?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
refs API: make files_copy_or_rename_ref() et al not set errno
[git/debian.git] / pretty.c
blob73b5ead5099d3efdb3e1999865ab868da5053ea6
1 #include "cache.h"
2 #include "config.h"
3 #include "commit.h"
4 #include "utf8.h"
5 #include "diff.h"
6 #include "revision.h"
7 #include "string-list.h"
8 #include "mailmap.h"
9 #include "log-tree.h"
10 #include "notes.h"
11 #include "color.h"
12 #include "reflog-walk.h"
13 #include "gpg-interface.h"
14 #include "trailer.h"
15 #include "run-command.h"
16
17 static char *user_format;
18 static struct cmt_fmt_map {
19 const char *name;
20 enum cmit_fmt format;
21 int is_tformat;
22 int expand_tabs_in_log;
23 int is_alias;
24 enum date_mode_type default_date_mode_type;
25 const char *user_format;
26 } *commit_formats;
27 static size_t builtin_formats_len;
28 static size_t commit_formats_len;
29 static size_t commit_formats_alloc;
30 static struct cmt_fmt_map *find_commit_format(const char *sought);
31
32 int commit_format_is_empty(enum cmit_fmt fmt)
33 {
34 return fmt == CMIT_FMT_USERFORMAT && !*user_format;
35 }
36
37 static void save_user_format(struct rev_info *rev, const char *cp, int is_tformat)
38 {
39 free(user_format);
40 user_format = xstrdup(cp);
41 if (is_tformat)
42 rev->use_terminator = 1;
43 rev->commit_format = CMIT_FMT_USERFORMAT;
44 }
45
46 static int git_pretty_formats_config(const char *var, const char *value, void *cb)
47 {
48 struct cmt_fmt_map *commit_format = NULL;
49 const char *name;
50 const char *fmt;
51 int i;
52
53 if (!skip_prefix(var, "pretty.", &name))
54 return 0;
55
56 for (i = 0; i < builtin_formats_len; i++) {
57 if (!strcmp(commit_formats[i].name, name))
58 return 0;
59 }
60
61 for (i = builtin_formats_len; i < commit_formats_len; i++) {
62 if (!strcmp(commit_formats[i].name, name)) {
63 commit_format = &commit_formats[i];
64 break;
65 }
66 }
67
68 if (!commit_format) {
69 ALLOC_GROW(commit_formats, commit_formats_len+1,
70 commit_formats_alloc);
71 commit_format = &commit_formats[commit_formats_len];
72 memset(commit_format, 0, sizeof(*commit_format));
73 commit_formats_len++;
74 }
75
76 commit_format->name = xstrdup(name);
77 commit_format->format = CMIT_FMT_USERFORMAT;
78 if (git_config_string(&fmt, var, value))
79 return -1;
80
81 if (skip_prefix(fmt, "format:", &fmt))
82 commit_format->is_tformat = 0;
83 else if (skip_prefix(fmt, "tformat:", &fmt) || strchr(fmt, '%'))
84 commit_format->is_tformat = 1;
85 else
86 commit_format->is_alias = 1;
87 commit_format->user_format = fmt;
88
89 return 0;
90 }
91
92 static void setup_commit_formats(void)
93 {
94 struct cmt_fmt_map builtin_formats[] = {
95 { "raw", CMIT_FMT_RAW, 0, 0 },
96 { "medium", CMIT_FMT_MEDIUM, 0, 8 },
97 { "short", CMIT_FMT_SHORT, 0, 0 },
98 { "email", CMIT_FMT_EMAIL, 0, 0 },
99 { "mboxrd", CMIT_FMT_MBOXRD, 0, 0 },
100 { "fuller", CMIT_FMT_FULLER, 0, 8 },
101 { "full", CMIT_FMT_FULL, 0, 8 },
102 { "oneline", CMIT_FMT_ONELINE, 1, 0 },
103 { "reference", CMIT_FMT_USERFORMAT, 1, 0,
104 0, DATE_SHORT, "%C(auto)%h (%s, %ad)" },
105 /*
106 * Please update $__git_log_pretty_formats in
107 * git-completion.bash when you add new formats.
108 */
109 };
110 commit_formats_len = ARRAY_SIZE(builtin_formats);
111 builtin_formats_len = commit_formats_len;
112 ALLOC_GROW(commit_formats, commit_formats_len, commit_formats_alloc);
113 COPY_ARRAY(commit_formats, builtin_formats,
114 ARRAY_SIZE(builtin_formats));
115
116 git_config(git_pretty_formats_config, NULL);
117 }
118
119 static struct cmt_fmt_map *find_commit_format_recursive(const char *sought,
120 const char *original,
121 int num_redirections)
122 {
123 struct cmt_fmt_map *found = NULL;
124 size_t found_match_len = 0;
125 int i;
126
127 if (num_redirections >= commit_formats_len)
128 die("invalid --pretty format: "
129 "'%s' references an alias which points to itself",
130 original);
131
132 for (i = 0; i < commit_formats_len; i++) {
133 size_t match_len;
134
135 if (!starts_with(commit_formats[i].name, sought))
136 continue;
137
138 match_len = strlen(commit_formats[i].name);
139 if (found == NULL || found_match_len > match_len) {
140 found = &commit_formats[i];
141 found_match_len = match_len;
142 }
143 }
144
145 if (found && found->is_alias) {
146 found = find_commit_format_recursive(found->user_format,
147 original,
148 num_redirections+1);
149 }
150
151 return found;
152 }
153
154 static struct cmt_fmt_map *find_commit_format(const char *sought)
155 {
156 if (!commit_formats)
157 setup_commit_formats();
158
159 return find_commit_format_recursive(sought, sought, 0);
160 }
161
162 void get_commit_format(const char *arg, struct rev_info *rev)
163 {
164 struct cmt_fmt_map *commit_format;
165
166 rev->use_terminator = 0;
167 if (!arg) {
168 rev->commit_format = CMIT_FMT_DEFAULT;
169 return;
170 }
171 if (skip_prefix(arg, "format:", &arg)) {
172 save_user_format(rev, arg, 0);
173 return;
174 }
175
176 if (!*arg || skip_prefix(arg, "tformat:", &arg) || strchr(arg, '%')) {
177 save_user_format(rev, arg, 1);
178 return;
179 }
180
181 commit_format = find_commit_format(arg);
182 if (!commit_format)
183 die("invalid --pretty format: %s", arg);
184
185 rev->commit_format = commit_format->format;
186 rev->use_terminator = commit_format->is_tformat;
187 rev->expand_tabs_in_log_default = commit_format->expand_tabs_in_log;
188 if (!rev->date_mode_explicit && commit_format->default_date_mode_type)
189 rev->date_mode.type = commit_format->default_date_mode_type;
190 if (commit_format->format == CMIT_FMT_USERFORMAT) {
191 save_user_format(rev, commit_format->user_format,
192 commit_format->is_tformat);
193 }
194 }
195
196 /*
197 * Generic support for pretty-printing the header
198 */
199 static int get_one_line(const char *msg)
200 {
201 int ret = 0;
202
203 for (;;) {
204 char c = *msg++;
205 if (!c)
206 break;
207 ret++;
208 if (c == '\n')
209 break;
210 }
211 return ret;
212 }
213
214 /* High bit set, or ISO-2022-INT */
215 static int non_ascii(int ch)
216 {
217 return !isascii(ch) || ch == '\033';
218 }
219
220 int has_non_ascii(const char *s)
221 {
222 int ch;
223 if (!s)
224 return 0;
225 while ((ch = *s++) != '\0') {
226 if (non_ascii(ch))
227 return 1;
228 }
229 return 0;
230 }
231
232 static int is_rfc822_special(char ch)
233 {
234 switch (ch) {
235 case '(':
236 case ')':
237 case '<':
238 case '>':
239 case '[':
240 case ']':
241 case ':':
242 case ';':
243 case '@':
244 case ',':
245 case '.':
246 case '"':
247 case '\\':
248 return 1;
249 default:
250 return 0;
251 }
252 }
253
254 static int needs_rfc822_quoting(const char *s, int len)
255 {
256 int i;
257 for (i = 0; i < len; i++)
258 if (is_rfc822_special(s[i]))
259 return 1;
260 return 0;
261 }
262
263 static int last_line_length(struct strbuf *sb)
264 {
265 int i;
266
267 /* How many bytes are already used on the last line? */
268 for (i = sb->len - 1; i >= 0; i--)
269 if (sb->buf[i] == '\n')
270 break;
271 return sb->len - (i + 1);
272 }
273
274 static void add_rfc822_quoted(struct strbuf *out, const char *s, int len)
275 {
276 int i;
277
278 /* just a guess, we may have to also backslash-quote */
279 strbuf_grow(out, len + 2);
280
281 strbuf_addch(out, '"');
282 for (i = 0; i < len; i++) {
283 switch (s[i]) {
284 case '"':
285 case '\\':
286 strbuf_addch(out, '\\');
287 /* fall through */
288 default:
289 strbuf_addch(out, s[i]);
290 }
291 }
292 strbuf_addch(out, '"');
293 }
294
295 enum rfc2047_type {
296 RFC2047_SUBJECT,
297 RFC2047_ADDRESS
298 };
299
300 static int is_rfc2047_special(char ch, enum rfc2047_type type)
301 {
302 /*
303 * rfc2047, section 4.2:
304 *
305 * 8-bit values which correspond to printable ASCII characters other
306 * than "=", "?", and "_" (underscore), MAY be represented as those
307 * characters. (But see section 5 for restrictions.) In
308 * particular, SPACE and TAB MUST NOT be represented as themselves
309 * within encoded words.
310 */
311
312 /*
313 * rule out non-ASCII characters and non-printable characters (the
314 * non-ASCII check should be redundant as isprint() is not localized
315 * and only knows about ASCII, but be defensive about that)
316 */
317 if (non_ascii(ch) || !isprint(ch))
318 return 1;
319
320 /*
321 * rule out special printable characters (' ' should be the only
322 * whitespace character considered printable, but be defensive and use
323 * isspace())
324 */
325 if (isspace(ch) || ch == '=' || ch == '?' || ch == '_')
326 return 1;
327
328 /*
329 * rfc2047, section 5.3:
330 *
331 * As a replacement for a 'word' entity within a 'phrase', for example,
332 * one that precedes an address in a From, To, or Cc header. The ABNF
333 * definition for 'phrase' from RFC 822 thus becomes:
334 *
335 * phrase = 1*( encoded-word / word )
336 *
337 * In this case the set of characters that may be used in a "Q"-encoded
338 * 'encoded-word' is restricted to: <upper and lower case ASCII
339 * letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
340 * (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
341 * 'phrase' MUST be separated from any adjacent 'word', 'text' or
342 * 'special' by 'linear-white-space'.
343 */
344
345 if (type != RFC2047_ADDRESS)
346 return 0;
347
348 /* '=' and '_' are special cases and have been checked above */
349 return !(isalnum(ch) || ch == '!' || ch == '*' || ch == '+' || ch == '-' || ch == '/');
350 }
351
352 static int needs_rfc2047_encoding(const char *line, int len)
353 {
354 int i;
355
356 for (i = 0; i < len; i++) {
357 int ch = line[i];
358 if (non_ascii(ch) || ch == '\n')
359 return 1;
360 if ((i + 1 < len) && (ch == '=' && line[i+1] == '?'))
361 return 1;
362 }
363
364 return 0;
365 }
366
367 static void add_rfc2047(struct strbuf *sb, const char *line, size_t len,
368 const char *encoding, enum rfc2047_type type)
369 {
370 static const int max_encoded_length = 76; /* per rfc2047 */
371 int i;
372 int line_len = last_line_length(sb);
373
374 strbuf_grow(sb, len * 3 + strlen(encoding) + 100);
375 strbuf_addf(sb, "=?%s?q?", encoding);
376 line_len += strlen(encoding) + 5; /* 5 for =??q? */
377
378 while (len) {
379 /*
380 * RFC 2047, section 5 (3):
381 *
382 * Each 'encoded-word' MUST represent an integral number of
383 * characters. A multi-octet character may not be split across
384 * adjacent 'encoded- word's.
385 */
386 const unsigned char *p = (const unsigned char *)line;
387 int chrlen = mbs_chrlen(&line, &len, encoding);
388 int is_special = (chrlen > 1) || is_rfc2047_special(*p, type);
389
390 /* "=%02X" * chrlen, or the byte itself */
391 const char *encoded_fmt = is_special ? "=%02X" : "%c";
392 int encoded_len = is_special ? 3 * chrlen : 1;
393
394 /*
395 * According to RFC 2047, we could encode the special character
396 * ' ' (space) with '_' (underscore) for readability. But many
397 * programs do not understand this and just leave the
398 * underscore in place. Thus, we do nothing special here, which
399 * causes ' ' to be encoded as '=20', avoiding this problem.
400 */
401
402 if (line_len + encoded_len + 2 > max_encoded_length) {
403 /* It won't fit with trailing "?=" --- break the line */
404 strbuf_addf(sb, "?=\n =?%s?q?", encoding);
405 line_len = strlen(encoding) + 5 + 1; /* =??q? plus SP */
406 }
407
408 for (i = 0; i < chrlen; i++)
409 strbuf_addf(sb, encoded_fmt, p[i]);
410 line_len += encoded_len;
411 }
412 strbuf_addstr(sb, "?=");
413 }
414
415 const char *show_ident_date(const struct ident_split *ident,
416 const struct date_mode *mode)
417 {
418 timestamp_t date = 0;
419 long tz = 0;
420
421 if (ident->date_begin && ident->date_end)
422 date = parse_timestamp(ident->date_begin, NULL, 10);
423 if (date_overflows(date))
424 date = 0;
425 else {
426 if (ident->tz_begin && ident->tz_end)
427 tz = strtol(ident->tz_begin, NULL, 10);
428 if (tz >= INT_MAX || tz <= INT_MIN)
429 tz = 0;
430 }
431 return show_date(date, tz, mode);
432 }
433
434 void pp_user_info(struct pretty_print_context *pp,
435 const char *what, struct strbuf *sb,
436 const char *line, const char *encoding)
437 {
438 struct ident_split ident;
439 char *line_end;
440 const char *mailbuf, *namebuf;
441 size_t namelen, maillen;
442 int max_length = 78; /* per rfc2822 */
443
444 if (pp->fmt == CMIT_FMT_ONELINE)
445 return;
446
447 line_end = strchrnul(line, '\n');
448 if (split_ident_line(&ident, line, line_end - line))
449 return;
450
451 mailbuf = ident.mail_begin;
452 maillen = ident.mail_end - ident.mail_begin;
453 namebuf = ident.name_begin;
454 namelen = ident.name_end - ident.name_begin;
455
456 if (pp->mailmap)
457 map_user(pp->mailmap, &mailbuf, &maillen, &namebuf, &namelen);
458
459 if (cmit_fmt_is_mail(pp->fmt)) {
460 if (pp->from_ident && ident_cmp(pp->from_ident, &ident)) {
461 struct strbuf buf = STRBUF_INIT;
462
463 strbuf_addstr(&buf, "From: ");
464 strbuf_add(&buf, namebuf, namelen);
465 strbuf_addstr(&buf, " <");
466 strbuf_add(&buf, mailbuf, maillen);
467 strbuf_addstr(&buf, ">\n");
468 string_list_append(&pp->in_body_headers,
469 strbuf_detach(&buf, NULL));
470
471 mailbuf = pp->from_ident->mail_begin;
472 maillen = pp->from_ident->mail_end - mailbuf;
473 namebuf = pp->from_ident->name_begin;
474 namelen = pp->from_ident->name_end - namebuf;
475 }
476
477 strbuf_addstr(sb, "From: ");
478 if (pp->encode_email_headers &&
479 needs_rfc2047_encoding(namebuf, namelen)) {
480 add_rfc2047(sb, namebuf, namelen,
481 encoding, RFC2047_ADDRESS);
482 max_length = 76; /* per rfc2047 */
483 } else if (needs_rfc822_quoting(namebuf, namelen)) {
484 struct strbuf quoted = STRBUF_INIT;
485 add_rfc822_quoted(&quoted, namebuf, namelen);
486 strbuf_add_wrapped_bytes(sb, quoted.buf, quoted.len,
487 -6, 1, max_length);
488 strbuf_release(&quoted);
489 } else {
490 strbuf_add_wrapped_bytes(sb, namebuf, namelen,
491 -6, 1, max_length);
492 }
493
494 if (max_length <
495 last_line_length(sb) + strlen(" <") + maillen + strlen(">"))
496 strbuf_addch(sb, '\n');
497 strbuf_addf(sb, " <%.*s>\n", (int)maillen, mailbuf);
498 } else {
499 strbuf_addf(sb, "%s: %.*s%.*s <%.*s>\n", what,
500 (pp->fmt == CMIT_FMT_FULLER) ? 4 : 0, " ",
501 (int)namelen, namebuf, (int)maillen, mailbuf);
502 }
503
504 switch (pp->fmt) {
505 case CMIT_FMT_MEDIUM:
506 strbuf_addf(sb, "Date: %s\n",
507 show_ident_date(&ident, &pp->date_mode));
508 break;
509 case CMIT_FMT_EMAIL:
510 case CMIT_FMT_MBOXRD:
511 strbuf_addf(sb, "Date: %s\n",
512 show_ident_date(&ident, DATE_MODE(RFC2822)));
513 break;
514 case CMIT_FMT_FULLER:
515 strbuf_addf(sb, "%sDate: %s\n", what,
516 show_ident_date(&ident, &pp->date_mode));
517 break;
518 default:
519 /* notin' */
520 break;
521 }
522 }
523
524 static int is_blank_line(const char *line, int *len_p)
525 {
526 int len = *len_p;
527 while (len && isspace(line[len - 1]))
528 len--;
529 *len_p = len;
530 return !len;
531 }
532
533 const char *skip_blank_lines(const char *msg)
534 {
535 for (;;) {
536 int linelen = get_one_line(msg);
537 int ll = linelen;
538 if (!linelen)
539 break;
540 if (!is_blank_line(msg, &ll))
541 break;
542 msg += linelen;
543 }
544 return msg;
545 }
546
547 static void add_merge_info(const struct pretty_print_context *pp,
548 struct strbuf *sb, const struct commit *commit)
549 {
550 struct commit_list *parent = commit->parents;
551
552 if ((pp->fmt == CMIT_FMT_ONELINE) || (cmit_fmt_is_mail(pp->fmt)) ||
553 !parent || !parent->next)
554 return;
555
556 strbuf_addstr(sb, "Merge:");
557
558 while (parent) {
559 struct object_id *oidp = &parent->item->object.oid;
560 strbuf_addch(sb, ' ');
561 if (pp->abbrev)
562 strbuf_add_unique_abbrev(sb, oidp, pp->abbrev);
563 else
564 strbuf_addstr(sb, oid_to_hex(oidp));
565 parent = parent->next;
566 }
567 strbuf_addch(sb, '\n');
568 }
569
570 static char *get_header(const char *msg, const char *key)
571 {
572 size_t len;
573 const char *v = find_commit_header(msg, key, &len);
574 return v ? xmemdupz(v, len) : NULL;
575 }
576
577 static char *replace_encoding_header(char *buf, const char *encoding)
578 {
579 struct strbuf tmp = STRBUF_INIT;
580 size_t start, len;
581 char *cp = buf;
582
583 /* guess if there is an encoding header before a \n\n */
584 while (!starts_with(cp, "encoding ")) {
585 cp = strchr(cp, '\n');
586 if (!cp || *++cp == '\n')
587 return buf;
588 }
589 start = cp - buf;
590 cp = strchr(cp, '\n');
591 if (!cp)
592 return buf; /* should not happen but be defensive */
593 len = cp + 1 - (buf + start);
594
595 strbuf_attach(&tmp, buf, strlen(buf), strlen(buf) + 1);
596 if (is_encoding_utf8(encoding)) {
597 /* we have re-coded to UTF-8; drop the header */
598 strbuf_remove(&tmp, start, len);
599 } else {
600 /* just replaces XXXX in 'encoding XXXX\n' */
601 strbuf_splice(&tmp, start + strlen("encoding "),
602 len - strlen("encoding \n"),
603 encoding, strlen(encoding));
604 }
605 return strbuf_detach(&tmp, NULL);
606 }
607
608 const char *repo_logmsg_reencode(struct repository *r,
609 const struct commit *commit,
610 char **commit_encoding,
611 const char *output_encoding)
612 {
613 static const char *utf8 = "UTF-8";
614 const char *use_encoding;
615 char *encoding;
616 const char *msg = repo_get_commit_buffer(r, commit, NULL);
617 char *out;
618
619 if (!output_encoding || !*output_encoding) {
620 if (commit_encoding)
621 *commit_encoding = get_header(msg, "encoding");
622 return msg;
623 }
624 encoding = get_header(msg, "encoding");
625 if (commit_encoding)
626 *commit_encoding = encoding;
627 use_encoding = encoding ? encoding : utf8;
628 if (same_encoding(use_encoding, output_encoding)) {
629 /*
630 * No encoding work to be done. If we have no encoding header
631 * at all, then there's nothing to do, and we can return the
632 * message verbatim (whether newly allocated or not).
633 */
634 if (!encoding)
635 return msg;
636
637 /*
638 * Otherwise, we still want to munge the encoding header in the
639 * result, which will be done by modifying the buffer. If we
640 * are using a fresh copy, we can reuse it. But if we are using
641 * the cached copy from get_commit_buffer, we need to duplicate it
642 * to avoid munging the cached copy.
643 */
644 if (msg == get_cached_commit_buffer(r, commit, NULL))
645 out = xstrdup(msg);
646 else
647 out = (char *)msg;
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)
658 repo_unuse_commit_buffer(r, commit, 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 */
665 if (out)
666 out = replace_encoding_header(out, output_encoding);
667
668 if (!commit_encoding)
669 free(encoding);
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 if (!out) {
675 warning("unable to reencode commit to '%s'", output_encoding);
676 return msg;
677 }
678 return out;
679 }
680
681 static int mailmap_name(const char **email, size_t *email_len,
682 const char **name, size_t *name_len)
683 {
684 static struct string_list *mail_map;
685 if (!mail_map) {
686 CALLOC_ARRAY(mail_map, 1);
687 read_mailmap(mail_map);
688 }
689 return mail_map->nr && map_user(mail_map, email, email_len, name, name_len);
690 }
691
692 static size_t format_person_part(struct strbuf *sb, char part,
693 const char *msg, int len,
694 const struct date_mode *dmode)
695 {
696 /* currently all placeholders have same length */
697 const int placeholder_len = 2;
698 struct ident_split s;
699 const char *name, *mail;
700 size_t maillen, namelen;
701
702 if (split_ident_line(&s, msg, len) < 0)
703 goto skip;
704
705 name = s.name_begin;
706 namelen = s.name_end - s.name_begin;
707 mail = s.mail_begin;
708 maillen = s.mail_end - s.mail_begin;
709
710 if (part == 'N' || part == 'E' || part == 'L') /* mailmap lookup */
711 mailmap_name(&mail, &maillen, &name, &namelen);
712 if (part == 'n' || part == 'N') { /* name */
713 strbuf_add(sb, name, namelen);
714 return placeholder_len;
715 }
716 if (part == 'e' || part == 'E') { /* email */
717 strbuf_add(sb, mail, maillen);
718 return placeholder_len;
719 }
720 if (part == 'l' || part == 'L') { /* local-part */
721 const char *at = memchr(mail, '@', maillen);
722 if (at)
723 maillen = at - mail;
724 strbuf_add(sb, mail, maillen);
725 return placeholder_len;
726 }
727
728 if (!s.date_begin)
729 goto skip;
730
731 if (part == 't') { /* date, UNIX timestamp */
732 strbuf_add(sb, s.date_begin, s.date_end - s.date_begin);
733 return placeholder_len;
734 }
735
736 switch (part) {
737 case 'd': /* date */
738 strbuf_addstr(sb, show_ident_date(&s, dmode));
739 return placeholder_len;
740 case 'D': /* date, RFC2822 style */
741 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(RFC2822)));
742 return placeholder_len;
743 case 'r': /* date, relative */
744 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(RELATIVE)));
745 return placeholder_len;
746 case 'i': /* date, ISO 8601-like */
747 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(ISO8601)));
748 return placeholder_len;
749 case 'I': /* date, ISO 8601 strict */
750 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(ISO8601_STRICT)));
751 return placeholder_len;
752 case 'h': /* date, human */
753 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(HUMAN)));
754 return placeholder_len;
755 case 's':
756 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(SHORT)));
757 return placeholder_len;
758 }
759
760 skip:
761 /*
762 * reading from either a bogus commit, or a reflog entry with
763 * %gn, %ge, etc.; 'sb' cannot be updated, but we still need
764 * to compute a valid return value.
765 */
766 if (part == 'n' || part == 'e' || part == 't' || part == 'd'
767 || part == 'D' || part == 'r' || part == 'i')
768 return placeholder_len;
769
770 return 0; /* unknown placeholder */
771 }
772
773 struct chunk {
774 size_t off;
775 size_t len;
776 };
777
778 enum flush_type {
779 no_flush,
780 flush_right,
781 flush_left,
782 flush_left_and_steal,
783 flush_both
784 };
785
786 enum trunc_type {
787 trunc_none,
788 trunc_left,
789 trunc_middle,
790 trunc_right
791 };
792
793 struct format_commit_context {
794 struct repository *repository;
795 const struct commit *commit;
796 const struct pretty_print_context *pretty_ctx;
797 unsigned commit_header_parsed:1;
798 unsigned commit_message_parsed:1;
799 struct signature_check signature_check;
800 enum flush_type flush_type;
801 enum trunc_type truncate;
802 const char *message;
803 char *commit_encoding;
804 size_t width, indent1, indent2;
805 int auto_color;
806 int padding;
807
808 /* These offsets are relative to the start of the commit message. */
809 struct chunk author;
810 struct chunk committer;
811 size_t message_off;
812 size_t subject_off;
813 size_t body_off;
814
815 /* The following ones are relative to the result struct strbuf. */
816 size_t wrap_start;
817 };
818
819 static void parse_commit_header(struct format_commit_context *context)
820 {
821 const char *msg = context->message;
822 int i;
823
824 for (i = 0; msg[i]; i++) {
825 const char *name;
826 int eol;
827 for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
828 ; /* do nothing */
829
830 if (i == eol) {
831 break;
832 } else if (skip_prefix(msg + i, "author ", &name)) {
833 context->author.off = name - msg;
834 context->author.len = msg + eol - name;
835 } else if (skip_prefix(msg + i, "committer ", &name)) {
836 context->committer.off = name - msg;
837 context->committer.len = msg + eol - name;
838 }
839 i = eol;
840 }
841 context->message_off = i;
842 context->commit_header_parsed = 1;
843 }
844
845 static int istitlechar(char c)
846 {
847 return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') ||
848 (c >= '0' && c <= '9') || c == '.' || c == '_';
849 }
850
851 void format_sanitized_subject(struct strbuf *sb, const char *msg, size_t len)
852 {
853 size_t trimlen;
854 size_t start_len = sb->len;
855 int space = 2;
856 int i;
857
858 for (i = 0; i < len; i++) {
859 if (istitlechar(msg[i])) {
860 if (space == 1)
861 strbuf_addch(sb, '-');
862 space = 0;
863 strbuf_addch(sb, msg[i]);
864 if (msg[i] == '.')
865 while (msg[i+1] == '.')
866 i++;
867 } else
868 space |= 1;
869 }
870
871 /* trim any trailing '.' or '-' characters */
872 trimlen = 0;
873 while (sb->len - trimlen > start_len &&
874 (sb->buf[sb->len - 1 - trimlen] == '.'
875 || sb->buf[sb->len - 1 - trimlen] == '-'))
876 trimlen++;
877 strbuf_remove(sb, sb->len - trimlen, trimlen);
878 }
879
880 const char *format_subject(struct strbuf *sb, const char *msg,
881 const char *line_separator)
882 {
883 int first = 1;
884
885 for (;;) {
886 const char *line = msg;
887 int linelen = get_one_line(line);
888
889 msg += linelen;
890 if (!linelen || is_blank_line(line, &linelen))
891 break;
892
893 if (!sb)
894 continue;
895 strbuf_grow(sb, linelen + 2);
896 if (!first)
897 strbuf_addstr(sb, line_separator);
898 strbuf_add(sb, line, linelen);
899 first = 0;
900 }
901 return msg;
902 }
903
904 static void parse_commit_message(struct format_commit_context *c)
905 {
906 const char *msg = c->message + c->message_off;
907 const char *start = c->message;
908
909 msg = skip_blank_lines(msg);
910 c->subject_off = msg - start;
911
912 msg = format_subject(NULL, msg, NULL);
913 msg = skip_blank_lines(msg);
914 c->body_off = msg - start;
915
916 c->commit_message_parsed = 1;
917 }
918
919 static void strbuf_wrap(struct strbuf *sb, size_t pos,
920 size_t width, size_t indent1, size_t indent2)
921 {
922 struct strbuf tmp = STRBUF_INIT;
923
924 if (pos)
925 strbuf_add(&tmp, sb->buf, pos);
926 strbuf_add_wrapped_text(&tmp, sb->buf + pos,
927 (int) indent1, (int) indent2, (int) width);
928 strbuf_swap(&tmp, sb);
929 strbuf_release(&tmp);
930 }
931
932 static void rewrap_message_tail(struct strbuf *sb,
933 struct format_commit_context *c,
934 size_t new_width, size_t new_indent1,
935 size_t new_indent2)
936 {
937 if (c->width == new_width && c->indent1 == new_indent1 &&
938 c->indent2 == new_indent2)
939 return;
940 if (c->wrap_start < sb->len)
941 strbuf_wrap(sb, c->wrap_start, c->width, c->indent1, c->indent2);
942 c->wrap_start = sb->len;
943 c->width = new_width;
944 c->indent1 = new_indent1;
945 c->indent2 = new_indent2;
946 }
947
948 static int format_reflog_person(struct strbuf *sb,
949 char part,
950 struct reflog_walk_info *log,
951 const struct date_mode *dmode)
952 {
953 const char *ident;
954
955 if (!log)
956 return 2;
957
958 ident = get_reflog_ident(log);
959 if (!ident)
960 return 2;
961
962 return format_person_part(sb, part, ident, strlen(ident), dmode);
963 }
964
965 static size_t parse_color(struct strbuf *sb, /* in UTF-8 */
966 const char *placeholder,
967 struct format_commit_context *c)
968 {
969 const char *rest = placeholder;
970 const char *basic_color = NULL;
971
972 if (placeholder[1] == '(') {
973 const char *begin = placeholder + 2;
974 const char *end = strchr(begin, ')');
975 char color[COLOR_MAXLEN];
976
977 if (!end)
978 return 0;
979
980 if (skip_prefix(begin, "auto,", &begin)) {
981 if (!want_color(c->pretty_ctx->color))
982 return end - placeholder + 1;
983 } else if (skip_prefix(begin, "always,", &begin)) {
984 /* nothing to do; we do not respect want_color at all */
985 } else {
986 /* the default is the same as "auto" */
987 if (!want_color(c->pretty_ctx->color))
988 return end - placeholder + 1;
989 }
990
991 if (color_parse_mem(begin, end - begin, color) < 0)
992 die(_("unable to parse --pretty format"));
993 strbuf_addstr(sb, color);
994 return end - placeholder + 1;
995 }
996
997 /*
998 * We handle things like "%C(red)" above; for historical reasons, there
999 * are a few colors that can be specified without parentheses (and
1000 * they cannot support things like "auto" or "always" at all).
1001 */
1002 if (skip_prefix(placeholder + 1, "red", &rest))
1003 basic_color = GIT_COLOR_RED;
1004 else if (skip_prefix(placeholder + 1, "green", &rest))
1005 basic_color = GIT_COLOR_GREEN;
1006 else if (skip_prefix(placeholder + 1, "blue", &rest))
1007 basic_color = GIT_COLOR_BLUE;
1008 else if (skip_prefix(placeholder + 1, "reset", &rest))
1009 basic_color = GIT_COLOR_RESET;
1010
1011 if (basic_color && want_color(c->pretty_ctx->color))
1012 strbuf_addstr(sb, basic_color);
1013
1014 return rest - placeholder;
1015 }
1016
1017 static size_t parse_padding_placeholder(const char *placeholder,
1018 struct format_commit_context *c)
1019 {
1020 const char *ch = placeholder;
1021 enum flush_type flush_type;
1022 int to_column = 0;
1023
1024 switch (*ch++) {
1025 case '<':
1026 flush_type = flush_right;
1027 break;
1028 case '>':
1029 if (*ch == '<') {
1030 flush_type = flush_both;
1031 ch++;
1032 } else if (*ch == '>') {
1033 flush_type = flush_left_and_steal;
1034 ch++;
1035 } else
1036 flush_type = flush_left;
1037 break;
1038 default:
1039 return 0;
1040 }
1041
1042 /* the next value means "wide enough to that column" */
1043 if (*ch == '|') {
1044 to_column = 1;
1045 ch++;
1046 }
1047
1048 if (*ch == '(') {
1049 const char *start = ch + 1;
1050 const char *end = start + strcspn(start, ",)");
1051 char *next;
1052 int width;
1053 if (!end || end == start)
1054 return 0;
1055 width = strtol(start, &next, 10);
1056 if (next == start || width == 0)
1057 return 0;
1058 if (width < 0) {
1059 if (to_column)
1060 width += term_columns();
1061 if (width < 0)
1062 return 0;
1063 }
1064 c->padding = to_column ? -width : width;
1065 c->flush_type = flush_type;
1066
1067 if (*end == ',') {
1068 start = end + 1;
1069 end = strchr(start, ')');
1070 if (!end || end == start)
1071 return 0;
1072 if (starts_with(start, "trunc)"))
1073 c->truncate = trunc_right;
1074 else if (starts_with(start, "ltrunc)"))
1075 c->truncate = trunc_left;
1076 else if (starts_with(start, "mtrunc)"))
1077 c->truncate = trunc_middle;
1078 else
1079 return 0;
1080 } else
1081 c->truncate = trunc_none;
1082
1083 return end - placeholder + 1;
1084 }
1085 return 0;
1086 }
1087
1088 static int match_placeholder_arg_value(const char *to_parse, const char *candidate,
1089 const char **end, const char **valuestart,
1090 size_t *valuelen)
1091 {
1092 const char *p;
1093
1094 if (!(skip_prefix(to_parse, candidate, &p)))
1095 return 0;
1096 if (valuestart) {
1097 if (*p == '=') {
1098 *valuestart = p + 1;
1099 *valuelen = strcspn(*valuestart, ",)");
1100 p = *valuestart + *valuelen;
1101 } else {
1102 if (*p != ',' && *p != ')')
1103 return 0;
1104 *valuestart = NULL;
1105 *valuelen = 0;
1106 }
1107 }
1108 if (*p == ',') {
1109 *end = p + 1;
1110 return 1;
1111 }
1112 if (*p == ')') {
1113 *end = p;
1114 return 1;
1115 }
1116 return 0;
1117 }
1118
1119 static int match_placeholder_bool_arg(const char *to_parse, const char *candidate,
1120 const char **end, int *val)
1121 {
1122 const char *argval;
1123 char *strval;
1124 size_t arglen;
1125 int v;
1126
1127 if (!match_placeholder_arg_value(to_parse, candidate, end, &argval, &arglen))
1128 return 0;
1129
1130 if (!argval) {
1131 *val = 1;
1132 return 1;
1133 }
1134
1135 strval = xstrndup(argval, arglen);
1136 v = git_parse_maybe_bool(strval);
1137 free(strval);
1138
1139 if (v == -1)
1140 return 0;
1141
1142 *val = v;
1143
1144 return 1;
1145 }
1146
1147 static int format_trailer_match_cb(const struct strbuf *key, void *ud)
1148 {
1149 const struct string_list *list = ud;
1150 const struct string_list_item *item;
1151
1152 for_each_string_list_item (item, list) {
1153 if (key->len == (uintptr_t)item->util &&
1154 !strncasecmp(item->string, key->buf, key->len))
1155 return 1;
1156 }
1157 return 0;
1158 }
1159
1160 int format_set_trailers_options(struct process_trailer_options *opts,
1161 struct string_list *filter_list,
1162 struct strbuf *sepbuf,
1163 struct strbuf *kvsepbuf,
1164 const char **arg,
1165 char **invalid_arg)
1166 {
1167 for (;;) {
1168 const char *argval;
1169 size_t arglen;
1170
1171 if (**arg == ')')
1172 break;
1173
1174 if (match_placeholder_arg_value(*arg, "key", arg, &argval, &arglen)) {
1175 uintptr_t len = arglen;
1176
1177 if (!argval)
1178 return -1;
1179
1180 if (len && argval[len - 1] == ':')
1181 len--;
1182 string_list_append(filter_list, argval)->util = (char *)len;
1183
1184 opts->filter = format_trailer_match_cb;
1185 opts->filter_data = filter_list;
1186 opts->only_trailers = 1;
1187 } else if (match_placeholder_arg_value(*arg, "separator", arg, &argval, &arglen)) {
1188 char *fmt;
1189
1190 strbuf_reset(sepbuf);
1191 fmt = xstrndup(argval, arglen);
1192 strbuf_expand(sepbuf, fmt, strbuf_expand_literal_cb, NULL);
1193 free(fmt);
1194 opts->separator = sepbuf;
1195 } else if (match_placeholder_arg_value(*arg, "key_value_separator", arg, &argval, &arglen)) {
1196 char *fmt;
1197
1198 strbuf_reset(kvsepbuf);
1199 fmt = xstrndup(argval, arglen);
1200 strbuf_expand(kvsepbuf, fmt, strbuf_expand_literal_cb, NULL);
1201 free(fmt);
1202 opts->key_value_separator = kvsepbuf;
1203 } else if (!match_placeholder_bool_arg(*arg, "only", arg, &opts->only_trailers) &&
1204 !match_placeholder_bool_arg(*arg, "unfold", arg, &opts->unfold) &&
1205 !match_placeholder_bool_arg(*arg, "keyonly", arg, &opts->key_only) &&
1206 !match_placeholder_bool_arg(*arg, "valueonly", arg, &opts->value_only)) {
1207 if (invalid_arg) {
1208 size_t len = strcspn(*arg, ",)");
1209 *invalid_arg = xstrndup(*arg, len);
1210 }
1211 return -1;
1212 }
1213 }
1214 return 0;
1215 }
1216
1217 static size_t parse_describe_args(const char *start, struct strvec *args)
1218 {
1219 const char *options[] = { "match", "exclude" };
1220 const char *arg = start;
1221
1222 for (;;) {
1223 const char *matched = NULL;
1224 const char *argval;
1225 size_t arglen = 0;
1226 int i;
1227
1228 for (i = 0; i < ARRAY_SIZE(options); i++) {
1229 if (match_placeholder_arg_value(arg, options[i], &arg,
1230 &argval, &arglen)) {
1231 matched = options[i];
1232 break;
1233 }
1234 }
1235 if (!matched)
1236 break;
1237
1238 if (!arglen)
1239 return 0;
1240 strvec_pushf(args, "--%s=%.*s", matched, (int)arglen, argval);
1241 }
1242 return arg - start;
1243 }
1244
1245 static size_t format_commit_one(struct strbuf *sb, /* in UTF-8 */
1246 const char *placeholder,
1247 void *context)
1248 {
1249 struct format_commit_context *c = context;
1250 const struct commit *commit = c->commit;
1251 const char *msg = c->message;
1252 struct commit_list *p;
1253 const char *arg, *eol;
1254 size_t res;
1255 char **slot;
1256
1257 /* these are independent of the commit */
1258 res = strbuf_expand_literal_cb(sb, placeholder, NULL);
1259 if (res)
1260 return res;
1261
1262 switch (placeholder[0]) {
1263 case 'C':
1264 if (starts_with(placeholder + 1, "(auto)")) {
1265 c->auto_color = want_color(c->pretty_ctx->color);
1266 if (c->auto_color && sb->len)
1267 strbuf_addstr(sb, GIT_COLOR_RESET);
1268 return 7; /* consumed 7 bytes, "C(auto)" */
1269 } else {
1270 int ret = parse_color(sb, placeholder, c);
1271 if (ret)
1272 c->auto_color = 0;
1273 /*
1274 * Otherwise, we decided to treat %C<unknown>
1275 * as a literal string, and the previous
1276 * %C(auto) is still valid.
1277 */
1278 return ret;
1279 }
1280 case 'w':
1281 if (placeholder[1] == '(') {
1282 unsigned long width = 0, indent1 = 0, indent2 = 0;
1283 char *next;
1284 const char *start = placeholder + 2;
1285 const char *end = strchr(start, ')');
1286 if (!end)
1287 return 0;
1288 if (end > start) {
1289 width = strtoul(start, &next, 10);
1290 if (*next == ',') {
1291 indent1 = strtoul(next + 1, &next, 10);
1292 if (*next == ',') {
1293 indent2 = strtoul(next + 1,
1294 &next, 10);
1295 }
1296 }
1297 if (*next != ')')
1298 return 0;
1299 }
1300 rewrap_message_tail(sb, c, width, indent1, indent2);
1301 return end - placeholder + 1;
1302 } else
1303 return 0;
1304
1305 case '<':
1306 case '>':
1307 return parse_padding_placeholder(placeholder, c);
1308 }
1309
1310 if (skip_prefix(placeholder, "(describe", &arg)) {
1311 struct child_process cmd = CHILD_PROCESS_INIT;
1312 struct strbuf out = STRBUF_INIT;
1313 struct strbuf err = STRBUF_INIT;
1314 struct pretty_print_describe_status *describe_status;
1315
1316 describe_status = c->pretty_ctx->describe_status;
1317 if (describe_status) {
1318 if (!describe_status->max_invocations)
1319 return 0;
1320 describe_status->max_invocations--;
1321 }
1322
1323 cmd.git_cmd = 1;
1324 strvec_push(&cmd.args, "describe");
1325
1326 if (*arg == ':') {
1327 arg++;
1328 arg += parse_describe_args(arg, &cmd.args);
1329 }
1330
1331 if (*arg != ')') {
1332 child_process_clear(&cmd);
1333 return 0;
1334 }
1335
1336 strvec_push(&cmd.args, oid_to_hex(&commit->object.oid));
1337 pipe_command(&cmd, NULL, 0, &out, 0, &err, 0);
1338 strbuf_rtrim(&out);
1339 strbuf_addbuf(sb, &out);
1340 strbuf_release(&out);
1341 strbuf_release(&err);
1342 return arg - placeholder + 1;
1343 }
1344
1345 /* these depend on the commit */
1346 if (!commit->object.parsed)
1347 parse_object(the_repository, &commit->object.oid);
1348
1349 switch (placeholder[0]) {
1350 case 'H': /* commit hash */
1351 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_COMMIT));
1352 strbuf_addstr(sb, oid_to_hex(&commit->object.oid));
1353 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_RESET));
1354 return 1;
1355 case 'h': /* abbreviated commit hash */
1356 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_COMMIT));
1357 strbuf_add_unique_abbrev(sb, &commit->object.oid,
1358 c->pretty_ctx->abbrev);
1359 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_RESET));
1360 return 1;
1361 case 'T': /* tree hash */
1362 strbuf_addstr(sb, oid_to_hex(get_commit_tree_oid(commit)));
1363 return 1;
1364 case 't': /* abbreviated tree hash */
1365 strbuf_add_unique_abbrev(sb,
1366 get_commit_tree_oid(commit),
1367 c->pretty_ctx->abbrev);
1368 return 1;
1369 case 'P': /* parent hashes */
1370 for (p = commit->parents; p; p = p->next) {
1371 if (p != commit->parents)
1372 strbuf_addch(sb, ' ');
1373 strbuf_addstr(sb, oid_to_hex(&p->item->object.oid));
1374 }
1375 return 1;
1376 case 'p': /* abbreviated parent hashes */
1377 for (p = commit->parents; p; p = p->next) {
1378 if (p != commit->parents)
1379 strbuf_addch(sb, ' ');
1380 strbuf_add_unique_abbrev(sb, &p->item->object.oid,
1381 c->pretty_ctx->abbrev);
1382 }
1383 return 1;
1384 case 'm': /* left/right/bottom */
1385 strbuf_addstr(sb, get_revision_mark(NULL, commit));
1386 return 1;
1387 case 'd':
1388 format_decorations(sb, commit, c->auto_color);
1389 return 1;
1390 case 'D':
1391 format_decorations_extended(sb, commit, c->auto_color, "", ", ", "");
1392 return 1;
1393 case 'S': /* tag/branch like --source */
1394 if (!(c->pretty_ctx->rev && c->pretty_ctx->rev->sources))
1395 return 0;
1396 slot = revision_sources_at(c->pretty_ctx->rev->sources, commit);
1397 if (!(slot && *slot))
1398 return 0;
1399 strbuf_addstr(sb, *slot);
1400 return 1;
1401 case 'g': /* reflog info */
1402 switch(placeholder[1]) {
1403 case 'd': /* reflog selector */
1404 case 'D':
1405 if (c->pretty_ctx->reflog_info)
1406 get_reflog_selector(sb,
1407 c->pretty_ctx->reflog_info,
1408 &c->pretty_ctx->date_mode,
1409 c->pretty_ctx->date_mode_explicit,
1410 (placeholder[1] == 'd'));
1411 return 2;
1412 case 's': /* reflog message */
1413 if (c->pretty_ctx->reflog_info)
1414 get_reflog_message(sb, c->pretty_ctx->reflog_info);
1415 return 2;
1416 case 'n':
1417 case 'N':
1418 case 'e':
1419 case 'E':
1420 return format_reflog_person(sb,
1421 placeholder[1],
1422 c->pretty_ctx->reflog_info,
1423 &c->pretty_ctx->date_mode);
1424 }
1425 return 0; /* unknown %g placeholder */
1426 case 'N':
1427 if (c->pretty_ctx->notes_message) {
1428 strbuf_addstr(sb, c->pretty_ctx->notes_message);
1429 return 1;
1430 }
1431 return 0;
1432 }
1433
1434 if (placeholder[0] == 'G') {
1435 if (!c->signature_check.result)
1436 check_commit_signature(c->commit, &(c->signature_check));
1437 switch (placeholder[1]) {
1438 case 'G':
1439 if (c->signature_check.gpg_output)
1440 strbuf_addstr(sb, c->signature_check.gpg_output);
1441 break;
1442 case '?':
1443 switch (c->signature_check.result) {
1444 case 'G':
1445 switch (c->signature_check.trust_level) {
1446 case TRUST_UNDEFINED:
1447 case TRUST_NEVER:
1448 strbuf_addch(sb, 'U');
1449 break;
1450 default:
1451 strbuf_addch(sb, 'G');
1452 break;
1453 }
1454 break;
1455 case 'B':
1456 case 'E':
1457 case 'N':
1458 case 'X':
1459 case 'Y':
1460 case 'R':
1461 strbuf_addch(sb, c->signature_check.result);
1462 }
1463 break;
1464 case 'S':
1465 if (c->signature_check.signer)
1466 strbuf_addstr(sb, c->signature_check.signer);
1467 break;
1468 case 'K':
1469 if (c->signature_check.key)
1470 strbuf_addstr(sb, c->signature_check.key);
1471 break;
1472 case 'F':
1473 if (c->signature_check.fingerprint)
1474 strbuf_addstr(sb, c->signature_check.fingerprint);
1475 break;
1476 case 'P':
1477 if (c->signature_check.primary_key_fingerprint)
1478 strbuf_addstr(sb, c->signature_check.primary_key_fingerprint);
1479 break;
1480 case 'T':
1481 switch (c->signature_check.trust_level) {
1482 case TRUST_UNDEFINED:
1483 strbuf_addstr(sb, "undefined");
1484 break;
1485 case TRUST_NEVER:
1486 strbuf_addstr(sb, "never");
1487 break;
1488 case TRUST_MARGINAL:
1489 strbuf_addstr(sb, "marginal");
1490 break;
1491 case TRUST_FULLY:
1492 strbuf_addstr(sb, "fully");
1493 break;
1494 case TRUST_ULTIMATE:
1495 strbuf_addstr(sb, "ultimate");
1496 break;
1497 }
1498 break;
1499 default:
1500 return 0;
1501 }
1502 return 2;
1503 }
1504
1505 /* For the rest we have to parse the commit header. */
1506 if (!c->commit_header_parsed) {
1507 msg = c->message =
1508 repo_logmsg_reencode(c->repository, commit,
1509 &c->commit_encoding, "UTF-8");
1510 parse_commit_header(c);
1511 }
1512
1513 switch (placeholder[0]) {
1514 case 'a': /* author ... */
1515 return format_person_part(sb, placeholder[1],
1516 msg + c->author.off, c->author.len,
1517 &c->pretty_ctx->date_mode);
1518 case 'c': /* committer ... */
1519 return format_person_part(sb, placeholder[1],
1520 msg + c->committer.off, c->committer.len,
1521 &c->pretty_ctx->date_mode);
1522 case 'e': /* encoding */
1523 if (c->commit_encoding)
1524 strbuf_addstr(sb, c->commit_encoding);
1525 return 1;
1526 case 'B': /* raw body */
1527 /* message_off is always left at the initial newline */
1528 strbuf_addstr(sb, msg + c->message_off + 1);
1529 return 1;
1530 }
1531
1532 /* Now we need to parse the commit message. */
1533 if (!c->commit_message_parsed)
1534 parse_commit_message(c);
1535
1536 switch (placeholder[0]) {
1537 case 's': /* subject */
1538 format_subject(sb, msg + c->subject_off, " ");
1539 return 1;
1540 case 'f': /* sanitized subject */
1541 eol = strchrnul(msg + c->subject_off, '\n');
1542 format_sanitized_subject(sb, msg + c->subject_off, eol - (msg + c->subject_off));
1543 return 1;
1544 case 'b': /* body */
1545 strbuf_addstr(sb, msg + c->body_off);
1546 return 1;
1547 }
1548
1549 if (skip_prefix(placeholder, "(trailers", &arg)) {
1550 struct process_trailer_options opts = PROCESS_TRAILER_OPTIONS_INIT;
1551 struct string_list filter_list = STRING_LIST_INIT_NODUP;
1552 struct strbuf sepbuf = STRBUF_INIT;
1553 struct strbuf kvsepbuf = STRBUF_INIT;
1554 size_t ret = 0;
1555
1556 opts.no_divider = 1;
1557
1558 if (*arg == ':') {
1559 arg++;
1560 if (format_set_trailers_options(&opts, &filter_list, &sepbuf, &kvsepbuf, &arg, NULL))
1561 goto trailer_out;
1562 }
1563 if (*arg == ')') {
1564 format_trailers_from_commit(sb, msg + c->subject_off, &opts);
1565 ret = arg - placeholder + 1;
1566 }
1567 trailer_out:
1568 string_list_clear(&filter_list, 0);
1569 strbuf_release(&sepbuf);
1570 return ret;
1571 }
1572
1573 return 0; /* unknown placeholder */
1574 }
1575
1576 static size_t format_and_pad_commit(struct strbuf *sb, /* in UTF-8 */
1577 const char *placeholder,
1578 struct format_commit_context *c)
1579 {
1580 struct strbuf local_sb = STRBUF_INIT;
1581 int total_consumed = 0, len, padding = c->padding;
1582 if (padding < 0) {
1583 const char *start = strrchr(sb->buf, '\n');
1584 int occupied;
1585 if (!start)
1586 start = sb->buf;
1587 occupied = utf8_strnwidth(start, -1, 1);
1588 occupied += c->pretty_ctx->graph_width;
1589 padding = (-padding) - occupied;
1590 }
1591 while (1) {
1592 int modifier = *placeholder == 'C';
1593 int consumed = format_commit_one(&local_sb, placeholder, c);
1594 total_consumed += consumed;
1595
1596 if (!modifier)
1597 break;
1598
1599 placeholder += consumed;
1600 if (*placeholder != '%')
1601 break;
1602 placeholder++;
1603 total_consumed++;
1604 }
1605 len = utf8_strnwidth(local_sb.buf, -1, 1);
1606
1607 if (c->flush_type == flush_left_and_steal) {
1608 const char *ch = sb->buf + sb->len - 1;
1609 while (len > padding && ch > sb->buf) {
1610 const char *p;
1611 if (*ch == ' ') {
1612 ch--;
1613 padding++;
1614 continue;
1615 }
1616 /* check for trailing ansi sequences */
1617 if (*ch != 'm')
1618 break;
1619 p = ch - 1;
1620 while (ch - p < 10 && *p != '\033')
1621 p--;
1622 if (*p != '\033' ||
1623 ch + 1 - p != display_mode_esc_sequence_len(p))
1624 break;
1625 /*
1626 * got a good ansi sequence, put it back to
1627 * local_sb as we're cutting sb
1628 */
1629 strbuf_insert(&local_sb, 0, p, ch + 1 - p);
1630 ch = p - 1;
1631 }
1632 strbuf_setlen(sb, ch + 1 - sb->buf);
1633 c->flush_type = flush_left;
1634 }
1635
1636 if (len > padding) {
1637 switch (c->truncate) {
1638 case trunc_left:
1639 strbuf_utf8_replace(&local_sb,
1640 0, len - (padding - 2),
1641 "..");
1642 break;
1643 case trunc_middle:
1644 strbuf_utf8_replace(&local_sb,
1645 padding / 2 - 1,
1646 len - (padding - 2),
1647 "..");
1648 break;
1649 case trunc_right:
1650 strbuf_utf8_replace(&local_sb,
1651 padding - 2, len - (padding - 2),
1652 "..");
1653 break;
1654 case trunc_none:
1655 break;
1656 }
1657 strbuf_addbuf(sb, &local_sb);
1658 } else {
1659 int sb_len = sb->len, offset = 0;
1660 if (c->flush_type == flush_left)
1661 offset = padding - len;
1662 else if (c->flush_type == flush_both)
1663 offset = (padding - len) / 2;
1664 /*
1665 * we calculate padding in columns, now
1666 * convert it back to chars
1667 */
1668 padding = padding - len + local_sb.len;
1669 strbuf_addchars(sb, ' ', padding);
1670 memcpy(sb->buf + sb_len + offset, local_sb.buf,
1671 local_sb.len);
1672 }
1673 strbuf_release(&local_sb);
1674 c->flush_type = no_flush;
1675 return total_consumed;
1676 }
1677
1678 static size_t format_commit_item(struct strbuf *sb, /* in UTF-8 */
1679 const char *placeholder,
1680 void *context)
1681 {
1682 int consumed;
1683 size_t orig_len;
1684 enum {
1685 NO_MAGIC,
1686 ADD_LF_BEFORE_NON_EMPTY,
1687 DEL_LF_BEFORE_EMPTY,
1688 ADD_SP_BEFORE_NON_EMPTY
1689 } magic = NO_MAGIC;
1690
1691 switch (placeholder[0]) {
1692 case '-':
1693 magic = DEL_LF_BEFORE_EMPTY;
1694 break;
1695 case '+':
1696 magic = ADD_LF_BEFORE_NON_EMPTY;
1697 break;
1698 case ' ':
1699 magic = ADD_SP_BEFORE_NON_EMPTY;
1700 break;
1701 default:
1702 break;
1703 }
1704 if (magic != NO_MAGIC)
1705 placeholder++;
1706
1707 orig_len = sb->len;
1708 if (((struct format_commit_context *)context)->flush_type != no_flush)
1709 consumed = format_and_pad_commit(sb, placeholder, context);
1710 else
1711 consumed = format_commit_one(sb, placeholder, context);
1712 if (magic == NO_MAGIC)
1713 return consumed;
1714
1715 if ((orig_len == sb->len) && magic == DEL_LF_BEFORE_EMPTY) {
1716 while (sb->len && sb->buf[sb->len - 1] == '\n')
1717 strbuf_setlen(sb, sb->len - 1);
1718 } else if (orig_len != sb->len) {
1719 if (magic == ADD_LF_BEFORE_NON_EMPTY)
1720 strbuf_insertstr(sb, orig_len, "\n");
1721 else if (magic == ADD_SP_BEFORE_NON_EMPTY)
1722 strbuf_insertstr(sb, orig_len, " ");
1723 }
1724 return consumed + 1;
1725 }
1726
1727 static size_t userformat_want_item(struct strbuf *sb, const char *placeholder,
1728 void *context)
1729 {
1730 struct userformat_want *w = context;
1731
1732 if (*placeholder == '+' || *placeholder == '-' || *placeholder == ' ')
1733 placeholder++;
1734
1735 switch (*placeholder) {
1736 case 'N':
1737 w->notes = 1;
1738 break;
1739 case 'S':
1740 w->source = 1;
1741 break;
1742 case 'd':
1743 case 'D':
1744 w->decorate = 1;
1745 break;
1746 }
1747 return 0;
1748 }
1749
1750 void userformat_find_requirements(const char *fmt, struct userformat_want *w)
1751 {
1752 struct strbuf dummy = STRBUF_INIT;
1753
1754 if (!fmt) {
1755 if (!user_format)
1756 return;
1757 fmt = user_format;
1758 }
1759 strbuf_expand(&dummy, fmt, userformat_want_item, w);
1760 strbuf_release(&dummy);
1761 }
1762
1763 void repo_format_commit_message(struct repository *r,
1764 const struct commit *commit,
1765 const char *format, struct strbuf *sb,
1766 const struct pretty_print_context *pretty_ctx)
1767 {
1768 struct format_commit_context context = {
1769 .repository = r,
1770 .commit = commit,
1771 .pretty_ctx = pretty_ctx,
1772 .wrap_start = sb->len
1773 };
1774 const char *output_enc = pretty_ctx->output_encoding;
1775 const char *utf8 = "UTF-8";
1776
1777 strbuf_expand(sb, format, format_commit_item, &context);
1778 rewrap_message_tail(sb, &context, 0, 0, 0);
1779
1780 /*
1781 * Convert output to an actual output encoding; note that
1782 * format_commit_item() will always use UTF-8, so we don't
1783 * have to bother if that's what the output wants.
1784 */
1785 if (output_enc) {
1786 if (same_encoding(utf8, output_enc))
1787 output_enc = NULL;
1788 } else {
1789 if (context.commit_encoding &&
1790 !same_encoding(context.commit_encoding, utf8))
1791 output_enc = context.commit_encoding;
1792 }
1793
1794 if (output_enc) {
1795 size_t outsz;
1796 char *out = reencode_string_len(sb->buf, sb->len,
1797 output_enc, utf8, &outsz);
1798 if (out)
1799 strbuf_attach(sb, out, outsz, outsz + 1);
1800 }
1801
1802 free(context.commit_encoding);
1803 repo_unuse_commit_buffer(r, commit, context.message);
1804 }
1805
1806 static void pp_header(struct pretty_print_context *pp,
1807 const char *encoding,
1808 const struct commit *commit,
1809 const char **msg_p,
1810 struct strbuf *sb)
1811 {
1812 int parents_shown = 0;
1813
1814 for (;;) {
1815 const char *name, *line = *msg_p;
1816 int linelen = get_one_line(*msg_p);
1817
1818 if (!linelen)
1819 return;
1820 *msg_p += linelen;
1821
1822 if (linelen == 1)
1823 /* End of header */
1824 return;
1825
1826 if (pp->fmt == CMIT_FMT_RAW) {
1827 strbuf_add(sb, line, linelen);
1828 continue;
1829 }
1830
1831 if (starts_with(line, "parent ")) {
1832 if (linelen != the_hash_algo->hexsz + 8)
1833 die("bad parent line in commit");
1834 continue;
1835 }
1836
1837 if (!parents_shown) {
1838 unsigned num = commit_list_count(commit->parents);
1839 /* with enough slop */
1840 strbuf_grow(sb, num * (GIT_MAX_HEXSZ + 10) + 20);
1841 add_merge_info(pp, sb, commit);
1842 parents_shown = 1;
1843 }
1844
1845 /*
1846 * MEDIUM == DEFAULT shows only author with dates.
1847 * FULL shows both authors but not dates.
1848 * FULLER shows both authors and dates.
1849 */
1850 if (skip_prefix(line, "author ", &name)) {
1851 strbuf_grow(sb, linelen + 80);
1852 pp_user_info(pp, "Author", sb, name, encoding);
1853 }
1854 if (skip_prefix(line, "committer ", &name) &&
1855 (pp->fmt == CMIT_FMT_FULL || pp->fmt == CMIT_FMT_FULLER)) {
1856 strbuf_grow(sb, linelen + 80);
1857 pp_user_info(pp, "Commit", sb, name, encoding);
1858 }
1859 }
1860 }
1861
1862 void pp_title_line(struct pretty_print_context *pp,
1863 const char **msg_p,
1864 struct strbuf *sb,
1865 const char *encoding,
1866 int need_8bit_cte)
1867 {
1868 static const int max_length = 78; /* per rfc2047 */
1869 struct strbuf title;
1870
1871 strbuf_init(&title, 80);
1872 *msg_p = format_subject(&title, *msg_p,
1873 pp->preserve_subject ? "\n" : " ");
1874
1875 strbuf_grow(sb, title.len + 1024);
1876 if (pp->print_email_subject) {
1877 if (pp->rev)
1878 fmt_output_email_subject(sb, pp->rev);
1879 if (pp->encode_email_headers &&
1880 needs_rfc2047_encoding(title.buf, title.len))
1881 add_rfc2047(sb, title.buf, title.len,
1882 encoding, RFC2047_SUBJECT);
1883 else
1884 strbuf_add_wrapped_bytes(sb, title.buf, title.len,
1885 -last_line_length(sb), 1, max_length);
1886 } else {
1887 strbuf_addbuf(sb, &title);
1888 }
1889 strbuf_addch(sb, '\n');
1890
1891 if (need_8bit_cte == 0) {
1892 int i;
1893 for (i = 0; i < pp->in_body_headers.nr; i++) {
1894 if (has_non_ascii(pp->in_body_headers.items[i].string)) {
1895 need_8bit_cte = 1;
1896 break;
1897 }
1898 }
1899 }
1900
1901 if (need_8bit_cte > 0) {
1902 const char *header_fmt =
1903 "MIME-Version: 1.0\n"
1904 "Content-Type: text/plain; charset=%s\n"
1905 "Content-Transfer-Encoding: 8bit\n";
1906 strbuf_addf(sb, header_fmt, encoding);
1907 }
1908 if (pp->after_subject) {
1909 strbuf_addstr(sb, pp->after_subject);
1910 }
1911 if (cmit_fmt_is_mail(pp->fmt)) {
1912 strbuf_addch(sb, '\n');
1913 }
1914
1915 if (pp->in_body_headers.nr) {
1916 int i;
1917 for (i = 0; i < pp->in_body_headers.nr; i++) {
1918 strbuf_addstr(sb, pp->in_body_headers.items[i].string);
1919 free(pp->in_body_headers.items[i].string);
1920 }
1921 string_list_clear(&pp->in_body_headers, 0);
1922 strbuf_addch(sb, '\n');
1923 }
1924
1925 strbuf_release(&title);
1926 }
1927
1928 static int pp_utf8_width(const char *start, const char *end)
1929 {
1930 int width = 0;
1931 size_t remain = end - start;
1932
1933 while (remain) {
1934 int n = utf8_width(&start, &remain);
1935 if (n < 0 || !start)
1936 return -1;
1937 width += n;
1938 }
1939 return width;
1940 }
1941
1942 static void strbuf_add_tabexpand(struct strbuf *sb, int tabwidth,
1943 const char *line, int linelen)
1944 {
1945 const char *tab;
1946
1947 while ((tab = memchr(line, '\t', linelen)) != NULL) {
1948 int width = pp_utf8_width(line, tab);
1949
1950 /*
1951 * If it wasn't well-formed utf8, or it
1952 * had characters with badly defined
1953 * width (control characters etc), just
1954 * give up on trying to align things.
1955 */
1956 if (width < 0)
1957 break;
1958
1959 /* Output the data .. */
1960 strbuf_add(sb, line, tab - line);
1961
1962 /* .. and the de-tabified tab */
1963 strbuf_addchars(sb, ' ', tabwidth - (width % tabwidth));
1964
1965 /* Skip over the printed part .. */
1966 linelen -= tab + 1 - line;
1967 line = tab + 1;
1968 }
1969
1970 /*
1971 * Print out everything after the last tab without
1972 * worrying about width - there's nothing more to
1973 * align.
1974 */
1975 strbuf_add(sb, line, linelen);
1976 }
1977
1978 /*
1979 * pp_handle_indent() prints out the intendation, and
1980 * the whole line (without the final newline), after
1981 * de-tabifying.
1982 */
1983 static void pp_handle_indent(struct pretty_print_context *pp,
1984 struct strbuf *sb, int indent,
1985 const char *line, int linelen)
1986 {
1987 strbuf_addchars(sb, ' ', indent);
1988 if (pp->expand_tabs_in_log)
1989 strbuf_add_tabexpand(sb, pp->expand_tabs_in_log, line, linelen);
1990 else
1991 strbuf_add(sb, line, linelen);
1992 }
1993
1994 static int is_mboxrd_from(const char *line, int len)
1995 {
1996 /*
1997 * a line matching /^From $/ here would only have len == 4
1998 * at this point because is_empty_line would've trimmed all
1999 * trailing space
2000 */
2001 return len > 4 && starts_with(line + strspn(line, ">"), "From ");
2002 }
2003
2004 void pp_remainder(struct pretty_print_context *pp,
2005 const char **msg_p,
2006 struct strbuf *sb,
2007 int indent)
2008 {
2009 int first = 1;
2010 for (;;) {
2011 const char *line = *msg_p;
2012 int linelen = get_one_line(line);
2013 *msg_p += linelen;
2014
2015 if (!linelen)
2016 break;
2017
2018 if (is_blank_line(line, &linelen)) {
2019 if (first)
2020 continue;
2021 if (pp->fmt == CMIT_FMT_SHORT)
2022 break;
2023 }
2024 first = 0;
2025
2026 strbuf_grow(sb, linelen + indent + 20);
2027 if (indent)
2028 pp_handle_indent(pp, sb, indent, line, linelen);
2029 else if (pp->expand_tabs_in_log)
2030 strbuf_add_tabexpand(sb, pp->expand_tabs_in_log,
2031 line, linelen);
2032 else {
2033 if (pp->fmt == CMIT_FMT_MBOXRD &&
2034 is_mboxrd_from(line, linelen))
2035 strbuf_addch(sb, '>');
2036
2037 strbuf_add(sb, line, linelen);
2038 }
2039 strbuf_addch(sb, '\n');
2040 }
2041 }
2042
2043 void pretty_print_commit(struct pretty_print_context *pp,
2044 const struct commit *commit,
2045 struct strbuf *sb)
2046 {
2047 unsigned long beginning_of_body;
2048 int indent = 4;
2049 const char *msg;
2050 const char *reencoded;
2051 const char *encoding;
2052 int need_8bit_cte = pp->need_8bit_cte;
2053
2054 if (pp->fmt == CMIT_FMT_USERFORMAT) {
2055 format_commit_message(commit, user_format, sb, pp);
2056 return;
2057 }
2058
2059 encoding = get_log_output_encoding();
2060 msg = reencoded = logmsg_reencode(commit, NULL, encoding);
2061
2062 if (pp->fmt == CMIT_FMT_ONELINE || cmit_fmt_is_mail(pp->fmt))
2063 indent = 0;
2064
2065 /*
2066 * We need to check and emit Content-type: to mark it
2067 * as 8-bit if we haven't done so.
2068 */
2069 if (cmit_fmt_is_mail(pp->fmt) && need_8bit_cte == 0) {
2070 int i, ch, in_body;
2071
2072 for (in_body = i = 0; (ch = msg[i]); i++) {
2073 if (!in_body) {
2074 /* author could be non 7-bit ASCII but
2075 * the log may be so; skip over the
2076 * header part first.
2077 */
2078 if (ch == '\n' && msg[i+1] == '\n')
2079 in_body = 1;
2080 }
2081 else if (non_ascii(ch)) {
2082 need_8bit_cte = 1;
2083 break;
2084 }
2085 }
2086 }
2087
2088 pp_header(pp, encoding, commit, &msg, sb);
2089 if (pp->fmt != CMIT_FMT_ONELINE && !pp->print_email_subject) {
2090 strbuf_addch(sb, '\n');
2091 }
2092
2093 /* Skip excess blank lines at the beginning of body, if any... */
2094 msg = skip_blank_lines(msg);
2095
2096 /* These formats treat the title line specially. */
2097 if (pp->fmt == CMIT_FMT_ONELINE || cmit_fmt_is_mail(pp->fmt))
2098 pp_title_line(pp, &msg, sb, encoding, need_8bit_cte);
2099
2100 beginning_of_body = sb->len;
2101 if (pp->fmt != CMIT_FMT_ONELINE)
2102 pp_remainder(pp, &msg, sb, indent);
2103 strbuf_rtrim(sb);
2104
2105 /* Make sure there is an EOLN for the non-oneline case */
2106 if (pp->fmt != CMIT_FMT_ONELINE)
2107 strbuf_addch(sb, '\n');
2108
2109 /*
2110 * The caller may append additional body text in e-mail
2111 * format. Make sure we did not strip the blank line
2112 * between the header and the body.
2113 */
2114 if (cmit_fmt_is_mail(pp->fmt) && sb->len <= beginning_of_body)
2115 strbuf_addch(sb, '\n');
2116
2117 unuse_commit_buffer(commit, reencoded);
2118 }
2119
2120 void pp_commit_easy(enum cmit_fmt fmt, const struct commit *commit,
2121 struct strbuf *sb)
2122 {
2123 struct pretty_print_context pp = {0};
2124 pp.fmt = fmt;
2125 pretty_print_commit(&pp, commit, sb);
2126 }
git for Debian