Abstract SHA1 function usage

Refactor SHA1 interfaces and centralize the hash function. This will
allow easier introduction of different hash function later on.

Author: AdamMajer

cmd/hook.go

@@ -376,7 +376,9 @@ Gitea or set your environment appropriately.`, "")
 		oldCommitIDs[count] = string(fields[0])
 		newCommitIDs[count] = string(fields[1])
 		refFullNames[count] = git.RefName(fields[2])
-		if refFullNames[count] == git.BranchPrefix+"master" && newCommitIDs[count] != git.EmptySHA && count == total {
+
+		commitID, _ := git.IDFromString(newCommitIDs[count])
+		if refFullNames[count] == git.BranchPrefix+"master" && !commitID.IsZero() && count == total {
 			masterPushed = true
 		}
 		count++
@@ -669,7 +671,8 @@ Gitea or set your environment appropriately.`, "")
 		if err != nil {
 			return err
 		}
-		if rs.OldOID != git.EmptySHA {
+		commitID, _ := git.IDFromString(rs.OldOID)
+		if !commitID.IsZero() {
 			err = writeDataPktLine(ctx, os.Stdout, []byte("option old-oid "+rs.OldOID))
 			if err != nil {
 				return err

models/git/branch_test.go

@@ -30,7 +30,7 @@ func TestAddDeletedBranch(t *testing.T) {
 	assert.True(t, secondBranch.IsDeleted)
 
 	commit := &git.Commit{
-		ID:            git.MustIDFromString(secondBranch.CommitID),
+		ID:            repo.ObjectFormat.MustIDFromString(secondBranch.CommitID),
 		CommitMessage: secondBranch.CommitMessage,
 		Committer: &git.Signature{
 			When: secondBranch.CommitTime.AsLocalTime(),

models/git/commit_status.go

@@ -114,7 +114,8 @@ WHEN NOT MATCHED
 
 // GetNextCommitStatusIndex retried 3 times to generate a resource index
 func GetNextCommitStatusIndex(ctx context.Context, repoID int64, sha string) (int64, error) {
-	if !git.IsValidSHAPattern(sha) {
+	_, err := git.IDFromString(sha)
+	if err != nil {
 		return 0, git.ErrInvalidSHA{SHA: sha}
 	}
 
@@ -425,7 +426,7 @@ func FindRepoRecentCommitStatusContexts(ctx context.Context, repoID int64, befor
 type NewCommitStatusOptions struct {
 	Repo         *repo_model.Repository
 	Creator      *user_model.User
-	SHA          string
+	SHA          git.ObjectID
 	CommitStatus *CommitStatus
 }
 
@@ -440,26 +441,22 @@ func NewCommitStatus(ctx context.Context, opts NewCommitStatusOptions) error {
 		return fmt.Errorf("NewCommitStatus[%s, %s]: no user specified", repoPath, opts.SHA)
 	}
 
-	if _, err := git.NewIDFromString(opts.SHA); err != nil {
-		return fmt.Errorf("NewCommitStatus[%s, %s]: invalid sha: %w", repoPath, opts.SHA, err)
-	}
-
 	ctx, committer, err := db.TxContext(ctx)
 	if err != nil {
 		return fmt.Errorf("NewCommitStatus[repo_id: %d, user_id: %d, sha: %s]: %w", opts.Repo.ID, opts.Creator.ID, opts.SHA, err)
 	}
 	defer committer.Close()
 
 	// Get the next Status Index
-	idx, err := GetNextCommitStatusIndex(ctx, opts.Repo.ID, opts.SHA)
+	idx, err := GetNextCommitStatusIndex(ctx, opts.Repo.ID, opts.SHA.String())
 	if err != nil {
 		return fmt.Errorf("generate commit status index failed: %w", err)
 	}
 
 	opts.CommitStatus.Description = strings.TrimSpace(opts.CommitStatus.Description)
 	opts.CommitStatus.Context = strings.TrimSpace(opts.CommitStatus.Context)
 	opts.CommitStatus.TargetURL = strings.TrimSpace(opts.CommitStatus.TargetURL)
-	opts.CommitStatus.SHA = opts.SHA
+	opts.CommitStatus.SHA = opts.SHA.String()
 	opts.CommitStatus.CreatorID = opts.Creator.ID
 	opts.CommitStatus.RepoID = opts.Repo.ID
 	opts.CommitStatus.Index = idx

models/repo/repo.go

@@ -17,6 +17,7 @@ import (
 	"code.gitea.io/gitea/models/unit"
 	user_model "code.gitea.io/gitea/models/user"
 	"code.gitea.io/gitea/modules/base"
+	"code.gitea.io/gitea/modules/git"
 	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/markup"
 	"code.gitea.io/gitea/modules/setting"
@@ -179,6 +180,7 @@ type Repository struct {
 	IsFsckEnabled                   bool               `xorm:"NOT NULL DEFAULT true"`
 	CloseIssuesViaCommitInAnyBranch bool               `xorm:"NOT NULL DEFAULT false"`
 	Topics                          []string           `xorm:"TEXT JSON"`
+	ObjectFormat                    git.ObjectFormat   `xorm:"-"`
 
 	TrustModel TrustModelType
 
@@ -274,6 +276,8 @@ func (repo *Repository) AfterLoad() {
 	repo.NumOpenMilestones = repo.NumMilestones - repo.NumClosedMilestones
 	repo.NumOpenProjects = repo.NumProjects - repo.NumClosedProjects
 	repo.NumOpenActionRuns = repo.NumActionRuns - repo.NumClosedActionRuns
+
+	repo.ObjectFormat = git.ObjectFormatFromID(git.Sha1)
 }
 
 // LoadAttributes loads attributes of the repository.
@@ -313,7 +317,7 @@ func (repo *Repository) HTMLURL() string {
 // CommitLink make link to by commit full ID
 // note: won't check whether it's an right id
 func (repo *Repository) CommitLink(commitID string) (result string) {
-	if commitID == "" || commitID == "0000000000000000000000000000000000000000" {
+	if git.IsEmptyCommitID(commitID) {
 		result = ""
 	} else {
 		result = repo.Link() + "/commit/" + url.PathEscape(commitID)

modules/context/api.go

@@ -308,6 +308,12 @@ func RepoRefForAPI(next http.Handler) http.Handler {
 			return
 		}
 
+		objectFormat, err := ctx.Repo.GitRepo.GetObjectFormat()
+		if err != nil {
+			ctx.Error(http.StatusInternalServerError, "GetCommit", err)
+			return
+		}
+
 		if ref := ctx.FormTrim("ref"); len(ref) > 0 {
 			commit, err := ctx.Repo.GitRepo.GetCommit(ref)
 			if err != nil {
@@ -325,7 +331,6 @@ func RepoRefForAPI(next http.Handler) http.Handler {
 			return
 		}
 
-		var err error
 		refName := getRefName(ctx.Base, ctx.Repo, RepoRefAny)
 
 		if ctx.Repo.GitRepo.IsBranchExist(refName) {
@@ -342,7 +347,7 @@ func RepoRefForAPI(next http.Handler) http.Handler {
 				return
 			}
 			ctx.Repo.CommitID = ctx.Repo.Commit.ID.String()
-		} else if len(refName) == git.SHAFullLength {
+		} else if len(refName) == objectFormat.FullLength() {
 			ctx.Repo.CommitID = refName
 			ctx.Repo.Commit, err = ctx.Repo.GitRepo.GetCommit(refName)
 			if err != nil {

modules/context/repo.go

@@ -825,7 +825,9 @@ func getRefName(ctx *Base, repo *Repository, pathType RepoRefType) string {
 		}
 		// For legacy and API support only full commit sha
 		parts := strings.Split(path, "/")
-		if len(parts) > 0 && len(parts[0]) == git.SHAFullLength {
+		objectFormat, _ := repo.GitRepo.GetObjectFormat()
+
+		if len(parts) > 0 && len(parts[0]) == objectFormat.FullLength() {
 			repo.TreePath = strings.Join(parts[1:], "/")
 			return parts[0]
 		}
@@ -869,7 +871,9 @@ func getRefName(ctx *Base, repo *Repository, pathType RepoRefType) string {
 		return getRefNameFromPath(ctx, repo, path, repo.GitRepo.IsTagExist)
 	case RepoRefCommit:
 		parts := strings.Split(path, "/")
-		if len(parts) > 0 && len(parts[0]) >= 7 && len(parts[0]) <= git.SHAFullLength {
+		objectFormat, _ := repo.GitRepo.GetObjectFormat()
+
+		if len(parts) > 0 && len(parts[0]) >= 7 && len(parts[0]) <= objectFormat.FullLength() {
 			repo.TreePath = strings.Join(parts[1:], "/")
 			return parts[0]
 		}
@@ -929,6 +933,12 @@ func RepoRefByType(refType RepoRefType, ignoreNotExistErr ...bool) func(*Context
 			}
 		}
 
+		objectFormat, err := ctx.Repo.GitRepo.GetObjectFormat()
+		if err != nil {
+			log.Error("Cannot determine objectFormat for repository: %w", err)
+			ctx.Repo.Repository.MarkAsBrokenEmpty()
+		}
+
 		// Get default branch.
 		if len(ctx.Params("*")) == 0 {
 			refName = ctx.Repo.Repository.DefaultBranch
@@ -995,7 +1005,7 @@ func RepoRefByType(refType RepoRefType, ignoreNotExistErr ...bool) func(*Context
 					return cancel
 				}
 				ctx.Repo.CommitID = ctx.Repo.Commit.ID.String()
-			} else if len(refName) >= 7 && len(refName) <= git.SHAFullLength {
+			} else if len(refName) >= 7 && len(refName) <= objectFormat.FullLength() {
 				ctx.Repo.IsViewCommit = true
 				ctx.Repo.CommitID = refName
 
@@ -1005,7 +1015,7 @@ func RepoRefByType(refType RepoRefType, ignoreNotExistErr ...bool) func(*Context
 					return cancel
 				}
 				// If short commit ID add canonical link header
-				if len(refName) < git.SHAFullLength {
+				if len(refName) < objectFormat.FullLength() {
 					ctx.RespHeader().Set("Link", fmt.Sprintf("<%s>; rel=\"canonical\"",
 						util.URLJoin(setting.AppURL, strings.Replace(ctx.Req.URL.RequestURI(), util.PathEscapeSegments(refName), url.PathEscape(ctx.Repo.Commit.ID.String()), 1))))
 				}

modules/git/batch_reader.go

@@ -148,7 +148,7 @@ func CatFileBatch(ctx context.Context, repoPath string) (WriteCloserError, *bufi
 // ReadBatchLine reads the header line from cat-file --batch
 // We expect:
 // <sha> SP <type> SP <size> LF
-// sha is a 40byte not 20byte here
+// sha is a hex encoded here
 func ReadBatchLine(rd *bufio.Reader) (sha []byte, typ string, size int64, err error) {
 	typ, err = rd.ReadString('\n')
 	if err != nil {
@@ -251,20 +251,19 @@ headerLoop:
 }
 
 // git tree files are a list:
-// <mode-in-ascii> SP <fname> NUL <20-byte SHA>
+// <mode-in-ascii> SP <fname> NUL <binary Hash>
 //
 // Unfortunately this 20-byte notation is somewhat in conflict to all other git tools
-// Therefore we need some method to convert these 20-byte SHAs to a 40-byte SHA
+// Therefore we need some method to convert these binary hashes to hex hashes
 
-// constant hextable to help quickly convert between 20byte and 40byte hashes
+// constant hextable to help quickly convert between binary and hex representation
 const hextable = "0123456789abcdef"
 
-// To40ByteSHA converts a 20-byte SHA into a 40-byte sha. Input and output can be the
-// same 40 byte slice to support in place conversion without allocations.
+// BinToHexHeash converts a binary Hash into a hex encoded one. Input and output can be the
+// same byte slice to support in place conversion without allocations.
 // This is at least 100x quicker that hex.EncodeToString
-// NB This requires that out is a 40-byte slice
-func To40ByteSHA(sha, out []byte) []byte {
-	for i := 19; i >= 0; i-- {
+func BinToHex(objectFormat ObjectFormat, sha, out []byte) []byte {
+	for i := objectFormat.FullLength()/2 - 1; i >= 0; i-- {
 		v := sha[i]
 		vhi, vlo := v>>4, v&0x0f
 		shi, slo := hextable[vhi], hextable[vlo]
@@ -278,10 +277,10 @@ func To40ByteSHA(sha, out []byte) []byte {
 // It is recommended therefore to pass in an fnameBuf large enough to avoid almost all allocations
 //
 // Each line is composed of:
-// <mode-in-ascii-dropping-initial-zeros> SP <fname> NUL <20-byte SHA>
+// <mode-in-ascii-dropping-initial-zeros> SP <fname> NUL <binary HASH>
 //
-// We don't attempt to convert the 20-byte SHA to 40-byte SHA to save a lot of time
-func ParseTreeLine(rd *bufio.Reader, modeBuf, fnameBuf, shaBuf []byte) (mode, fname, sha []byte, n int, err error) {
+// We don't attempt to convert the raw HASH to save a lot of time
+func ParseTreeLine(objectFormat ObjectFormat, rd *bufio.Reader, modeBuf, fnameBuf, shaBuf []byte) (mode, fname, sha []byte, n int, err error) {
 	var readBytes []byte
 
 	// Read the Mode & fname
@@ -324,11 +323,12 @@ func ParseTreeLine(rd *bufio.Reader, modeBuf, fnameBuf, shaBuf []byte) (mode, fn
 	fnameBuf = fnameBuf[:len(fnameBuf)-1]
 	fname = fnameBuf
 
-	// Deal with the 20-byte SHA
+	// Deal with the binary hash
 	idx = 0
-	for idx < 20 {
+	len := objectFormat.FullLength() / 2
+	for idx < len {
 		var read int
-		read, err = rd.Read(shaBuf[idx:20])
+		read, err = rd.Read(shaBuf[idx:len])
 		n += read
 		if err != nil {
 			return mode, fname, sha, n, err

modules/git/blame.go

@@ -10,8 +10,6 @@ import (
 	"fmt"
 	"io"
 	"os"
-	"regexp"
-	"strings"
 
 	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/util"
@@ -33,14 +31,13 @@ type BlameReader struct {
 	done           chan error
 	lastSha        *string
 	ignoreRevsFile *string
+	objectFormat   ObjectFormat
 }
 
 func (r *BlameReader) UsesIgnoreRevs() bool {
 	return r.ignoreRevsFile != nil
 }
 
-var shaLineRegex = regexp.MustCompile("^([a-z0-9]{40})")
-
 // NextPart returns next part of blame (sequential code lines with the same commit)
 func (r *BlameReader) NextPart() (*BlamePart, error) {
 	var blamePart *BlamePart
@@ -52,6 +49,7 @@ func (r *BlameReader) NextPart() (*BlamePart, error) {
 		}
 	}
 
+	const previousHeader = "previous "
 	var lineBytes []byte
 	var isPrefix bool
 	var err error
@@ -67,21 +65,22 @@ func (r *BlameReader) NextPart() (*BlamePart, error) {
 			continue
 		}
 
-		line := string(lineBytes)
-
-		lines := shaLineRegex.FindStringSubmatch(line)
-		if lines != nil {
-			sha1 := lines[1]
+		var objectID string
+		objectFormatLength := r.objectFormat.FullLength()
 
+		if len(lineBytes) > objectFormatLength && lineBytes[objectFormatLength] == ' ' && r.objectFormat.IsValid(string(lineBytes[0:objectFormatLength])) {
+			objectID = string(lineBytes[0:objectFormatLength])
+		}
+		if len(objectID) > 0 {
 			if blamePart == nil {
 				blamePart = &BlamePart{
-					Sha:   sha1,
+					Sha:   objectID,
 					Lines: make([]string, 0),
 				}
 			}
 
-			if blamePart.Sha != sha1 {
-				r.lastSha = &sha1
+			if blamePart.Sha != objectID {
+				r.lastSha = &objectID
 				// need to munch to end of line...
 				for isPrefix {
 					_, isPrefix, err = r.bufferedReader.ReadLine()
@@ -91,12 +90,13 @@ func (r *BlameReader) NextPart() (*BlamePart, error) {
 				}
 				return blamePart, nil
 			}
-		} else if line[0] == '\t' {
-			blamePart.Lines = append(blamePart.Lines, line[1:])
-		} else if strings.HasPrefix(line, "previous ") {
-			parts := strings.SplitN(line[len("previous "):], " ", 2)
-			blamePart.PreviousSha = parts[0]
-			blamePart.PreviousPath = parts[1]
+		} else if lineBytes[0] == '\t' {
+			blamePart.Lines = append(blamePart.Lines, string(lineBytes[1:]))
+		} else if bytes.HasPrefix(lineBytes, []byte(previousHeader)) {
+			offset := len(previousHeader) // already includes a space
+			blamePart.PreviousSha = string(lineBytes[offset : offset+objectFormatLength])
+			offset += objectFormatLength + 1 // +1 for space
+			blamePart.PreviousPath = string(lineBytes[offset:])
 		}
 
 		// need to munch to end of line...
@@ -126,7 +126,7 @@ func (r *BlameReader) Close() error {
 }
 
 // CreateBlameReader creates reader for given repository, commit and file
-func CreateBlameReader(ctx context.Context, repoPath string, commit *Commit, file string, bypassBlameIgnore bool) (*BlameReader, error) {
+func CreateBlameReader(ctx context.Context, objectFormat ObjectFormat, repoPath string, commit *Commit, file string, bypassBlameIgnore bool) (*BlameReader, error) {
 	var ignoreRevsFile *string
 	if CheckGitVersionAtLeast("2.23") == nil && !bypassBlameIgnore {
 		ignoreRevsFile = tryCreateBlameIgnoreRevsFile(commit)
@@ -175,6 +175,7 @@ func CreateBlameReader(ctx context.Context, repoPath string, commit *Commit, fil
 		bufferedReader: bufferedReader,
 		done:           done,
 		ignoreRevsFile: ignoreRevsFile,
+		objectFormat:   objectFormat,
 	}, nil
 }
 

modules/git/blame_test.go

@@ -39,7 +39,7 @@ func TestReadingBlameOutput(t *testing.T) {
 		}
 
 		for _, bypass := range []bool{false, true} {
-			blameReader, err := CreateBlameReader(ctx, "./tests/repos/repo5_pulls", commit, "README.md", bypass)
+			blameReader, err := CreateBlameReader(ctx, &Sha1ObjectFormat{}, "./tests/repos/repo5_pulls", commit, "README.md", bypass)
 			assert.NoError(t, err)
 			assert.NotNil(t, blameReader)
 			defer blameReader.Close()
@@ -122,7 +122,7 @@ func TestReadingBlameOutput(t *testing.T) {
 			commit, err := repo.GetCommit(c.CommitID)
 			assert.NoError(t, err)
 
-			blameReader, err := CreateBlameReader(ctx, "./tests/repos/repo6_blame", commit, "blame.txt", c.Bypass)
+			blameReader, err := CreateBlameReader(ctx, repo.objectFormat, "./tests/repos/repo6_blame", commit, "blame.txt", c.Bypass)
 			assert.NoError(t, err)
 			assert.NotNil(t, blameReader)
 			defer blameReader.Close()

modules/git/blob_gogit.go

@@ -14,7 +14,7 @@ import (
 
 // Blob represents a Git object.
 type Blob struct {
-	ID SHA1
+	ID ObjectID
 
 	gogitEncodedObj plumbing.EncodedObject
 	name            string

modules/git/blob_nogogit.go

@@ -16,7 +16,7 @@ import (
 
 // Blob represents a Git object.
 type Blob struct {
-	ID SHA1
+	ID ObjectID
 
 	gotSize bool
 	size    int64