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>
Kirill Smelkov committed
Mar 7, 2013 at 14:55 UTC
6cd3c0532772749bcf6c4688f34c8ecc65ecb655
4 files changed
+77
-25
pretty.c
+22
-12
@@ -345,7 +345,7 @@ static int needs_rfc2047_encoding(const char *line, int len,
345
return 0;
346
}
347
348
-static void add_rfc2047(struct strbuf *sb, const char *line, int len,
348
+static void add_rfc2047(struct strbuf *sb, const char *line, size_t len,
349
const char *encoding, enum rfc2047_type type)
350
{
351
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,
355
strbuf_grow(sb, len * 3 + strlen(encoding) + 100);
356
strbuf_addf(sb, "=?%s?q?", encoding);
357
line_len += strlen(encoding) + 5; /* 5 for =??q? */
358
- for (i = 0; i < len; i++) {
359
- unsigned ch = line[i] & 0xFF;
360
- int is_special = is_rfc2047_special(ch, type);
358
+
359
+ while (len) {
360
+ /*
361
+ * RFC 2047, section 5 (3):
362
+ *
363
+ * Each 'encoded-word' MUST represent an integral number of
364
+ * characters. A multi-octet character may not be split across
365
+ * adjacent 'encoded- word's.
366
+ */
367
+ const unsigned char *p = (const unsigned char *)line;
368
+ int chrlen = mbs_chrlen(&line, &len, encoding);
369
+ int is_special = (chrlen > 1) || is_rfc2047_special(*p, type);
370
+
371
+ /* "=%02X" * chrlen, or the byte itself */
372
+ const char *encoded_fmt = is_special ? "=%02X" : "%c";
373
+ int encoded_len = is_special ? 3 * chrlen : 1;
374
375
/*
376
* 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,
380
* causes ' ' to be encoded as '=20', avoiding this problem.
381
*/
382
370
- if (line_len + 2 + (is_special ? 3 : 1) > max_encoded_length) {
383
+ if (line_len + encoded_len + 2 > max_encoded_length) {
384
+ /* It won't fit with trailing "?=" --- break the line */
385
strbuf_addf(sb, "?=\n =?%s?q?", encoding);
386
line_len = strlen(encoding) + 5 + 1; /* =??q? plus SP */
387
}
388
375
- if (is_special) {
376
- strbuf_addf(sb, "=%02X", ch);
377
- line_len += 3;
378
- } else {
379
- strbuf_addch(sb, ch);
380
- line_len++;
381
- }
389
+ for (i = 0; i < chrlen; i++)
390
+ strbuf_addf(sb, encoded_fmt, p[i]);
391
+ line_len += encoded_len;
392
}
393
strbuf_addstr(sb, "?=");
394
}
t/t4014-format-patch.sh
+14
-13
@@ -810,25 +810,26 @@ Subject: [PATCH] =?UTF-8?q?f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f?=
810
=?UTF-8?q?=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar?=
811
=?UTF-8?q?=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20?=
812
=?UTF-8?q?bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6?=
813
- =?UTF-8?q?=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3?=
814
- =?UTF-8?q?=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6?=
815
- =?UTF-8?q?=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3?=
816
- =?UTF-8?q?=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f?=
813
+ =?UTF-8?q?=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6?=
814
+ =?UTF-8?q?=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f?=
815
=?UTF-8?q?=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar?=
816
=?UTF-8?q?=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20?=
817
=?UTF-8?q?bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6?=
820
- =?UTF-8?q?=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3?=
821
- =?UTF-8?q?=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6?=
822
- =?UTF-8?q?=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3?=
823
- =?UTF-8?q?=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f?=
818
+ =?UTF-8?q?=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6?=
819
+ =?UTF-8?q?=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f?=
820
=?UTF-8?q?=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar?=
821
=?UTF-8?q?=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20?=
822
=?UTF-8?q?bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6?=
827
- =?UTF-8?q?=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3?=
828
- =?UTF-8?q?=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6?=
829
- =?UTF-8?q?=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3?=
830
- =?UTF-8?q?=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f?=
831
- =?UTF-8?q?=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar?=
823
+ =?UTF-8?q?=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6?=
824
+ =?UTF-8?q?=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f?=
825
+ =?UTF-8?q?=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar?=
826
+ =?UTF-8?q?=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20?=
827
+ =?UTF-8?q?bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6?=
828
+ =?UTF-8?q?=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6?=
829
+ =?UTF-8?q?=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f?=
830
+ =?UTF-8?q?=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar?=
831
+ =?UTF-8?q?=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20bar=20f=C3=B6=C3=B6=20?=
832
+ =?UTF-8?q?bar?=
833
EOF
834
test_expect_success 'format-patch wraps extremely long subject (rfc2047)' '
835
rm -rf patches/ &&
utf8.c
+39
@@ -495,3 +495,42 @@ char *reencode_string(const char *in, const char *out_encoding, const char *in_e
495
return out;
496
}
497
#endif
498
+
499
+/*
500
+ * Returns first character length in bytes for multi-byte `text` according to
501
+ * `encoding`.
502
+ *
503
+ * - The `text` pointer is updated to point at the next character.
504
+ * - When `remainder_p` is not NULL, on entry `*remainder_p` is how much bytes
505
+ * we can consume from text, and on exit `*remainder_p` is reduced by returned
506
+ * character length. Otherwise `text` is treated as limited by NUL.
507
+ */
508
+int mbs_chrlen(const char **text, size_t *remainder_p, const char *encoding)
509
+{
510
+ int chrlen;
511
+ const char *p = *text;
512
+ size_t r = (remainder_p ? *remainder_p : SIZE_MAX);
513
+
514
+ if (r < 1)
515
+ return 0;
516
+
517
+ if (is_encoding_utf8(encoding)) {
518
+ pick_one_utf8_char(&p, &r);
519
+
520
+ chrlen = p ? (p - *text)
521
+ : 1 /* not valid UTF-8 -> raw byte sequence */;
522
+ }
523
+ else {
524
+ /*
525
+ * TODO use iconv to decode one char and obtain its chrlen
526
+ * for now, let's treat encodings != UTF-8 as one-byte
527
+ */
528
+ chrlen = 1;
529
+ }
530
+
531
+ *text += chrlen;
532
+ if (remainder_p)
533
+ *remainder_p -= chrlen;
534
+
535
+ return chrlen;
536
+}
utf8.h
+2
@@ -21,4 +21,6 @@ char *reencode_string(const char *in, const char *out_encoding, const char *in_e
21
#define reencode_string(a,b,c) NULL
22
#endif
23
24
+int mbs_chrlen(const char **text, size_t *remainder_p, const char *encoding);
25
+
26
#endif