pretty: save commit encoding from logmsg_reencode if the caller needs it
[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
WP
42
43 if (prefixcmp(var, "pretty."))
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);
70 if (!prefixcmp(fmt, "format:") || !prefixcmp(fmt, "tformat:")) {
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
118 if (prefixcmp(commit_formats[i].name, sought))
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 }
4da45bef 154 if (!prefixcmp(arg, "format:") || !prefixcmp(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
6bf13944
JK
396void pp_user_info(const struct pretty_print_context *pp,
397 const char *what, struct strbuf *sb,
398 const char *line, const char *encoding)
93fc05eb 399{
dffd325f
AP
400 struct strbuf name;
401 struct strbuf mail;
3c020bd5 402 struct ident_split ident;
dffd325f 403 int linelen;
3c020bd5 404 char *line_end, *date;
dffd325f
AP
405 const char *mailbuf, *namebuf;
406 size_t namelen, maillen;
41dd00ba 407 int max_length = 78; /* per rfc2822 */
93fc05eb
JS
408 unsigned long time;
409 int tz;
93fc05eb 410
6bf13944 411 if (pp->fmt == CMIT_FMT_ONELINE)
93fc05eb 412 return;
3c020bd5
AP
413
414 line_end = strchr(line, '\n');
415 if (!line_end) {
416 line_end = strchr(line, '\0');
417 if (!line_end)
418 return;
419 }
420
421 linelen = ++line_end - line;
422 if (split_ident_line(&ident, line, linelen))
93fc05eb 423 return;
3c020bd5 424
dffd325f
AP
425
426 mailbuf = ident.mail_begin;
427 maillen = ident.mail_end - ident.mail_begin;
428 namebuf = ident.name_begin;
429 namelen = ident.name_end - ident.name_begin;
430
431 if (pp->mailmap)
432 map_user(pp->mailmap, &mailbuf, &maillen, &namebuf, &namelen);
433
434 strbuf_init(&mail, 0);
435 strbuf_init(&name, 0);
436
437 strbuf_add(&mail, mailbuf, maillen);
438 strbuf_add(&name, namebuf, namelen);
439
440 namelen = name.len + mail.len + 3; /* ' ' + '<' + '>' */
3c020bd5 441 time = strtoul(ident.date_begin, &date, 10);
93fc05eb
JS
442 tz = strtol(date, NULL, 10);
443
6bf13944 444 if (pp->fmt == CMIT_FMT_EMAIL) {
93fc05eb 445 strbuf_addstr(sb, "From: ");
dffd325f
AP
446 if (needs_rfc2047_encoding(name.buf, name.len, RFC2047_ADDRESS)) {
447 add_rfc2047(sb, name.buf, name.len,
448 encoding, RFC2047_ADDRESS);
41dd00ba 449 max_length = 76; /* per rfc2047 */
dffd325f 450 } else if (needs_rfc822_quoting(name.buf, name.len)) {
4d03c18a 451 struct strbuf quoted = STRBUF_INIT;
dffd325f 452 add_rfc822_quoted(&quoted, name.buf, name.len);
41dd00ba
JS
453 strbuf_add_wrapped_bytes(sb, quoted.buf, quoted.len,
454 -6, 1, max_length);
4d03c18a 455 strbuf_release(&quoted);
41dd00ba 456 } else {
dffd325f
AP
457 strbuf_add_wrapped_bytes(sb, name.buf, name.len,
458 -6, 1, max_length);
4d03c18a 459 }
dffd325f 460 if (namelen - name.len + last_line_length(sb) > max_length)
990f6e30 461 strbuf_addch(sb, '\n');
dffd325f
AP
462
463 strbuf_addf(sb, " <%s>\n", mail.buf);
93fc05eb 464 } else {
dffd325f 465 strbuf_addf(sb, "%s: %.*s%s <%s>\n", what,
6bf13944 466 (pp->fmt == CMIT_FMT_FULLER) ? 4 : 0,
dffd325f 467 " ", name.buf, mail.buf);
93fc05eb 468 }
dffd325f
AP
469
470 strbuf_release(&mail);
471 strbuf_release(&name);
472
6bf13944 473 switch (pp->fmt) {
93fc05eb 474 case CMIT_FMT_MEDIUM:
6bf13944 475 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, pp->date_mode));
93fc05eb
JS
476 break;
477 case CMIT_FMT_EMAIL:
478 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, DATE_RFC2822));
479 break;
480 case CMIT_FMT_FULLER:
6bf13944 481 strbuf_addf(sb, "%sDate: %s\n", what, show_date(time, tz, pp->date_mode));
93fc05eb
JS
482 break;
483 default:
484 /* notin' */
485 break;
486 }
487}
488
489static int is_empty_line(const char *line, int *len_p)
490{
491 int len = *len_p;
492 while (len && isspace(line[len-1]))
493 len--;
494 *len_p = len;
495 return !len;
496}
497
a0109668
RS
498static const char *skip_empty_lines(const char *msg)
499{
500 for (;;) {
501 int linelen = get_one_line(msg);
502 int ll = linelen;
503 if (!linelen)
504 break;
505 if (!is_empty_line(msg, &ll))
506 break;
507 msg += linelen;
508 }
509 return msg;
510}
511
6bf13944
JK
512static void add_merge_info(const struct pretty_print_context *pp,
513 struct strbuf *sb, const struct commit *commit)
93fc05eb
JS
514{
515 struct commit_list *parent = commit->parents;
516
6bf13944 517 if ((pp->fmt == CMIT_FMT_ONELINE) || (pp->fmt == CMIT_FMT_EMAIL) ||
93fc05eb
JS
518 !parent || !parent->next)
519 return;
520
521 strbuf_addstr(sb, "Merge:");
522
523 while (parent) {
524 struct commit *p = parent->item;
525 const char *hex = NULL;
6bf13944
JK
526 if (pp->abbrev)
527 hex = find_unique_abbrev(p->object.sha1, pp->abbrev);
93fc05eb
JS
528 if (!hex)
529 hex = sha1_to_hex(p->object.sha1);
93fc05eb
JS
530 parent = parent->next;
531
7fcda920 532 strbuf_addf(sb, " %s", hex);
93fc05eb
JS
533 }
534 strbuf_addch(sb, '\n');
535}
536
dd0d388c
JK
537static char *get_header(const struct commit *commit, const char *msg,
538 const char *key)
93fc05eb
JS
539{
540 int key_len = strlen(key);
dd0d388c 541 const char *line = msg;
93fc05eb 542
a9c7a8a8 543 while (line) {
93fc05eb
JS
544 const char *eol = strchr(line, '\n'), *next;
545
546 if (line == eol)
547 return NULL;
548 if (!eol) {
a9c7a8a8
JK
549 warning("malformed commit (header is missing newline): %s",
550 sha1_to_hex(commit->object.sha1));
93fc05eb
JS
551 eol = line + strlen(line);
552 next = NULL;
553 } else
554 next = eol + 1;
555 if (eol - line > key_len &&
556 !strncmp(line, key, key_len) &&
557 line[key_len] == ' ') {
558 return xmemdupz(line + key_len + 1, eol - line - key_len - 1);
559 }
560 line = next;
561 }
a9c7a8a8 562 return NULL;
93fc05eb
JS
563}
564
565static char *replace_encoding_header(char *buf, const char *encoding)
566{
f285a2d7 567 struct strbuf tmp = STRBUF_INIT;
93fc05eb
JS
568 size_t start, len;
569 char *cp = buf;
570
571 /* guess if there is an encoding header before a \n\n */
572 while (strncmp(cp, "encoding ", strlen("encoding "))) {
573 cp = strchr(cp, '\n');
574 if (!cp || *++cp == '\n')
575 return buf;
576 }
577 start = cp - buf;
578 cp = strchr(cp, '\n');
579 if (!cp)
580 return buf; /* should not happen but be defensive */
581 len = cp + 1 - (buf + start);
582
93fc05eb
JS
583 strbuf_attach(&tmp, buf, strlen(buf), strlen(buf) + 1);
584 if (is_encoding_utf8(encoding)) {
585 /* we have re-coded to UTF-8; drop the header */
586 strbuf_remove(&tmp, start, len);
587 } else {
588 /* just replaces XXXX in 'encoding XXXX\n' */
589 strbuf_splice(&tmp, start + strlen("encoding "),
590 len - strlen("encoding \n"),
591 encoding, strlen(encoding));
592 }
593 return strbuf_detach(&tmp, NULL);
594}
595
177b29dc 596char *logmsg_reencode(const struct commit *commit,
5a10d236 597 char **commit_encoding,
177b29dc 598 const char *output_encoding)
93fc05eb 599{
330db18c 600 static const char *utf8 = "UTF-8";
93fc05eb
JS
601 const char *use_encoding;
602 char *encoding;
dd0d388c 603 char *msg = commit->buffer;
93fc05eb
JS
604 char *out;
605
be5c9fb9
JK
606 if (!msg) {
607 enum object_type type;
608 unsigned long size;
609
610 msg = read_sha1_file(commit->object.sha1, &type, &size);
611 if (!msg)
612 die("Cannot read commit object %s",
613 sha1_to_hex(commit->object.sha1));
614 if (type != OBJ_COMMIT)
615 die("Expected commit for '%s', got %s",
616 sha1_to_hex(commit->object.sha1), typename(type));
617 }
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
642 * the cached copy from commit->buffer, we need to duplicate it
643 * to avoid munging commit->buffer.
644 */
645 out = msg;
646 if (out == commit->buffer)
647 out = xstrdup(out);
648 }
649 else {
650 /*
651 * There's actual encoding work to do. Do the reencoding, which
652 * still leaves the header to be replaced in the next step. At
653 * this point, we are done with msg. If we allocated a fresh
654 * copy, we can free it.
655 */
656 out = reencode_string(msg, output_encoding, use_encoding);
657 if (out && msg != commit->buffer)
658 free(msg);
659 }
660
661 /*
662 * This replacement actually consumes the buffer we hand it, so we do
663 * not have to worry about freeing the old "out" here.
664 */
93fc05eb
JS
665 if (out)
666 out = replace_encoding_header(out, output_encoding);
667
5a10d236
NTND
668 if (!commit_encoding)
669 free(encoding);
dd0d388c
JK
670 /*
671 * If the re-encoding failed, out might be NULL here; in that
672 * case we just return the commit message verbatim.
673 */
674 return out ? out : msg;
675}
676
677void logmsg_free(char *msg, const struct commit *commit)
678{
679 if (msg != commit->buffer)
680 free(msg);
93fc05eb
JS
681}
682
ea02ffa3
AP
683static int mailmap_name(const char **email, size_t *email_len,
684 const char **name, size_t *name_len)
e0cbc397 685{
c455c87c 686 static struct string_list *mail_map;
e0cbc397
JS
687 if (!mail_map) {
688 mail_map = xcalloc(1, sizeof(*mail_map));
d551a488 689 read_mailmap(mail_map, NULL);
e0cbc397 690 }
d20d654f 691 return mail_map->nr && map_user(mail_map, email, email_len, name, name_len);
e0cbc397
JS
692}
693
c3a670de 694static size_t format_person_part(struct strbuf *sb, char part,
d36f8679 695 const char *msg, int len, enum date_mode dmode)
93fc05eb 696{
c3a670de
MC
697 /* currently all placeholders have same length */
698 const int placeholder_len = 2;
4b340cfa 699 int tz;
c3a670de 700 unsigned long date = 0;
4b340cfa 701 struct ident_split s;
ea02ffa3
AP
702 const char *name, *mail;
703 size_t maillen, namelen;
93fc05eb 704
4b340cfa 705 if (split_ident_line(&s, msg, len) < 0)
c3a670de
MC
706 goto skip;
707
ea02ffa3
AP
708 name = s.name_begin;
709 namelen = s.name_end - s.name_begin;
710 mail = s.mail_begin;
711 maillen = s.mail_end - s.mail_begin;
712
713 if (part == 'N' || part == 'E') /* mailmap lookup */
714 mailmap_name(&mail, &maillen, &name, &namelen);
e0cbc397 715 if (part == 'n' || part == 'N') { /* name */
ea02ffa3 716 strbuf_add(sb, name, namelen);
c3a670de 717 return placeholder_len;
cde75e59 718 }
d20d654f 719 if (part == 'e' || part == 'E') { /* email */
ea02ffa3 720 strbuf_add(sb, mail, maillen);
c3a670de 721 return placeholder_len;
cde75e59 722 }
93fc05eb 723
4b340cfa 724 if (!s.date_begin)
c3a670de 725 goto skip;
93fc05eb 726
4b340cfa
JH
727 date = strtoul(s.date_begin, NULL, 10);
728
cde75e59 729 if (part == 't') { /* date, UNIX timestamp */
4b340cfa 730 strbuf_add(sb, s.date_begin, s.date_end - s.date_begin);
c3a670de 731 return placeholder_len;
cde75e59 732 }
93fc05eb
JS
733
734 /* parse tz */
4b340cfa
JH
735 tz = strtoul(s.tz_begin + 1, NULL, 10);
736 if (*s.tz_begin == '-')
737 tz = -tz;
93fc05eb 738
cde75e59
RS
739 switch (part) {
740 case 'd': /* date */
d36f8679 741 strbuf_addstr(sb, show_date(date, tz, dmode));
c3a670de 742 return placeholder_len;
cde75e59
RS
743 case 'D': /* date, RFC2822 style */
744 strbuf_addstr(sb, show_date(date, tz, DATE_RFC2822));
c3a670de 745 return placeholder_len;
cde75e59
RS
746 case 'r': /* date, relative */
747 strbuf_addstr(sb, show_date(date, tz, DATE_RELATIVE));
c3a670de 748 return placeholder_len;
cde75e59
RS
749 case 'i': /* date, ISO 8601 */
750 strbuf_addstr(sb, show_date(date, tz, DATE_ISO8601));
c3a670de 751 return placeholder_len;
cde75e59 752 }
c3a670de
MC
753
754skip:
755 /*
4b340cfa
JH
756 * reading from either a bogus commit, or a reflog entry with
757 * %gn, %ge, etc.; 'sb' cannot be updated, but we still need
758 * to compute a valid return value.
c3a670de
MC
759 */
760 if (part == 'n' || part == 'e' || part == 't' || part == 'd'
761 || part == 'D' || part == 'r' || part == 'i')
762 return placeholder_len;
763
764 return 0; /* unknown placeholder */
93fc05eb
JS
765}
766
f29d5958
RS
767struct chunk {
768 size_t off;
769 size_t len;
770};
771
772struct format_commit_context {
773 const struct commit *commit;
dd2e794a 774 const struct pretty_print_context *pretty_ctx;
f53bd743
RS
775 unsigned commit_header_parsed:1;
776 unsigned commit_message_parsed:1;
ffb6d7d5 777 struct signature_check signature_check;
177b29dc 778 char *message;
02edd56b 779 size_t width, indent1, indent2;
f29d5958
RS
780
781 /* These offsets are relative to the start of the commit message. */
f29d5958
RS
782 struct chunk author;
783 struct chunk committer;
784 struct chunk encoding;
f53bd743
RS
785 size_t message_off;
786 size_t subject_off;
f29d5958 787 size_t body_off;
b9c62321
RS
788
789 /* The following ones are relative to the result struct strbuf. */
790 struct chunk abbrev_commit_hash;
791 struct chunk abbrev_tree_hash;
792 struct chunk abbrev_parent_hashes;
02edd56b 793 size_t wrap_start;
f29d5958
RS
794};
795
b9c62321
RS
796static int add_again(struct strbuf *sb, struct chunk *chunk)
797{
798 if (chunk->len) {
799 strbuf_adddup(sb, chunk->off, chunk->len);
800 return 1;
801 }
802
803 /*
804 * We haven't seen this chunk before. Our caller is surely
805 * going to add it the hard way now. Remember the most likely
806 * start of the to-be-added chunk: the current end of the
807 * struct strbuf.
808 */
809 chunk->off = sb->len;
810 return 0;
811}
812
f29d5958 813static void parse_commit_header(struct format_commit_context *context)
93fc05eb 814{
177b29dc 815 const char *msg = context->message;
93fc05eb 816 int i;
f29d5958 817
f53bd743 818 for (i = 0; msg[i]; i++) {
f29d5958
RS
819 int eol;
820 for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
821 ; /* do nothing */
822
f29d5958 823 if (i == eol) {
f53bd743 824 break;
f29d5958
RS
825 } else if (!prefixcmp(msg + i, "author ")) {
826 context->author.off = i + 7;
827 context->author.len = eol - i - 7;
828 } else if (!prefixcmp(msg + i, "committer ")) {
829 context->committer.off = i + 10;
830 context->committer.len = eol - i - 10;
831 } else if (!prefixcmp(msg + i, "encoding ")) {
832 context->encoding.off = i + 9;
833 context->encoding.len = eol - i - 9;
834 }
835 i = eol;
836 }
f53bd743 837 context->message_off = i;
f29d5958
RS
838 context->commit_header_parsed = 1;
839}
840
46d164b0
SB
841static int istitlechar(char c)
842{
843 return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') ||
844 (c >= '0' && c <= '9') || c == '.' || c == '_';
845}
846
847static void format_sanitized_subject(struct strbuf *sb, const char *msg)
848{
849 size_t trimlen;
871d21d4 850 size_t start_len = sb->len;
46d164b0
SB
851 int space = 2;
852
853 for (; *msg && *msg != '\n'; msg++) {
854 if (istitlechar(*msg)) {
855 if (space == 1)
856 strbuf_addch(sb, '-');
857 space = 0;
858 strbuf_addch(sb, *msg);
859 if (*msg == '.')
860 while (*(msg+1) == '.')
861 msg++;
862 } else
863 space |= 1;
864 }
865
866 /* trim any trailing '.' or '-' characters */
867 trimlen = 0;
871d21d4
SB
868 while (sb->len - trimlen > start_len &&
869 (sb->buf[sb->len - 1 - trimlen] == '.'
870 || sb->buf[sb->len - 1 - trimlen] == '-'))
46d164b0
SB
871 trimlen++;
872 strbuf_remove(sb, sb->len - trimlen, trimlen);
873}
874
cec08717
RS
875const char *format_subject(struct strbuf *sb, const char *msg,
876 const char *line_separator)
88c44735
RS
877{
878 int first = 1;
879
880 for (;;) {
881 const char *line = msg;
882 int linelen = get_one_line(line);
883
884 msg += linelen;
885 if (!linelen || is_empty_line(line, &linelen))
886 break;
887
f53bd743
RS
888 if (!sb)
889 continue;
88c44735
RS
890 strbuf_grow(sb, linelen + 2);
891 if (!first)
892 strbuf_addstr(sb, line_separator);
893 strbuf_add(sb, line, linelen);
894 first = 0;
895 }
896 return msg;
897}
898
f53bd743
RS
899static void parse_commit_message(struct format_commit_context *c)
900{
177b29dc
PN
901 const char *msg = c->message + c->message_off;
902 const char *start = c->message;
f53bd743
RS
903
904 msg = skip_empty_lines(msg);
905 c->subject_off = msg - start;
906
907 msg = format_subject(NULL, msg, NULL);
908 msg = skip_empty_lines(msg);
909 c->body_off = msg - start;
910
911 c->commit_message_parsed = 1;
912}
913
3b3d443f
RS
914static void format_decoration(struct strbuf *sb, const struct commit *commit)
915{
916 struct name_decoration *d;
917 const char *prefix = " (";
918
33e7018c 919 load_ref_decorations(DECORATE_SHORT_REFS);
3b3d443f
RS
920 d = lookup_decoration(&name_decoration, &commit->object);
921 while (d) {
922 strbuf_addstr(sb, prefix);
923 prefix = ", ";
924 strbuf_addstr(sb, d->name);
925 d = d->next;
926 }
927 if (prefix[0] == ',')
928 strbuf_addch(sb, ')');
929}
930
02edd56b
RS
931static void strbuf_wrap(struct strbuf *sb, size_t pos,
932 size_t width, size_t indent1, size_t indent2)
933{
934 struct strbuf tmp = STRBUF_INIT;
935
936 if (pos)
937 strbuf_add(&tmp, sb->buf, pos);
938 strbuf_add_wrapped_text(&tmp, sb->buf + pos,
939 (int) indent1, (int) indent2, (int) width);
940 strbuf_swap(&tmp, sb);
941 strbuf_release(&tmp);
942}
943
944static void rewrap_message_tail(struct strbuf *sb,
945 struct format_commit_context *c,
946 size_t new_width, size_t new_indent1,
947 size_t new_indent2)
948{
949 if (c->width == new_width && c->indent1 == new_indent1 &&
950 c->indent2 == new_indent2)
951 return;
32ca4249 952 if (c->wrap_start < sb->len)
02edd56b
RS
953 strbuf_wrap(sb, c->wrap_start, c->width, c->indent1, c->indent2);
954 c->wrap_start = sb->len;
955 c->width = new_width;
956 c->indent1 = new_indent1;
957 c->indent2 = new_indent2;
958}
959
cd1957f5
JK
960static int format_reflog_person(struct strbuf *sb,
961 char part,
962 struct reflog_walk_info *log,
963 enum date_mode dmode)
964{
965 const char *ident;
966
967 if (!log)
968 return 2;
969
970 ident = get_reflog_ident(log);
971 if (!ident)
972 return 2;
973
974 return format_person_part(sb, part, ident, strlen(ident), dmode);
975}
976
9fa708da
JH
977static size_t format_commit_one(struct strbuf *sb, const char *placeholder,
978 void *context)
f29d5958
RS
979{
980 struct format_commit_context *c = context;
981 const struct commit *commit = c->commit;
177b29dc 982 const char *msg = c->message;
f29d5958 983 struct commit_list *p;
42c8c74c 984 int h1, h2;
93fc05eb 985
93fc05eb 986 /* these are independent of the commit */
cde75e59
RS
987 switch (placeholder[0]) {
988 case 'C':
c002922a 989 if (placeholder[1] == '(') {
30825178
JH
990 const char *begin = placeholder + 2;
991 const char *end = strchr(begin, ')');
c002922a 992 char color[COLOR_MAXLEN];
30825178 993
c002922a
JK
994 if (!end)
995 return 0;
8a692d27 996 if (!prefixcmp(begin, "auto,")) {
30825178
JH
997 if (!want_color(c->pretty_ctx->color))
998 return end - placeholder + 1;
999 begin += 5;
1000 }
1001 color_parse_mem(begin,
1002 end - begin,
c002922a
JK
1003 "--pretty format", color);
1004 strbuf_addstr(sb, color);
1005 return end - placeholder + 1;
1006 }
c3a670de 1007 if (!prefixcmp(placeholder + 1, "red")) {
dc6ebd4c 1008 strbuf_addstr(sb, GIT_COLOR_RED);
c3a670de
MC
1009 return 4;
1010 } else if (!prefixcmp(placeholder + 1, "green")) {
dc6ebd4c 1011 strbuf_addstr(sb, GIT_COLOR_GREEN);
c3a670de
MC
1012 return 6;
1013 } else if (!prefixcmp(placeholder + 1, "blue")) {
dc6ebd4c 1014 strbuf_addstr(sb, GIT_COLOR_BLUE);
c3a670de
MC
1015 return 5;
1016 } else if (!prefixcmp(placeholder + 1, "reset")) {
dc6ebd4c 1017 strbuf_addstr(sb, GIT_COLOR_RESET);
c3a670de
MC
1018 return 6;
1019 } else
1020 return 0;
cde75e59
RS
1021 case 'n': /* newline */
1022 strbuf_addch(sb, '\n');
c3a670de 1023 return 1;
42c8c74c
GS
1024 case 'x':
1025 /* %x00 == NUL, %x0a == LF, etc. */
1026 if (0 <= (h1 = hexval_table[0xff & placeholder[1]]) &&
1027 h1 <= 16 &&
1028 0 <= (h2 = hexval_table[0xff & placeholder[2]]) &&
1029 h2 <= 16) {
1030 strbuf_addch(sb, (h1<<4)|h2);
1031 return 3;
1032 } else
1033 return 0;
02edd56b
RS
1034 case 'w':
1035 if (placeholder[1] == '(') {
1036 unsigned long width = 0, indent1 = 0, indent2 = 0;
1037 char *next;
1038 const char *start = placeholder + 2;
1039 const char *end = strchr(start, ')');
1040 if (!end)
1041 return 0;
1042 if (end > start) {
1043 width = strtoul(start, &next, 10);
1044 if (*next == ',') {
1045 indent1 = strtoul(next + 1, &next, 10);
1046 if (*next == ',') {
1047 indent2 = strtoul(next + 1,
1048 &next, 10);
1049 }
1050 }
1051 if (*next != ')')
1052 return 0;
1053 }
1054 rewrap_message_tail(sb, c, width, indent1, indent2);
1055 return end - placeholder + 1;
1056 } else
1057 return 0;
cde75e59 1058 }
93fc05eb
JS
1059
1060 /* these depend on the commit */
1061 if (!commit->object.parsed)
1062 parse_object(commit->object.sha1);
93fc05eb 1063
cde75e59
RS
1064 switch (placeholder[0]) {
1065 case 'H': /* commit hash */
1066 strbuf_addstr(sb, sha1_to_hex(commit->object.sha1));
c3a670de 1067 return 1;
cde75e59 1068 case 'h': /* abbreviated commit hash */
b9c62321 1069 if (add_again(sb, &c->abbrev_commit_hash))
c3a670de 1070 return 1;
cde75e59 1071 strbuf_addstr(sb, find_unique_abbrev(commit->object.sha1,
c1977021 1072 c->pretty_ctx->abbrev));
b9c62321 1073 c->abbrev_commit_hash.len = sb->len - c->abbrev_commit_hash.off;
c3a670de 1074 return 1;
cde75e59
RS
1075 case 'T': /* tree hash */
1076 strbuf_addstr(sb, sha1_to_hex(commit->tree->object.sha1));
c3a670de 1077 return 1;
cde75e59 1078 case 't': /* abbreviated tree hash */
b9c62321 1079 if (add_again(sb, &c->abbrev_tree_hash))
c3a670de 1080 return 1;
cde75e59 1081 strbuf_addstr(sb, find_unique_abbrev(commit->tree->object.sha1,
c1977021 1082 c->pretty_ctx->abbrev));
b9c62321 1083 c->abbrev_tree_hash.len = sb->len - c->abbrev_tree_hash.off;
c3a670de 1084 return 1;
cde75e59
RS
1085 case 'P': /* parent hashes */
1086 for (p = commit->parents; p; p = p->next) {
1087 if (p != commit->parents)
1088 strbuf_addch(sb, ' ');
1089 strbuf_addstr(sb, sha1_to_hex(p->item->object.sha1));
1090 }
c3a670de 1091 return 1;
cde75e59 1092 case 'p': /* abbreviated parent hashes */
b9c62321 1093 if (add_again(sb, &c->abbrev_parent_hashes))
c3a670de 1094 return 1;
cde75e59
RS
1095 for (p = commit->parents; p; p = p->next) {
1096 if (p != commit->parents)
1097 strbuf_addch(sb, ' ');
1098 strbuf_addstr(sb, find_unique_abbrev(
c1977021
WP
1099 p->item->object.sha1,
1100 c->pretty_ctx->abbrev));
cde75e59 1101 }
b9c62321
RS
1102 c->abbrev_parent_hashes.len = sb->len -
1103 c->abbrev_parent_hashes.off;
c3a670de 1104 return 1;
cde75e59 1105 case 'm': /* left/right/bottom */
1df2d656 1106 strbuf_addstr(sb, get_revision_mark(NULL, commit));
c3a670de 1107 return 1;
3b3d443f
RS
1108 case 'd':
1109 format_decoration(sb, commit);
1110 return 1;
8f8f5476
TR
1111 case 'g': /* reflog info */
1112 switch(placeholder[1]) {
1113 case 'd': /* reflog selector */
1114 case 'D':
1115 if (c->pretty_ctx->reflog_info)
1116 get_reflog_selector(sb,
1117 c->pretty_ctx->reflog_info,
1118 c->pretty_ctx->date_mode,
55ccf85a 1119 c->pretty_ctx->date_mode_explicit,
8f8f5476
TR
1120 (placeholder[1] == 'd'));
1121 return 2;
1122 case 's': /* reflog message */
1123 if (c->pretty_ctx->reflog_info)
1124 get_reflog_message(sb, c->pretty_ctx->reflog_info);
1125 return 2;
cd1957f5
JK
1126 case 'n':
1127 case 'N':
1128 case 'e':
1129 case 'E':
1130 return format_reflog_person(sb,
1131 placeholder[1],
1132 c->pretty_ctx->reflog_info,
1133 c->pretty_ctx->date_mode);
8f8f5476
TR
1134 }
1135 return 0; /* unknown %g placeholder */
8b208f02 1136 case 'N':
ddf333f6
JH
1137 if (c->pretty_ctx->notes_message) {
1138 strbuf_addstr(sb, c->pretty_ctx->notes_message);
5b163603
JG
1139 return 1;
1140 }
1141 return 0;
cde75e59
RS
1142 }
1143
f6667c5e 1144 if (placeholder[0] == 'G') {
ffb6d7d5
SG
1145 if (!c->signature_check.result)
1146 check_commit_signature(c->commit, &(c->signature_check));
f6667c5e
JH
1147 switch (placeholder[1]) {
1148 case 'G':
ffb6d7d5
SG
1149 if (c->signature_check.gpg_output)
1150 strbuf_addstr(sb, c->signature_check.gpg_output);
f6667c5e
JH
1151 break;
1152 case '?':
ffb6d7d5 1153 switch (c->signature_check.result) {
f6667c5e
JH
1154 case 'G':
1155 case 'B':
e290c4b9
SG
1156 case 'U':
1157 case 'N':
ffb6d7d5 1158 strbuf_addch(sb, c->signature_check.result);
f6667c5e
JH
1159 }
1160 break;
1161 case 'S':
ffb6d7d5
SG
1162 if (c->signature_check.signer)
1163 strbuf_addstr(sb, c->signature_check.signer);
f6667c5e 1164 break;
0174eeaa 1165 case 'K':
ffb6d7d5
SG
1166 if (c->signature_check.key)
1167 strbuf_addstr(sb, c->signature_check.key);
0174eeaa 1168 break;
f6667c5e
JH
1169 }
1170 return 2;
1171 }
1172
1173
cde75e59 1174 /* For the rest we have to parse the commit header. */
f29d5958
RS
1175 if (!c->commit_header_parsed)
1176 parse_commit_header(c);
93fc05eb 1177
f29d5958 1178 switch (placeholder[0]) {
c3a670de
MC
1179 case 'a': /* author ... */
1180 return format_person_part(sb, placeholder[1],
d36f8679 1181 msg + c->author.off, c->author.len,
dd2e794a 1182 c->pretty_ctx->date_mode);
c3a670de
MC
1183 case 'c': /* committer ... */
1184 return format_person_part(sb, placeholder[1],
d36f8679 1185 msg + c->committer.off, c->committer.len,
dd2e794a 1186 c->pretty_ctx->date_mode);
c3a670de 1187 case 'e': /* encoding */
f29d5958 1188 strbuf_add(sb, msg + c->encoding.off, c->encoding.len);
c3a670de 1189 return 1;
1367b12a
EB
1190 case 'B': /* raw body */
1191 /* message_off is always left at the initial newline */
1192 strbuf_addstr(sb, msg + c->message_off + 1);
1193 return 1;
f53bd743
RS
1194 }
1195
1196 /* Now we need to parse the commit message. */
1197 if (!c->commit_message_parsed)
1198 parse_commit_message(c);
1199
1200 switch (placeholder[0]) {
1201 case 's': /* subject */
1202 format_subject(sb, msg + c->subject_off, " ");
1203 return 1;
46d164b0
SB
1204 case 'f': /* sanitized subject */
1205 format_sanitized_subject(sb, msg + c->subject_off);
1206 return 1;
c3a670de 1207 case 'b': /* body */
f29d5958 1208 strbuf_addstr(sb, msg + c->body_off);
c3a670de 1209 return 1;
93fc05eb 1210 }
c3a670de 1211 return 0; /* unknown placeholder */
cde75e59
RS
1212}
1213
9fa708da
JH
1214static size_t format_commit_item(struct strbuf *sb, const char *placeholder,
1215 void *context)
1216{
1217 int consumed;
1218 size_t orig_len;
1219 enum {
1220 NO_MAGIC,
1221 ADD_LF_BEFORE_NON_EMPTY,
1222 DEL_LF_BEFORE_EMPTY,
223a923c 1223 ADD_SP_BEFORE_NON_EMPTY
9fa708da
JH
1224 } magic = NO_MAGIC;
1225
1226 switch (placeholder[0]) {
1227 case '-':
1228 magic = DEL_LF_BEFORE_EMPTY;
1229 break;
1230 case '+':
1231 magic = ADD_LF_BEFORE_NON_EMPTY;
1232 break;
7b88176e
MG
1233 case ' ':
1234 magic = ADD_SP_BEFORE_NON_EMPTY;
1235 break;
9fa708da
JH
1236 default:
1237 break;
1238 }
1239 if (magic != NO_MAGIC)
1240 placeholder++;
1241
1242 orig_len = sb->len;
1243 consumed = format_commit_one(sb, placeholder, context);
1244 if (magic == NO_MAGIC)
1245 return consumed;
1246
1247 if ((orig_len == sb->len) && magic == DEL_LF_BEFORE_EMPTY) {
1248 while (sb->len && sb->buf[sb->len - 1] == '\n')
1249 strbuf_setlen(sb, sb->len - 1);
7b88176e
MG
1250 } else if (orig_len != sb->len) {
1251 if (magic == ADD_LF_BEFORE_NON_EMPTY)
1252 strbuf_insert(sb, orig_len, "\n", 1);
1253 else if (magic == ADD_SP_BEFORE_NON_EMPTY)
1254 strbuf_insert(sb, orig_len, " ", 1);
9fa708da
JH
1255 }
1256 return consumed + 1;
1257}
1258
5b163603
JG
1259static size_t userformat_want_item(struct strbuf *sb, const char *placeholder,
1260 void *context)
1261{
1262 struct userformat_want *w = context;
1263
7b88176e 1264 if (*placeholder == '+' || *placeholder == '-' || *placeholder == ' ')
5b163603
JG
1265 placeholder++;
1266
1267 switch (*placeholder) {
1268 case 'N':
1269 w->notes = 1;
1270 break;
1271 }
1272 return 0;
1273}
1274
1275void userformat_find_requirements(const char *fmt, struct userformat_want *w)
1276{
1277 struct strbuf dummy = STRBUF_INIT;
1278
1279 if (!fmt) {
1280 if (!user_format)
1281 return;
1282 fmt = user_format;
1283 }
a6253d10 1284 strbuf_expand(&dummy, fmt, userformat_want_item, w);
5b163603
JG
1285 strbuf_release(&dummy);
1286}
1287
cde75e59 1288void format_commit_message(const struct commit *commit,
7f98ebc8 1289 const char *format, struct strbuf *sb,
dd2e794a 1290 const struct pretty_print_context *pretty_ctx)
cde75e59 1291{
f29d5958 1292 struct format_commit_context context;
177b29dc 1293 const char *output_enc = pretty_ctx->output_encoding;
f29d5958
RS
1294
1295 memset(&context, 0, sizeof(context));
1296 context.commit = commit;
dd2e794a 1297 context.pretty_ctx = pretty_ctx;
02edd56b 1298 context.wrap_start = sb->len;
5a10d236 1299 context.message = logmsg_reencode(commit, NULL, output_enc);
177b29dc 1300
c3a670de 1301 strbuf_expand(sb, format, format_commit_item, &context);
02edd56b 1302 rewrap_message_tail(sb, &context, 0, 0, 0);
177b29dc 1303
dd0d388c 1304 logmsg_free(context.message, commit);
ffb6d7d5
SG
1305 free(context.signature_check.gpg_output);
1306 free(context.signature_check.signer);
93fc05eb
JS
1307}
1308
6bf13944 1309static void pp_header(const struct pretty_print_context *pp,
93fc05eb
JS
1310 const char *encoding,
1311 const struct commit *commit,
1312 const char **msg_p,
1313 struct strbuf *sb)
1314{
1315 int parents_shown = 0;
1316
1317 for (;;) {
1318 const char *line = *msg_p;
1319 int linelen = get_one_line(*msg_p);
1320
1321 if (!linelen)
1322 return;
1323 *msg_p += linelen;
1324
1325 if (linelen == 1)
1326 /* End of header */
1327 return;
1328
6bf13944 1329 if (pp->fmt == CMIT_FMT_RAW) {
93fc05eb
JS
1330 strbuf_add(sb, line, linelen);
1331 continue;
1332 }
1333
8a692d27 1334 if (!prefixcmp(line, "parent ")) {
93fc05eb
JS
1335 if (linelen != 48)
1336 die("bad parent line in commit");
1337 continue;
1338 }
1339
1340 if (!parents_shown) {
1341 struct commit_list *parent;
1342 int num;
1343 for (parent = commit->parents, num = 0;
1344 parent;
1345 parent = parent->next, num++)
1346 ;
1347 /* with enough slop */
1348 strbuf_grow(sb, num * 50 + 20);
6bf13944 1349 add_merge_info(pp, sb, commit);
93fc05eb
JS
1350 parents_shown = 1;
1351 }
1352
1353 /*
1354 * MEDIUM == DEFAULT shows only author with dates.
1355 * FULL shows both authors but not dates.
1356 * FULLER shows both authors and dates.
1357 */
8a692d27 1358 if (!prefixcmp(line, "author ")) {
93fc05eb 1359 strbuf_grow(sb, linelen + 80);
6bf13944 1360 pp_user_info(pp, "Author", sb, line + 7, encoding);
93fc05eb 1361 }
8a692d27 1362 if (!prefixcmp(line, "committer ") &&
6bf13944 1363 (pp->fmt == CMIT_FMT_FULL || pp->fmt == CMIT_FMT_FULLER)) {
93fc05eb 1364 strbuf_grow(sb, linelen + 80);
6bf13944 1365 pp_user_info(pp, "Commit", sb, line + 10, encoding);
93fc05eb
JS
1366 }
1367 }
1368}
1369
6bf13944 1370void pp_title_line(const struct pretty_print_context *pp,
b02bd65f
DB
1371 const char **msg_p,
1372 struct strbuf *sb,
b02bd65f 1373 const char *encoding,
267123b4 1374 int need_8bit_cte)
93fc05eb 1375{
41dd00ba 1376 static const int max_length = 78; /* per rfc2047 */
93fc05eb
JS
1377 struct strbuf title;
1378
1379 strbuf_init(&title, 80);
9553d2b2
JK
1380 *msg_p = format_subject(&title, *msg_p,
1381 pp->preserve_subject ? "\n" : " ");
93fc05eb
JS
1382
1383 strbuf_grow(sb, title.len + 1024);
6bf13944
JK
1384 if (pp->subject) {
1385 strbuf_addstr(sb, pp->subject);
41dd00ba
JS
1386 if (needs_rfc2047_encoding(title.buf, title.len, RFC2047_SUBJECT))
1387 add_rfc2047(sb, title.buf, title.len,
1388 encoding, RFC2047_SUBJECT);
1389 else
1390 strbuf_add_wrapped_bytes(sb, title.buf, title.len,
1391 -last_line_length(sb), 1, max_length);
93fc05eb
JS
1392 } else {
1393 strbuf_addbuf(sb, &title);
1394 }
1395 strbuf_addch(sb, '\n');
1396
6bf4f1b4 1397 if (need_8bit_cte > 0) {
93fc05eb
JS
1398 const char *header_fmt =
1399 "MIME-Version: 1.0\n"
1400 "Content-Type: text/plain; charset=%s\n"
1401 "Content-Transfer-Encoding: 8bit\n";
1402 strbuf_addf(sb, header_fmt, encoding);
1403 }
6bf13944
JK
1404 if (pp->after_subject) {
1405 strbuf_addstr(sb, pp->after_subject);
93fc05eb 1406 }
6bf13944 1407 if (pp->fmt == CMIT_FMT_EMAIL) {
93fc05eb
JS
1408 strbuf_addch(sb, '\n');
1409 }
1410 strbuf_release(&title);
1411}
1412
6bf13944 1413void pp_remainder(const struct pretty_print_context *pp,
b02bd65f
DB
1414 const char **msg_p,
1415 struct strbuf *sb,
1416 int indent)
93fc05eb
JS
1417{
1418 int first = 1;
1419 for (;;) {
1420 const char *line = *msg_p;
1421 int linelen = get_one_line(line);
1422 *msg_p += linelen;
1423
1424 if (!linelen)
1425 break;
1426
1427 if (is_empty_line(line, &linelen)) {
1428 if (first)
1429 continue;
6bf13944 1430 if (pp->fmt == CMIT_FMT_SHORT)
93fc05eb
JS
1431 break;
1432 }
1433 first = 0;
1434
1435 strbuf_grow(sb, linelen + indent + 20);
1436 if (indent) {
1437 memset(sb->buf + sb->len, ' ', indent);
1438 strbuf_setlen(sb, sb->len + indent);
1439 }
1440 strbuf_add(sb, line, linelen);
1441 strbuf_addch(sb, '\n');
1442 }
1443}
1444
6bf13944
JK
1445void pretty_print_commit(const struct pretty_print_context *pp,
1446 const struct commit *commit,
1447 struct strbuf *sb)
93fc05eb
JS
1448{
1449 unsigned long beginning_of_body;
1450 int indent = 4;
dd0d388c 1451 const char *msg;
93fc05eb
JS
1452 char *reencoded;
1453 const char *encoding;
6bf13944 1454 int need_8bit_cte = pp->need_8bit_cte;
93fc05eb 1455
6bf13944
JK
1456 if (pp->fmt == CMIT_FMT_USERFORMAT) {
1457 format_commit_message(commit, user_format, sb, pp);
93fc05eb
JS
1458 return;
1459 }
1460
e297cf5a 1461 encoding = get_log_output_encoding();
5a10d236 1462 msg = reencoded = logmsg_reencode(commit, NULL, encoding);
93fc05eb 1463
6bf13944 1464 if (pp->fmt == CMIT_FMT_ONELINE || pp->fmt == CMIT_FMT_EMAIL)
93fc05eb
JS
1465 indent = 0;
1466
6bf4f1b4
JH
1467 /*
1468 * We need to check and emit Content-type: to mark it
1469 * as 8-bit if we haven't done so.
93fc05eb 1470 */
6bf13944 1471 if (pp->fmt == CMIT_FMT_EMAIL && need_8bit_cte == 0) {
93fc05eb
JS
1472 int i, ch, in_body;
1473
1474 for (in_body = i = 0; (ch = msg[i]); i++) {
1475 if (!in_body) {
1476 /* author could be non 7-bit ASCII but
1477 * the log may be so; skip over the
1478 * header part first.
1479 */
1480 if (ch == '\n' && msg[i+1] == '\n')
1481 in_body = 1;
1482 }
1483 else if (non_ascii(ch)) {
6bf4f1b4 1484 need_8bit_cte = 1;
93fc05eb
JS
1485 break;
1486 }
1487 }
1488 }
1489
6bf13944
JK
1490 pp_header(pp, encoding, commit, &msg, sb);
1491 if (pp->fmt != CMIT_FMT_ONELINE && !pp->subject) {
93fc05eb
JS
1492 strbuf_addch(sb, '\n');
1493 }
1494
1495 /* Skip excess blank lines at the beginning of body, if any... */
a0109668 1496 msg = skip_empty_lines(msg);
93fc05eb
JS
1497
1498 /* These formats treat the title line specially. */
6bf13944
JK
1499 if (pp->fmt == CMIT_FMT_ONELINE || pp->fmt == CMIT_FMT_EMAIL)
1500 pp_title_line(pp, &msg, sb, encoding, need_8bit_cte);
93fc05eb
JS
1501
1502 beginning_of_body = sb->len;
6bf13944
JK
1503 if (pp->fmt != CMIT_FMT_ONELINE)
1504 pp_remainder(pp, &msg, sb, indent);
93fc05eb
JS
1505 strbuf_rtrim(sb);
1506
1507 /* Make sure there is an EOLN for the non-oneline case */
6bf13944 1508 if (pp->fmt != CMIT_FMT_ONELINE)
93fc05eb
JS
1509 strbuf_addch(sb, '\n');
1510
1511 /*
1512 * The caller may append additional body text in e-mail
1513 * format. Make sure we did not strip the blank line
1514 * between the header and the body.
1515 */
6bf13944 1516 if (pp->fmt == CMIT_FMT_EMAIL && sb->len <= beginning_of_body)
93fc05eb 1517 strbuf_addch(sb, '\n');
a97a7468 1518
dd0d388c 1519 logmsg_free(reencoded, commit);
93fc05eb 1520}
8b8a5374
JK
1521
1522void pp_commit_easy(enum cmit_fmt fmt, const struct commit *commit,
1523 struct strbuf *sb)
1524{
1525 struct pretty_print_context pp = {0};
6bf13944
JK
1526 pp.fmt = fmt;
1527 pretty_print_commit(&pp, commit, sb);
8b8a5374 1528}