Refactor parseSignatureFromCommitLine (#29054) (#29108)

Backport #29054. Fix #28840

This backport is for 1.21 only and it is different from the change in
1.22: this backport still accept the legacy date format to avoid
breaking.

modules/git/repo.go

@@ -236,7 +236,7 @@ func GetLatestCommitTime(ctx context.Context, repoPath string) (time.Time, error
 		return time.Time{}, err
 	}
 	commitTime := strings.TrimSpace(stdout)
-	return time.Parse(GitTimeLayout, commitTime)
+	return time.Parse("Mon Jan _2 15:04:05 2006 -0700", commitTime)
 }
 
 // DivergeObject represents commit count diverging commits

modules/git/repo_tag.go

@@ -183,11 +183,7 @@ func parseTagRef(ref map[string]string) (tag *Tag, err error) {
 		}
 	}
 
-	tag.Tagger, err = newSignatureFromCommitline([]byte(ref["creator"]))
+	tag.Tagger = parseSignatureFromCommitLine(ref["creator"])
-	if err != nil {
-		return nil, fmt.Errorf("parse tagger: %w", err)
-	}
-
 	tag.Message = ref["contents"]
 	// strip PGP signature if present in contents field
 	pgpStart := strings.Index(tag.Message, beginpgp)

modules/git/repo_tag_test.go

@@ -228,7 +228,7 @@ func TestRepository_parseTagRef(t *testing.T) {
 				ID:        MustIDFromString("ab23e4b7f4cd0caafe0174c0e7ef6d651ba72889"),
 				Object:    MustIDFromString("ab23e4b7f4cd0caafe0174c0e7ef6d651ba72889"),
 				Type:      "commit",
-				Tagger:    parseAuthorLine(t, "Foo Bar <foo@bar.com> 1565789218 +0300"),
+				Tagger:    parseSignatureFromCommitLine("Foo Bar <foo@bar.com> 1565789218 +0300"),
 				Message:   "Add changelog of v1.9.1 (#7859)\n\n* add changelog of v1.9.1\n* Update CHANGELOG.md\n",
 				Signature: nil,
 			},
@@ -257,7 +257,7 @@ func TestRepository_parseTagRef(t *testing.T) {
 				ID:        MustIDFromString("8c68a1f06fc59c655b7e3905b159d761e91c53c9"),
 				Object:    MustIDFromString("3325fd8a973321fd59455492976c042dde3fd1ca"),
 				Type:      "tag",
-				Tagger:    parseAuthorLine(t, "Foo Bar <foo@bar.com> 1565789218 +0300"),
+				Tagger:    parseSignatureFromCommitLine("Foo Bar <foo@bar.com> 1565789218 +0300"),
 				Message:   "Add changelog of v1.9.1 (#7859)\n\n* add changelog of v1.9.1\n* Update CHANGELOG.md\n",
 				Signature: nil,
 			},
@@ -315,7 +315,7 @@ qbHDASXl
 				ID:      MustIDFromString("8c68a1f06fc59c655b7e3905b159d761e91c53c9"),
 				Object:  MustIDFromString("3325fd8a973321fd59455492976c042dde3fd1ca"),
 				Type:    "tag",
-				Tagger:  parseAuthorLine(t, "Foo Bar <foo@bar.com> 1565789218 +0300"),
+				Tagger:  parseSignatureFromCommitLine("Foo Bar <foo@bar.com> 1565789218 +0300"),
 				Message: "Add changelog of v1.9.1 (#7859)\n\n* add changelog of v1.9.1\n* Update CHANGELOG.md",
 				Signature: &CommitGPGSignature{
 					Signature: `-----BEGIN PGP SIGNATURE-----
@@ -364,14 +364,3 @@ Add changelog of v1.9.1 (#7859)
 		})
 	}
 }
-
-func parseAuthorLine(t *testing.T, committer string) *Signature {
-	t.Helper()
-
-	sig, err := newSignatureFromCommitline([]byte(committer))
-	if err != nil {
-		t.Fatalf("parse author line '%s': %v", committer, err)
-	}
-
-	return sig
-}

modules/git/signature.go

@@ -4,7 +4,46 @@
 
 package git
 
-const (
+import (
-	// GitTimeLayout is the (default) time layout used by git.
+	"strconv"
-	GitTimeLayout = "Mon Jan _2 15:04:05 2006 -0700"
+	"strings"
+	"time"
+
+	"code.gitea.io/gitea/modules/log"
 )
+
+// Helper to get a signature from the commit line, which looks like:
+//
+//	full name <user@example.com> 1378823654 +0200
+//
+// Haven't found the official reference for the standard format yet.
+// This function never fails, if the "line" can't be parsed, it returns a default Signature with "zero" time.
+func parseSignatureFromCommitLine(line string) *Signature {
+	sig := &Signature{}
+	s1, sx, ok1 := strings.Cut(line, " <")
+	s2, s3, ok2 := strings.Cut(sx, "> ")
+	if !ok1 || !ok2 {
+		sig.Name = line
+		return sig
+	}
+	sig.Name, sig.Email = s1, s2
+
+	if strings.Count(s3, " ") == 1 {
+		ts, tz, _ := strings.Cut(s3, " ")
+		seconds, _ := strconv.ParseInt(ts, 10, 64)
+		if tzTime, err := time.Parse("-0700", tz); err == nil {
+			sig.When = time.Unix(seconds, 0).In(tzTime.Location())
+		}
+	} else {
+		// the old gitea code tried to parse the date in a few different formats, but it's not clear why.
+		// according to public document, only the standard format "timestamp timezone" could be found, so drop other formats.
+		log.Error("suspicious commit line format: %q", line)
+		for _, fmt := range []string{"Mon Jan _2 15:04:05 2006 -0700"} {
+			if t, err := time.Parse(fmt, s3); err == nil {
+				sig.When = t
+				break
+			}
+		}
+	}
+	return sig
+}

modules/git/signature_gogit.go

@@ -7,52 +7,8 @@
 package git
 
 import (
-	"bytes"
-	"strconv"
-	"strings"
-	"time"
-
 	"github.com/go-git/go-git/v5/plumbing/object"
 )
 
 // Signature represents the Author or Committer information.
 type Signature = object.Signature
-
-// Helper to get a signature from the commit line, which looks like these:
-//
-//	author Patrick Gundlach <gundlach@speedata.de> 1378823654 +0200
-//	author Patrick Gundlach <gundlach@speedata.de> Thu, 07 Apr 2005 22:13:13 +0200
-//
-// but without the "author " at the beginning (this method should)
-// be used for author and committer.
-//
-// FIXME: include timezone for timestamp!
-func newSignatureFromCommitline(line []byte) (_ *Signature, err error) {
-	sig := new(Signature)
-	emailStart := bytes.IndexByte(line, '<')
-	if emailStart > 0 { // Empty name has already occurred, even if it shouldn't
-		sig.Name = strings.TrimSpace(string(line[:emailStart-1]))
-	}
-	emailEnd := bytes.IndexByte(line, '>')
-	sig.Email = string(line[emailStart+1 : emailEnd])
-
-	// Check date format.
-	if len(line) > emailEnd+2 {
-		firstChar := line[emailEnd+2]
-		if firstChar >= 48 && firstChar <= 57 {
-			timestop := bytes.IndexByte(line[emailEnd+2:], ' ')
-			timestring := string(line[emailEnd+2 : emailEnd+2+timestop])
-			seconds, _ := strconv.ParseInt(timestring, 10, 64)
-			sig.When = time.Unix(seconds, 0)
-		} else {
-			sig.When, err = time.Parse(GitTimeLayout, string(line[emailEnd+2:]))
-			if err != nil {
-				return nil, err
-			}
-		}
-	} else {
-		// Fall back to unix 0 time
-		sig.When = time.Unix(0, 0)
-	}
-	return sig, nil
-}

modules/git/signature_nogogit.go

@@ -7,21 +7,17 @@
 package git
 
 import (
-	"bytes"
 	"fmt"
-	"strconv"
-	"strings"
 	"time"
+
+	"code.gitea.io/gitea/modules/util"
 )
 
-// Signature represents the Author or Committer information.
+// Signature represents the Author, Committer or Tagger information.
 type Signature struct {
-	// Name represents a person name. It is an arbitrary string.
+	Name  string    // the committer name, it can be anything
-	Name string
+	Email string    // the committer email, it can be anything
-	// Email is an email, but it cannot be assumed to be well-formed.
+	When  time.Time // the timestamp of the signature
-	Email string
-	// When is the timestamp of the signature.
-	When time.Time
 }
 
 func (s *Signature) String() string {
@@ -30,71 +26,5 @@ func (s *Signature) String() string {
 
 // Decode decodes a byte array representing a signature to signature
 func (s *Signature) Decode(b []byte) {
-	sig, _ := newSignatureFromCommitline(b)
+	*s = *parseSignatureFromCommitLine(util.UnsafeBytesToString(b))
-	s.Email = sig.Email
-	s.Name = sig.Name
-	s.When = sig.When
-}
-
-// Helper to get a signature from the commit line, which looks like these:
-//
-//	author Patrick Gundlach <gundlach@speedata.de> 1378823654 +0200
-//	author Patrick Gundlach <gundlach@speedata.de> Thu, 07 Apr 2005 22:13:13 +0200
-//
-// but without the "author " at the beginning (this method should)
-// be used for author and committer.
-// FIXME: there are a lot of "return sig, err" (but the err is also nil), that's the old behavior, to avoid breaking
-func newSignatureFromCommitline(line []byte) (sig *Signature, err error) {
-	sig = new(Signature)
-	emailStart := bytes.LastIndexByte(line, '<')
-	emailEnd := bytes.LastIndexByte(line, '>')
-	if emailStart == -1 || emailEnd == -1 || emailEnd < emailStart {
-		return sig, err
-	}
-
-	if emailStart > 0 { // Empty name has already occurred, even if it shouldn't
-		sig.Name = strings.TrimSpace(string(line[:emailStart-1]))
-	}
-	sig.Email = string(line[emailStart+1 : emailEnd])
-
-	hasTime := emailEnd+2 < len(line)
-	if !hasTime {
-		return sig, err
-	}
-
-	// Check date format.
-	firstChar := line[emailEnd+2]
-	if firstChar >= 48 && firstChar <= 57 {
-		idx := bytes.IndexByte(line[emailEnd+2:], ' ')
-		if idx < 0 {
-			return sig, err
-		}
-
-		timestring := string(line[emailEnd+2 : emailEnd+2+idx])
-		seconds, _ := strconv.ParseInt(timestring, 10, 64)
-		sig.When = time.Unix(seconds, 0)
-
-		idx += emailEnd + 3
-		if idx >= len(line) || idx+5 > len(line) {
-			return sig, err
-		}
-
-		timezone := string(line[idx : idx+5])
-		tzhours, err1 := strconv.ParseInt(timezone[0:3], 10, 64)
-		tzmins, err2 := strconv.ParseInt(timezone[3:], 10, 64)
-		if err1 != nil || err2 != nil {
-			return sig, err
-		}
-		if tzhours < 0 {
-			tzmins *= -1
-		}
-		tz := time.FixedZone("", int(tzhours*60*60+tzmins*60))
-		sig.When = sig.When.In(tz)
-	} else {
-		sig.When, err = time.Parse(GitTimeLayout, string(line[emailEnd+2:]))
-		if err != nil {
-			return sig, err
-		}
-	}
-	return sig, err
 }

modules/git/signature_test.go

@@ -0,0 +1,47 @@
+// Copyright 2019 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package git
+
+import (
+	"testing"
+	"time"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestParseSignatureFromCommitLine(t *testing.T) {
+	tests := []struct {
+		line string
+		want *Signature
+	}{
+		{
+			line: "a b <c@d.com> 12345 +0100",
+			want: &Signature{
+				Name:  "a b",
+				Email: "c@d.com",
+				When:  time.Unix(12345, 0).In(time.FixedZone("", 3600)),
+			},
+		},
+		{
+			line: "bad line",
+			want: &Signature{Name: "bad line"},
+		},
+		{
+			line: "bad < line",
+			want: &Signature{Name: "bad < line"},
+		},
+		{
+			line: "bad > line",
+			want: &Signature{Name: "bad > line"},
+		},
+		{
+			line: "bad-line <name@example.com>",
+			want: &Signature{Name: "bad-line <name@example.com>"},
+		},
+	}
+	for _, test := range tests {
+		got := parseSignatureFromCommitLine(test.line)
+		assert.EqualValues(t, test.want, got)
+	}
+}

modules/git/tag.go

@@ -7,6 +7,8 @@ import (
 	"bytes"
 	"sort"
 	"strings"
+
+	"code.gitea.io/gitea/modules/util"
 )
 
 const (
@@ -57,11 +59,7 @@ l:
 				// A commit can have one or more parents
 				tag.Type = string(line[spacepos+1:])
 			case "tagger":
-				sig, err := newSignatureFromCommitline(line[spacepos+1:])
+				tag.Tagger = parseSignatureFromCommitLine(util.UnsafeBytesToString(line[spacepos+1:]))
-				if err != nil {
-					return nil, err
-				}
-				tag.Tagger = sig
 			}
 			nextline += eol + 1
 		case eol == 0: