format-patch: RFC 2047 says multi-octet character may not be split

Even though an earlier attempt (bafc478..41dd00bad) cleaned
up RFC 2047 encoding, pretty.c::add_rfc2047() still decides
where to split the output line by going through the input
one byte at a time, and potentially splits a character in
the middle.  A subject line may end up showing like this:

     ".... fö?? bar".   (instead of  ".... föö bar".)

if split incorrectly.

RFC 2047, section 5 (3) explicitly forbids such beaviour

    Each 'encoded-word' MUST represent an integral number of
    characters.  A multi-octet character may not be split across
    adjacent 'encoded- word's.

that means that e.g. for

    Subject: .... föö bar

encoding

    Subject: =?UTF-8?q?....=20f=C3=B6=C3=B6?=
     =?UTF-8?q?=20bar?=

is correct, and

    Subject: =?UTF-8?q?....=20f=C3=B6=C3?=      <-- NOTE ö is broken here
     =?UTF-8?q?=B6=20bar?=

is not, because "ö" character UTF-8 encoding C3 B6 is split here across
adjacent encoded words.

To fix the problem, make the loop grab one _character_ at a time and
determine its output length to see where to break the output line.  Note
that this version only knows about UTF-8, but the logic to grab one
character is abstracted out in mbs_chrlen() function to make it possible
to extend it to other encodings with the help of iconv in the future.

Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>
Signed-off-by: Junio C Hamano <gitster@pobox.com>

pretty.c

@@ -345,7 +345,7 @@ static int needs_rfc2047_encoding(const char *line, int len,
 	return 0;
 }
 
-static void add_rfc2047(struct strbuf *sb, const char *line, int len,
+static void add_rfc2047(struct strbuf *sb, const char *line, size_t len,
 		       const char *encoding, enum rfc2047_type type)
 {
 	static const int max_encoded_length = 76; /* per rfc2047 */
@@ -355,9 +355,22 @@ static void add_rfc2047(struct strbuf *sb, const char *line, int len,
 	strbuf_grow(sb, len * 3 + strlen(encoding) + 100);
 	strbuf_addf(sb, "=?%s?q?", encoding);
 	line_len += strlen(encoding) + 5; /* 5 for =??q? */
-	for (i = 0; i < len; i++) {
-		unsigned ch = line[i] & 0xFF;
-		int is_special = is_rfc2047_special(ch, type);
+
+	while (len) {
+		/*
+		 * RFC 2047, section 5 (3):
+		 *
+		 * Each 'encoded-word' MUST represent an integral number of
+		 * characters.  A multi-octet character may not be split across
+		 * adjacent 'encoded- word's.
+		 */
+		const unsigned char *p = (const unsigned char *)line;
+		int chrlen = mbs_chrlen(&line, &len, encoding);
+		int is_special = (chrlen > 1) || is_rfc2047_special(*p, type);
+
+		/* "=%02X" * chrlen, or the byte itself */
+		const char *encoded_fmt = is_special ? "=%02X"    : "%c";
+		int	    encoded_len = is_special ? 3 * chrlen : 1;
 
 		/*
 		 * According to RFC 2047, we could encode the special character
@@ -367,18 +380,15 @@ static void add_rfc2047(struct strbuf *sb, const char *line, int len,
 		 * causes ' ' to be encoded as '=20', avoiding this problem.
 		 */
 
-		if (line_len + 2 + (is_special ? 3 : 1) > max_encoded_length) {
+		if (line_len + encoded_len + 2 > max_encoded_length) {
+			/* It won't fit with trailing "?=" --- break the line */
 			strbuf_addf(sb, "?=\n =?%s?q?", encoding);
 			line_len = strlen(encoding) + 5 + 1; /* =??q? plus SP */
 		}
 
-		if (is_special) {
-			strbuf_addf(sb, "=%02X", ch);
-			line_len += 3;
-		} else {
-			strbuf_addch(sb, ch);
-			line_len++;
-		}
+		for (i = 0; i < chrlen; i++)
+			strbuf_addf(sb, encoded_fmt, p[i]);
+		line_len += encoded_len;
 	}
 	strbuf_addstr(sb, "?=");
 }

t/t4014-format-patch.sh

@@ -810,25 +810,26 @@ Subject: [PATCH] =?UTF-8?q?f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f?=
  =?UTF-8?q?=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar?=
  =?UTF-8?q?=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20?=
  =?UTF-8?q?bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6?=
- =?UTF-8?q?=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3?=
- =?UTF-8?q?=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6?=
- =?UTF-8?q?=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3?=
- =?UTF-8?q?=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f?=
+ =?UTF-8?q?=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6?=
+ =?UTF-8?q?=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f?=
  =?UTF-8?q?=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar?=
  =?UTF-8?q?=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20?=
  =?UTF-8?q?bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6?=
- =?UTF-8?q?=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3?=
- =?UTF-8?q?=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6?=
- =?UTF-8?q?=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3?=
- =?UTF-8?q?=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f?=
+ =?UTF-8?q?=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6?=
+ =?UTF-8?q?=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f?=
  =?UTF-8?q?=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar?=
  =?UTF-8?q?=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20?=
  =?UTF-8?q?bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6?=
- =?UTF-8?q?=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3?=
- =?UTF-8?q?=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6?=
- =?UTF-8?q?=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3?=
- =?UTF-8?q?=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f?=
- =?UTF-8?q?=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar?=
+ =?UTF-8?q?=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6?=
+ =?UTF-8?q?=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f?=
+ =?UTF-8?q?=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar?=
+ =?UTF-8?q?=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20?=
+ =?UTF-8?q?bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6?=
+ =?UTF-8?q?=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6?=
+ =?UTF-8?q?=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f?=
+ =?UTF-8?q?=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar?=
+ =?UTF-8?q?=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20?=
+ =?UTF-8?q?bar?=
 EOF
 test_expect_success 'format-patch wraps extremely long subject (rfc2047)' '
 	rm -rf patches/ &&

utf8.c

@@ -495,3 +495,42 @@ char *reencode_string(const char *in, const char *out_encoding, const char *in_e
 	return out;
 }
 #endif
+
+/*
+ * Returns first character length in bytes for multi-byte `text` according to
+ * `encoding`.
+ *
+ * - The `text` pointer is updated to point at the next character.
+ * - When `remainder_p` is not NULL, on entry `*remainder_p` is how much bytes
+ *   we can consume from text, and on exit `*remainder_p` is reduced by returned
+ *   character length. Otherwise `text` is treated as limited by NUL.
+ */
+int mbs_chrlen(const char **text, size_t *remainder_p, const char *encoding)
+{
+	int chrlen;
+	const char *p = *text;
+	size_t r = (remainder_p ? *remainder_p : SIZE_MAX);
+
+	if (r < 1)
+		return 0;
+
+	if (is_encoding_utf8(encoding)) {
+		pick_one_utf8_char(&p, &r);
+
+		chrlen = p ? (p - *text)
+			   : 1 /* not valid UTF-8 -> raw byte sequence */;
+	}
+	else {
+		/*
+		 * TODO use iconv to decode one char and obtain its chrlen
+		 * for now, let's treat encodings != UTF-8 as one-byte
+		 */
+		chrlen = 1;
+	}
+
+	*text += chrlen;
+	if (remainder_p)
+		*remainder_p -= chrlen;
+
+	return chrlen;
+}

utf8.h

@@ -21,4 +21,6 @@ char *reencode_string(const char *in, const char *out_encoding, const char *in_e
 #define reencode_string(a,b,c) NULL
 #endif
 
+int mbs_chrlen(const char **text, size_t *remainder_p, const char *encoding);
+
 #endif