Abstract hash function usage

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

Author: AdamMajer

cmd/hook.go

@@ -376,7 +376,12 @@ 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 {
+
+		hash, err := git.HashFromString(newCommitIDs[count])
+		if err != nil {
+			return err
+		}
+		if refFullNames[count] == git.BranchPrefix+"master" && newCommitIDs[count] != hash.Type().Empty().String() && count == total {
 			masterPushed = true
 		}
 		count++
@@ -669,7 +674,11 @@ Gitea or set your environment appropriately.`, "")
 		if err != nil {
 			return err
 		}
-		if rs.OldOID != git.EmptySHA {
+		hash, err := git.HashFromString(rs.OldOID)
+		if err != nil {
+			return err
+		}
+		if rs.OldOID != hash.Type().Empty().String() {
 			err = writeDataPktLine(ctx, os.Stdout, []byte("option old-oid "+rs.OldOID))
 			if err != nil {
 				return err

models/git/branch_test.go

@@ -21,7 +21,9 @@ func TestAddDeletedBranch(t *testing.T) {
 	assert.NoError(t, unittest.PrepareTestDatabase())
 	repo := unittest.AssertExistsAndLoadBean(t, &repo_model.Repository{ID: 1})
 	firstBranch := unittest.AssertExistsAndLoadBean(t, &git_model.Branch{ID: 1})
+	hash, err := git.HashTypeFromString(repo.HashType)
 
+	assert.NoError(t, err)
 	assert.True(t, firstBranch.IsDeleted)
 	assert.NoError(t, git_model.AddDeletedBranch(db.DefaultContext, repo.ID, firstBranch.Name, firstBranch.DeletedByID))
 	assert.NoError(t, git_model.AddDeletedBranch(db.DefaultContext, repo.ID, "branch2", int64(1)))
@@ -30,14 +32,14 @@ func TestAddDeletedBranch(t *testing.T) {
 	assert.True(t, secondBranch.IsDeleted)
 
 	commit := &git.Commit{
-		ID:            git.MustIDFromString(secondBranch.CommitID),
+		ID:            hash.MustIDFromString(secondBranch.CommitID),
 		CommitMessage: secondBranch.CommitMessage,
 		Committer: &git.Signature{
 			When: secondBranch.CommitTime.AsLocalTime(),
 		},
 	}
 
-	err := git_model.UpdateBranch(db.DefaultContext, repo.ID, secondBranch.PusherID, secondBranch.Name, commit)
+	err = git_model.UpdateBranch(db.DefaultContext, repo.ID, secondBranch.PusherID, secondBranch.Name, commit)
 	assert.NoError(t, err)
 }
 

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.HashFromString(sha)
+	if err != nil {
 		return 0, git.ErrInvalidSHA{SHA: sha}
 	}
 
@@ -423,41 +424,37 @@ func FindRepoRecentCommitStatusContexts(ctx context.Context, repoID int64, befor
 type NewCommitStatusOptions struct {
 	Repo         *repo_model.Repository
 	Creator      *user_model.User
-	SHA          string
+	SHA          git.Hash
 	CommitStatus *CommitStatus
 }
 
 // NewCommitStatus save commit statuses into database
 func NewCommitStatus(ctx context.Context, opts NewCommitStatusOptions) error {
 	if opts.Repo == nil {
-		return fmt.Errorf("NewCommitStatus[nil, %s]: no repository specified", opts.SHA)
+		return fmt.Errorf("NewCommitStatus[nil, %s]: no repository specified", opts.SHA.String())
 	}
 
 	repoPath := opts.Repo.RepoPath()
 	if opts.Creator == nil {
-		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)
+		return fmt.Errorf("NewCommitStatus[%s, %s]: no user specified", repoPath, opts.SHA.String())
 	}
 
 	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)
+		return fmt.Errorf("NewCommitStatus[repo_id: %d, user_id: %d, sha: %s]: %w", opts.Repo.ID, opts.Creator.ID, opts.SHA.String(), 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"`
+	HashType                        string
 
 	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.HashType = "sha1"
 }
 
 // LoadAttributes loads attributes of the repository.
@@ -313,7 +317,8 @@ 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" {
+	hash, _ := git.HashTypeFromString(repo.HashType)
+	if commitID == "" || commitID == hash.Empty().String() {
 		result = ""
 	} else {
 		result = repo.Link() + "/commit/" + url.PathEscape(commitID)

modules/context/api.go

@@ -342,7 +342,7 @@ 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) == ctx.Repo.GitRepo.Hash.FullLength() {
 			ctx.Repo.CommitID = refName
 			ctx.Repo.Commit, err = ctx.Repo.GitRepo.GetCommit(refName)
 			if err != nil {

modules/context/repo.go

@@ -826,7 +826,7 @@ 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 {
+		if len(parts) > 0 && len(parts[0]) == repo.GitRepo.Hash.FullLength() {
 			repo.TreePath = strings.Join(parts[1:], "/")
 			return parts[0]
 		}
@@ -870,7 +870,7 @@ 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 {
+		if len(parts) > 0 && len(parts[0]) >= 7 && len(parts[0]) <= repo.GitRepo.Hash.FullLength() {
 			repo.TreePath = strings.Join(parts[1:], "/")
 			return parts[0]
 		}
@@ -996,7 +996,7 @@ 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) <= ctx.Repo.GitRepo.Hash.FullLength() {
 				ctx.Repo.IsViewCommit = true
 				ctx.Repo.CommitID = refName
 
@@ -1006,7 +1006,7 @@ 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) < ctx.Repo.GitRepo.Hash.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 BinToHexHash(hash HashType, sha, out []byte) []byte {
+	for i := hash.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(hash HashType, 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 {
+	hashLen := hash.FullLength() / 2
+	for idx < hashLen {
 		var read int
-		read, err = rd.Read(shaBuf[idx:20])
+		read, err = rd.Read(shaBuf[idx:hashLen])
 		n += read
 		if err != nil {
 			return mode, fname, sha, n, err

modules/git/blame.go

@@ -10,7 +10,6 @@ import (
 	"fmt"
 	"io"
 	"os"
-	"regexp"
 
 	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/util"
@@ -30,14 +29,13 @@ type BlameReader struct {
 	done           chan error
 	lastSha        *string
 	ignoreRevsFile *string
+	hash           HashType
 }
 
 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
@@ -61,16 +59,19 @@ func (r *BlameReader) NextPart() (*BlamePart, error) {
 			continue
 		}
 
-		lines := shaLineRegex.FindSubmatch(line)
-		if lines != nil {
-			sha1 := string(lines[1])
+		var hash string
+		hashLen := r.hash.FullLength()
 
+		if len(line) > hashLen && line[hashLen] == ' ' && r.hash.IsValid(string(line[0:hashLen])) {
+			hash = string(line[0:hashLen])
+		}
+		if len(hash) > 0 {
 			if blamePart == nil {
-				blamePart = &BlamePart{sha1, make([]string, 0)}
+				blamePart = &BlamePart{hash, make([]string, 0)}
 			}
 
-			if blamePart.Sha != sha1 {
-				r.lastSha = &sha1
+			if blamePart.Sha != hash {
+				r.lastSha = &hash
 				// need to munch to end of line...
 				for isPrefix {
 					_, isPrefix, err = r.bufferedReader.ReadLine()
@@ -113,7 +114,7 @@ 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, hash HashType, repoPath string, commit *Commit, file string, bypassBlameIgnore bool) (*BlameReader, error) {
 	var ignoreRevsFile *string
 	if CheckGitVersionAtLeast("2.23") == nil && !bypassBlameIgnore {
 		ignoreRevsFile = tryCreateBlameIgnoreRevsFile(commit)
@@ -162,6 +163,7 @@ func CreateBlameReader(ctx context.Context, repoPath string, commit *Commit, fil
 		bufferedReader: bufferedReader,
 		done:           done,
 		ignoreRevsFile: ignoreRevsFile,
+		hash:           hash,
 	}, nil
 }
 

modules/git/blame_test.go

@@ -37,7 +37,7 @@ 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, &Sha1HashType{}, "./tests/repos/repo5_pulls", commit, "README.md", bypass)
 			assert.NoError(t, err)
 			assert.NotNil(t, blameReader)
 			defer blameReader.Close()
@@ -118,7 +118,7 @@ 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.Hash, "./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 Hash
 
 	gogitEncodedObj plumbing.EncodedObject
 	name            string