pretty: simplify input line length calculation in pp_user_info()
[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
9dbe7c3d
RS
396static const char *show_ident_date(const struct ident_split *ident,
397 enum date_mode mode)
398{
399 unsigned long date = 0;
400 int tz = 0;
401
402 if (ident->date_begin && ident->date_end)
403 date = strtoul(ident->date_begin, NULL, 10);
404 if (ident->tz_begin && ident->tz_end)
405 tz = strtol(ident->tz_begin, NULL, 10);
406 return show_date(date, tz, mode);
407}
408
6bf13944
JK
409void pp_user_info(const struct pretty_print_context *pp,
410 const char *what, struct strbuf *sb,
411 const char *line, const char *encoding)
93fc05eb 412{
dffd325f
AP
413 struct strbuf name;
414 struct strbuf mail;
3c020bd5 415 struct ident_split ident;
9dbe7c3d 416 char *line_end;
dffd325f
AP
417 const char *mailbuf, *namebuf;
418 size_t namelen, maillen;
41dd00ba 419 int max_length = 78; /* per rfc2822 */
93fc05eb 420
6bf13944 421 if (pp->fmt == CMIT_FMT_ONELINE)
93fc05eb 422 return;
3c020bd5 423
30e77bcb
RS
424 line_end = strchrnul(line, '\n');
425 if (split_ident_line(&ident, line, line_end - line))
93fc05eb 426 return;
3c020bd5 427
dffd325f
AP
428 mailbuf = ident.mail_begin;
429 maillen = ident.mail_end - ident.mail_begin;
430 namebuf = ident.name_begin;
431 namelen = ident.name_end - ident.name_begin;
432
433 if (pp->mailmap)
434 map_user(pp->mailmap, &mailbuf, &maillen, &namebuf, &namelen);
435
436 strbuf_init(&mail, 0);
437 strbuf_init(&name, 0);
438
439 strbuf_add(&mail, mailbuf, maillen);
440 strbuf_add(&name, namebuf, namelen);
441
442 namelen = name.len + mail.len + 3; /* ' ' + '<' + '>' */
93fc05eb 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:
9dbe7c3d
RS
475 strbuf_addf(sb, "Date: %s\n",
476 show_ident_date(&ident, pp->date_mode));
93fc05eb
JS
477 break;
478 case CMIT_FMT_EMAIL:
9dbe7c3d
RS
479 strbuf_addf(sb, "Date: %s\n",
480 show_ident_date(&ident, DATE_RFC2822));
93fc05eb
JS
481 break;
482 case CMIT_FMT_FULLER:
9dbe7c3d
RS
483 strbuf_addf(sb, "%sDate: %s\n", what,
484 show_ident_date(&ident, pp->date_mode));
93fc05eb
JS
485 break;
486 default:
487 /* notin' */
488 break;
489 }
490}
491
492static int is_empty_line(const char *line, int *len_p)
493{
494 int len = *len_p;
495 while (len && isspace(line[len-1]))
496 len--;
497 *len_p = len;
498 return !len;
499}
500
a0109668
RS
501static const char *skip_empty_lines(const char *msg)
502{
503 for (;;) {
504 int linelen = get_one_line(msg);
505 int ll = linelen;
506 if (!linelen)
507 break;
508 if (!is_empty_line(msg, &ll))
509 break;
510 msg += linelen;
511 }
512 return msg;
513}
514
6bf13944
JK
515static void add_merge_info(const struct pretty_print_context *pp,
516 struct strbuf *sb, const struct commit *commit)
93fc05eb
JS
517{
518 struct commit_list *parent = commit->parents;
519
6bf13944 520 if ((pp->fmt == CMIT_FMT_ONELINE) || (pp->fmt == CMIT_FMT_EMAIL) ||
93fc05eb
JS
521 !parent || !parent->next)
522 return;
523
524 strbuf_addstr(sb, "Merge:");
525
526 while (parent) {
527 struct commit *p = parent->item;
528 const char *hex = NULL;
6bf13944
JK
529 if (pp->abbrev)
530 hex = find_unique_abbrev(p->object.sha1, pp->abbrev);
93fc05eb
JS
531 if (!hex)
532 hex = sha1_to_hex(p->object.sha1);
93fc05eb
JS
533 parent = parent->next;
534
7fcda920 535 strbuf_addf(sb, " %s", hex);
93fc05eb
JS
536 }
537 strbuf_addch(sb, '\n');
538}
539
dd0d388c
JK
540static char *get_header(const struct commit *commit, const char *msg,
541 const char *key)
93fc05eb
JS
542{
543 int key_len = strlen(key);
dd0d388c 544 const char *line = msg;
93fc05eb 545
a9c7a8a8 546 while (line) {
93fc05eb
JS
547 const char *eol = strchr(line, '\n'), *next;
548
549 if (line == eol)
550 return NULL;
551 if (!eol) {
a9c7a8a8
JK
552 warning("malformed commit (header is missing newline): %s",
553 sha1_to_hex(commit->object.sha1));
93fc05eb
JS
554 eol = line + strlen(line);
555 next = NULL;
556 } else
557 next = eol + 1;
558 if (eol - line > key_len &&
559 !strncmp(line, key, key_len) &&
560 line[key_len] == ' ') {
561 return xmemdupz(line + key_len + 1, eol - line - key_len - 1);
562 }
563 line = next;
564 }
a9c7a8a8 565 return NULL;
93fc05eb
JS
566}
567
568static char *replace_encoding_header(char *buf, const char *encoding)
569{
f285a2d7 570 struct strbuf tmp = STRBUF_INIT;
93fc05eb
JS
571 size_t start, len;
572 char *cp = buf;
573
574 /* guess if there is an encoding header before a \n\n */
575 while (strncmp(cp, "encoding ", strlen("encoding "))) {
576 cp = strchr(cp, '\n');
577 if (!cp || *++cp == '\n')
578 return buf;
579 }
580 start = cp - buf;
581 cp = strchr(cp, '\n');
582 if (!cp)
583 return buf; /* should not happen but be defensive */
584 len = cp + 1 - (buf + start);
585
93fc05eb
JS
586 strbuf_attach(&tmp, buf, strlen(buf), strlen(buf) + 1);
587 if (is_encoding_utf8(encoding)) {
588 /* we have re-coded to UTF-8; drop the header */
589 strbuf_remove(&tmp, start, len);
590 } else {
591 /* just replaces XXXX in 'encoding XXXX\n' */
592 strbuf_splice(&tmp, start + strlen("encoding "),
593 len - strlen("encoding \n"),
594 encoding, strlen(encoding));
595 }
596 return strbuf_detach(&tmp, NULL);
597}
598
177b29dc
PN
599char *logmsg_reencode(const struct commit *commit,
600 const char *output_encoding)
93fc05eb 601{
330db18c 602 static const char *utf8 = "UTF-8";
93fc05eb
JS
603 const char *use_encoding;
604 char *encoding;
dd0d388c 605 char *msg = commit->buffer;
93fc05eb
JS
606 char *out;
607
be5c9fb9
JK
608 if (!msg) {
609 enum object_type type;
610 unsigned long size;
611
612 msg = read_sha1_file(commit->object.sha1, &type, &size);
613 if (!msg)
614 die("Cannot read commit object %s",
615 sha1_to_hex(commit->object.sha1));
616 if (type != OBJ_COMMIT)
617 die("Expected commit for '%s', got %s",
618 sha1_to_hex(commit->object.sha1), typename(type));
619 }
620
ec008076 621 if (!output_encoding || !*output_encoding)
dd0d388c
JK
622 return msg;
623 encoding = get_header(commit, msg, "encoding");
93fc05eb 624 use_encoding = encoding ? encoding : utf8;
be5c9fb9
JK
625 if (same_encoding(use_encoding, output_encoding)) {
626 /*
627 * No encoding work to be done. If we have no encoding header
628 * at all, then there's nothing to do, and we can return the
629 * message verbatim (whether newly allocated or not).
630 */
631 if (!encoding)
632 return msg;
633
634 /*
635 * Otherwise, we still want to munge the encoding header in the
636 * result, which will be done by modifying the buffer. If we
637 * are using a fresh copy, we can reuse it. But if we are using
638 * the cached copy from commit->buffer, we need to duplicate it
639 * to avoid munging commit->buffer.
640 */
641 out = msg;
642 if (out == commit->buffer)
643 out = xstrdup(out);
644 }
645 else {
646 /*
647 * There's actual encoding work to do. Do the reencoding, which
648 * still leaves the header to be replaced in the next step. At
649 * this point, we are done with msg. If we allocated a fresh
650 * copy, we can free it.
651 */
652 out = reencode_string(msg, output_encoding, use_encoding);
653 if (out && msg != commit->buffer)
654 free(msg);
655 }
656
657 /*
658 * This replacement actually consumes the buffer we hand it, so we do
659 * not have to worry about freeing the old "out" here.
660 */
93fc05eb
JS
661 if (out)
662 out = replace_encoding_header(out, output_encoding);
663
664 free(encoding);
dd0d388c
JK
665 /*
666 * If the re-encoding failed, out might be NULL here; in that
667 * case we just return the commit message verbatim.
668 */
669 return out ? out : msg;
670}
671
672void logmsg_free(char *msg, const struct commit *commit)
673{
674 if (msg != commit->buffer)
675 free(msg);
93fc05eb
JS
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
758struct format_commit_context {
759 const struct commit *commit;
dd2e794a 760 const struct pretty_print_context *pretty_ctx;
f53bd743
RS
761 unsigned commit_header_parsed:1;
762 unsigned commit_message_parsed:1;
f6667c5e
JH
763 unsigned commit_signature_parsed:1;
764 struct {
765 char *gpg_output;
4a868fd6 766 char *gpg_status;
f6667c5e
JH
767 char good_bad;
768 char *signer;
0174eeaa 769 char *key;
f6667c5e 770 } signature;
177b29dc 771 char *message;
02edd56b 772 size_t width, indent1, indent2;
f29d5958
RS
773
774 /* These offsets are relative to the start of the commit message. */
f29d5958
RS
775 struct chunk author;
776 struct chunk committer;
777 struct chunk encoding;
f53bd743
RS
778 size_t message_off;
779 size_t subject_off;
f29d5958 780 size_t body_off;
b9c62321
RS
781
782 /* The following ones are relative to the result struct strbuf. */
783 struct chunk abbrev_commit_hash;
784 struct chunk abbrev_tree_hash;
785 struct chunk abbrev_parent_hashes;
02edd56b 786 size_t wrap_start;
f29d5958
RS
787};
788
b9c62321
RS
789static int add_again(struct strbuf *sb, struct chunk *chunk)
790{
791 if (chunk->len) {
792 strbuf_adddup(sb, chunk->off, chunk->len);
793 return 1;
794 }
795
796 /*
797 * We haven't seen this chunk before. Our caller is surely
798 * going to add it the hard way now. Remember the most likely
799 * start of the to-be-added chunk: the current end of the
800 * struct strbuf.
801 */
802 chunk->off = sb->len;
803 return 0;
804}
805
f29d5958 806static void parse_commit_header(struct format_commit_context *context)
93fc05eb 807{
177b29dc 808 const char *msg = context->message;
93fc05eb 809 int i;
f29d5958 810
f53bd743 811 for (i = 0; msg[i]; i++) {
f29d5958
RS
812 int eol;
813 for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
814 ; /* do nothing */
815
f29d5958 816 if (i == eol) {
f53bd743 817 break;
f29d5958
RS
818 } else if (!prefixcmp(msg + i, "author ")) {
819 context->author.off = i + 7;
820 context->author.len = eol - i - 7;
821 } else if (!prefixcmp(msg + i, "committer ")) {
822 context->committer.off = i + 10;
823 context->committer.len = eol - i - 10;
824 } else if (!prefixcmp(msg + i, "encoding ")) {
825 context->encoding.off = i + 9;
826 context->encoding.len = eol - i - 9;
827 }
828 i = eol;
829 }
f53bd743 830 context->message_off = i;
f29d5958
RS
831 context->commit_header_parsed = 1;
832}
833
46d164b0
SB
834static int istitlechar(char c)
835{
836 return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') ||
837 (c >= '0' && c <= '9') || c == '.' || c == '_';
838}
839
840static void format_sanitized_subject(struct strbuf *sb, const char *msg)
841{
842 size_t trimlen;
871d21d4 843 size_t start_len = sb->len;
46d164b0
SB
844 int space = 2;
845
846 for (; *msg && *msg != '\n'; msg++) {
847 if (istitlechar(*msg)) {
848 if (space == 1)
849 strbuf_addch(sb, '-');
850 space = 0;
851 strbuf_addch(sb, *msg);
852 if (*msg == '.')
853 while (*(msg+1) == '.')
854 msg++;
855 } else
856 space |= 1;
857 }
858
859 /* trim any trailing '.' or '-' characters */
860 trimlen = 0;
871d21d4
SB
861 while (sb->len - trimlen > start_len &&
862 (sb->buf[sb->len - 1 - trimlen] == '.'
863 || sb->buf[sb->len - 1 - trimlen] == '-'))
46d164b0
SB
864 trimlen++;
865 strbuf_remove(sb, sb->len - trimlen, trimlen);
866}
867
cec08717
RS
868const char *format_subject(struct strbuf *sb, const char *msg,
869 const char *line_separator)
88c44735
RS
870{
871 int first = 1;
872
873 for (;;) {
874 const char *line = msg;
875 int linelen = get_one_line(line);
876
877 msg += linelen;
878 if (!linelen || is_empty_line(line, &linelen))
879 break;
880
f53bd743
RS
881 if (!sb)
882 continue;
88c44735
RS
883 strbuf_grow(sb, linelen + 2);
884 if (!first)
885 strbuf_addstr(sb, line_separator);
886 strbuf_add(sb, line, linelen);
887 first = 0;
888 }
889 return msg;
890}
891
f53bd743
RS
892static void parse_commit_message(struct format_commit_context *c)
893{
177b29dc
PN
894 const char *msg = c->message + c->message_off;
895 const char *start = c->message;
f53bd743
RS
896
897 msg = skip_empty_lines(msg);
898 c->subject_off = msg - start;
899
900 msg = format_subject(NULL, msg, NULL);
901 msg = skip_empty_lines(msg);
902 c->body_off = msg - start;
903
904 c->commit_message_parsed = 1;
905}
906
3b3d443f
RS
907static void format_decoration(struct strbuf *sb, const struct commit *commit)
908{
909 struct name_decoration *d;
910 const char *prefix = " (";
911
33e7018c 912 load_ref_decorations(DECORATE_SHORT_REFS);
3b3d443f
RS
913 d = lookup_decoration(&name_decoration, &commit->object);
914 while (d) {
915 strbuf_addstr(sb, prefix);
916 prefix = ", ";
917 strbuf_addstr(sb, d->name);
918 d = d->next;
919 }
920 if (prefix[0] == ',')
921 strbuf_addch(sb, ')');
922}
923
02edd56b
RS
924static void strbuf_wrap(struct strbuf *sb, size_t pos,
925 size_t width, size_t indent1, size_t indent2)
926{
927 struct strbuf tmp = STRBUF_INIT;
928
929 if (pos)
930 strbuf_add(&tmp, sb->buf, pos);
931 strbuf_add_wrapped_text(&tmp, sb->buf + pos,
932 (int) indent1, (int) indent2, (int) width);
933 strbuf_swap(&tmp, sb);
934 strbuf_release(&tmp);
935}
936
937static void rewrap_message_tail(struct strbuf *sb,
938 struct format_commit_context *c,
939 size_t new_width, size_t new_indent1,
940 size_t new_indent2)
941{
942 if (c->width == new_width && c->indent1 == new_indent1 &&
943 c->indent2 == new_indent2)
944 return;
32ca4249 945 if (c->wrap_start < sb->len)
02edd56b
RS
946 strbuf_wrap(sb, c->wrap_start, c->width, c->indent1, c->indent2);
947 c->wrap_start = sb->len;
948 c->width = new_width;
949 c->indent1 = new_indent1;
950 c->indent2 = new_indent2;
951}
952
f6667c5e
JH
953static struct {
954 char result;
955 const char *check;
956} signature_check[] = {
4a868fd6
MG
957 { 'G', "\n[GNUPG:] GOODSIG " },
958 { 'B', "\n[GNUPG:] BADSIG " },
f6667c5e
JH
959};
960
961static void parse_signature_lines(struct format_commit_context *ctx)
962{
4a868fd6 963 const char *buf = ctx->signature.gpg_status;
f6667c5e
JH
964 int i;
965
966 for (i = 0; i < ARRAY_SIZE(signature_check); i++) {
967 const char *found = strstr(buf, signature_check[i].check);
968 const char *next;
969 if (!found)
970 continue;
971 ctx->signature.good_bad = signature_check[i].result;
972 found += strlen(signature_check[i].check);
0174eeaa
MG
973 ctx->signature.key = xmemdupz(found, 16);
974 found += 17;
f6667c5e
JH
975 next = strchrnul(found, '\n');
976 ctx->signature.signer = xmemdupz(found, next - found);
977 break;
978 }
979}
980
981static void parse_commit_signature(struct format_commit_context *ctx)
982{
983 struct strbuf payload = STRBUF_INIT;
984 struct strbuf signature = STRBUF_INIT;
985 struct strbuf gpg_output = STRBUF_INIT;
4a868fd6 986 struct strbuf gpg_status = STRBUF_INIT;
f6667c5e
JH
987 int status;
988
989 ctx->commit_signature_parsed = 1;
990
991 if (parse_signed_commit(ctx->commit->object.sha1,
992 &payload, &signature) <= 0)
993 goto out;
994 status = verify_signed_buffer(payload.buf, payload.len,
995 signature.buf, signature.len,
4a868fd6 996 &gpg_output, &gpg_status);
f6667c5e
JH
997 if (status && !gpg_output.len)
998 goto out;
999 ctx->signature.gpg_output = strbuf_detach(&gpg_output, NULL);
4a868fd6 1000 ctx->signature.gpg_status = strbuf_detach(&gpg_status, NULL);
f6667c5e
JH
1001 parse_signature_lines(ctx);
1002
1003 out:
4a868fd6 1004 strbuf_release(&gpg_status);
f6667c5e
JH
1005 strbuf_release(&gpg_output);
1006 strbuf_release(&payload);
1007 strbuf_release(&signature);
1008}
1009
1010
cd1957f5
JK
1011static int format_reflog_person(struct strbuf *sb,
1012 char part,
1013 struct reflog_walk_info *log,
1014 enum date_mode dmode)
1015{
1016 const char *ident;
1017
1018 if (!log)
1019 return 2;
1020
1021 ident = get_reflog_ident(log);
1022 if (!ident)
1023 return 2;
1024
1025 return format_person_part(sb, part, ident, strlen(ident), dmode);
1026}
1027
9fa708da
JH
1028static size_t format_commit_one(struct strbuf *sb, const char *placeholder,
1029 void *context)
f29d5958
RS
1030{
1031 struct format_commit_context *c = context;
1032 const struct commit *commit = c->commit;
177b29dc 1033 const char *msg = c->message;
f29d5958 1034 struct commit_list *p;
42c8c74c 1035 int h1, h2;
93fc05eb 1036
93fc05eb 1037 /* these are independent of the commit */
cde75e59
RS
1038 switch (placeholder[0]) {
1039 case 'C':
c002922a 1040 if (placeholder[1] == '(') {
30825178
JH
1041 const char *begin = placeholder + 2;
1042 const char *end = strchr(begin, ')');
c002922a 1043 char color[COLOR_MAXLEN];
30825178 1044
c002922a
JK
1045 if (!end)
1046 return 0;
8a692d27 1047 if (!prefixcmp(begin, "auto,")) {
30825178
JH
1048 if (!want_color(c->pretty_ctx->color))
1049 return end - placeholder + 1;
1050 begin += 5;
1051 }
1052 color_parse_mem(begin,
1053 end - begin,
c002922a
JK
1054 "--pretty format", color);
1055 strbuf_addstr(sb, color);
1056 return end - placeholder + 1;
1057 }
c3a670de 1058 if (!prefixcmp(placeholder + 1, "red")) {
dc6ebd4c 1059 strbuf_addstr(sb, GIT_COLOR_RED);
c3a670de
MC
1060 return 4;
1061 } else if (!prefixcmp(placeholder + 1, "green")) {
dc6ebd4c 1062 strbuf_addstr(sb, GIT_COLOR_GREEN);
c3a670de
MC
1063 return 6;
1064 } else if (!prefixcmp(placeholder + 1, "blue")) {
dc6ebd4c 1065 strbuf_addstr(sb, GIT_COLOR_BLUE);
c3a670de
MC
1066 return 5;
1067 } else if (!prefixcmp(placeholder + 1, "reset")) {
dc6ebd4c 1068 strbuf_addstr(sb, GIT_COLOR_RESET);
c3a670de
MC
1069 return 6;
1070 } else
1071 return 0;
cde75e59
RS
1072 case 'n': /* newline */
1073 strbuf_addch(sb, '\n');
c3a670de 1074 return 1;
42c8c74c
GS
1075 case 'x':
1076 /* %x00 == NUL, %x0a == LF, etc. */
1077 if (0 <= (h1 = hexval_table[0xff & placeholder[1]]) &&
1078 h1 <= 16 &&
1079 0 <= (h2 = hexval_table[0xff & placeholder[2]]) &&
1080 h2 <= 16) {
1081 strbuf_addch(sb, (h1<<4)|h2);
1082 return 3;
1083 } else
1084 return 0;
02edd56b
RS
1085 case 'w':
1086 if (placeholder[1] == '(') {
1087 unsigned long width = 0, indent1 = 0, indent2 = 0;
1088 char *next;
1089 const char *start = placeholder + 2;
1090 const char *end = strchr(start, ')');
1091 if (!end)
1092 return 0;
1093 if (end > start) {
1094 width = strtoul(start, &next, 10);
1095 if (*next == ',') {
1096 indent1 = strtoul(next + 1, &next, 10);
1097 if (*next == ',') {
1098 indent2 = strtoul(next + 1,
1099 &next, 10);
1100 }
1101 }
1102 if (*next != ')')
1103 return 0;
1104 }
1105 rewrap_message_tail(sb, c, width, indent1, indent2);
1106 return end - placeholder + 1;
1107 } else
1108 return 0;
cde75e59 1109 }
93fc05eb
JS
1110
1111 /* these depend on the commit */
1112 if (!commit->object.parsed)
1113 parse_object(commit->object.sha1);
93fc05eb 1114
cde75e59
RS
1115 switch (placeholder[0]) {
1116 case 'H': /* commit hash */
1117 strbuf_addstr(sb, sha1_to_hex(commit->object.sha1));
c3a670de 1118 return 1;
cde75e59 1119 case 'h': /* abbreviated commit hash */
b9c62321 1120 if (add_again(sb, &c->abbrev_commit_hash))
c3a670de 1121 return 1;
cde75e59 1122 strbuf_addstr(sb, find_unique_abbrev(commit->object.sha1,
c1977021 1123 c->pretty_ctx->abbrev));
b9c62321 1124 c->abbrev_commit_hash.len = sb->len - c->abbrev_commit_hash.off;
c3a670de 1125 return 1;
cde75e59
RS
1126 case 'T': /* tree hash */
1127 strbuf_addstr(sb, sha1_to_hex(commit->tree->object.sha1));
c3a670de 1128 return 1;
cde75e59 1129 case 't': /* abbreviated tree hash */
b9c62321 1130 if (add_again(sb, &c->abbrev_tree_hash))
c3a670de 1131 return 1;
cde75e59 1132 strbuf_addstr(sb, find_unique_abbrev(commit->tree->object.sha1,
c1977021 1133 c->pretty_ctx->abbrev));
b9c62321 1134 c->abbrev_tree_hash.len = sb->len - c->abbrev_tree_hash.off;
c3a670de 1135 return 1;
cde75e59
RS
1136 case 'P': /* parent hashes */
1137 for (p = commit->parents; p; p = p->next) {
1138 if (p != commit->parents)
1139 strbuf_addch(sb, ' ');
1140 strbuf_addstr(sb, sha1_to_hex(p->item->object.sha1));
1141 }
c3a670de 1142 return 1;
cde75e59 1143 case 'p': /* abbreviated parent hashes */
b9c62321 1144 if (add_again(sb, &c->abbrev_parent_hashes))
c3a670de 1145 return 1;
cde75e59
RS
1146 for (p = commit->parents; p; p = p->next) {
1147 if (p != commit->parents)
1148 strbuf_addch(sb, ' ');
1149 strbuf_addstr(sb, find_unique_abbrev(
c1977021
WP
1150 p->item->object.sha1,
1151 c->pretty_ctx->abbrev));
cde75e59 1152 }
b9c62321
RS
1153 c->abbrev_parent_hashes.len = sb->len -
1154 c->abbrev_parent_hashes.off;
c3a670de 1155 return 1;
cde75e59 1156 case 'm': /* left/right/bottom */
1df2d656 1157 strbuf_addstr(sb, get_revision_mark(NULL, commit));
c3a670de 1158 return 1;
3b3d443f
RS
1159 case 'd':
1160 format_decoration(sb, commit);
1161 return 1;
8f8f5476
TR
1162 case 'g': /* reflog info */
1163 switch(placeholder[1]) {
1164 case 'd': /* reflog selector */
1165 case 'D':
1166 if (c->pretty_ctx->reflog_info)
1167 get_reflog_selector(sb,
1168 c->pretty_ctx->reflog_info,
1169 c->pretty_ctx->date_mode,
55ccf85a 1170 c->pretty_ctx->date_mode_explicit,
8f8f5476
TR
1171 (placeholder[1] == 'd'));
1172 return 2;
1173 case 's': /* reflog message */
1174 if (c->pretty_ctx->reflog_info)
1175 get_reflog_message(sb, c->pretty_ctx->reflog_info);
1176 return 2;
cd1957f5
JK
1177 case 'n':
1178 case 'N':
1179 case 'e':
1180 case 'E':
1181 return format_reflog_person(sb,
1182 placeholder[1],
1183 c->pretty_ctx->reflog_info,
1184 c->pretty_ctx->date_mode);
8f8f5476
TR
1185 }
1186 return 0; /* unknown %g placeholder */
8b208f02 1187 case 'N':
ddf333f6
JH
1188 if (c->pretty_ctx->notes_message) {
1189 strbuf_addstr(sb, c->pretty_ctx->notes_message);
5b163603
JG
1190 return 1;
1191 }
1192 return 0;
cde75e59
RS
1193 }
1194
f6667c5e
JH
1195 if (placeholder[0] == 'G') {
1196 if (!c->commit_signature_parsed)
1197 parse_commit_signature(c);
1198 switch (placeholder[1]) {
1199 case 'G':
1200 if (c->signature.gpg_output)
1201 strbuf_addstr(sb, c->signature.gpg_output);
1202 break;
1203 case '?':
1204 switch (c->signature.good_bad) {
1205 case 'G':
1206 case 'B':
1207 strbuf_addch(sb, c->signature.good_bad);
1208 }
1209 break;
1210 case 'S':
1211 if (c->signature.signer)
1212 strbuf_addstr(sb, c->signature.signer);
1213 break;
0174eeaa
MG
1214 case 'K':
1215 if (c->signature.key)
1216 strbuf_addstr(sb, c->signature.key);
1217 break;
f6667c5e
JH
1218 }
1219 return 2;
1220 }
1221
1222
cde75e59 1223 /* For the rest we have to parse the commit header. */
f29d5958
RS
1224 if (!c->commit_header_parsed)
1225 parse_commit_header(c);
93fc05eb 1226
f29d5958 1227 switch (placeholder[0]) {
c3a670de
MC
1228 case 'a': /* author ... */
1229 return format_person_part(sb, placeholder[1],
d36f8679 1230 msg + c->author.off, c->author.len,
dd2e794a 1231 c->pretty_ctx->date_mode);
c3a670de
MC
1232 case 'c': /* committer ... */
1233 return format_person_part(sb, placeholder[1],
d36f8679 1234 msg + c->committer.off, c->committer.len,
dd2e794a 1235 c->pretty_ctx->date_mode);
c3a670de 1236 case 'e': /* encoding */
f29d5958 1237 strbuf_add(sb, msg + c->encoding.off, c->encoding.len);
c3a670de 1238 return 1;
1367b12a
EB
1239 case 'B': /* raw body */
1240 /* message_off is always left at the initial newline */
1241 strbuf_addstr(sb, msg + c->message_off + 1);
1242 return 1;
f53bd743
RS
1243 }
1244
1245 /* Now we need to parse the commit message. */
1246 if (!c->commit_message_parsed)
1247 parse_commit_message(c);
1248
1249 switch (placeholder[0]) {
1250 case 's': /* subject */
1251 format_subject(sb, msg + c->subject_off, " ");
1252 return 1;
46d164b0
SB
1253 case 'f': /* sanitized subject */
1254 format_sanitized_subject(sb, msg + c->subject_off);
1255 return 1;
c3a670de 1256 case 'b': /* body */
f29d5958 1257 strbuf_addstr(sb, msg + c->body_off);
c3a670de 1258 return 1;
93fc05eb 1259 }
c3a670de 1260 return 0; /* unknown placeholder */
cde75e59
RS
1261}
1262
9fa708da
JH
1263static size_t format_commit_item(struct strbuf *sb, const char *placeholder,
1264 void *context)
1265{
1266 int consumed;
1267 size_t orig_len;
1268 enum {
1269 NO_MAGIC,
1270 ADD_LF_BEFORE_NON_EMPTY,
1271 DEL_LF_BEFORE_EMPTY,
223a923c 1272 ADD_SP_BEFORE_NON_EMPTY
9fa708da
JH
1273 } magic = NO_MAGIC;
1274
1275 switch (placeholder[0]) {
1276 case '-':
1277 magic = DEL_LF_BEFORE_EMPTY;
1278 break;
1279 case '+':
1280 magic = ADD_LF_BEFORE_NON_EMPTY;
1281 break;
7b88176e
MG
1282 case ' ':
1283 magic = ADD_SP_BEFORE_NON_EMPTY;
1284 break;
9fa708da
JH
1285 default:
1286 break;
1287 }
1288 if (magic != NO_MAGIC)
1289 placeholder++;
1290
1291 orig_len = sb->len;
1292 consumed = format_commit_one(sb, placeholder, context);
1293 if (magic == NO_MAGIC)
1294 return consumed;
1295
1296 if ((orig_len == sb->len) && magic == DEL_LF_BEFORE_EMPTY) {
1297 while (sb->len && sb->buf[sb->len - 1] == '\n')
1298 strbuf_setlen(sb, sb->len - 1);
7b88176e
MG
1299 } else if (orig_len != sb->len) {
1300 if (magic == ADD_LF_BEFORE_NON_EMPTY)
1301 strbuf_insert(sb, orig_len, "\n", 1);
1302 else if (magic == ADD_SP_BEFORE_NON_EMPTY)
1303 strbuf_insert(sb, orig_len, " ", 1);
9fa708da
JH
1304 }
1305 return consumed + 1;
1306}
1307
5b163603
JG
1308static size_t userformat_want_item(struct strbuf *sb, const char *placeholder,
1309 void *context)
1310{
1311 struct userformat_want *w = context;
1312
7b88176e 1313 if (*placeholder == '+' || *placeholder == '-' || *placeholder == ' ')
5b163603
JG
1314 placeholder++;
1315
1316 switch (*placeholder) {
1317 case 'N':
1318 w->notes = 1;
1319 break;
1320 }
1321 return 0;
1322}
1323
1324void userformat_find_requirements(const char *fmt, struct userformat_want *w)
1325{
1326 struct strbuf dummy = STRBUF_INIT;
1327
1328 if (!fmt) {
1329 if (!user_format)
1330 return;
1331 fmt = user_format;
1332 }
a6253d10 1333 strbuf_expand(&dummy, fmt, userformat_want_item, w);
5b163603
JG
1334 strbuf_release(&dummy);
1335}
1336
cde75e59 1337void format_commit_message(const struct commit *commit,
7f98ebc8 1338 const char *format, struct strbuf *sb,
dd2e794a 1339 const struct pretty_print_context *pretty_ctx)
cde75e59 1340{
f29d5958 1341 struct format_commit_context context;
177b29dc 1342 const char *output_enc = pretty_ctx->output_encoding;
f29d5958
RS
1343
1344 memset(&context, 0, sizeof(context));
1345 context.commit = commit;
dd2e794a 1346 context.pretty_ctx = pretty_ctx;
02edd56b 1347 context.wrap_start = sb->len;
ec008076 1348 context.message = logmsg_reencode(commit, output_enc);
177b29dc 1349
c3a670de 1350 strbuf_expand(sb, format, format_commit_item, &context);
02edd56b 1351 rewrap_message_tail(sb, &context, 0, 0, 0);
177b29dc 1352
dd0d388c 1353 logmsg_free(context.message, commit);
f6667c5e
JH
1354 free(context.signature.gpg_output);
1355 free(context.signature.signer);
93fc05eb
JS
1356}
1357
6bf13944 1358static void pp_header(const struct pretty_print_context *pp,
93fc05eb
JS
1359 const char *encoding,
1360 const struct commit *commit,
1361 const char **msg_p,
1362 struct strbuf *sb)
1363{
1364 int parents_shown = 0;
1365
1366 for (;;) {
1367 const char *line = *msg_p;
1368 int linelen = get_one_line(*msg_p);
1369
1370 if (!linelen)
1371 return;
1372 *msg_p += linelen;
1373
1374 if (linelen == 1)
1375 /* End of header */
1376 return;
1377
6bf13944 1378 if (pp->fmt == CMIT_FMT_RAW) {
93fc05eb
JS
1379 strbuf_add(sb, line, linelen);
1380 continue;
1381 }
1382
8a692d27 1383 if (!prefixcmp(line, "parent ")) {
93fc05eb
JS
1384 if (linelen != 48)
1385 die("bad parent line in commit");
1386 continue;
1387 }
1388
1389 if (!parents_shown) {
1390 struct commit_list *parent;
1391 int num;
1392 for (parent = commit->parents, num = 0;
1393 parent;
1394 parent = parent->next, num++)
1395 ;
1396 /* with enough slop */
1397 strbuf_grow(sb, num * 50 + 20);
6bf13944 1398 add_merge_info(pp, sb, commit);
93fc05eb
JS
1399 parents_shown = 1;
1400 }
1401
1402 /*
1403 * MEDIUM == DEFAULT shows only author with dates.
1404 * FULL shows both authors but not dates.
1405 * FULLER shows both authors and dates.
1406 */
8a692d27 1407 if (!prefixcmp(line, "author ")) {
93fc05eb 1408 strbuf_grow(sb, linelen + 80);
6bf13944 1409 pp_user_info(pp, "Author", sb, line + 7, encoding);
93fc05eb 1410 }
8a692d27 1411 if (!prefixcmp(line, "committer ") &&
6bf13944 1412 (pp->fmt == CMIT_FMT_FULL || pp->fmt == CMIT_FMT_FULLER)) {
93fc05eb 1413 strbuf_grow(sb, linelen + 80);
6bf13944 1414 pp_user_info(pp, "Commit", sb, line + 10, encoding);
93fc05eb
JS
1415 }
1416 }
1417}
1418
6bf13944 1419void pp_title_line(const struct pretty_print_context *pp,
b02bd65f
DB
1420 const char **msg_p,
1421 struct strbuf *sb,
b02bd65f 1422 const char *encoding,
267123b4 1423 int need_8bit_cte)
93fc05eb 1424{
41dd00ba 1425 static const int max_length = 78; /* per rfc2047 */
93fc05eb
JS
1426 struct strbuf title;
1427
1428 strbuf_init(&title, 80);
9553d2b2
JK
1429 *msg_p = format_subject(&title, *msg_p,
1430 pp->preserve_subject ? "\n" : " ");
93fc05eb
JS
1431
1432 strbuf_grow(sb, title.len + 1024);
6bf13944
JK
1433 if (pp->subject) {
1434 strbuf_addstr(sb, pp->subject);
41dd00ba
JS
1435 if (needs_rfc2047_encoding(title.buf, title.len, RFC2047_SUBJECT))
1436 add_rfc2047(sb, title.buf, title.len,
1437 encoding, RFC2047_SUBJECT);
1438 else
1439 strbuf_add_wrapped_bytes(sb, title.buf, title.len,
1440 -last_line_length(sb), 1, max_length);
93fc05eb
JS
1441 } else {
1442 strbuf_addbuf(sb, &title);
1443 }
1444 strbuf_addch(sb, '\n');
1445
6bf4f1b4 1446 if (need_8bit_cte > 0) {
93fc05eb
JS
1447 const char *header_fmt =
1448 "MIME-Version: 1.0\n"
1449 "Content-Type: text/plain; charset=%s\n"
1450 "Content-Transfer-Encoding: 8bit\n";
1451 strbuf_addf(sb, header_fmt, encoding);
1452 }
6bf13944
JK
1453 if (pp->after_subject) {
1454 strbuf_addstr(sb, pp->after_subject);
93fc05eb 1455 }
6bf13944 1456 if (pp->fmt == CMIT_FMT_EMAIL) {
93fc05eb
JS
1457 strbuf_addch(sb, '\n');
1458 }
1459 strbuf_release(&title);
1460}
1461
6bf13944 1462void pp_remainder(const struct pretty_print_context *pp,
b02bd65f
DB
1463 const char **msg_p,
1464 struct strbuf *sb,
1465 int indent)
93fc05eb
JS
1466{
1467 int first = 1;
1468 for (;;) {
1469 const char *line = *msg_p;
1470 int linelen = get_one_line(line);
1471 *msg_p += linelen;
1472
1473 if (!linelen)
1474 break;
1475
1476 if (is_empty_line(line, &linelen)) {
1477 if (first)
1478 continue;
6bf13944 1479 if (pp->fmt == CMIT_FMT_SHORT)
93fc05eb
JS
1480 break;
1481 }
1482 first = 0;
1483
1484 strbuf_grow(sb, linelen + indent + 20);
1485 if (indent) {
1486 memset(sb->buf + sb->len, ' ', indent);
1487 strbuf_setlen(sb, sb->len + indent);
1488 }
1489 strbuf_add(sb, line, linelen);
1490 strbuf_addch(sb, '\n');
1491 }
1492}
1493
6bf13944
JK
1494void pretty_print_commit(const struct pretty_print_context *pp,
1495 const struct commit *commit,
1496 struct strbuf *sb)
93fc05eb
JS
1497{
1498 unsigned long beginning_of_body;
1499 int indent = 4;
dd0d388c 1500 const char *msg;
93fc05eb
JS
1501 char *reencoded;
1502 const char *encoding;
6bf13944 1503 int need_8bit_cte = pp->need_8bit_cte;
93fc05eb 1504
6bf13944
JK
1505 if (pp->fmt == CMIT_FMT_USERFORMAT) {
1506 format_commit_message(commit, user_format, sb, pp);
93fc05eb
JS
1507 return;
1508 }
1509
e297cf5a 1510 encoding = get_log_output_encoding();
dd0d388c 1511 msg = reencoded = logmsg_reencode(commit, encoding);
93fc05eb 1512
6bf13944 1513 if (pp->fmt == CMIT_FMT_ONELINE || pp->fmt == CMIT_FMT_EMAIL)
93fc05eb
JS
1514 indent = 0;
1515
6bf4f1b4
JH
1516 /*
1517 * We need to check and emit Content-type: to mark it
1518 * as 8-bit if we haven't done so.
93fc05eb 1519 */
6bf13944 1520 if (pp->fmt == CMIT_FMT_EMAIL && need_8bit_cte == 0) {
93fc05eb
JS
1521 int i, ch, in_body;
1522
1523 for (in_body = i = 0; (ch = msg[i]); i++) {
1524 if (!in_body) {
1525 /* author could be non 7-bit ASCII but
1526 * the log may be so; skip over the
1527 * header part first.
1528 */
1529 if (ch == '\n' && msg[i+1] == '\n')
1530 in_body = 1;
1531 }
1532 else if (non_ascii(ch)) {
6bf4f1b4 1533 need_8bit_cte = 1;
93fc05eb
JS
1534 break;
1535 }
1536 }
1537 }
1538
6bf13944
JK
1539 pp_header(pp, encoding, commit, &msg, sb);
1540 if (pp->fmt != CMIT_FMT_ONELINE && !pp->subject) {
93fc05eb
JS
1541 strbuf_addch(sb, '\n');
1542 }
1543
1544 /* Skip excess blank lines at the beginning of body, if any... */
a0109668 1545 msg = skip_empty_lines(msg);
93fc05eb
JS
1546
1547 /* These formats treat the title line specially. */
6bf13944
JK
1548 if (pp->fmt == CMIT_FMT_ONELINE || pp->fmt == CMIT_FMT_EMAIL)
1549 pp_title_line(pp, &msg, sb, encoding, need_8bit_cte);
93fc05eb
JS
1550
1551 beginning_of_body = sb->len;
6bf13944
JK
1552 if (pp->fmt != CMIT_FMT_ONELINE)
1553 pp_remainder(pp, &msg, sb, indent);
93fc05eb
JS
1554 strbuf_rtrim(sb);
1555
1556 /* Make sure there is an EOLN for the non-oneline case */
6bf13944 1557 if (pp->fmt != CMIT_FMT_ONELINE)
93fc05eb
JS
1558 strbuf_addch(sb, '\n');
1559
1560 /*
1561 * The caller may append additional body text in e-mail
1562 * format. Make sure we did not strip the blank line
1563 * between the header and the body.
1564 */
6bf13944 1565 if (pp->fmt == CMIT_FMT_EMAIL && sb->len <= beginning_of_body)
93fc05eb 1566 strbuf_addch(sb, '\n');
a97a7468 1567
dd0d388c 1568 logmsg_free(reencoded, commit);
93fc05eb 1569}
8b8a5374
JK
1570
1571void pp_commit_easy(enum cmit_fmt fmt, const struct commit *commit,
1572 struct strbuf *sb)
1573{
1574 struct pretty_print_context pp = {0};
6bf13944
JK
1575 pp.fmt = fmt;
1576 pretty_print_commit(&pp, commit, sb);
8b8a5374 1577}