTX\n\x00" +
-			"ENV\nfeefie\x00fofum\x00" +
-			"DET\ntype\x00nasaldemons\x00",
+		}, "bb761979683e7c268e967eb5bcdedaa7fa5d1d472b0826b00b69acafbaad7ee6",
 		},
 	}
 	for _, test := range tests {
-		var buf bytes.Buffer
-		Unit{Proto: test.unit}.Digest(&buf)
-		got := buf.String()
+		got := Unit{Proto: test.unit}.Digest()
 		if got != test.want {
 			t.Errorf("Digest: got %q, want %q\nInput: %+v", got, test.want, test.unit)
 		}

kythe/go/platform/kcd/memdb/memdb.go

@@ -188,7 +188,7 @@ func (db *DB) WriteUnit(_ context.Context, revision, corpus, formatKey string, u
 	if err != nil {
 		return "", err
 	}
-	digest := kcd.UnitDigest(unit)
+	digest := unit.Digest()
 	if db.Unit == nil {
 		db.Unit = make(map[string]Unit)
 	}

kythe/go/platform/kzip/kzip.go

@@ -513,9 +513,7 @@ var toJSON = &jsonpb.Marshaler{OrigName: true}
 func (w *Writer) AddUnit(cu *apb.CompilationUnit, index *apb.IndexedCompilation_Index) (string, error) {
 	unit := kythe.Unit{Proto: cu}
 	unit.Canonicalize()
-	hash := sha256.New()
-	unit.Digest(hash)
-	digest := hex.EncodeToString(hash.Sum(nil))
+	digest := unit.Digest()
 
 	w.mu.Lock()
 	defer w.mu.Unlock()