ParsePatchHeader: Copy functionality of git mailinfo's cleanup_subject

Primarily to get rid of [PATCH] at the front, but while we're here
just be generally compatible with `git am`:

 * Remove `re` and variations
 * Remove whitespace
 * Remove anything in brackets

But only at the very beginning of the subject.

Inspired by
https://github.com/git/git/blob/master/mailinfo.c:cleanup_subject()

Signed-off-by: George Dunlap <[email protected]>

Author: George Dunlap

gitdiff/patch_header.go

@@ -160,10 +160,13 @@ func ParsePatchDate(s string) (time.Time, error) {
 // formats used by git diff, git log, and git show and the UNIX mailbox format
 // used by git format-patch.
 //
-// ParsePatchHeader makes no assumptions about the format of the patch title or
-// message other than trimming whitespace and condensing blank lines. In
-// particular, it does not remove the extra content that git format-patch adds
-// to make emailed patches friendlier, like subject prefixes or commit stats.
+// If ParsePatchHeader detect that it is handling an email, it will
+// remove extra content at the beginning of the title line, such as
+// `[PATCH]` or `Re:` in the same way that `git mailinfo` does.  (`git
+// mailinfo` is the core part of `git am` that pulls information out
+// of an individual mail.)  Unline `git mailinfo`, it does not at the
+// moment remove commit states or other extraneous matter after a
+// `---` line.
 func ParsePatchHeader(s string) (*PatchHeader, error) {
 	r := bufio.NewReader(strings.NewReader(s))
 
@@ -359,7 +362,8 @@ func parseHeaderMail(mailLine string, r io.Reader) (*PatchHeader, error) {
 		h.AuthorDate = d
 	}
 
-	h.Title = msg.Header.Get("Subject")
+	subject := msg.Header.Get("Subject")
+	h.Title = cleanupSubject(subject)
 
 	s := bufio.NewScanner(msg.Body)
 	h.Body = scanMessageBody(s, "")
@@ -369,3 +373,61 @@ func parseHeaderMail(mailLine string, r io.Reader) (*PatchHeader, error) {
 
 	return h, nil
 }
+
+func cleanupSubject(s string) string {
+	r := []rune(s)
+
+	// This is meant to be compatible with
+	// https://github.com/git/git/blob/master/mailinfo.c:cleanup_subject().
+	// If compatibility with `git am` drifts, go there to see if there
+	// are any updates.
+
+	var at int
+	for at = 0; at < len(r); {
+		switch r[at] {
+		case 'r', 'R':
+			// Detect re:, Re:, rE: and RE:
+			if at+3 > len(r) {
+				break
+			}
+			if (r[at+1] == 'e' || r[at+1] == 'E') &&
+				r[at+2] == ':' {
+				at += 3
+				continue
+			}
+
+		case ' ', '\t', ':':
+			// Delete whitespace and duplicate ':' characters
+			at++
+			continue
+
+		case '[':
+			// Look for closing parenthesis
+			var closepos int
+			for j := at + 1; j < len(r); j++ {
+				if r[j] == ']' {
+					closepos = j
+					break
+				}
+			}
+
+			if closepos == 0 {
+				break
+			}
+
+			at = closepos + 1
+			continue
+		}
+
+		// Only loop if we actually removed something
+		break
+	}
+
+	if at > 0 {
+		for i := 0; i < len(r)-at; i++ {
+			r[i] = r[at+i]
+		}
+	}
+
+	return string(r[:len(r)-at])
+}

gitdiff/patch_header_test.go

@@ -236,7 +236,7 @@ Another body line.
 				SHA:        expectedSHA,
 				Author:     expectedIdentity,
 				AuthorDate: expectedDate,
-				Title:      "[PATCH] " + expectedTitle,
+				Title:      expectedTitle,
 				Body:       expectedBody,
 			},
 		},
@@ -348,3 +348,43 @@ func assertPatchIdentity(t *testing.T, kind string, exp, act *PatchIdentity) {
 		t.Errorf("incorrect parsed %s, expected %+v, bot got %+v", kind, exp, act)
 	}
 }
+
+func TestCleanupSubject(t *testing.T) {
+	exp := "A sample commit to test header parsing"
+	tests := map[string]string{
+		"plain":        "A sample commit to test header parsing",
+		"patch":        "[PATCH] A sample commit to test header parsing",
+		"patchv5":      "[PATCH v5] A sample commit to test header parsing",
+		"patchrfc":     "[PATCH RFC] A sample commit to test header parsing",
+		"patchnospace": "[PATCH]A sample commit to test header parsing",
+		"space":        "   A sample commit to test header parsing",
+		"re":           "re: A sample commit to test header parsing",
+		"Re":           "Re: A sample commit to test header parsing",
+		"RE":           "rE: A sample commit to test header parsing",
+		"rere":         "re: re: A sample commit to test header parsing",
+	}
+
+	for name, test := range tests {
+		act := cleanupSubject(test)
+		if act != exp {
+			t.Errorf("%s: Incorrect cleanup of %s: got %s, wanted %s",
+				name, test, act, exp)
+		}
+	}
+
+	moretests := map[string]struct {
+		in, exp string
+	}{
+		"Reimplement":       {"Reimplement something", "Reimplement something"},
+		"patch-reimplement": {"[PATCH v5] Reimplement something", "Reimplement something"},
+		"Openbracket":       {"[Just to annoy people", "[Just to annoy people"},
+	}
+
+	for name, test := range moretests {
+		act := cleanupSubject(test.in)
+		if act != test.exp {
+			t.Errorf("%s: Incorrect cleanup of %s: got %s, wanted %s",
+				name, test.in, act, test.exp)
+		}
+	}
+}