pkg/git: extract relevant email addresses from commits

Update #501

2 files changed, 111 insertions, 9 deletions

diff --git a/pkg/git/git.go b/pkg/git/git.go
index 59715ed07..899930bff 100644
--- a/pkg/git/git.go
+++ b/pkg/git/git.go
@@ -131,9 +131,11 @@ func initRepo(dir string) error {
 }
 
 type Commit struct {
-	Hash  string
-	Title string
-	Date  time.Time
+	Hash   string
+	Title  string
+	Author string
+	CC     []string
+	Date   time.Time
 }
 
 // HeadCommit returns info about the HEAD commit of the current branch of git repository in dir.
@@ -142,22 +144,49 @@ func HeadCommit(dir string) (*Commit, error) {
 }
 
 func GetCommit(dir, commit string) (*Commit, error) {
-	output, err := runSandboxed(dir, "git", "log", "--pretty=format:%H%n%s%n%ad", "-n", "1", commit)
+	output, err := runSandboxed(dir, "git", "log", "--format=%H%n%s%n%ae%n%ad%n%b", "-n", "1", commit)
 	if err != nil {
 		return nil, err
 	}
+	return parseCommit(output)
+}
+
+func parseCommit(output []byte) (*Commit, error) {
 	lines := bytes.Split(output, []byte{'\n'})
-	if len(lines) != 3 || len(lines[0]) != 40 {
+	if len(lines) < 4 || len(lines[0]) != 40 {
 		return nil, fmt.Errorf("unexpected git log output: %q", output)
 	}
-	date, err := time.Parse(DateFormat, string(lines[2]))
+	date, err := time.Parse(DateFormat, string(lines[3]))
 	if err != nil {
 		return nil, fmt.Errorf("failed to parse date in git log output: %v\n%q", err, output)
 	}
+	cc := make(map[string]bool)
+	cc[strings.ToLower(string(lines[2]))] = true
+	for _, line := range lines[4:] {
+		for _, re := range ccRes {
+			matches := re.FindSubmatchIndex(line)
+			if matches == nil {
+				continue
+			}
+			addr, err := mail.ParseAddress(string(line[matches[2]:matches[3]]))
+			if err != nil {
+				break
+			}
+			cc[strings.ToLower(addr.Address)] = true
+			break
+		}
+	}
+	sortedCC := make([]string, 0, len(cc))
+	for addr := range cc {
+		sortedCC = append(sortedCC, addr)
+	}
+	sort.Strings(sortedCC)
 	com := &Commit{
-		Hash:  string(lines[0]),
-		Title: string(lines[1]),
-		Date:  date,
+		Hash:   string(lines[0]),
+		Title:  string(lines[1]),
+		Author: string(lines[2]),
+		CC:     sortedCC,
+		Date:   date,
 	}
 	return com, nil
 }
@@ -456,4 +485,12 @@ var (
 	gitBranchRe  = regexp.MustCompile("^[a-zA-Z0-9-_/.]{2,200}$")
 	gitHashRe    = regexp.MustCompile("^[a-f0-9]+$")
 	releaseTagRe = regexp.MustCompile(`^v([0-9]+).([0-9]+)(?:\.([0-9]+))?$`)
+	ccRes        = []*regexp.Regexp{
+		regexp.MustCompile(`^Reviewed\-.*: (.*)$`),
+		regexp.MustCompile(`^[A-Za-z-]+\-and\-[Rr]eviewed\-.*: (.*)$`),
+		regexp.MustCompile(`^Acked\-.*: (.*)$`),
+		regexp.MustCompile(`^[A-Za-z-]+\-and\-[Aa]cked\-.*: (.*)$`),
+		regexp.MustCompile(`^Tested\-.*: (.*)$`),
+		regexp.MustCompile(`^[A-Za-z-]+\-and\-[Tt]ested\-.*: (.*)$`),
+	}
 )
diff --git a/pkg/git/git_test.go b/pkg/git/git_test.go
index 66159107f..86f033acf 100644
--- a/pkg/git/git_test.go
+++ b/pkg/git/git_test.go
@@ -7,8 +7,73 @@ import (
 	"reflect"
 	"strings"
 	"testing"
+	"time"
 )
 
+func TestParseCommit(t *testing.T) {
+	tests := map[string]*Commit{
+		`2075b16e32c26e4031b9fd3cbe26c54676a8fcb5
+rbtree: include rcu.h
+foobar@foobar.de
+Fri May 11 16:02:14 2018 -0700
+Since commit c1adf20052d8 ("Introduce rb_replace_node_rcu()")
+rbtree_augmented.h uses RCU related data structures but does not include
+the header file.  It works as long as it gets somehow included before
+that and fails otherwise.
+
+Link: http://lkml.kernel.org/r/20180504103159.19938-1-bigeasy@linutronix.de
+Signed-off-by: Foo Bad Baz <another@email.de>
+Reviewed-by: <yetanother@email.org>
+Cc: Unrelated Guy <somewhere@email.com>
+Acked-by: Subsystem reviewer <Subsystem@reviewer.com>
+Reported-and-tested-by: and@me.com
+Reported-and-Tested-by: Name-name <name@name.com>
+Tested-by: Must be correct <mustbe@correct.com>
+Signed-off-by: Linux Master <linux@linux-foundation.org>
+`: &Commit{
+			Hash:   "2075b16e32c26e4031b9fd3cbe26c54676a8fcb5",
+			Title:  "rbtree: include rcu.h",
+			Author: "foobar@foobar.de",
+			CC: []string{
+				"and@me.com",
+				"foobar@foobar.de",
+				"mustbe@correct.com",
+				"name@name.com",
+				"subsystem@reviewer.com",
+				"yetanother@email.org",
+			},
+			Date: time.Date(2018, 5, 11, 16, 02, 14, 0, time.FixedZone("", -7*60*60)),
+		},
+	}
+	for input, com := range tests {
+		res, err := parseCommit([]byte(input))
+		if err != nil && com != nil {
+			t.Fatalf("want %+v, got error: %v", com, err)
+		}
+		if err == nil && com == nil {
+			t.Fatalf("want error, got commit %+v", res)
+		}
+		if com == nil {
+			continue
+		}
+		if com.Hash != res.Hash {
+			t.Fatalf("want hash %q, got %q", com.Hash, res.Hash)
+		}
+		if com.Title != res.Title {
+			t.Fatalf("want title %q, got %q", com.Title, res.Title)
+		}
+		if com.Author != res.Author {
+			t.Fatalf("want author %q, got %q", com.Author, res.Author)
+		}
+		if !reflect.DeepEqual(com.CC, res.CC) {
+			t.Fatalf("want CC %q, got %q", com.CC, res.CC)
+		}
+		if !com.Date.Equal(res.Date) {
+			t.Fatalf("want date %v, got %v", com.Date, res.Date)
+		}
+	}
+}
+
 func TestCanonicalizeCommit(t *testing.T) {
 	tests := map[string]string{
 		"foo bar":                     "foo bar",