ParseBatchHeader(): make function public and change its signature

Author: mhagger

git/batch_header.go

@@ -0,0 +1,47 @@
+package git
+
+import (
+	"fmt"
+	"strconv"
+	"strings"
+
+	"github.com/github/git-sizer/counts"
+)
+
+type BatchHeader struct {
+	OID        OID
+	ObjectType ObjectType
+	ObjectSize counts.Count32
+}
+
+var missingHeader = BatchHeader{
+	ObjectType: "missing",
+}
+
+// Parse a `cat-file --batch[-check]` output header line (including
+// the trailing LF). `spec`, if not "", is used in error messages.
+func ParseBatchHeader(spec string, header string) (BatchHeader, error) {
+	header = header[:len(header)-1]
+	words := strings.Split(header, " ")
+	if words[len(words)-1] == "missing" {
+		if spec == "" {
+			spec = words[0]
+		}
+		return missingHeader, fmt.Errorf("missing object %s", spec)
+	}
+
+	oid, err := NewOID(words[0])
+	if err != nil {
+		return missingHeader, err
+	}
+
+	size, err := strconv.ParseUint(words[2], 10, 0)
+	if err != nil {
+		return missingHeader, err
+	}
+	return BatchHeader{
+		OID:        oid,
+		ObjectType: ObjectType(words[1]),
+		ObjectSize: counts.NewCount32(size),
+	}, nil
+}

git/batch_obj_iter.go

@@ -2,14 +2,9 @@ package git
 
 import (
 	"bufio"
-	"fmt"
 	"io"
 	"os"
 	"os/exec"
-	"strconv"
-	"strings"
-
-	"github.com/github/git-sizer/counts"
 )
 
 // BatchObjectIter iterates over objects whose names are fed into its
@@ -55,23 +50,23 @@ func (repo *Repository) NewBatchObjectIter() (*BatchObjectIter, io.WriteCloser,
 
 // Next returns the next object: its OID, type, size, and contents.
 // When no more data are available, it returns an `io.EOF` error.
-func (iter *BatchObjectIter) Next() (OID, ObjectType, counts.Count32, []byte, error) {
+func (iter *BatchObjectIter) Next() (BatchHeader, []byte, error) {
 	header, err := iter.f.ReadString('\n')
 	if err != nil {
-		return OID{}, "", 0, nil, err
+		return missingHeader, nil, err
 	}
-	oid, objectType, objectSize, err := parseBatchHeader("", header)
+	obj, err := ParseBatchHeader("", header)
 	if err != nil {
-		return OID{}, "", 0, nil, err
+		return missingHeader, nil, err
 	}
 	// +1 for LF:
-	data := make([]byte, objectSize+1)
+	data := make([]byte, obj.ObjectSize+1)
 	_, err = io.ReadFull(iter.f, data)
 	if err != nil {
-		return OID{}, "", 0, nil, err
+		return missingHeader, nil, err
 	}
 	data = data[:len(data)-1]
-	return oid, objectType, objectSize, data, nil
+	return obj, data, nil
 }
 
 // Close closes the iterator and frees up resources. If any iterator
@@ -84,27 +79,3 @@ func (iter *BatchObjectIter) Close() error {
 	}
 	return err
 }
-
-// Parse a `cat-file --batch[-check]` output header line (including
-// the trailing LF). `spec`, if not "", is used in error messages.
-func parseBatchHeader(spec string, header string) (OID, ObjectType, counts.Count32, error) {
-	header = header[:len(header)-1]
-	words := strings.Split(header, " ")
-	if words[len(words)-1] == "missing" {
-		if spec == "" {
-			spec = words[0]
-		}
-		return OID{}, "missing", 0, fmt.Errorf("missing object %s", spec)
-	}
-
-	oid, err := NewOID(words[0])
-	if err != nil {
-		return OID{}, "missing", 0, err
-	}
-
-	size, err := strconv.ParseUint(words[2], 10, 0)
-	if err != nil {
-		return OID{}, "missing", 0, err
-	}
-	return oid, ObjectType(words[1]), counts.NewCount32(size), nil
-}

git/obj_iter.go

@@ -6,8 +6,6 @@ import (
 	"io"
 	"os"
 	"os/exec"
-
-	"github.com/github/git-sizer/counts"
 )
 
 // ObjectIter iterates over objects in a Git repository.
@@ -105,13 +103,13 @@ func (repo *Repository) NewObjectIter(
 
 // Next returns the next object: its OID, type, and size. When no more
 // data are available, it returns an `io.EOF` error.
-func (iter *ObjectIter) Next() (OID, ObjectType, counts.Count32, error) {
+func (iter *ObjectIter) Next() (BatchHeader, error) {
 	line, err := iter.f.ReadString('\n')
 	if err != nil {
-		return OID{}, "", 0, err
+		return missingHeader, err
 	}
 
-	return parseBatchHeader("", line)
+	return ParseBatchHeader("", line)
 }
 
 // Close closes the iterator and frees up resources.

sizes/graph.go

@@ -192,25 +192,25 @@ func ScanRepositoryUsingGraph(
 
 	progressMeter.Start("Processing blobs: %d")
 	for {
-		oid, objectType, objectSize, err := iter.Next()
+		obj, err := iter.Next()
 		if err != nil {
 			if err != io.EOF {
 				return HistorySize{}, err
 			}
 			break
 		}
-		switch objectType {
+		switch obj.ObjectType {
 		case "blob":
 			progressMeter.Inc()
-			graph.RegisterBlob(oid, objectSize)
+			graph.RegisterBlob(obj.OID, obj.ObjectSize)
 		case "tree":
-			trees = append(trees, ObjectHeader{oid, objectSize})
+			trees = append(trees, ObjectHeader{obj.OID, obj.ObjectSize})
 		case "commit":
-			commits = append(commits, CommitHeader{ObjectHeader{oid, objectSize}, git.NullOID})
+			commits = append(commits, CommitHeader{ObjectHeader{obj.OID, obj.ObjectSize}, git.NullOID})
 		case "tag":
-			tags = append(tags, ObjectHeader{oid, objectSize})
+			tags = append(tags, ObjectHeader{obj.OID, obj.ObjectSize})
 		default:
-			return HistorySize{}, fmt.Errorf("unexpected object type: %s", objectType)
+			return HistorySize{}, fmt.Errorf("unexpected object type: %s", obj.ObjectType)
 		}
 	}
 	progressMeter.Done()
@@ -286,22 +286,22 @@ func ScanRepositoryUsingGraph(
 
 	progressMeter.Start("Processing trees: %d")
 	for range trees {
-		oid, objectType, _, data, err := objectIter.Next()
+		obj, data, err := objectIter.Next()
 		if err != nil {
 			if err != io.EOF {
 				return HistorySize{}, err
 			}
 			return HistorySize{}, errors.New("fewer trees read than expected")
 		}
-		if objectType != "tree" {
-			return HistorySize{}, fmt.Errorf("expected tree; read %#v", objectType)
+		if obj.ObjectType != "tree" {
+			return HistorySize{}, fmt.Errorf("expected tree; read %#v", obj.ObjectType)
 		}
 		progressMeter.Inc()
-		tree, err := git.ParseTree(oid, data)
+		tree, err := git.ParseTree(obj.OID, data)
 		if err != nil {
 			return HistorySize{}, err
 		}
-		err = graph.RegisterTree(oid, tree)
+		err = graph.RegisterTree(obj.OID, tree)
 		if err != nil {
 			return HistorySize{}, err
 		}
@@ -313,26 +313,26 @@ func ScanRepositoryUsingGraph(
 	// time:
 	progressMeter.Start("Processing commits: %d")
 	for i := len(commits); i > 0; i-- {
-		oid, objectType, _, data, err := objectIter.Next()
+		obj, data, err := objectIter.Next()
 		if err != nil {
 			if err != io.EOF {
 				return HistorySize{}, err
 			}
 			return HistorySize{}, errors.New("fewer commits read than expected")
 		}
-		if objectType != "commit" {
-			return HistorySize{}, fmt.Errorf("expected commit; read %#v", objectType)
+		if obj.ObjectType != "commit" {
+			return HistorySize{}, fmt.Errorf("expected commit; read %#v", obj.ObjectType)
 		}
-		commit, err := git.ParseCommit(oid, data)
+		commit, err := git.ParseCommit(obj.OID, data)
 		if err != nil {
 			return HistorySize{}, err
 		}
-		if oid != commits[i-1].oid {
+		if obj.OID != commits[i-1].oid {
 			panic("commits not read in same order as requested")
 		}
 		commits[i-1].tree = commit.Tree
 		progressMeter.Inc()
-		graph.RegisterCommit(oid, commit)
+		graph.RegisterCommit(obj.OID, commit)
 	}
 	progressMeter.Done()
 
@@ -349,22 +349,22 @@ func ScanRepositoryUsingGraph(
 
 	progressMeter.Start("Processing annotated tags: %d")
 	for range tags {
-		oid, objectType, _, data, err := objectIter.Next()
+		obj, data, err := objectIter.Next()
 		if err != nil {
 			if err != io.EOF {
 				return HistorySize{}, err
 			}
 			return HistorySize{}, errors.New("fewer tags read than expected")
 		}
-		if objectType != "tag" {
-			return HistorySize{}, fmt.Errorf("expected tag; read %#v", objectType)
+		if obj.ObjectType != "tag" {
+			return HistorySize{}, fmt.Errorf("expected tag; read %#v", obj.ObjectType)
 		}
-		tag, err := git.ParseTag(oid, data)
+		tag, err := git.ParseTag(obj.OID, data)
 		if err != nil {
 			return HistorySize{}, err
 		}
 		progressMeter.Inc()
-		graph.RegisterTag(oid, tag)
+		graph.RegisterTag(obj.OID, tag)
 	}
 	progressMeter.Done()