Decode quoted-printable UTF8 in email subjects (#25)

Author: jmcampanini

gitdiff/patch_header.go

@@ -5,6 +5,7 @@ import (
 	"errors"
 	"fmt"
 	"io"
+	"mime/quotedprintable"
 	"net/mail"
 	"strconv"
 	"strings"
@@ -457,5 +458,30 @@ func parseSubject(s string) (string, string) {
 		break
 	}
 
-	return s[:at], s[at:]
+	return s[:at], decodeSubject(s[at:])
+}
+
+// Decodes a subject line. Currently only supports quoted-printable UTF-8. This format is the result
+// of a `git format-patch` when the commit title has a non-ASCII character (i.e. an emoji).
+// See for reference: https://stackoverflow.com/questions/27695749/gmail-api-not-respecting-utf-encoding-in-subject
+func decodeSubject(encoded string) string {
+	if !strings.HasPrefix(encoded, "=?UTF-8?q?") {
+		// not UTF-8 encoded
+		return encoded
+	}
+
+	// If the subject is too long, `git format-patch` may produce a subject line across
+	// multiple lines. When parsed, this can look like the following:
+	// <UTF8-prefix><first-line> <UTF8-prefix><second-line>
+	payload := " " + encoded
+	payload = strings.ReplaceAll(payload, " =?UTF-8?q?", "")
+	payload = strings.ReplaceAll(payload, "?=", "")
+
+	decoded, err := io.ReadAll(quotedprintable.NewReader(strings.NewReader(payload)))
+	if err != nil {
+		// if err, abort decoding and return original subject
+		return encoded
+	}
+
+	return string(decoded)
 }

gitdiff/patch_header_test.go

@@ -138,6 +138,8 @@ func TestParsePatchHeader(t *testing.T) {
 	}
 	expectedDate := time.Date(2020, 04, 11, 15, 21, 23, 0, time.FixedZone("PDT", -7*60*60))
 	expectedTitle := "A sample commit to test header parsing"
+	expectedEmojiOneLineTitle := "🤖 Enabling auto-merging"
+	expectedEmojiMultiLineTitle := "[IA64] Put ia64 config files on the Uwe Kleine-König diet"
 	expectedBody := "The medium format shows the body, which\nmay wrap on to multiple lines.\n\nAnother body line."
 	expectedBodyAppendix := "CC: Joe Smith <[email protected]>"
 
@@ -267,6 +269,45 @@ Another body line.
 				Body:       expectedBody,
 			},
 		},
+		"mailboxEmojiOneLine": {
+			Input: `From 61f5cd90bed4d204ee3feb3aa41ee91d4734855b Mon Sep 17 00:00:00 2001
+From: Morton Haypenny <[email protected]>
+Date: Sat, 11 Apr 2020 15:21:23 -0700
+Subject: [PATCH] =?UTF-8?q?=F0=9F=A4=96=20Enabling=20auto-merging?=
+
+The medium format shows the body, which
+may wrap on to multiple lines.
+
+Another body line.
+`,
+			Header: PatchHeader{
+				SHA:        expectedSHA,
+				Author:     expectedIdentity,
+				AuthorDate: expectedDate,
+				Title:      expectedEmojiOneLineTitle,
+				Body:       expectedBody,
+			},
+		},
+		"mailboxEmojiMultiLine": {
+			Input: `From 61f5cd90bed4d204ee3feb3aa41ee91d4734855b Mon Sep 17 00:00:00 2001
+From: Morton Haypenny <[email protected]>
+Date: Sat, 11 Apr 2020 15:21:23 -0700
+Subject: [PATCH] =?UTF-8?q?[IA64]=20Put=20ia64=20config=20files=20on=20the=20?=
+ =?UTF-8?q?Uwe=20Kleine-K=C3=B6nig=20diet?=
+
+The medium format shows the body, which
+may wrap on to multiple lines.
+
+Another body line.
+`,
+			Header: PatchHeader{
+				SHA:        expectedSHA,
+				Author:     expectedIdentity,
+				AuthorDate: expectedDate,
+				Title:      expectedEmojiMultiLineTitle,
+				Body:       expectedBody,
+			},
+		},
 		"mailboxAppendix": {
 			Input: `From 61f5cd90bed4d204ee3feb3aa41ee91d4734855b Mon Sep 17 00:00:00 2001
 From: Morton Haypenny <[email protected]>