From 0a9e63317526b9efa3d74de5df0c990e2916c463 Mon Sep 17 00:00:00 2001
From: Sameer <11097096+sameer@users.noreply.github.com>
Date: Tue, 18 Dec 2018 14:48:19 -0500
Subject: [PATCH] Cache RawData for Commit, Tag, & Tree, fixes #6

---
 commit.go   | 70 +++++++++++++++++++++++++++++------------------------
 git_test.go | 70 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 tag.go      | 36 ++++++++++++++++-----------
 tree.go     | 31 +++++++++++++++---------
 4 files changed, 150 insertions(+), 57 deletions(-)

diff --git a/commit.go b/commit.go
index c0810c5..ebe2a2f 100644
--- a/commit.go
+++ b/commit.go
@@ -7,6 +7,7 @@ import (
 	"errors"
 	"fmt"
 	"strconv"
+	"sync"
 
 	cid "github.com/ipfs/go-cid"
 	node "github.com/ipfs/go-ipld-format"
@@ -14,8 +15,8 @@ import (
 
 type Commit struct {
 	DataSize  string      `json:"-"`
-	GitTree   cid.Cid    `json:"tree"`
-	Parents   []cid.Cid  `json:"parents"`
+	GitTree   cid.Cid     `json:"tree"`
+	Parents   []cid.Cid   `json:"parents"`
 	Message   string      `json:"message"`
 	Author    *PersonInfo `json:"author"`
 	Committer *PersonInfo `json:"committer"`
@@ -27,6 +28,9 @@ type Commit struct {
 	Other []string `json:"other,omitempty"`
 
 	cid cid.Cid
+
+	rawData     []byte
+	rawDataOnce sync.Once
 }
 
 type PersonInfo struct {
@@ -80,7 +84,7 @@ func (pi *PersonInfo) resolve(p []string) (interface{}, []string, error) {
 }
 
 type MergeTag struct {
-	Object cid.Cid    `json:"object"`
+	Object cid.Cid     `json:"object"`
 	Type   string      `json:"type"`
 	Tag    string      `json:"tag"`
 	Tagger *PersonInfo `json:"tagger"`
@@ -118,34 +122,38 @@ func (c *Commit) Loggable() map[string]interface{} {
 }
 
 func (c *Commit) RawData() []byte {
-	buf := new(bytes.Buffer)
-	fmt.Fprintf(buf, "commit %s\x00", c.DataSize)
-	fmt.Fprintf(buf, "tree %s\n", hex.EncodeToString(cidToSha(c.GitTree)))
-	for _, p := range c.Parents {
-		fmt.Fprintf(buf, "parent %s\n", hex.EncodeToString(cidToSha(p)))
-	}
-	fmt.Fprintf(buf, "author %s\n", c.Author.String())
-	fmt.Fprintf(buf, "committer %s\n", c.Committer.String())
-	if len(c.Encoding) > 0 {
-		fmt.Fprintf(buf, "encoding %s\n", c.Encoding)
-	}
-	for _, mtag := range c.MergeTag {
-		fmt.Fprintf(buf, "mergetag object %s\n", hex.EncodeToString(cidToSha(mtag.Object)))
-		fmt.Fprintf(buf, " type %s\n", mtag.Type)
-		fmt.Fprintf(buf, " tag %s\n", mtag.Tag)
-		fmt.Fprintf(buf, " tagger %s\n \n", mtag.Tagger.String())
-		fmt.Fprintf(buf, "%s", mtag.Text)
-	}
-	if c.Sig != nil {
-		fmt.Fprintln(buf, "gpgsig -----BEGIN PGP SIGNATURE-----")
-		fmt.Fprint(buf, c.Sig.Text)
-		fmt.Fprintln(buf, " -----END PGP SIGNATURE-----")
-	}
-	for _, line := range c.Other {
-		fmt.Fprintln(buf, line)
-	}
-	fmt.Fprintf(buf, "\n%s", c.Message)
-	return buf.Bytes()
+	c.rawDataOnce.Do(func() {
+		buf := new(bytes.Buffer)
+		fmt.Fprintf(buf, "commit %s\x00", c.DataSize)
+		fmt.Fprintf(buf, "tree %s\n", hex.EncodeToString(cidToSha(c.GitTree)))
+		for _, p := range c.Parents {
+			fmt.Fprintf(buf, "parent %s\n", hex.EncodeToString(cidToSha(p)))
+		}
+		fmt.Fprintf(buf, "author %s\n", c.Author.String())
+		fmt.Fprintf(buf, "committer %s\n", c.Committer.String())
+		if len(c.Encoding) > 0 {
+			fmt.Fprintf(buf, "encoding %s\n", c.Encoding)
+		}
+		for _, mtag := range c.MergeTag {
+			fmt.Fprintf(buf, "mergetag object %s\n", hex.EncodeToString(cidToSha(mtag.Object)))
+			fmt.Fprintf(buf, " type %s\n", mtag.Type)
+			fmt.Fprintf(buf, " tag %s\n", mtag.Tag)
+			fmt.Fprintf(buf, " tagger %s\n \n", mtag.Tagger.String())
+			fmt.Fprintf(buf, "%s", mtag.Text)
+		}
+		if c.Sig != nil {
+			fmt.Fprintln(buf, "gpgsig -----BEGIN PGP SIGNATURE-----")
+			fmt.Fprint(buf, c.Sig.Text)
+			fmt.Fprintln(buf, " -----END PGP SIGNATURE-----")
+		}
+		for _, line := range c.Other {
+			fmt.Fprintln(buf, line)
+		}
+		fmt.Fprintf(buf, "\n%s", c.Message)
+		c.rawData = buf.Bytes()
+	})
+
+	return c.rawData
 }
 
 func (c *Commit) Resolve(path []string) (interface{}, []string, error) {
diff --git a/git_test.go b/git_test.go
index b0e0b7e..51a8031 100644
--- a/git_test.go
+++ b/git_test.go
@@ -7,6 +7,7 @@ import (
 	"io"
 	"os"
 	"path/filepath"
+	"reflect"
 	"strings"
 	"testing"
 
@@ -171,6 +172,7 @@ func testNode(t *testing.T, nd node.Node) error {
 
 		/*s, _ := commit.Size()
 		assert.Equal(t, len(commit.RawData()), int(s))*/ //TODO: Known breakage
+		assert(t, reflect.DeepEqual(commit.RawData(), commit.RawData()))
 		assert(t, commit.GitTree.Defined())
 		assert(t, commit.Links() != nil)
 		assert(t, commit.Loggable()["type"] == "git_commit")
@@ -228,6 +230,7 @@ func testNode(t *testing.T, nd node.Node) error {
 		}
 
 		assert(t, tag.Type == "commit" || tag.Type == "tree" || tag.Type == "blob" || tag.Type == "tag")
+		assert(t, reflect.DeepEqual(tag.RawData(), tag.RawData()))
 		assert(t, tag.Object.Defined())
 		assert(t, tag.Loggable()["type"] == "git_tag")
 		assert(t, tag.Tree("", -1) != nil)
@@ -243,6 +246,7 @@ func testNode(t *testing.T, nd node.Node) error {
 			t.Fatalf("Tree is not a tree")
 		}
 
+		assert(t, reflect.DeepEqual(tree.RawData(), tree.RawData()))
 		assert(t, tree.entries != nil)
 		assert(t, tree.Tree("", 0) == nil)
 	}
@@ -345,3 +349,69 @@ func assert(t *testing.T, ok bool) {
 		t.Fatal("Assertion failed")
 	}
 }
+
+func BenchmarkRawData(b *testing.B) {
+	for i := 0; i < b.N; i++ {
+		err := filepath.Walk(".git/objects", func(path string, info os.FileInfo, err error) error {
+			if err != nil {
+				return nil
+			}
+			if info.IsDir() {
+				return nil
+			}
+
+			parts := strings.Split(path, "/")
+			if dir := parts[len(parts)-2]; dir == "info" || dir == "pack" {
+				return nil
+			}
+
+			fi, err := os.Open(path)
+			if err != nil {
+				return err
+			}
+
+			thing, err := ParseCompressedObject(fi)
+			if err != nil {
+				return err
+			}
+			thing.RawData()
+			return nil
+		})
+		if err != nil {
+			b.Fatal(err)
+		}
+	}
+}
+
+func BenchmarkCid(b *testing.B) {
+	for i := 0; i < b.N; i++ {
+		err := filepath.Walk(".git/objects", func(path string, info os.FileInfo, err error) error {
+			if err != nil {
+				return nil
+			}
+			if info.IsDir() {
+				return nil
+			}
+
+			parts := strings.Split(path, "/")
+			if dir := parts[len(parts)-2]; dir == "info" || dir == "pack" {
+				return nil
+			}
+
+			fi, err := os.Open(path)
+			if err != nil {
+				return err
+			}
+
+			thing, err := ParseCompressedObject(fi)
+			if err != nil {
+				return err
+			}
+			thing.Cid()
+			return nil
+		})
+		if err != nil {
+			b.Fatal(err)
+		}
+	}
+}
diff --git a/tag.go b/tag.go
index 4bee947..0655c3d 100644
--- a/tag.go
+++ b/tag.go
@@ -3,15 +3,16 @@ package ipldgit
 import (
 	"bytes"
 	"encoding/hex"
+	"errors"
 	"fmt"
+	"sync"
 
-	"errors"
 	cid "github.com/ipfs/go-cid"
 	node "github.com/ipfs/go-ipld-format"
 )
 
 type Tag struct {
-	Object   cid.Cid    `json:"object"`
+	Object   cid.Cid     `json:"object"`
 	Type     string      `json:"type"`
 	Tag      string      `json:"tag"`
 	Tagger   *PersonInfo `json:"tagger"`
@@ -19,6 +20,9 @@ type Tag struct {
 	dataSize string
 
 	cid cid.Cid
+
+	rawData     []byte
+	rawDataOnce sync.Once
 }
 
 func (t *Tag) Cid() cid.Cid {
@@ -41,18 +45,22 @@ func (t *Tag) Loggable() map[string]interface{} {
 }
 
 func (t *Tag) RawData() []byte {
-	buf := new(bytes.Buffer)
-	fmt.Fprintf(buf, "tag %s\x00", t.dataSize)
-	fmt.Fprintf(buf, "object %s\n", hex.EncodeToString(cidToSha(t.Object)))
-	fmt.Fprintf(buf, "type %s\n", t.Type)
-	fmt.Fprintf(buf, "tag %s\n", t.Tag)
-	if t.Tagger != nil {
-		fmt.Fprintf(buf, "tagger %s\n", t.Tagger.String())
-	}
-	if t.Message != "" {
-		fmt.Fprintf(buf, "\n%s", t.Message)
-	}
-	return buf.Bytes()
+	t.rawDataOnce.Do(func() {
+		buf := new(bytes.Buffer)
+		fmt.Fprintf(buf, "tag %s\x00", t.dataSize)
+		fmt.Fprintf(buf, "object %s\n", hex.EncodeToString(cidToSha(t.Object)))
+		fmt.Fprintf(buf, "type %s\n", t.Type)
+		fmt.Fprintf(buf, "tag %s\n", t.Tag)
+		if t.Tagger != nil {
+			fmt.Fprintf(buf, "tagger %s\n", t.Tagger.String())
+		}
+		if t.Message != "" {
+			fmt.Fprintf(buf, "\n%s", t.Message)
+		}
+		t.rawData = buf.Bytes()
+	})
+
+	return t.rawData
 }
 
 func (t *Tag) Resolve(path []string) (interface{}, []string, error) {
diff --git a/tree.go b/tree.go
index 5907383..fb49249 100644
--- a/tree.go
+++ b/tree.go
@@ -3,24 +3,27 @@ package ipldgit
 import (
 	"bytes"
 	"encoding/json"
+	"errors"
 	"fmt"
 	"io"
+	"sync"
 
-	"errors"
 	cid "github.com/ipfs/go-cid"
 	node "github.com/ipfs/go-ipld-format"
 )
 
 type Tree struct {
-	entries map[string]*TreeEntry
-	size    int
-	order   []string
-	cid     cid.Cid
+	entries     map[string]*TreeEntry
+	size        int
+	order       []string
+	cid         cid.Cid
+	rawData     []byte
+	rawDataOnce sync.Once
 }
 
 type TreeEntry struct {
 	name string
-	Mode string   `json:"mode"`
+	Mode string  `json:"mode"`
 	Hash cid.Cid `json:"hash"`
 }
 
@@ -95,13 +98,17 @@ func (t *Tree) Loggable() map[string]interface{} {
 }
 
 func (t *Tree) RawData() []byte {
-	buf := new(bytes.Buffer)
+	t.rawDataOnce.Do(func() {
+		buf := new(bytes.Buffer)
 
-	fmt.Fprintf(buf, "tree %d\x00", t.size)
-	for _, s := range t.order {
-		t.entries[s].WriteTo(buf)
-	}
-	return buf.Bytes()
+		fmt.Fprintf(buf, "tree %d\x00", t.size)
+		for _, s := range t.order {
+			t.entries[s].WriteTo(buf)
+		}
+		t.rawData = buf.Bytes()
+	})
+
+	return t.rawData
 }
 
 func (t *Tree) Resolve(p []string) (interface{}, []string, error) {