convert logmsg_reencode to get_commit_buffer
[git/git.git] / pretty.c
CommitLineData
93fc05eb
JS
1#include "cache.h"
2#include "commit.h"
93fc05eb
JS
3#include "utf8.h"
4#include "diff.h"
5#include "revision.h"
c455c87c 6#include "string-list.h"
e0cbc397 7#include "mailmap.h"
3b3d443f 8#include "log-tree.h"
a97a7468 9#include "notes.h"
c002922a 10#include "color.h"
8f8f5476 11#include "reflog-walk.h"
f6667c5e 12#include "gpg-interface.h"
93fc05eb 13
93fc05eb 14static char *user_format;
40957891
WP
15static struct cmt_fmt_map {
16 const char *name;
17 enum cmit_fmt format;
18 int is_tformat;
2d7671ef
WP
19 int is_alias;
20 const char *user_format;
40957891 21} *commit_formats;
8028184e 22static size_t builtin_formats_len;
40957891 23static size_t commit_formats_len;
8028184e 24static size_t commit_formats_alloc;
40957891 25static struct cmt_fmt_map *find_commit_format(const char *sought);
93fc05eb 26
36407548
NS
27static void save_user_format(struct rev_info *rev, const char *cp, int is_tformat)
28{
29 free(user_format);
30 user_format = xstrdup(cp);
31 if (is_tformat)
32 rev->use_terminator = 1;
33 rev->commit_format = CMIT_FMT_USERFORMAT;
34}
35
8028184e 36static int git_pretty_formats_config(const char *var, const char *value, void *cb)
93fc05eb 37{
8028184e
WP
38 struct cmt_fmt_map *commit_format = NULL;
39 const char *name;
40 const char *fmt;
93fc05eb 41 int i;
8028184e 42
59556548 43 if (!starts_with(var, "pretty."))
8028184e
WP
44 return 0;
45
46 name = var + strlen("pretty.");
47 for (i = 0; i < builtin_formats_len; i++) {
48 if (!strcmp(commit_formats[i].name, name))
49 return 0;
50 }
51
52 for (i = builtin_formats_len; i < commit_formats_len; i++) {
53 if (!strcmp(commit_formats[i].name, name)) {
54 commit_format = &commit_formats[i];
55 break;
56 }
57 }
58
59 if (!commit_format) {
60 ALLOC_GROW(commit_formats, commit_formats_len+1,
61 commit_formats_alloc);
62 commit_format = &commit_formats[commit_formats_len];
95a2618f 63 memset(commit_format, 0, sizeof(*commit_format));
8028184e
WP
64 commit_formats_len++;
65 }
66
67 commit_format->name = xstrdup(name);
68 commit_format->format = CMIT_FMT_USERFORMAT;
69 git_config_string(&fmt, var, value);
59556548 70 if (starts_with(fmt, "format:") || starts_with(fmt, "tformat:")) {
8028184e
WP
71 commit_format->is_tformat = fmt[0] == 't';
72 fmt = strchr(fmt, ':') + 1;
73 } else if (strchr(fmt, '%'))
74 commit_format->is_tformat = 1;
75 else
76 commit_format->is_alias = 1;
77 commit_format->user_format = fmt;
78
79 return 0;
80}
81
40957891 82static void setup_commit_formats(void)
93fc05eb 83{
40957891
WP
84 struct cmt_fmt_map builtin_formats[] = {
85 { "raw", CMIT_FMT_RAW, 0 },
86 { "medium", CMIT_FMT_MEDIUM, 0 },
87 { "short", CMIT_FMT_SHORT, 0 },
88 { "email", CMIT_FMT_EMAIL, 0 },
89 { "fuller", CMIT_FMT_FULLER, 0 },
90 { "full", CMIT_FMT_FULL, 0 },
91 { "oneline", CMIT_FMT_ONELINE, 1 }
4da45bef 92 };
40957891 93 commit_formats_len = ARRAY_SIZE(builtin_formats);
8028184e
WP
94 builtin_formats_len = commit_formats_len;
95 ALLOC_GROW(commit_formats, commit_formats_len, commit_formats_alloc);
40957891
WP
96 memcpy(commit_formats, builtin_formats,
97 sizeof(*builtin_formats)*ARRAY_SIZE(builtin_formats));
8028184e
WP
98
99 git_config(git_pretty_formats_config, NULL);
40957891
WP
100}
101
2d7671ef
WP
102static struct cmt_fmt_map *find_commit_format_recursive(const char *sought,
103 const char *original,
104 int num_redirections)
40957891
WP
105{
106 struct cmt_fmt_map *found = NULL;
107 size_t found_match_len = 0;
108 int i;
109
2d7671ef
WP
110 if (num_redirections >= commit_formats_len)
111 die("invalid --pretty format: "
112 "'%s' references an alias which points to itself",
113 original);
40957891
WP
114
115 for (i = 0; i < commit_formats_len; i++) {
116 size_t match_len;
117
59556548 118 if (!starts_with(commit_formats[i].name, sought))
40957891
WP
119 continue;
120
121 match_len = strlen(commit_formats[i].name);
122 if (found == NULL || found_match_len > match_len) {
123 found = &commit_formats[i];
124 found_match_len = match_len;
125 }
126 }
2d7671ef
WP
127
128 if (found && found->is_alias) {
129 found = find_commit_format_recursive(found->user_format,
130 original,
131 num_redirections+1);
132 }
133
40957891
WP
134 return found;
135}
136
2d7671ef
WP
137static struct cmt_fmt_map *find_commit_format(const char *sought)
138{
139 if (!commit_formats)
140 setup_commit_formats();
141
142 return find_commit_format_recursive(sought, sought, 0);
143}
144
40957891
WP
145void get_commit_format(const char *arg, struct rev_info *rev)
146{
147 struct cmt_fmt_map *commit_format;
93fc05eb 148
4da45bef
JH
149 rev->use_terminator = 0;
150 if (!arg || !*arg) {
151 rev->commit_format = CMIT_FMT_DEFAULT;
152 return;
153 }
59556548 154 if (starts_with(arg, "format:") || starts_with(arg, "tformat:")) {
36407548 155 save_user_format(rev, strchr(arg, ':') + 1, arg[0] == 't');
4da45bef 156 return;
93fc05eb 157 }
40957891 158
36407548
NS
159 if (strchr(arg, '%')) {
160 save_user_format(rev, arg, 1);
161 return;
162 }
93fc05eb 163
40957891
WP
164 commit_format = find_commit_format(arg);
165 if (!commit_format)
166 die("invalid --pretty format: %s", arg);
167
168 rev->commit_format = commit_format->format;
169 rev->use_terminator = commit_format->is_tformat;
8028184e
WP
170 if (commit_format->format == CMIT_FMT_USERFORMAT) {
171 save_user_format(rev, commit_format->user_format,
172 commit_format->is_tformat);
173 }
93fc05eb
JS
174}
175
176/*
177 * Generic support for pretty-printing the header
178 */
179static int get_one_line(const char *msg)
180{
181 int ret = 0;
182
183 for (;;) {
184 char c = *msg++;
185 if (!c)
186 break;
187 ret++;
188 if (c == '\n')
189 break;
190 }
191 return ret;
192}
193
194/* High bit set, or ISO-2022-INT */
cc571142 195static int non_ascii(int ch)
93fc05eb 196{
c2e9364a 197 return !isascii(ch) || ch == '\033';
93fc05eb
JS
198}
199
28e9cf65
JS
200int has_non_ascii(const char *s)
201{
202 int ch;
203 if (!s)
204 return 0;
205 while ((ch = *s++) != '\0') {
206 if (non_ascii(ch))
207 return 1;
208 }
209 return 0;
210}
211
4d03c18a
JK
212static int is_rfc822_special(char ch)
213{
214 switch (ch) {
215 case '(':
216 case ')':
217 case '<':
218 case '>':
219 case '[':
220 case ']':
221 case ':':
222 case ';':
223 case '@':
224 case ',':
225 case '.':
226 case '"':
227 case '\\':
228 return 1;
229 default:
230 return 0;
231 }
232}
233
41dd00ba 234static int needs_rfc822_quoting(const char *s, int len)
4d03c18a
JK
235{
236 int i;
237 for (i = 0; i < len; i++)
238 if (is_rfc822_special(s[i]))
239 return 1;
240 return 0;
241}
242
f9b7204b
JS
243static int last_line_length(struct strbuf *sb)
244{
245 int i;
246
247 /* How many bytes are already used on the last line? */
248 for (i = sb->len - 1; i >= 0; i--)
249 if (sb->buf[i] == '\n')
250 break;
251 return sb->len - (i + 1);
252}
253
4d03c18a
JK
254static void add_rfc822_quoted(struct strbuf *out, const char *s, int len)
255{
256 int i;
257
258 /* just a guess, we may have to also backslash-quote */
259 strbuf_grow(out, len + 2);
260
261 strbuf_addch(out, '"');
262 for (i = 0; i < len; i++) {
263 switch (s[i]) {
264 case '"':
265 case '\\':
266 strbuf_addch(out, '\\');
267 /* fall through */
268 default:
269 strbuf_addch(out, s[i]);
270 }
271 }
272 strbuf_addch(out, '"');
273}
274
0fcec2ce
JS
275enum rfc2047_type {
276 RFC2047_SUBJECT,
277 RFC2047_ADDRESS,
278};
279
280static int is_rfc2047_special(char ch, enum rfc2047_type type)
93fc05eb 281{
0fcec2ce
JS
282 /*
283 * rfc2047, section 4.2:
284 *
285 * 8-bit values which correspond to printable ASCII characters other
286 * than "=", "?", and "_" (underscore), MAY be represented as those
287 * characters. (But see section 5 for restrictions.) In
288 * particular, SPACE and TAB MUST NOT be represented as themselves
289 * within encoded words.
290 */
291
292 /*
293 * rule out non-ASCII characters and non-printable characters (the
294 * non-ASCII check should be redundant as isprint() is not localized
295 * and only knows about ASCII, but be defensive about that)
296 */
297 if (non_ascii(ch) || !isprint(ch))
298 return 1;
299
300 /*
301 * rule out special printable characters (' ' should be the only
302 * whitespace character considered printable, but be defensive and use
303 * isspace())
304 */
305 if (isspace(ch) || ch == '=' || ch == '?' || ch == '_')
94f6cdf6
JS
306 return 1;
307
0fcec2ce
JS
308 /*
309 * rfc2047, section 5.3:
310 *
311 * As a replacement for a 'word' entity within a 'phrase', for example,
312 * one that precedes an address in a From, To, or Cc header. The ABNF
313 * definition for 'phrase' from RFC 822 thus becomes:
314 *
315 * phrase = 1*( encoded-word / word )
316 *
317 * In this case the set of characters that may be used in a "Q"-encoded
318 * 'encoded-word' is restricted to: <upper and lower case ASCII
319 * letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
320 * (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
321 * 'phrase' MUST be separated from any adjacent 'word', 'text' or
322 * 'special' by 'linear-white-space'.
323 */
324
325 if (type != RFC2047_ADDRESS)
326 return 0;
327
328 /* '=' and '_' are special cases and have been checked above */
329 return !(isalnum(ch) || ch == '!' || ch == '*' || ch == '+' || ch == '-' || ch == '/');
93fc05eb
JS
330}
331
41dd00ba
JS
332static int needs_rfc2047_encoding(const char *line, int len,
333 enum rfc2047_type type)
93fc05eb 334{
a1f6baa5 335 int i;
93fc05eb
JS
336
337 for (i = 0; i < len; i++) {
338 int ch = line[i];
c22e7de3 339 if (non_ascii(ch) || ch == '\n')
41dd00ba 340 return 1;
93fc05eb 341 if ((i + 1 < len) && (ch == '=' && line[i+1] == '?'))
41dd00ba 342 return 1;
93fc05eb 343 }
93fc05eb 344
41dd00ba
JS
345 return 0;
346}
347
6cd3c053 348static void add_rfc2047(struct strbuf *sb, const char *line, size_t len,
41dd00ba
JS
349 const char *encoding, enum rfc2047_type type)
350{
351 static const int max_encoded_length = 76; /* per rfc2047 */
352 int i;
353 int line_len = last_line_length(sb);
354
93fc05eb
JS
355 strbuf_grow(sb, len * 3 + strlen(encoding) + 100);
356 strbuf_addf(sb, "=?%s?q?", encoding);
a1f6baa5 357 line_len += strlen(encoding) + 5; /* 5 for =??q? */
6cd3c053
KS
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;
94f6cdf6
JS
374
375 /*
376 * According to RFC 2047, we could encode the special character
377 * ' ' (space) with '_' (underscore) for readability. But many
378 * programs do not understand this and just leave the
379 * underscore in place. Thus, we do nothing special here, which
380 * causes ' ' to be encoded as '=20', avoiding this problem.
381 */
a1f6baa5 382
6cd3c053
KS
383 if (line_len + encoded_len + 2 > max_encoded_length) {
384 /* It won't fit with trailing "?=" --- break the line */
a1f6baa5
JK
385 strbuf_addf(sb, "?=\n =?%s?q?", encoding);
386 line_len = strlen(encoding) + 5 + 1; /* =??q? plus SP */
387 }
388
6cd3c053
KS
389 for (i = 0; i < chrlen; i++)
390 strbuf_addf(sb, encoded_fmt, p[i]);
391 line_len += encoded_len;
93fc05eb 392 }
93fc05eb
JS
393 strbuf_addstr(sb, "?=");
394}
395
9dbe7c3d
RS
396static const char *show_ident_date(const struct ident_split *ident,
397 enum date_mode mode)
398{
399 unsigned long date = 0;
3f419d45 400 long tz = 0;
9dbe7c3d
RS
401
402 if (ident->date_begin && ident->date_end)
403 date = strtoul(ident->date_begin, NULL, 10);
1dca155f
JK
404 if (date_overflows(date))
405 date = 0;
406 else {
407 if (ident->tz_begin && ident->tz_end)
408 tz = strtol(ident->tz_begin, NULL, 10);
3f419d45 409 if (tz >= INT_MAX || tz <= INT_MIN)
1dca155f
JK
410 tz = 0;
411 }
9dbe7c3d
RS
412 return show_date(date, tz, mode);
413}
414
10f2fbff 415void pp_user_info(struct pretty_print_context *pp,
6bf13944
JK
416 const char *what, struct strbuf *sb,
417 const char *line, const char *encoding)
93fc05eb 418{
3c020bd5 419 struct ident_split ident;
9dbe7c3d 420 char *line_end;
dffd325f
AP
421 const char *mailbuf, *namebuf;
422 size_t namelen, maillen;
41dd00ba 423 int max_length = 78; /* per rfc2822 */
93fc05eb 424
6bf13944 425 if (pp->fmt == CMIT_FMT_ONELINE)
93fc05eb 426 return;
3c020bd5 427
30e77bcb
RS
428 line_end = strchrnul(line, '\n');
429 if (split_ident_line(&ident, line, line_end - line))
93fc05eb 430 return;
3c020bd5 431
dffd325f
AP
432 mailbuf = ident.mail_begin;
433 maillen = ident.mail_end - ident.mail_begin;
434 namebuf = ident.name_begin;
435 namelen = ident.name_end - ident.name_begin;
436
437 if (pp->mailmap)
438 map_user(pp->mailmap, &mailbuf, &maillen, &namebuf, &namelen);
439
6bf13944 440 if (pp->fmt == CMIT_FMT_EMAIL) {
662cc30c 441 if (pp->from_ident && ident_cmp(pp->from_ident, &ident)) {
a9080475
JK
442 struct strbuf buf = STRBUF_INIT;
443
444 strbuf_addstr(&buf, "From: ");
445 strbuf_add(&buf, namebuf, namelen);
446 strbuf_addstr(&buf, " <");
447 strbuf_add(&buf, mailbuf, maillen);
448 strbuf_addstr(&buf, ">\n");
449 string_list_append(&pp->in_body_headers,
450 strbuf_detach(&buf, NULL));
451
452 mailbuf = pp->from_ident->mail_begin;
453 maillen = pp->from_ident->mail_end - mailbuf;
454 namebuf = pp->from_ident->name_begin;
455 namelen = pp->from_ident->name_end - namebuf;
456 }
457
93fc05eb 458 strbuf_addstr(sb, "From: ");
a0511b39
RS
459 if (needs_rfc2047_encoding(namebuf, namelen, RFC2047_ADDRESS)) {
460 add_rfc2047(sb, namebuf, namelen,
dffd325f 461 encoding, RFC2047_ADDRESS);
41dd00ba 462 max_length = 76; /* per rfc2047 */
a0511b39 463 } else if (needs_rfc822_quoting(namebuf, namelen)) {
4d03c18a 464 struct strbuf quoted = STRBUF_INIT;
a0511b39 465 add_rfc822_quoted(&quoted, namebuf, namelen);
41dd00ba
JS
466 strbuf_add_wrapped_bytes(sb, quoted.buf, quoted.len,
467 -6, 1, max_length);
4d03c18a 468 strbuf_release(&quoted);
41dd00ba 469 } else {
a0511b39 470 strbuf_add_wrapped_bytes(sb, namebuf, namelen,
dffd325f 471 -6, 1, max_length);
4d03c18a 472 }
dffd325f 473
97a17e77
RS
474 if (max_length <
475 last_line_length(sb) + strlen(" <") + maillen + strlen(">"))
476 strbuf_addch(sb, '\n');
a0511b39 477 strbuf_addf(sb, " <%.*s>\n", (int)maillen, mailbuf);
93fc05eb 478 } else {
a0511b39
RS
479 strbuf_addf(sb, "%s: %.*s%.*s <%.*s>\n", what,
480 (pp->fmt == CMIT_FMT_FULLER) ? 4 : 0, " ",
481 (int)namelen, namebuf, (int)maillen, mailbuf);
93fc05eb 482 }
dffd325f 483
6bf13944 484 switch (pp->fmt) {
93fc05eb 485 case CMIT_FMT_MEDIUM:
9dbe7c3d
RS
486 strbuf_addf(sb, "Date: %s\n",
487 show_ident_date(&ident, pp->date_mode));
93fc05eb
JS
488 break;
489 case CMIT_FMT_EMAIL:
9dbe7c3d
RS
490 strbuf_addf(sb, "Date: %s\n",
491 show_ident_date(&ident, DATE_RFC2822));
93fc05eb
JS
492 break;
493 case CMIT_FMT_FULLER:
9dbe7c3d
RS
494 strbuf_addf(sb, "%sDate: %s\n", what,
495 show_ident_date(&ident, pp->date_mode));
93fc05eb
JS
496 break;
497 default:
498 /* notin' */
499 break;
500 }
501}
502
503static int is_empty_line(const char *line, int *len_p)
504{
505 int len = *len_p;
35b2fa5b 506 while (len && isspace(line[len - 1]))
93fc05eb
JS
507 len--;
508 *len_p = len;
509 return !len;
510}
511
a0109668
RS
512static const char *skip_empty_lines(const char *msg)
513{
514 for (;;) {
515 int linelen = get_one_line(msg);
516 int ll = linelen;
517 if (!linelen)
518 break;
519 if (!is_empty_line(msg, &ll))
520 break;
521 msg += linelen;
522 }
523 return msg;
524}
525
6bf13944
JK
526static void add_merge_info(const struct pretty_print_context *pp,
527 struct strbuf *sb, const struct commit *commit)
93fc05eb
JS
528{
529 struct commit_list *parent = commit->parents;
530
6bf13944 531 if ((pp->fmt == CMIT_FMT_ONELINE) || (pp->fmt == CMIT_FMT_EMAIL) ||
93fc05eb
JS
532 !parent || !parent->next)
533 return;
534
535 strbuf_addstr(sb, "Merge:");
536
537 while (parent) {
538 struct commit *p = parent->item;
539 const char *hex = NULL;
6bf13944
JK
540 if (pp->abbrev)
541 hex = find_unique_abbrev(p->object.sha1, pp->abbrev);
93fc05eb
JS
542 if (!hex)
543 hex = sha1_to_hex(p->object.sha1);
93fc05eb
JS
544 parent = parent->next;
545
7fcda920 546 strbuf_addf(sb, " %s", hex);
93fc05eb
JS
547 }
548 strbuf_addch(sb, '\n');
549}
550
dd0d388c
JK
551static char *get_header(const struct commit *commit, const char *msg,
552 const char *key)
93fc05eb
JS
553{
554 int key_len = strlen(key);
dd0d388c 555 const char *line = msg;
93fc05eb 556
a9c7a8a8 557 while (line) {
2c5495f7 558 const char *eol = strchrnul(line, '\n'), *next;
93fc05eb
JS
559
560 if (line == eol)
561 return NULL;
2c5495f7 562 if (!*eol) {
a9c7a8a8
JK
563 warning("malformed commit (header is missing newline): %s",
564 sha1_to_hex(commit->object.sha1));
93fc05eb
JS
565 next = NULL;
566 } else
567 next = eol + 1;
568 if (eol - line > key_len &&
569 !strncmp(line, key, key_len) &&
570 line[key_len] == ' ') {
571 return xmemdupz(line + key_len + 1, eol - line - key_len - 1);
572 }
573 line = next;
574 }
a9c7a8a8 575 return NULL;
93fc05eb
JS
576}
577
578static char *replace_encoding_header(char *buf, const char *encoding)
579{
f285a2d7 580 struct strbuf tmp = STRBUF_INIT;
93fc05eb
JS
581 size_t start, len;
582 char *cp = buf;
583
584 /* guess if there is an encoding header before a \n\n */
585 while (strncmp(cp, "encoding ", strlen("encoding "))) {
586 cp = strchr(cp, '\n');
587 if (!cp || *++cp == '\n')
588 return buf;
589 }
590 start = cp - buf;
591 cp = strchr(cp, '\n');
592 if (!cp)
593 return buf; /* should not happen but be defensive */
594 len = cp + 1 - (buf + start);
595
93fc05eb
JS
596 strbuf_attach(&tmp, buf, strlen(buf), strlen(buf) + 1);
597 if (is_encoding_utf8(encoding)) {
598 /* we have re-coded to UTF-8; drop the header */
599 strbuf_remove(&tmp, start, len);
600 } else {
601 /* just replaces XXXX in 'encoding XXXX\n' */
602 strbuf_splice(&tmp, start + strlen("encoding "),
603 len - strlen("encoding \n"),
604 encoding, strlen(encoding));
605 }
606 return strbuf_detach(&tmp, NULL);
607}
608
b000c59b
JK
609const char *logmsg_reencode(const struct commit *commit,
610 char **commit_encoding,
611 const char *output_encoding)
93fc05eb 612{
330db18c 613 static const char *utf8 = "UTF-8";
93fc05eb
JS
614 const char *use_encoding;
615 char *encoding;
b66103c3 616 const char *msg = get_commit_buffer(commit);
93fc05eb
JS
617 char *out;
618
5a10d236
NTND
619 if (!output_encoding || !*output_encoding) {
620 if (commit_encoding)
621 *commit_encoding =
622 get_header(commit, msg, "encoding");
dd0d388c 623 return msg;
5a10d236 624 }
dd0d388c 625 encoding = get_header(commit, msg, "encoding");
5a10d236
NTND
626 if (commit_encoding)
627 *commit_encoding = encoding;
93fc05eb 628 use_encoding = encoding ? encoding : utf8;
be5c9fb9
JK
629 if (same_encoding(use_encoding, output_encoding)) {
630 /*
631 * No encoding work to be done. If we have no encoding header
632 * at all, then there's nothing to do, and we can return the
633 * message verbatim (whether newly allocated or not).
634 */
635 if (!encoding)
636 return msg;
637
638 /*
639 * Otherwise, we still want to munge the encoding header in the
640 * result, which will be done by modifying the buffer. If we
641 * are using a fresh copy, we can reuse it. But if we are using
b66103c3
JK
642 * the cached copy from get_commit_buffer, we need to duplicate it
643 * to avoid munging the cached copy.
be5c9fb9 644 */
b66103c3
JK
645 if (msg == get_cached_commit_buffer(commit))
646 out = xstrdup(msg);
647 else
648 out = (char *)msg;
be5c9fb9
JK
649 }
650 else {
651 /*
652 * There's actual encoding work to do. Do the reencoding, which
653 * still leaves the header to be replaced in the next step. At
654 * this point, we are done with msg. If we allocated a fresh
655 * copy, we can free it.
656 */
657 out = reencode_string(msg, output_encoding, use_encoding);
b66103c3
JK
658 if (out)
659 unuse_commit_buffer(commit, msg);
be5c9fb9
JK
660 }
661
662 /*
663 * This replacement actually consumes the buffer we hand it, so we do
664 * not have to worry about freeing the old "out" here.
665 */
93fc05eb
JS
666 if (out)
667 out = replace_encoding_header(out, output_encoding);
668
5a10d236
NTND
669 if (!commit_encoding)
670 free(encoding);
dd0d388c
JK
671 /*
672 * If the re-encoding failed, out might be NULL here; in that
673 * case we just return the commit message verbatim.
674 */
675 return out ? out : msg;
676}
677
ea02ffa3
AP
678static int mailmap_name(const char **email, size_t *email_len,
679 const char **name, size_t *name_len)
e0cbc397 680{
c455c87c 681 static struct string_list *mail_map;
e0cbc397
JS
682 if (!mail_map) {
683 mail_map = xcalloc(1, sizeof(*mail_map));
d551a488 684 read_mailmap(mail_map, NULL);
e0cbc397 685 }
d20d654f 686 return mail_map->nr && map_user(mail_map, email, email_len, name, name_len);
e0cbc397
JS
687}
688
c3a670de 689static size_t format_person_part(struct strbuf *sb, char part,
d36f8679 690 const char *msg, int len, enum date_mode dmode)
93fc05eb 691{
c3a670de
MC
692 /* currently all placeholders have same length */
693 const int placeholder_len = 2;
4b340cfa 694 struct ident_split s;
ea02ffa3
AP
695 const char *name, *mail;
696 size_t maillen, namelen;
93fc05eb 697
4b340cfa 698 if (split_ident_line(&s, msg, len) < 0)
c3a670de
MC
699 goto skip;
700
ea02ffa3
AP
701 name = s.name_begin;
702 namelen = s.name_end - s.name_begin;
703 mail = s.mail_begin;
704 maillen = s.mail_end - s.mail_begin;
705
706 if (part == 'N' || part == 'E') /* mailmap lookup */
707 mailmap_name(&mail, &maillen, &name, &namelen);
e0cbc397 708 if (part == 'n' || part == 'N') { /* name */
ea02ffa3 709 strbuf_add(sb, name, namelen);
c3a670de 710 return placeholder_len;
cde75e59 711 }
d20d654f 712 if (part == 'e' || part == 'E') { /* email */
ea02ffa3 713 strbuf_add(sb, mail, maillen);
c3a670de 714 return placeholder_len;
cde75e59 715 }
93fc05eb 716
4b340cfa 717 if (!s.date_begin)
c3a670de 718 goto skip;
93fc05eb 719
cde75e59 720 if (part == 't') { /* date, UNIX timestamp */
4b340cfa 721 strbuf_add(sb, s.date_begin, s.date_end - s.date_begin);
c3a670de 722 return placeholder_len;
cde75e59 723 }
93fc05eb 724
cde75e59
RS
725 switch (part) {
726 case 'd': /* date */
9dbe7c3d 727 strbuf_addstr(sb, show_ident_date(&s, dmode));
c3a670de 728 return placeholder_len;
cde75e59 729 case 'D': /* date, RFC2822 style */
9dbe7c3d 730 strbuf_addstr(sb, show_ident_date(&s, DATE_RFC2822));
c3a670de 731 return placeholder_len;
cde75e59 732 case 'r': /* date, relative */
9dbe7c3d 733 strbuf_addstr(sb, show_ident_date(&s, DATE_RELATIVE));
c3a670de 734 return placeholder_len;
cde75e59 735 case 'i': /* date, ISO 8601 */
9dbe7c3d 736 strbuf_addstr(sb, show_ident_date(&s, DATE_ISO8601));
c3a670de 737 return placeholder_len;
cde75e59 738 }
c3a670de
MC
739
740skip:
741 /*
4b340cfa
JH
742 * reading from either a bogus commit, or a reflog entry with
743 * %gn, %ge, etc.; 'sb' cannot be updated, but we still need
744 * to compute a valid return value.
c3a670de
MC
745 */
746 if (part == 'n' || part == 'e' || part == 't' || part == 'd'
747 || part == 'D' || part == 'r' || part == 'i')
748 return placeholder_len;
749
750 return 0; /* unknown placeholder */
93fc05eb
JS
751}
752
f29d5958
RS
753struct chunk {
754 size_t off;
755 size_t len;
756};
757
a5752342
NTND
758enum flush_type {
759 no_flush,
760 flush_right,
761 flush_left,
1640632b 762 flush_left_and_steal,
a5752342
NTND
763 flush_both
764};
765
a7f01c6b
NTND
766enum trunc_type {
767 trunc_none,
768 trunc_left,
769 trunc_middle,
770 trunc_right
771};
772
f29d5958
RS
773struct format_commit_context {
774 const struct commit *commit;
dd2e794a 775 const struct pretty_print_context *pretty_ctx;
f53bd743
RS
776 unsigned commit_header_parsed:1;
777 unsigned commit_message_parsed:1;
ffb6d7d5 778 struct signature_check signature_check;
a5752342 779 enum flush_type flush_type;
a7f01c6b 780 enum trunc_type truncate;
b000c59b 781 const char *message;
0940a76d 782 char *commit_encoding;
02edd56b 783 size_t width, indent1, indent2;
a95f067e 784 int auto_color;
a5752342 785 int padding;
f29d5958
RS
786
787 /* These offsets are relative to the start of the commit message. */
f29d5958
RS
788 struct chunk author;
789 struct chunk committer;
f53bd743
RS
790 size_t message_off;
791 size_t subject_off;
f29d5958 792 size_t body_off;
b9c62321
RS
793
794 /* The following ones are relative to the result struct strbuf. */
795 struct chunk abbrev_commit_hash;
796 struct chunk abbrev_tree_hash;
797 struct chunk abbrev_parent_hashes;
02edd56b 798 size_t wrap_start;
f29d5958
RS
799};
800
b9c62321
RS
801static int add_again(struct strbuf *sb, struct chunk *chunk)
802{
803 if (chunk->len) {
804 strbuf_adddup(sb, chunk->off, chunk->len);
805 return 1;
806 }
807
808 /*
809 * We haven't seen this chunk before. Our caller is surely
810 * going to add it the hard way now. Remember the most likely
811 * start of the to-be-added chunk: the current end of the
812 * struct strbuf.
813 */
814 chunk->off = sb->len;
815 return 0;
816}
817
f29d5958 818static void parse_commit_header(struct format_commit_context *context)
93fc05eb 819{
177b29dc 820 const char *msg = context->message;
93fc05eb 821 int i;
f29d5958 822
f53bd743 823 for (i = 0; msg[i]; i++) {
f29d5958
RS
824 int eol;
825 for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
826 ; /* do nothing */
827
f29d5958 828 if (i == eol) {
f53bd743 829 break;
59556548 830 } else if (starts_with(msg + i, "author ")) {
f29d5958
RS
831 context->author.off = i + 7;
832 context->author.len = eol - i - 7;
59556548 833 } else if (starts_with(msg + i, "committer ")) {
f29d5958
RS
834 context->committer.off = i + 10;
835 context->committer.len = eol - i - 10;
f29d5958
RS
836 }
837 i = eol;
838 }
f53bd743 839 context->message_off = i;
f29d5958
RS
840 context->commit_header_parsed = 1;
841}
842
46d164b0
SB
843static int istitlechar(char c)
844{
845 return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') ||
846 (c >= '0' && c <= '9') || c == '.' || c == '_';
847}
848
849static void format_sanitized_subject(struct strbuf *sb, const char *msg)
850{
851 size_t trimlen;
871d21d4 852 size_t start_len = sb->len;
46d164b0
SB
853 int space = 2;
854
855 for (; *msg && *msg != '\n'; msg++) {
856 if (istitlechar(*msg)) {
857 if (space == 1)
858 strbuf_addch(sb, '-');
859 space = 0;
860 strbuf_addch(sb, *msg);
861 if (*msg == '.')
862 while (*(msg+1) == '.')
863 msg++;
864 } else
865 space |= 1;
866 }
867
868 /* trim any trailing '.' or '-' characters */
869 trimlen = 0;
871d21d4
SB
870 while (sb->len - trimlen > start_len &&
871 (sb->buf[sb->len - 1 - trimlen] == '.'
872 || sb->buf[sb->len - 1 - trimlen] == '-'))
46d164b0
SB
873 trimlen++;
874 strbuf_remove(sb, sb->len - trimlen, trimlen);
875}
876
cec08717
RS
877const char *format_subject(struct strbuf *sb, const char *msg,
878 const char *line_separator)
88c44735
RS
879{
880 int first = 1;
881
882 for (;;) {
883 const char *line = msg;
884 int linelen = get_one_line(line);
885
886 msg += linelen;
887 if (!linelen || is_empty_line(line, &linelen))
888 break;
889
f53bd743
RS
890 if (!sb)
891 continue;
88c44735
RS
892 strbuf_grow(sb, linelen + 2);
893 if (!first)
894 strbuf_addstr(sb, line_separator);
895 strbuf_add(sb, line, linelen);
896 first = 0;
897 }
898 return msg;
899}
900
f53bd743
RS
901static void parse_commit_message(struct format_commit_context *c)
902{
177b29dc
PN
903 const char *msg = c->message + c->message_off;
904 const char *start = c->message;
f53bd743
RS
905
906 msg = skip_empty_lines(msg);
907 c->subject_off = msg - start;
908
909 msg = format_subject(NULL, msg, NULL);
910 msg = skip_empty_lines(msg);
911 c->body_off = msg - start;
912
913 c->commit_message_parsed = 1;
914}
915
02edd56b
RS
916static void strbuf_wrap(struct strbuf *sb, size_t pos,
917 size_t width, size_t indent1, size_t indent2)
918{
919 struct strbuf tmp = STRBUF_INIT;
920
921 if (pos)
922 strbuf_add(&tmp, sb->buf, pos);
923 strbuf_add_wrapped_text(&tmp, sb->buf + pos,
924 (int) indent1, (int) indent2, (int) width);
925 strbuf_swap(&tmp, sb);
926 strbuf_release(&tmp);
927}
928
929static void rewrap_message_tail(struct strbuf *sb,
930 struct format_commit_context *c,
931 size_t new_width, size_t new_indent1,
932 size_t new_indent2)
933{
934 if (c->width == new_width && c->indent1 == new_indent1 &&
935 c->indent2 == new_indent2)
936 return;
32ca4249 937 if (c->wrap_start < sb->len)
02edd56b
RS
938 strbuf_wrap(sb, c->wrap_start, c->width, c->indent1, c->indent2);
939 c->wrap_start = sb->len;
940 c->width = new_width;
941 c->indent1 = new_indent1;
942 c->indent2 = new_indent2;
943}
944
cd1957f5
JK
945static int format_reflog_person(struct strbuf *sb,
946 char part,
947 struct reflog_walk_info *log,
948 enum date_mode dmode)
949{
950 const char *ident;
951
952 if (!log)
953 return 2;
954
955 ident = get_reflog_ident(log);
956 if (!ident)
957 return 2;
958
959 return format_person_part(sb, part, ident, strlen(ident), dmode);
960}
961
fcabc2d9
NTND
962static size_t parse_color(struct strbuf *sb, /* in UTF-8 */
963 const char *placeholder,
964 struct format_commit_context *c)
965{
966 if (placeholder[1] == '(') {
967 const char *begin = placeholder + 2;
968 const char *end = strchr(begin, ')');
969 char color[COLOR_MAXLEN];
970
971 if (!end)
972 return 0;
59556548 973 if (starts_with(begin, "auto,")) {
fcabc2d9
NTND
974 if (!want_color(c->pretty_ctx->color))
975 return end - placeholder + 1;
976 begin += 5;
977 }
978 color_parse_mem(begin,
979 end - begin,
980 "--pretty format", color);
981 strbuf_addstr(sb, color);
982 return end - placeholder + 1;
983 }
59556548 984 if (starts_with(placeholder + 1, "red")) {
fcabc2d9
NTND
985 strbuf_addstr(sb, GIT_COLOR_RED);
986 return 4;
59556548 987 } else if (starts_with(placeholder + 1, "green")) {
fcabc2d9
NTND
988 strbuf_addstr(sb, GIT_COLOR_GREEN);
989 return 6;
59556548 990 } else if (starts_with(placeholder + 1, "blue")) {
fcabc2d9
NTND
991 strbuf_addstr(sb, GIT_COLOR_BLUE);
992 return 5;
59556548 993 } else if (starts_with(placeholder + 1, "reset")) {
fcabc2d9
NTND
994 strbuf_addstr(sb, GIT_COLOR_RESET);
995 return 6;
996 } else
997 return 0;
998}
999
a5752342
NTND
1000static size_t parse_padding_placeholder(struct strbuf *sb,
1001 const char *placeholder,
1002 struct format_commit_context *c)
1003{
1004 const char *ch = placeholder;
1005 enum flush_type flush_type;
1006 int to_column = 0;
1007
1008 switch (*ch++) {
1009 case '<':
1010 flush_type = flush_right;
1011 break;
1012 case '>':
1013 if (*ch == '<') {
1014 flush_type = flush_both;
1015 ch++;
1640632b
NTND
1016 } else if (*ch == '>') {
1017 flush_type = flush_left_and_steal;
1018 ch++;
a5752342
NTND
1019 } else
1020 flush_type = flush_left;
1021 break;
1022 default:
1023 return 0;
1024 }
1025
1026 /* the next value means "wide enough to that column" */
1027 if (*ch == '|') {
1028 to_column = 1;
1029 ch++;
1030 }
1031
1032 if (*ch == '(') {
1033 const char *start = ch + 1;
a7f01c6b 1034 const char *end = start + strcspn(start, ",)");
a5752342
NTND
1035 char *next;
1036 int width;
1037 if (!end || end == start)
1038 return 0;
1039 width = strtoul(start, &next, 10);
1040 if (next == start || width == 0)
1041 return 0;
1042 c->padding = to_column ? -width : width;
1043 c->flush_type = flush_type;
a7f01c6b
NTND
1044
1045 if (*end == ',') {
1046 start = end + 1;
1047 end = strchr(start, ')');
1048 if (!end || end == start)
1049 return 0;
59556548 1050 if (starts_with(start, "trunc)"))
a7f01c6b 1051 c->truncate = trunc_right;
59556548 1052 else if (starts_with(start, "ltrunc)"))
a7f01c6b 1053 c->truncate = trunc_left;
59556548 1054 else if (starts_with(start, "mtrunc)"))
a7f01c6b
NTND
1055 c->truncate = trunc_middle;
1056 else
1057 return 0;
1058 } else
1059 c->truncate = trunc_none;
1060
a5752342
NTND
1061 return end - placeholder + 1;
1062 }
1063 return 0;
1064}
1065
7e77df39
NTND
1066static size_t format_commit_one(struct strbuf *sb, /* in UTF-8 */
1067 const char *placeholder,
9fa708da 1068 void *context)
f29d5958
RS
1069{
1070 struct format_commit_context *c = context;
1071 const struct commit *commit = c->commit;
177b29dc 1072 const char *msg = c->message;
f29d5958 1073 struct commit_list *p;
42c8c74c 1074 int h1, h2;
93fc05eb 1075
93fc05eb 1076 /* these are independent of the commit */
cde75e59
RS
1077 switch (placeholder[0]) {
1078 case 'C':
59556548 1079 if (starts_with(placeholder + 1, "(auto)")) {
a95f067e
NTND
1080 c->auto_color = 1;
1081 return 7; /* consumed 7 bytes, "C(auto)" */
1082 } else {
1083 int ret = parse_color(sb, placeholder, c);
1084 if (ret)
1085 c->auto_color = 0;
1086 /*
1087 * Otherwise, we decided to treat %C<unknown>
1088 * as a literal string, and the previous
1089 * %C(auto) is still valid.
1090 */
1091 return ret;
c002922a 1092 }
cde75e59
RS
1093 case 'n': /* newline */
1094 strbuf_addch(sb, '\n');
c3a670de 1095 return 1;
42c8c74c
GS
1096 case 'x':
1097 /* %x00 == NUL, %x0a == LF, etc. */
1098 if (0 <= (h1 = hexval_table[0xff & placeholder[1]]) &&
1099 h1 <= 16 &&
1100 0 <= (h2 = hexval_table[0xff & placeholder[2]]) &&
1101 h2 <= 16) {
1102 strbuf_addch(sb, (h1<<4)|h2);
1103 return 3;
1104 } else
1105 return 0;
02edd56b
RS
1106 case 'w':
1107 if (placeholder[1] == '(') {
1108 unsigned long width = 0, indent1 = 0, indent2 = 0;
1109 char *next;
1110 const char *start = placeholder + 2;
1111 const char *end = strchr(start, ')');
1112 if (!end)
1113 return 0;
1114 if (end > start) {
1115 width = strtoul(start, &next, 10);
1116 if (*next == ',') {
1117 indent1 = strtoul(next + 1, &next, 10);
1118 if (*next == ',') {
1119 indent2 = strtoul(next + 1,
1120 &next, 10);
1121 }
1122 }
1123 if (*next != ')')
1124 return 0;
1125 }
1126 rewrap_message_tail(sb, c, width, indent1, indent2);
1127 return end - placeholder + 1;
1128 } else
1129 return 0;
a5752342
NTND
1130
1131 case '<':
1132 case '>':
1133 return parse_padding_placeholder(sb, placeholder, c);
cde75e59 1134 }
93fc05eb
JS
1135
1136 /* these depend on the commit */
1137 if (!commit->object.parsed)
1138 parse_object(commit->object.sha1);
93fc05eb 1139
cde75e59
RS
1140 switch (placeholder[0]) {
1141 case 'H': /* commit hash */
a95f067e 1142 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_COMMIT));
cde75e59 1143 strbuf_addstr(sb, sha1_to_hex(commit->object.sha1));
a95f067e 1144 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_RESET));
c3a670de 1145 return 1;
cde75e59 1146 case 'h': /* abbreviated commit hash */
a95f067e
NTND
1147 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_COMMIT));
1148 if (add_again(sb, &c->abbrev_commit_hash)) {
1149 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_RESET));
c3a670de 1150 return 1;
a95f067e 1151 }
cde75e59 1152 strbuf_addstr(sb, find_unique_abbrev(commit->object.sha1,
c1977021 1153 c->pretty_ctx->abbrev));
a95f067e 1154 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_RESET));
b9c62321 1155 c->abbrev_commit_hash.len = sb->len - c->abbrev_commit_hash.off;
c3a670de 1156 return 1;
cde75e59
RS
1157 case 'T': /* tree hash */
1158 strbuf_addstr(sb, sha1_to_hex(commit->tree->object.sha1));
c3a670de 1159 return 1;
cde75e59 1160 case 't': /* abbreviated tree hash */
b9c62321 1161 if (add_again(sb, &c->abbrev_tree_hash))
c3a670de 1162 return 1;
cde75e59 1163 strbuf_addstr(sb, find_unique_abbrev(commit->tree->object.sha1,
c1977021 1164 c->pretty_ctx->abbrev));
b9c62321 1165 c->abbrev_tree_hash.len = sb->len - c->abbrev_tree_hash.off;
c3a670de 1166 return 1;
cde75e59
RS
1167 case 'P': /* parent hashes */
1168 for (p = commit->parents; p; p = p->next) {
1169 if (p != commit->parents)
1170 strbuf_addch(sb, ' ');
1171 strbuf_addstr(sb, sha1_to_hex(p->item->object.sha1));
1172 }
c3a670de 1173 return 1;
cde75e59 1174 case 'p': /* abbreviated parent hashes */
b9c62321 1175 if (add_again(sb, &c->abbrev_parent_hashes))
c3a670de 1176 return 1;
cde75e59
RS
1177 for (p = commit->parents; p; p = p->next) {
1178 if (p != commit->parents)
1179 strbuf_addch(sb, ' ');
1180 strbuf_addstr(sb, find_unique_abbrev(
c1977021
WP
1181 p->item->object.sha1,
1182 c->pretty_ctx->abbrev));
cde75e59 1183 }
b9c62321
RS
1184 c->abbrev_parent_hashes.len = sb->len -
1185 c->abbrev_parent_hashes.off;
c3a670de 1186 return 1;
cde75e59 1187 case 'm': /* left/right/bottom */
1df2d656 1188 strbuf_addstr(sb, get_revision_mark(NULL, commit));
c3a670de 1189 return 1;
3b3d443f 1190 case 'd':
9d3f002f 1191 load_ref_decorations(DECORATE_SHORT_REFS);
a95f067e 1192 format_decorations(sb, commit, c->auto_color);
3b3d443f 1193 return 1;
8f8f5476
TR
1194 case 'g': /* reflog info */
1195 switch(placeholder[1]) {
1196 case 'd': /* reflog selector */
1197 case 'D':
1198 if (c->pretty_ctx->reflog_info)
1199 get_reflog_selector(sb,
1200 c->pretty_ctx->reflog_info,
1201 c->pretty_ctx->date_mode,
55ccf85a 1202 c->pretty_ctx->date_mode_explicit,
8f8f5476
TR
1203 (placeholder[1] == 'd'));
1204 return 2;
1205 case 's': /* reflog message */
1206 if (c->pretty_ctx->reflog_info)
1207 get_reflog_message(sb, c->pretty_ctx->reflog_info);
1208 return 2;
cd1957f5
JK
1209 case 'n':
1210 case 'N':
1211 case 'e':
1212 case 'E':
1213 return format_reflog_person(sb,
1214 placeholder[1],
1215 c->pretty_ctx->reflog_info,
1216 c->pretty_ctx->date_mode);
8f8f5476
TR
1217 }
1218 return 0; /* unknown %g placeholder */
8b208f02 1219 case 'N':
ddf333f6
JH
1220 if (c->pretty_ctx->notes_message) {
1221 strbuf_addstr(sb, c->pretty_ctx->notes_message);
5b163603
JG
1222 return 1;
1223 }
1224 return 0;
cde75e59
RS
1225 }
1226
f6667c5e 1227 if (placeholder[0] == 'G') {
ffb6d7d5
SG
1228 if (!c->signature_check.result)
1229 check_commit_signature(c->commit, &(c->signature_check));
f6667c5e
JH
1230 switch (placeholder[1]) {
1231 case 'G':
ffb6d7d5
SG
1232 if (c->signature_check.gpg_output)
1233 strbuf_addstr(sb, c->signature_check.gpg_output);
f6667c5e
JH
1234 break;
1235 case '?':
ffb6d7d5 1236 switch (c->signature_check.result) {
f6667c5e
JH
1237 case 'G':
1238 case 'B':
e290c4b9
SG
1239 case 'U':
1240 case 'N':
ffb6d7d5 1241 strbuf_addch(sb, c->signature_check.result);
f6667c5e
JH
1242 }
1243 break;
1244 case 'S':
ffb6d7d5
SG
1245 if (c->signature_check.signer)
1246 strbuf_addstr(sb, c->signature_check.signer);
f6667c5e 1247 break;
0174eeaa 1248 case 'K':
ffb6d7d5
SG
1249 if (c->signature_check.key)
1250 strbuf_addstr(sb, c->signature_check.key);
0174eeaa 1251 break;
f6667c5e
JH
1252 }
1253 return 2;
1254 }
1255
1256
cde75e59 1257 /* For the rest we have to parse the commit header. */
f29d5958
RS
1258 if (!c->commit_header_parsed)
1259 parse_commit_header(c);
93fc05eb 1260
f29d5958 1261 switch (placeholder[0]) {
c3a670de
MC
1262 case 'a': /* author ... */
1263 return format_person_part(sb, placeholder[1],
d36f8679 1264 msg + c->author.off, c->author.len,
dd2e794a 1265 c->pretty_ctx->date_mode);
c3a670de
MC
1266 case 'c': /* committer ... */
1267 return format_person_part(sb, placeholder[1],
d36f8679 1268 msg + c->committer.off, c->committer.len,
dd2e794a 1269 c->pretty_ctx->date_mode);
c3a670de 1270 case 'e': /* encoding */
0940a76d
NTND
1271 if (c->commit_encoding)
1272 strbuf_addstr(sb, c->commit_encoding);
c3a670de 1273 return 1;
1367b12a
EB
1274 case 'B': /* raw body */
1275 /* message_off is always left at the initial newline */
1276 strbuf_addstr(sb, msg + c->message_off + 1);
1277 return 1;
f53bd743
RS
1278 }
1279
1280 /* Now we need to parse the commit message. */
1281 if (!c->commit_message_parsed)
1282 parse_commit_message(c);
1283
1284 switch (placeholder[0]) {
1285 case 's': /* subject */
1286 format_subject(sb, msg + c->subject_off, " ");
1287 return 1;
46d164b0
SB
1288 case 'f': /* sanitized subject */
1289 format_sanitized_subject(sb, msg + c->subject_off);
1290 return 1;
c3a670de 1291 case 'b': /* body */
f29d5958 1292 strbuf_addstr(sb, msg + c->body_off);
c3a670de 1293 return 1;
93fc05eb 1294 }
c3a670de 1295 return 0; /* unknown placeholder */
cde75e59
RS
1296}
1297
a5752342
NTND
1298static size_t format_and_pad_commit(struct strbuf *sb, /* in UTF-8 */
1299 const char *placeholder,
1300 struct format_commit_context *c)
1301{
1302 struct strbuf local_sb = STRBUF_INIT;
1303 int total_consumed = 0, len, padding = c->padding;
1304 if (padding < 0) {
1305 const char *start = strrchr(sb->buf, '\n');
1306 int occupied;
1307 if (!start)
1308 start = sb->buf;
1309 occupied = utf8_strnwidth(start, -1, 1);
1310 padding = (-padding) - occupied;
1311 }
1312 while (1) {
1313 int modifier = *placeholder == 'C';
1314 int consumed = format_commit_one(&local_sb, placeholder, c);
1315 total_consumed += consumed;
1316
1317 if (!modifier)
1318 break;
1319
1320 placeholder += consumed;
1321 if (*placeholder != '%')
1322 break;
1323 placeholder++;
1324 total_consumed++;
1325 }
1326 len = utf8_strnwidth(local_sb.buf, -1, 1);
1640632b
NTND
1327
1328 if (c->flush_type == flush_left_and_steal) {
1329 const char *ch = sb->buf + sb->len - 1;
1330 while (len > padding && ch > sb->buf) {
1331 const char *p;
1332 if (*ch == ' ') {
1333 ch--;
1334 padding++;
1335 continue;
1336 }
1337 /* check for trailing ansi sequences */
1338 if (*ch != 'm')
1339 break;
1340 p = ch - 1;
1341 while (ch - p < 10 && *p != '\033')
1342 p--;
1343 if (*p != '\033' ||
1344 ch + 1 - p != display_mode_esc_sequence_len(p))
1345 break;
1346 /*
1347 * got a good ansi sequence, put it back to
1348 * local_sb as we're cutting sb
1349 */
1350 strbuf_insert(&local_sb, 0, p, ch + 1 - p);
1351 ch = p - 1;
1352 }
1353 strbuf_setlen(sb, ch + 1 - sb->buf);
1354 c->flush_type = flush_left;
1355 }
1356
a7f01c6b
NTND
1357 if (len > padding) {
1358 switch (c->truncate) {
1359 case trunc_left:
1360 strbuf_utf8_replace(&local_sb,
1361 0, len - (padding - 2),
1362 "..");
1363 break;
1364 case trunc_middle:
1365 strbuf_utf8_replace(&local_sb,
1366 padding / 2 - 1,
1367 len - (padding - 2),
1368 "..");
1369 break;
1370 case trunc_right:
1371 strbuf_utf8_replace(&local_sb,
1372 padding - 2, len - (padding - 2),
1373 "..");
1374 break;
1375 case trunc_none:
1376 break;
1377 }
a5752342 1378 strbuf_addstr(sb, local_sb.buf);
a7f01c6b 1379 } else {
a5752342
NTND
1380 int sb_len = sb->len, offset = 0;
1381 if (c->flush_type == flush_left)
1382 offset = padding - len;
1383 else if (c->flush_type == flush_both)
1384 offset = (padding - len) / 2;
1385 /*
1386 * we calculate padding in columns, now
1387 * convert it back to chars
1388 */
1389 padding = padding - len + local_sb.len;
1390 strbuf_grow(sb, padding);
1391 strbuf_setlen(sb, sb_len + padding);
1392 memset(sb->buf + sb_len, ' ', sb->len - sb_len);
1393 memcpy(sb->buf + sb_len + offset, local_sb.buf,
1394 local_sb.len);
1395 }
1396 strbuf_release(&local_sb);
1397 c->flush_type = no_flush;
1398 return total_consumed;
1399}
1400
7e77df39
NTND
1401static size_t format_commit_item(struct strbuf *sb, /* in UTF-8 */
1402 const char *placeholder,
9fa708da
JH
1403 void *context)
1404{
1405 int consumed;
1406 size_t orig_len;
1407 enum {
1408 NO_MAGIC,
1409 ADD_LF_BEFORE_NON_EMPTY,
1410 DEL_LF_BEFORE_EMPTY,
223a923c 1411 ADD_SP_BEFORE_NON_EMPTY
9fa708da
JH
1412 } magic = NO_MAGIC;
1413
1414 switch (placeholder[0]) {
1415 case '-':
1416 magic = DEL_LF_BEFORE_EMPTY;
1417 break;
1418 case '+':
1419 magic = ADD_LF_BEFORE_NON_EMPTY;
1420 break;
7b88176e
MG
1421 case ' ':
1422 magic = ADD_SP_BEFORE_NON_EMPTY;
1423 break;
9fa708da
JH
1424 default:
1425 break;
1426 }
1427 if (magic != NO_MAGIC)
1428 placeholder++;
1429
1430 orig_len = sb->len;
a5752342
NTND
1431 if (((struct format_commit_context *)context)->flush_type != no_flush)
1432 consumed = format_and_pad_commit(sb, placeholder, context);
1433 else
1434 consumed = format_commit_one(sb, placeholder, context);
9fa708da
JH
1435 if (magic == NO_MAGIC)
1436 return consumed;
1437
1438 if ((orig_len == sb->len) && magic == DEL_LF_BEFORE_EMPTY) {
1439 while (sb->len && sb->buf[sb->len - 1] == '\n')
1440 strbuf_setlen(sb, sb->len - 1);
7b88176e
MG
1441 } else if (orig_len != sb->len) {
1442 if (magic == ADD_LF_BEFORE_NON_EMPTY)
1443 strbuf_insert(sb, orig_len, "\n", 1);
1444 else if (magic == ADD_SP_BEFORE_NON_EMPTY)
1445 strbuf_insert(sb, orig_len, " ", 1);
9fa708da
JH
1446 }
1447 return consumed + 1;
1448}
1449
5b163603
JG
1450static size_t userformat_want_item(struct strbuf *sb, const char *placeholder,
1451 void *context)
1452{
1453 struct userformat_want *w = context;
1454
7b88176e 1455 if (*placeholder == '+' || *placeholder == '-' || *placeholder == ' ')
5b163603
JG
1456 placeholder++;
1457
1458 switch (*placeholder) {
1459 case 'N':
1460 w->notes = 1;
1461 break;
1462 }
1463 return 0;
1464}
1465
1466void userformat_find_requirements(const char *fmt, struct userformat_want *w)
1467{
1468 struct strbuf dummy = STRBUF_INIT;
1469
1470 if (!fmt) {
1471 if (!user_format)
1472 return;
1473 fmt = user_format;
1474 }
a6253d10 1475 strbuf_expand(&dummy, fmt, userformat_want_item, w);
5b163603
JG
1476 strbuf_release(&dummy);
1477}
1478
cde75e59 1479void format_commit_message(const struct commit *commit,
7f98ebc8 1480 const char *format, struct strbuf *sb,
dd2e794a 1481 const struct pretty_print_context *pretty_ctx)
cde75e59 1482{
f29d5958 1483 struct format_commit_context context;
177b29dc 1484 const char *output_enc = pretty_ctx->output_encoding;
7e77df39 1485 const char *utf8 = "UTF-8";
f29d5958
RS
1486
1487 memset(&context, 0, sizeof(context));
1488 context.commit = commit;
dd2e794a 1489 context.pretty_ctx = pretty_ctx;
02edd56b 1490 context.wrap_start = sb->len;
0940a76d
NTND
1491 context.message = logmsg_reencode(commit,
1492 &context.commit_encoding,
1493 output_enc);
177b29dc 1494
c3a670de 1495 strbuf_expand(sb, format, format_commit_item, &context);
02edd56b 1496 rewrap_message_tail(sb, &context, 0, 0, 0);
177b29dc 1497
7e77df39
NTND
1498 if (output_enc) {
1499 if (same_encoding(utf8, output_enc))
1500 output_enc = NULL;
1501 } else {
1502 if (context.commit_encoding &&
1503 !same_encoding(context.commit_encoding, utf8))
1504 output_enc = context.commit_encoding;
1505 }
1506
1507 if (output_enc) {
1508 int outsz;
1509 char *out = reencode_string_len(sb->buf, sb->len,
1510 output_enc, utf8, &outsz);
1511 if (out)
1512 strbuf_attach(sb, out, outsz, outsz + 1);
1513 }
1514
0940a76d 1515 free(context.commit_encoding);
b66103c3 1516 unuse_commit_buffer(commit, context.message);
ffb6d7d5
SG
1517 free(context.signature_check.gpg_output);
1518 free(context.signature_check.signer);
93fc05eb
JS
1519}
1520
10f2fbff 1521static void pp_header(struct pretty_print_context *pp,
93fc05eb
JS
1522 const char *encoding,
1523 const struct commit *commit,
1524 const char **msg_p,
1525 struct strbuf *sb)
1526{
1527 int parents_shown = 0;
1528
1529 for (;;) {
1530 const char *line = *msg_p;
1531 int linelen = get_one_line(*msg_p);
1532
1533 if (!linelen)
1534 return;
1535 *msg_p += linelen;
1536
1537 if (linelen == 1)
1538 /* End of header */
1539 return;
1540
6bf13944 1541 if (pp->fmt == CMIT_FMT_RAW) {
93fc05eb
JS
1542 strbuf_add(sb, line, linelen);
1543 continue;
1544 }
1545
59556548 1546 if (starts_with(line, "parent ")) {
93fc05eb
JS
1547 if (linelen != 48)
1548 die("bad parent line in commit");
1549 continue;
1550 }
1551
1552 if (!parents_shown) {
1553 struct commit_list *parent;
1554 int num;
1555 for (parent = commit->parents, num = 0;
1556 parent;
1557 parent = parent->next, num++)
1558 ;
1559 /* with enough slop */
1560 strbuf_grow(sb, num * 50 + 20);
6bf13944 1561 add_merge_info(pp, sb, commit);
93fc05eb
JS
1562 parents_shown = 1;
1563 }
1564
1565 /*
1566 * MEDIUM == DEFAULT shows only author with dates.
1567 * FULL shows both authors but not dates.
1568 * FULLER shows both authors and dates.
1569 */
59556548 1570 if (starts_with(line, "author ")) {
93fc05eb 1571 strbuf_grow(sb, linelen + 80);
6bf13944 1572 pp_user_info(pp, "Author", sb, line + 7, encoding);
93fc05eb 1573 }
59556548 1574 if (starts_with(line, "committer ") &&
6bf13944 1575 (pp->fmt == CMIT_FMT_FULL || pp->fmt == CMIT_FMT_FULLER)) {
93fc05eb 1576 strbuf_grow(sb, linelen + 80);
6bf13944 1577 pp_user_info(pp, "Commit", sb, line + 10, encoding);
93fc05eb
JS
1578 }
1579 }
1580}
1581
10f2fbff 1582void pp_title_line(struct pretty_print_context *pp,
b02bd65f
DB
1583 const char **msg_p,
1584 struct strbuf *sb,
b02bd65f 1585 const char *encoding,
267123b4 1586 int need_8bit_cte)
93fc05eb 1587{
41dd00ba 1588 static const int max_length = 78; /* per rfc2047 */
93fc05eb
JS
1589 struct strbuf title;
1590
1591 strbuf_init(&title, 80);
9553d2b2
JK
1592 *msg_p = format_subject(&title, *msg_p,
1593 pp->preserve_subject ? "\n" : " ");
93fc05eb
JS
1594
1595 strbuf_grow(sb, title.len + 1024);
6bf13944
JK
1596 if (pp->subject) {
1597 strbuf_addstr(sb, pp->subject);
41dd00ba
JS
1598 if (needs_rfc2047_encoding(title.buf, title.len, RFC2047_SUBJECT))
1599 add_rfc2047(sb, title.buf, title.len,
1600 encoding, RFC2047_SUBJECT);
1601 else
1602 strbuf_add_wrapped_bytes(sb, title.buf, title.len,
1603 -last_line_length(sb), 1, max_length);
93fc05eb
JS
1604 } else {
1605 strbuf_addbuf(sb, &title);
1606 }
1607 strbuf_addch(sb, '\n');
1608
a9080475
JK
1609 if (need_8bit_cte == 0) {
1610 int i;
1611 for (i = 0; i < pp->in_body_headers.nr; i++) {
1612 if (has_non_ascii(pp->in_body_headers.items[i].string)) {
1613 need_8bit_cte = 1;
1614 break;
1615 }
1616 }
1617 }
1618
6bf4f1b4 1619 if (need_8bit_cte > 0) {
93fc05eb
JS
1620 const char *header_fmt =
1621 "MIME-Version: 1.0\n"
1622 "Content-Type: text/plain; charset=%s\n"
1623 "Content-Transfer-Encoding: 8bit\n";
1624 strbuf_addf(sb, header_fmt, encoding);
1625 }
6bf13944
JK
1626 if (pp->after_subject) {
1627 strbuf_addstr(sb, pp->after_subject);
93fc05eb 1628 }
6bf13944 1629 if (pp->fmt == CMIT_FMT_EMAIL) {
93fc05eb
JS
1630 strbuf_addch(sb, '\n');
1631 }
a9080475
JK
1632
1633 if (pp->in_body_headers.nr) {
1634 int i;
1635 for (i = 0; i < pp->in_body_headers.nr; i++) {
1636 strbuf_addstr(sb, pp->in_body_headers.items[i].string);
1637 free(pp->in_body_headers.items[i].string);
1638 }
1639 string_list_clear(&pp->in_body_headers, 0);
1640 strbuf_addch(sb, '\n');
1641 }
1642
93fc05eb
JS
1643 strbuf_release(&title);
1644}
1645
10f2fbff 1646void pp_remainder(struct pretty_print_context *pp,
b02bd65f
DB
1647 const char **msg_p,
1648 struct strbuf *sb,
1649 int indent)
93fc05eb
JS
1650{
1651 int first = 1;
1652 for (;;) {
1653 const char *line = *msg_p;
1654 int linelen = get_one_line(line);
1655 *msg_p += linelen;
1656
1657 if (!linelen)
1658 break;
1659
1660 if (is_empty_line(line, &linelen)) {
1661 if (first)
1662 continue;
6bf13944 1663 if (pp->fmt == CMIT_FMT_SHORT)
93fc05eb
JS
1664 break;
1665 }
1666 first = 0;
1667
1668 strbuf_grow(sb, linelen + indent + 20);
1669 if (indent) {
1670 memset(sb->buf + sb->len, ' ', indent);
1671 strbuf_setlen(sb, sb->len + indent);
1672 }
1673 strbuf_add(sb, line, linelen);
1674 strbuf_addch(sb, '\n');
1675 }
1676}
1677
10f2fbff 1678void pretty_print_commit(struct pretty_print_context *pp,
6bf13944
JK
1679 const struct commit *commit,
1680 struct strbuf *sb)
93fc05eb
JS
1681{
1682 unsigned long beginning_of_body;
1683 int indent = 4;
dd0d388c 1684 const char *msg;
b000c59b 1685 const char *reencoded;
93fc05eb 1686 const char *encoding;
6bf13944 1687 int need_8bit_cte = pp->need_8bit_cte;
93fc05eb 1688
6bf13944
JK
1689 if (pp->fmt == CMIT_FMT_USERFORMAT) {
1690 format_commit_message(commit, user_format, sb, pp);
93fc05eb
JS
1691 return;
1692 }
1693
e297cf5a 1694 encoding = get_log_output_encoding();
5a10d236 1695 msg = reencoded = logmsg_reencode(commit, NULL, encoding);
93fc05eb 1696
6bf13944 1697 if (pp->fmt == CMIT_FMT_ONELINE || pp->fmt == CMIT_FMT_EMAIL)
93fc05eb
JS
1698 indent = 0;
1699
6bf4f1b4
JH
1700 /*
1701 * We need to check and emit Content-type: to mark it
1702 * as 8-bit if we haven't done so.
93fc05eb 1703 */
6bf13944 1704 if (pp->fmt == CMIT_FMT_EMAIL && need_8bit_cte == 0) {
93fc05eb
JS
1705 int i, ch, in_body;
1706
1707 for (in_body = i = 0; (ch = msg[i]); i++) {
1708 if (!in_body) {
1709 /* author could be non 7-bit ASCII but
1710 * the log may be so; skip over the
1711 * header part first.
1712 */
1713 if (ch == '\n' && msg[i+1] == '\n')
1714 in_body = 1;
1715 }
1716 else if (non_ascii(ch)) {
6bf4f1b4 1717 need_8bit_cte = 1;
93fc05eb
JS
1718 break;
1719 }
1720 }
1721 }
1722
6bf13944
JK
1723 pp_header(pp, encoding, commit, &msg, sb);
1724 if (pp->fmt != CMIT_FMT_ONELINE && !pp->subject) {
93fc05eb
JS
1725 strbuf_addch(sb, '\n');
1726 }
1727
1728 /* Skip excess blank lines at the beginning of body, if any... */
a0109668 1729 msg = skip_empty_lines(msg);
93fc05eb
JS
1730
1731 /* These formats treat the title line specially. */
6bf13944
JK
1732 if (pp->fmt == CMIT_FMT_ONELINE || pp->fmt == CMIT_FMT_EMAIL)
1733 pp_title_line(pp, &msg, sb, encoding, need_8bit_cte);
93fc05eb
JS
1734
1735 beginning_of_body = sb->len;
6bf13944
JK
1736 if (pp->fmt != CMIT_FMT_ONELINE)
1737 pp_remainder(pp, &msg, sb, indent);
93fc05eb
JS
1738 strbuf_rtrim(sb);
1739
1740 /* Make sure there is an EOLN for the non-oneline case */
6bf13944 1741 if (pp->fmt != CMIT_FMT_ONELINE)
93fc05eb
JS
1742 strbuf_addch(sb, '\n');
1743
1744 /*
1745 * The caller may append additional body text in e-mail
1746 * format. Make sure we did not strip the blank line
1747 * between the header and the body.
1748 */
6bf13944 1749 if (pp->fmt == CMIT_FMT_EMAIL && sb->len <= beginning_of_body)
93fc05eb 1750 strbuf_addch(sb, '\n');
a97a7468 1751
b66103c3 1752 unuse_commit_buffer(commit, reencoded);
93fc05eb 1753}
8b8a5374
JK
1754
1755void pp_commit_easy(enum cmit_fmt fmt, const struct commit *commit,
1756 struct strbuf *sb)
1757{
1758 struct pretty_print_context pp = {0};
6bf13944
JK
1759 pp.fmt = fmt;
1760 pretty_print_commit(&pp, commit, sb);
8b8a5374 1761}