?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Sync with maint-2.32
[git/debian.git] / pretty.c
blob024edc9a5ac1b6f638a541902ec5c97490952a95
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 /*
18 * The limit for formatting directives, which enable the caller to append
19 * arbitrarily many bytes to the formatted buffer. This includes padding
20 * and wrapping formatters.
21 */
22 #define FORMATTING_LIMIT (16 * 1024)
23
24 static char *user_format;
25 static struct cmt_fmt_map {
26 const char *name;
27 enum cmit_fmt format;
28 int is_tformat;
29 int expand_tabs_in_log;
30 int is_alias;
31 enum date_mode_type default_date_mode_type;
32 const char *user_format;
33 } *commit_formats;
34 static size_t builtin_formats_len;
35 static size_t commit_formats_len;
36 static size_t commit_formats_alloc;
37 static struct cmt_fmt_map *find_commit_format(const char *sought);
38
39 int commit_format_is_empty(enum cmit_fmt fmt)
40 {
41 return fmt == CMIT_FMT_USERFORMAT && !*user_format;
42 }
43
44 static void save_user_format(struct rev_info *rev, const char *cp, int is_tformat)
45 {
46 free(user_format);
47 user_format = xstrdup(cp);
48 if (is_tformat)
49 rev->use_terminator = 1;
50 rev->commit_format = CMIT_FMT_USERFORMAT;
51 }
52
53 static int git_pretty_formats_config(const char *var, const char *value, void *cb)
54 {
55 struct cmt_fmt_map *commit_format = NULL;
56 const char *name;
57 const char *fmt;
58 int i;
59
60 if (!skip_prefix(var, "pretty.", &name))
61 return 0;
62
63 for (i = 0; i < builtin_formats_len; i++) {
64 if (!strcmp(commit_formats[i].name, name))
65 return 0;
66 }
67
68 for (i = builtin_formats_len; i < commit_formats_len; i++) {
69 if (!strcmp(commit_formats[i].name, name)) {
70 commit_format = &commit_formats[i];
71 break;
72 }
73 }
74
75 if (!commit_format) {
76 ALLOC_GROW(commit_formats, commit_formats_len+1,
77 commit_formats_alloc);
78 commit_format = &commit_formats[commit_formats_len];
79 memset(commit_format, 0, sizeof(*commit_format));
80 commit_formats_len++;
81 }
82
83 commit_format->name = xstrdup(name);
84 commit_format->format = CMIT_FMT_USERFORMAT;
85 if (git_config_string(&fmt, var, value))
86 return -1;
87
88 if (skip_prefix(fmt, "format:", &fmt))
89 commit_format->is_tformat = 0;
90 else if (skip_prefix(fmt, "tformat:", &fmt) || strchr(fmt, '%'))
91 commit_format->is_tformat = 1;
92 else
93 commit_format->is_alias = 1;
94 commit_format->user_format = fmt;
95
96 return 0;
97 }
98
99 static void setup_commit_formats(void)
100 {
101 struct cmt_fmt_map builtin_formats[] = {
102 { "raw", CMIT_FMT_RAW, 0, 0 },
103 { "medium", CMIT_FMT_MEDIUM, 0, 8 },
104 { "short", CMIT_FMT_SHORT, 0, 0 },
105 { "email", CMIT_FMT_EMAIL, 0, 0 },
106 { "mboxrd", CMIT_FMT_MBOXRD, 0, 0 },
107 { "fuller", CMIT_FMT_FULLER, 0, 8 },
108 { "full", CMIT_FMT_FULL, 0, 8 },
109 { "oneline", CMIT_FMT_ONELINE, 1, 0 },
110 { "reference", CMIT_FMT_USERFORMAT, 1, 0,
111 0, DATE_SHORT, "%C(auto)%h (%s, %ad)" },
112 /*
113 * Please update $__git_log_pretty_formats in
114 * git-completion.bash when you add new formats.
115 */
116 };
117 commit_formats_len = ARRAY_SIZE(builtin_formats);
118 builtin_formats_len = commit_formats_len;
119 ALLOC_GROW(commit_formats, commit_formats_len, commit_formats_alloc);
120 COPY_ARRAY(commit_formats, builtin_formats,
121 ARRAY_SIZE(builtin_formats));
122
123 git_config(git_pretty_formats_config, NULL);
124 }
125
126 static struct cmt_fmt_map *find_commit_format_recursive(const char *sought,
127 const char *original,
128 int num_redirections)
129 {
130 struct cmt_fmt_map *found = NULL;
131 size_t found_match_len = 0;
132 int i;
133
134 if (num_redirections >= commit_formats_len)
135 die("invalid --pretty format: "
136 "'%s' references an alias which points to itself",
137 original);
138
139 for (i = 0; i < commit_formats_len; i++) {
140 size_t match_len;
141
142 if (!starts_with(commit_formats[i].name, sought))
143 continue;
144
145 match_len = strlen(commit_formats[i].name);
146 if (found == NULL || found_match_len > match_len) {
147 found = &commit_formats[i];
148 found_match_len = match_len;
149 }
150 }
151
152 if (found && found->is_alias) {
153 found = find_commit_format_recursive(found->user_format,
154 original,
155 num_redirections+1);
156 }
157
158 return found;
159 }
160
161 static struct cmt_fmt_map *find_commit_format(const char *sought)
162 {
163 if (!commit_formats)
164 setup_commit_formats();
165
166 return find_commit_format_recursive(sought, sought, 0);
167 }
168
169 void get_commit_format(const char *arg, struct rev_info *rev)
170 {
171 struct cmt_fmt_map *commit_format;
172
173 rev->use_terminator = 0;
174 if (!arg) {
175 rev->commit_format = CMIT_FMT_DEFAULT;
176 return;
177 }
178 if (skip_prefix(arg, "format:", &arg)) {
179 save_user_format(rev, arg, 0);
180 return;
181 }
182
183 if (!*arg || skip_prefix(arg, "tformat:", &arg) || strchr(arg, '%')) {
184 save_user_format(rev, arg, 1);
185 return;
186 }
187
188 commit_format = find_commit_format(arg);
189 if (!commit_format)
190 die("invalid --pretty format: %s", arg);
191
192 rev->commit_format = commit_format->format;
193 rev->use_terminator = commit_format->is_tformat;
194 rev->expand_tabs_in_log_default = commit_format->expand_tabs_in_log;
195 if (!rev->date_mode_explicit && commit_format->default_date_mode_type)
196 rev->date_mode.type = commit_format->default_date_mode_type;
197 if (commit_format->format == CMIT_FMT_USERFORMAT) {
198 save_user_format(rev, commit_format->user_format,
199 commit_format->is_tformat);
200 }
201 }
202
203 /*
204 * Generic support for pretty-printing the header
205 */
206 static int get_one_line(const char *msg)
207 {
208 int ret = 0;
209
210 for (;;) {
211 char c = *msg++;
212 if (!c)
213 break;
214 ret++;
215 if (c == '\n')
216 break;
217 }
218 return ret;
219 }
220
221 /* High bit set, or ISO-2022-INT */
222 static int non_ascii(int ch)
223 {
224 return !isascii(ch) || ch == '\033';
225 }
226
227 int has_non_ascii(const char *s)
228 {
229 int ch;
230 if (!s)
231 return 0;
232 while ((ch = *s++) != '\0') {
233 if (non_ascii(ch))
234 return 1;
235 }
236 return 0;
237 }
238
239 static int is_rfc822_special(char ch)
240 {
241 switch (ch) {
242 case '(':
243 case ')':
244 case '<':
245 case '>':
246 case '[':
247 case ']':
248 case ':':
249 case ';':
250 case '@':
251 case ',':
252 case '.':
253 case '"':
254 case '\\':
255 return 1;
256 default:
257 return 0;
258 }
259 }
260
261 static int needs_rfc822_quoting(const char *s, int len)
262 {
263 int i;
264 for (i = 0; i < len; i++)
265 if (is_rfc822_special(s[i]))
266 return 1;
267 return 0;
268 }
269
270 static int last_line_length(struct strbuf *sb)
271 {
272 int i;
273
274 /* How many bytes are already used on the last line? */
275 for (i = sb->len - 1; i >= 0; i--)
276 if (sb->buf[i] == '\n')
277 break;
278 return sb->len - (i + 1);
279 }
280
281 static void add_rfc822_quoted(struct strbuf *out, const char *s, int len)
282 {
283 int i;
284
285 /* just a guess, we may have to also backslash-quote */
286 strbuf_grow(out, len + 2);
287
288 strbuf_addch(out, '"');
289 for (i = 0; i < len; i++) {
290 switch (s[i]) {
291 case '"':
292 case '\\':
293 strbuf_addch(out, '\\');
294 /* fall through */
295 default:
296 strbuf_addch(out, s[i]);
297 }
298 }
299 strbuf_addch(out, '"');
300 }
301
302 enum rfc2047_type {
303 RFC2047_SUBJECT,
304 RFC2047_ADDRESS
305 };
306
307 static int is_rfc2047_special(char ch, enum rfc2047_type type)
308 {
309 /*
310 * rfc2047, section 4.2:
311 *
312 * 8-bit values which correspond to printable ASCII characters other
313 * than "=", "?", and "_" (underscore), MAY be represented as those
314 * characters. (But see section 5 for restrictions.) In
315 * particular, SPACE and TAB MUST NOT be represented as themselves
316 * within encoded words.
317 */
318
319 /*
320 * rule out non-ASCII characters and non-printable characters (the
321 * non-ASCII check should be redundant as isprint() is not localized
322 * and only knows about ASCII, but be defensive about that)
323 */
324 if (non_ascii(ch) || !isprint(ch))
325 return 1;
326
327 /*
328 * rule out special printable characters (' ' should be the only
329 * whitespace character considered printable, but be defensive and use
330 * isspace())
331 */
332 if (isspace(ch) || ch == '=' || ch == '?' || ch == '_')
333 return 1;
334
335 /*
336 * rfc2047, section 5.3:
337 *
338 * As a replacement for a 'word' entity within a 'phrase', for example,
339 * one that precedes an address in a From, To, or Cc header. The ABNF
340 * definition for 'phrase' from RFC 822 thus becomes:
341 *
342 * phrase = 1*( encoded-word / word )
343 *
344 * In this case the set of characters that may be used in a "Q"-encoded
345 * 'encoded-word' is restricted to: <upper and lower case ASCII
346 * letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
347 * (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
348 * 'phrase' MUST be separated from any adjacent 'word', 'text' or
349 * 'special' by 'linear-white-space'.
350 */
351
352 if (type != RFC2047_ADDRESS)
353 return 0;
354
355 /* '=' and '_' are special cases and have been checked above */
356 return !(isalnum(ch) || ch == '!' || ch == '*' || ch == '+' || ch == '-' || ch == '/');
357 }
358
359 static int needs_rfc2047_encoding(const char *line, int len)
360 {
361 int i;
362
363 for (i = 0; i < len; i++) {
364 int ch = line[i];
365 if (non_ascii(ch) || ch == '\n')
366 return 1;
367 if ((i + 1 < len) && (ch == '=' && line[i+1] == '?'))
368 return 1;
369 }
370
371 return 0;
372 }
373
374 static void add_rfc2047(struct strbuf *sb, const char *line, size_t len,
375 const char *encoding, enum rfc2047_type type)
376 {
377 static const int max_encoded_length = 76; /* per rfc2047 */
378 int i;
379 int line_len = last_line_length(sb);
380
381 strbuf_grow(sb, len * 3 + strlen(encoding) + 100);
382 strbuf_addf(sb, "=?%s?q?", encoding);
383 line_len += strlen(encoding) + 5; /* 5 for =??q? */
384
385 while (len) {
386 /*
387 * RFC 2047, section 5 (3):
388 *
389 * Each 'encoded-word' MUST represent an integral number of
390 * characters. A multi-octet character may not be split across
391 * adjacent 'encoded- word's.
392 */
393 const unsigned char *p = (const unsigned char *)line;
394 int chrlen = mbs_chrlen(&line, &len, encoding);
395 int is_special = (chrlen > 1) || is_rfc2047_special(*p, type);
396
397 /* "=%02X" * chrlen, or the byte itself */
398 const char *encoded_fmt = is_special ? "=%02X" : "%c";
399 int encoded_len = is_special ? 3 * chrlen : 1;
400
401 /*
402 * According to RFC 2047, we could encode the special character
403 * ' ' (space) with '_' (underscore) for readability. But many
404 * programs do not understand this and just leave the
405 * underscore in place. Thus, we do nothing special here, which
406 * causes ' ' to be encoded as '=20', avoiding this problem.
407 */
408
409 if (line_len + encoded_len + 2 > max_encoded_length) {
410 /* It won't fit with trailing "?=" --- break the line */
411 strbuf_addf(sb, "?=\n =?%s?q?", encoding);
412 line_len = strlen(encoding) + 5 + 1; /* =??q? plus SP */
413 }
414
415 for (i = 0; i < chrlen; i++)
416 strbuf_addf(sb, encoded_fmt, p[i]);
417 line_len += encoded_len;
418 }
419 strbuf_addstr(sb, "?=");
420 }
421
422 const char *show_ident_date(const struct ident_split *ident,
423 const struct date_mode *mode)
424 {
425 timestamp_t date = 0;
426 long tz = 0;
427
428 if (ident->date_begin && ident->date_end)
429 date = parse_timestamp(ident->date_begin, NULL, 10);
430 if (date_overflows(date))
431 date = 0;
432 else {
433 if (ident->tz_begin && ident->tz_end)
434 tz = strtol(ident->tz_begin, NULL, 10);
435 if (tz >= INT_MAX || tz <= INT_MIN)
436 tz = 0;
437 }
438 return show_date(date, tz, mode);
439 }
440
441 void pp_user_info(struct pretty_print_context *pp,
442 const char *what, struct strbuf *sb,
443 const char *line, const char *encoding)
444 {
445 struct ident_split ident;
446 char *line_end;
447 const char *mailbuf, *namebuf;
448 size_t namelen, maillen;
449 int max_length = 78; /* per rfc2822 */
450
451 if (pp->fmt == CMIT_FMT_ONELINE)
452 return;
453
454 line_end = strchrnul(line, '\n');
455 if (split_ident_line(&ident, line, line_end - line))
456 return;
457
458 mailbuf = ident.mail_begin;
459 maillen = ident.mail_end - ident.mail_begin;
460 namebuf = ident.name_begin;
461 namelen = ident.name_end - ident.name_begin;
462
463 if (pp->mailmap)
464 map_user(pp->mailmap, &mailbuf, &maillen, &namebuf, &namelen);
465
466 if (cmit_fmt_is_mail(pp->fmt)) {
467 if (pp->from_ident && ident_cmp(pp->from_ident, &ident)) {
468 struct strbuf buf = STRBUF_INIT;
469
470 strbuf_addstr(&buf, "From: ");
471 strbuf_add(&buf, namebuf, namelen);
472 strbuf_addstr(&buf, " <");
473 strbuf_add(&buf, mailbuf, maillen);
474 strbuf_addstr(&buf, ">\n");
475 string_list_append(&pp->in_body_headers,
476 strbuf_detach(&buf, NULL));
477
478 mailbuf = pp->from_ident->mail_begin;
479 maillen = pp->from_ident->mail_end - mailbuf;
480 namebuf = pp->from_ident->name_begin;
481 namelen = pp->from_ident->name_end - namebuf;
482 }
483
484 strbuf_addstr(sb, "From: ");
485 if (pp->encode_email_headers &&
486 needs_rfc2047_encoding(namebuf, namelen)) {
487 add_rfc2047(sb, namebuf, namelen,
488 encoding, RFC2047_ADDRESS);
489 max_length = 76; /* per rfc2047 */
490 } else if (needs_rfc822_quoting(namebuf, namelen)) {
491 struct strbuf quoted = STRBUF_INIT;
492 add_rfc822_quoted(&quoted, namebuf, namelen);
493 strbuf_add_wrapped_bytes(sb, quoted.buf, quoted.len,
494 -6, 1, max_length);
495 strbuf_release(&quoted);
496 } else {
497 strbuf_add_wrapped_bytes(sb, namebuf, namelen,
498 -6, 1, max_length);
499 }
500
501 if (max_length <
502 last_line_length(sb) + strlen(" <") + maillen + strlen(">"))
503 strbuf_addch(sb, '\n');
504 strbuf_addf(sb, " <%.*s>\n", (int)maillen, mailbuf);
505 } else {
506 strbuf_addf(sb, "%s: %.*s%.*s <%.*s>\n", what,
507 (pp->fmt == CMIT_FMT_FULLER) ? 4 : 0, " ",
508 (int)namelen, namebuf, (int)maillen, mailbuf);
509 }
510
511 switch (pp->fmt) {
512 case CMIT_FMT_MEDIUM:
513 strbuf_addf(sb, "Date: %s\n",
514 show_ident_date(&ident, &pp->date_mode));
515 break;
516 case CMIT_FMT_EMAIL:
517 case CMIT_FMT_MBOXRD:
518 strbuf_addf(sb, "Date: %s\n",
519 show_ident_date(&ident, DATE_MODE(RFC2822)));
520 break;
521 case CMIT_FMT_FULLER:
522 strbuf_addf(sb, "%sDate: %s\n", what,
523 show_ident_date(&ident, &pp->date_mode));
524 break;
525 default:
526 /* notin' */
527 break;
528 }
529 }
530
531 static int is_blank_line(const char *line, int *len_p)
532 {
533 int len = *len_p;
534 while (len && isspace(line[len - 1]))
535 len--;
536 *len_p = len;
537 return !len;
538 }
539
540 const char *skip_blank_lines(const char *msg)
541 {
542 for (;;) {
543 int linelen = get_one_line(msg);
544 int ll = linelen;
545 if (!linelen)
546 break;
547 if (!is_blank_line(msg, &ll))
548 break;
549 msg += linelen;
550 }
551 return msg;
552 }
553
554 static void add_merge_info(const struct pretty_print_context *pp,
555 struct strbuf *sb, const struct commit *commit)
556 {
557 struct commit_list *parent = commit->parents;
558
559 if ((pp->fmt == CMIT_FMT_ONELINE) || (cmit_fmt_is_mail(pp->fmt)) ||
560 !parent || !parent->next)
561 return;
562
563 strbuf_addstr(sb, "Merge:");
564
565 while (parent) {
566 struct object_id *oidp = &parent->item->object.oid;
567 strbuf_addch(sb, ' ');
568 if (pp->abbrev)
569 strbuf_add_unique_abbrev(sb, oidp, pp->abbrev);
570 else
571 strbuf_addstr(sb, oid_to_hex(oidp));
572 parent = parent->next;
573 }
574 strbuf_addch(sb, '\n');
575 }
576
577 static char *get_header(const char *msg, const char *key)
578 {
579 size_t len;
580 const char *v = find_commit_header(msg, key, &len);
581 return v ? xmemdupz(v, len) : NULL;
582 }
583
584 static char *replace_encoding_header(char *buf, const char *encoding)
585 {
586 struct strbuf tmp = STRBUF_INIT;
587 size_t start, len;
588 char *cp = buf;
589
590 /* guess if there is an encoding header before a \n\n */
591 while (!starts_with(cp, "encoding ")) {
592 cp = strchr(cp, '\n');
593 if (!cp || *++cp == '\n')
594 return buf;
595 }
596 start = cp - buf;
597 cp = strchr(cp, '\n');
598 if (!cp)
599 return buf; /* should not happen but be defensive */
600 len = cp + 1 - (buf + start);
601
602 strbuf_attach(&tmp, buf, strlen(buf), strlen(buf) + 1);
603 if (is_encoding_utf8(encoding)) {
604 /* we have re-coded to UTF-8; drop the header */
605 strbuf_remove(&tmp, start, len);
606 } else {
607 /* just replaces XXXX in 'encoding XXXX\n' */
608 strbuf_splice(&tmp, start + strlen("encoding "),
609 len - strlen("encoding \n"),
610 encoding, strlen(encoding));
611 }
612 return strbuf_detach(&tmp, NULL);
613 }
614
615 const char *repo_logmsg_reencode(struct repository *r,
616 const struct commit *commit,
617 char **commit_encoding,
618 const char *output_encoding)
619 {
620 static const char *utf8 = "UTF-8";
621 const char *use_encoding;
622 char *encoding;
623 const char *msg = repo_get_commit_buffer(r, commit, NULL);
624 char *out;
625
626 if (!output_encoding || !*output_encoding) {
627 if (commit_encoding)
628 *commit_encoding = get_header(msg, "encoding");
629 return msg;
630 }
631 encoding = get_header(msg, "encoding");
632 if (commit_encoding)
633 *commit_encoding = encoding;
634 use_encoding = encoding ? encoding : utf8;
635 if (same_encoding(use_encoding, output_encoding)) {
636 /*
637 * No encoding work to be done. If we have no encoding header
638 * at all, then there's nothing to do, and we can return the
639 * message verbatim (whether newly allocated or not).
640 */
641 if (!encoding)
642 return msg;
643
644 /*
645 * Otherwise, we still want to munge the encoding header in the
646 * result, which will be done by modifying the buffer. If we
647 * are using a fresh copy, we can reuse it. But if we are using
648 * the cached copy from get_commit_buffer, we need to duplicate it
649 * to avoid munging the cached copy.
650 */
651 if (msg == get_cached_commit_buffer(r, commit, NULL))
652 out = xstrdup(msg);
653 else
654 out = (char *)msg;
655 }
656 else {
657 /*
658 * There's actual encoding work to do. Do the reencoding, which
659 * still leaves the header to be replaced in the next step. At
660 * this point, we are done with msg. If we allocated a fresh
661 * copy, we can free it.
662 */
663 out = reencode_string(msg, output_encoding, use_encoding);
664 if (out)
665 repo_unuse_commit_buffer(r, commit, msg);
666 }
667
668 /*
669 * This replacement actually consumes the buffer we hand it, so we do
670 * not have to worry about freeing the old "out" here.
671 */
672 if (out)
673 out = replace_encoding_header(out, output_encoding);
674
675 if (!commit_encoding)
676 free(encoding);
677 /*
678 * If the re-encoding failed, out might be NULL here; in that
679 * case we just return the commit message verbatim.
680 */
681 return out ? out : msg;
682 }
683
684 static int mailmap_name(const char **email, size_t *email_len,
685 const char **name, size_t *name_len)
686 {
687 static struct string_list *mail_map;
688 if (!mail_map) {
689 CALLOC_ARRAY(mail_map, 1);
690 read_mailmap(mail_map);
691 }
692 return mail_map->nr && map_user(mail_map, email, email_len, name, name_len);
693 }
694
695 static size_t format_person_part(struct strbuf *sb, char part,
696 const char *msg, int len,
697 const struct date_mode *dmode)
698 {
699 /* currently all placeholders have same length */
700 const int placeholder_len = 2;
701 struct ident_split s;
702 const char *name, *mail;
703 size_t maillen, namelen;
704
705 if (split_ident_line(&s, msg, len) < 0)
706 goto skip;
707
708 name = s.name_begin;
709 namelen = s.name_end - s.name_begin;
710 mail = s.mail_begin;
711 maillen = s.mail_end - s.mail_begin;
712
713 if (part == 'N' || part == 'E' || part == 'L') /* mailmap lookup */
714 mailmap_name(&mail, &maillen, &name, &namelen);
715 if (part == 'n' || part == 'N') { /* name */
716 strbuf_add(sb, name, namelen);
717 return placeholder_len;
718 }
719 if (part == 'e' || part == 'E') { /* email */
720 strbuf_add(sb, mail, maillen);
721 return placeholder_len;
722 }
723 if (part == 'l' || part == 'L') { /* local-part */
724 const char *at = memchr(mail, '@', maillen);
725 if (at)
726 maillen = at - mail;
727 strbuf_add(sb, mail, maillen);
728 return placeholder_len;
729 }
730
731 if (!s.date_begin)
732 goto skip;
733
734 if (part == 't') { /* date, UNIX timestamp */
735 strbuf_add(sb, s.date_begin, s.date_end - s.date_begin);
736 return placeholder_len;
737 }
738
739 switch (part) {
740 case 'd': /* date */
741 strbuf_addstr(sb, show_ident_date(&s, dmode));
742 return placeholder_len;
743 case 'D': /* date, RFC2822 style */
744 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(RFC2822)));
745 return placeholder_len;
746 case 'r': /* date, relative */
747 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(RELATIVE)));
748 return placeholder_len;
749 case 'i': /* date, ISO 8601-like */
750 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(ISO8601)));
751 return placeholder_len;
752 case 'I': /* date, ISO 8601 strict */
753 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(ISO8601_STRICT)));
754 return placeholder_len;
755 case 'h': /* date, human */
756 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(HUMAN)));
757 return placeholder_len;
758 case 's':
759 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(SHORT)));
760 return placeholder_len;
761 }
762
763 skip:
764 /*
765 * reading from either a bogus commit, or a reflog entry with
766 * %gn, %ge, etc.; 'sb' cannot be updated, but we still need
767 * to compute a valid return value.
768 */
769 if (part == 'n' || part == 'e' || part == 't' || part == 'd'
770 || part == 'D' || part == 'r' || part == 'i')
771 return placeholder_len;
772
773 return 0; /* unknown placeholder */
774 }
775
776 struct chunk {
777 size_t off;
778 size_t len;
779 };
780
781 enum flush_type {
782 no_flush,
783 flush_right,
784 flush_left,
785 flush_left_and_steal,
786 flush_both
787 };
788
789 enum trunc_type {
790 trunc_none,
791 trunc_left,
792 trunc_middle,
793 trunc_right
794 };
795
796 struct format_commit_context {
797 struct repository *repository;
798 const struct commit *commit;
799 const struct pretty_print_context *pretty_ctx;
800 unsigned commit_header_parsed:1;
801 unsigned commit_message_parsed:1;
802 struct signature_check signature_check;
803 enum flush_type flush_type;
804 enum trunc_type truncate;
805 const char *message;
806 char *commit_encoding;
807 size_t width, indent1, indent2;
808 int auto_color;
809 int padding;
810
811 /* These offsets are relative to the start of the commit message. */
812 struct chunk author;
813 struct chunk committer;
814 size_t message_off;
815 size_t subject_off;
816 size_t body_off;
817
818 /* The following ones are relative to the result struct strbuf. */
819 size_t wrap_start;
820 };
821
822 static void parse_commit_header(struct format_commit_context *context)
823 {
824 const char *msg = context->message;
825 int i;
826
827 for (i = 0; msg[i]; i++) {
828 const char *name;
829 int eol;
830 for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
831 ; /* do nothing */
832
833 if (i == eol) {
834 break;
835 } else if (skip_prefix(msg + i, "author ", &name)) {
836 context->author.off = name - msg;
837 context->author.len = msg + eol - name;
838 } else if (skip_prefix(msg + i, "committer ", &name)) {
839 context->committer.off = name - msg;
840 context->committer.len = msg + eol - name;
841 }
842 i = eol;
843 }
844 context->message_off = i;
845 context->commit_header_parsed = 1;
846 }
847
848 static int istitlechar(char c)
849 {
850 return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') ||
851 (c >= '0' && c <= '9') || c == '.' || c == '_';
852 }
853
854 void format_sanitized_subject(struct strbuf *sb, const char *msg, size_t len)
855 {
856 size_t trimlen;
857 size_t start_len = sb->len;
858 int space = 2;
859 int i;
860
861 for (i = 0; i < len; i++) {
862 if (istitlechar(msg[i])) {
863 if (space == 1)
864 strbuf_addch(sb, '-');
865 space = 0;
866 strbuf_addch(sb, msg[i]);
867 if (msg[i] == '.')
868 while (msg[i+1] == '.')
869 i++;
870 } else
871 space |= 1;
872 }
873
874 /* trim any trailing '.' or '-' characters */
875 trimlen = 0;
876 while (sb->len - trimlen > start_len &&
877 (sb->buf[sb->len - 1 - trimlen] == '.'
878 || sb->buf[sb->len - 1 - trimlen] == '-'))
879 trimlen++;
880 strbuf_remove(sb, sb->len - trimlen, trimlen);
881 }
882
883 const char *format_subject(struct strbuf *sb, const char *msg,
884 const char *line_separator)
885 {
886 int first = 1;
887
888 for (;;) {
889 const char *line = msg;
890 int linelen = get_one_line(line);
891
892 msg += linelen;
893 if (!linelen || is_blank_line(line, &linelen))
894 break;
895
896 if (!sb)
897 continue;
898 strbuf_grow(sb, linelen + 2);
899 if (!first)
900 strbuf_addstr(sb, line_separator);
901 strbuf_add(sb, line, linelen);
902 first = 0;
903 }
904 return msg;
905 }
906
907 static void parse_commit_message(struct format_commit_context *c)
908 {
909 const char *msg = c->message + c->message_off;
910 const char *start = c->message;
911
912 msg = skip_blank_lines(msg);
913 c->subject_off = msg - start;
914
915 msg = format_subject(NULL, msg, NULL);
916 msg = skip_blank_lines(msg);
917 c->body_off = msg - start;
918
919 c->commit_message_parsed = 1;
920 }
921
922 static void strbuf_wrap(struct strbuf *sb, size_t pos,
923 size_t width, size_t indent1, size_t indent2)
924 {
925 struct strbuf tmp = STRBUF_INIT;
926
927 if (pos)
928 strbuf_add(&tmp, sb->buf, pos);
929 strbuf_add_wrapped_text(&tmp, sb->buf + pos,
930 cast_size_t_to_int(indent1),
931 cast_size_t_to_int(indent2),
932 cast_size_t_to_int(width));
933 strbuf_swap(&tmp, sb);
934 strbuf_release(&tmp);
935 }
936
937 static 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;
945 if (c->wrap_start < sb->len)
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
953 static int format_reflog_person(struct strbuf *sb,
954 char part,
955 struct reflog_walk_info *log,
956 const struct date_mode *dmode)
957 {
958 const char *ident;
959
960 if (!log)
961 return 2;
962
963 ident = get_reflog_ident(log);
964 if (!ident)
965 return 2;
966
967 return format_person_part(sb, part, ident, strlen(ident), dmode);
968 }
969
970 static size_t parse_color(struct strbuf *sb, /* in UTF-8 */
971 const char *placeholder,
972 struct format_commit_context *c)
973 {
974 const char *rest = placeholder;
975 const char *basic_color = NULL;
976
977 if (placeholder[1] == '(') {
978 const char *begin = placeholder + 2;
979 const char *end = strchr(begin, ')');
980 char color[COLOR_MAXLEN];
981
982 if (!end)
983 return 0;
984
985 if (skip_prefix(begin, "auto,", &begin)) {
986 if (!want_color(c->pretty_ctx->color))
987 return end - placeholder + 1;
988 } else if (skip_prefix(begin, "always,", &begin)) {
989 /* nothing to do; we do not respect want_color at all */
990 } else {
991 /* the default is the same as "auto" */
992 if (!want_color(c->pretty_ctx->color))
993 return end - placeholder + 1;
994 }
995
996 if (color_parse_mem(begin, end - begin, color) < 0)
997 die(_("unable to parse --pretty format"));
998 strbuf_addstr(sb, color);
999 return end - placeholder + 1;
1000 }
1001
1002 /*
1003 * We handle things like "%C(red)" above; for historical reasons, there
1004 * are a few colors that can be specified without parentheses (and
1005 * they cannot support things like "auto" or "always" at all).
1006 */
1007 if (skip_prefix(placeholder + 1, "red", &rest))
1008 basic_color = GIT_COLOR_RED;
1009 else if (skip_prefix(placeholder + 1, "green", &rest))
1010 basic_color = GIT_COLOR_GREEN;
1011 else if (skip_prefix(placeholder + 1, "blue", &rest))
1012 basic_color = GIT_COLOR_BLUE;
1013 else if (skip_prefix(placeholder + 1, "reset", &rest))
1014 basic_color = GIT_COLOR_RESET;
1015
1016 if (basic_color && want_color(c->pretty_ctx->color))
1017 strbuf_addstr(sb, basic_color);
1018
1019 return rest - placeholder;
1020 }
1021
1022 static size_t parse_padding_placeholder(const char *placeholder,
1023 struct format_commit_context *c)
1024 {
1025 const char *ch = placeholder;
1026 enum flush_type flush_type;
1027 int to_column = 0;
1028
1029 switch (*ch++) {
1030 case '<':
1031 flush_type = flush_right;
1032 break;
1033 case '>':
1034 if (*ch == '<') {
1035 flush_type = flush_both;
1036 ch++;
1037 } else if (*ch == '>') {
1038 flush_type = flush_left_and_steal;
1039 ch++;
1040 } else
1041 flush_type = flush_left;
1042 break;
1043 default:
1044 return 0;
1045 }
1046
1047 /* the next value means "wide enough to that column" */
1048 if (*ch == '|') {
1049 to_column = 1;
1050 ch++;
1051 }
1052
1053 if (*ch == '(') {
1054 const char *start = ch + 1;
1055 const char *end = start + strcspn(start, ",)");
1056 char *next;
1057 int width;
1058 if (!*end || end == start)
1059 return 0;
1060 width = strtol(start, &next, 10);
1061
1062 /*
1063 * We need to limit the amount of padding, or otherwise this
1064 * would allow the user to pad the buffer by arbitrarily many
1065 * bytes and thus cause resource exhaustion.
1066 */
1067 if (width < -FORMATTING_LIMIT || width > FORMATTING_LIMIT)
1068 return 0;
1069
1070 if (next == start || width == 0)
1071 return 0;
1072 if (width < 0) {
1073 if (to_column)
1074 width += term_columns();
1075 if (width < 0)
1076 return 0;
1077 }
1078 c->padding = to_column ? -width : width;
1079 c->flush_type = flush_type;
1080
1081 if (*end == ',') {
1082 start = end + 1;
1083 end = strchr(start, ')');
1084 if (!end || end == start)
1085 return 0;
1086 if (starts_with(start, "trunc)"))
1087 c->truncate = trunc_right;
1088 else if (starts_with(start, "ltrunc)"))
1089 c->truncate = trunc_left;
1090 else if (starts_with(start, "mtrunc)"))
1091 c->truncate = trunc_middle;
1092 else
1093 return 0;
1094 } else
1095 c->truncate = trunc_none;
1096
1097 return end - placeholder + 1;
1098 }
1099 return 0;
1100 }
1101
1102 static int match_placeholder_arg_value(const char *to_parse, const char *candidate,
1103 const char **end, const char **valuestart,
1104 size_t *valuelen)
1105 {
1106 const char *p;
1107
1108 if (!(skip_prefix(to_parse, candidate, &p)))
1109 return 0;
1110 if (valuestart) {
1111 if (*p == '=') {
1112 *valuestart = p + 1;
1113 *valuelen = strcspn(*valuestart, ",)");
1114 p = *valuestart + *valuelen;
1115 } else {
1116 if (*p != ',' && *p != ')')
1117 return 0;
1118 *valuestart = NULL;
1119 *valuelen = 0;
1120 }
1121 }
1122 if (*p == ',') {
1123 *end = p + 1;
1124 return 1;
1125 }
1126 if (*p == ')') {
1127 *end = p;
1128 return 1;
1129 }
1130 return 0;
1131 }
1132
1133 static int match_placeholder_bool_arg(const char *to_parse, const char *candidate,
1134 const char **end, int *val)
1135 {
1136 const char *argval;
1137 char *strval;
1138 size_t arglen;
1139 int v;
1140
1141 if (!match_placeholder_arg_value(to_parse, candidate, end, &argval, &arglen))
1142 return 0;
1143
1144 if (!argval) {
1145 *val = 1;
1146 return 1;
1147 }
1148
1149 strval = xstrndup(argval, arglen);
1150 v = git_parse_maybe_bool(strval);
1151 free(strval);
1152
1153 if (v == -1)
1154 return 0;
1155
1156 *val = v;
1157
1158 return 1;
1159 }
1160
1161 static int format_trailer_match_cb(const struct strbuf *key, void *ud)
1162 {
1163 const struct string_list *list = ud;
1164 const struct string_list_item *item;
1165
1166 for_each_string_list_item (item, list) {
1167 if (key->len == (uintptr_t)item->util &&
1168 !strncasecmp(item->string, key->buf, key->len))
1169 return 1;
1170 }
1171 return 0;
1172 }
1173
1174 int format_set_trailers_options(struct process_trailer_options *opts,
1175 struct string_list *filter_list,
1176 struct strbuf *sepbuf,
1177 struct strbuf *kvsepbuf,
1178 const char **arg,
1179 char **invalid_arg)
1180 {
1181 for (;;) {
1182 const char *argval;
1183 size_t arglen;
1184
1185 if (**arg == ')')
1186 break;
1187
1188 if (match_placeholder_arg_value(*arg, "key", arg, &argval, &arglen)) {
1189 uintptr_t len = arglen;
1190
1191 if (!argval)
1192 return -1;
1193
1194 if (len && argval[len - 1] == ':')
1195 len--;
1196 string_list_append(filter_list, argval)->util = (char *)len;
1197
1198 opts->filter = format_trailer_match_cb;
1199 opts->filter_data = filter_list;
1200 opts->only_trailers = 1;
1201 } else if (match_placeholder_arg_value(*arg, "separator", arg, &argval, &arglen)) {
1202 char *fmt;
1203
1204 strbuf_reset(sepbuf);
1205 fmt = xstrndup(argval, arglen);
1206 strbuf_expand(sepbuf, fmt, strbuf_expand_literal_cb, NULL);
1207 free(fmt);
1208 opts->separator = sepbuf;
1209 } else if (match_placeholder_arg_value(*arg, "key_value_separator", arg, &argval, &arglen)) {
1210 char *fmt;
1211
1212 strbuf_reset(kvsepbuf);
1213 fmt = xstrndup(argval, arglen);
1214 strbuf_expand(kvsepbuf, fmt, strbuf_expand_literal_cb, NULL);
1215 free(fmt);
1216 opts->key_value_separator = kvsepbuf;
1217 } else if (!match_placeholder_bool_arg(*arg, "only", arg, &opts->only_trailers) &&
1218 !match_placeholder_bool_arg(*arg, "unfold", arg, &opts->unfold) &&
1219 !match_placeholder_bool_arg(*arg, "keyonly", arg, &opts->key_only) &&
1220 !match_placeholder_bool_arg(*arg, "valueonly", arg, &opts->value_only)) {
1221 if (invalid_arg) {
1222 size_t len = strcspn(*arg, ",)");
1223 *invalid_arg = xstrndup(*arg, len);
1224 }
1225 return -1;
1226 }
1227 }
1228 return 0;
1229 }
1230
1231 static size_t parse_describe_args(const char *start, struct strvec *args)
1232 {
1233 const char *options[] = { "match", "exclude" };
1234 const char *arg = start;
1235
1236 for (;;) {
1237 const char *matched = NULL;
1238 const char *argval;
1239 size_t arglen = 0;
1240 int i;
1241
1242 for (i = 0; i < ARRAY_SIZE(options); i++) {
1243 if (match_placeholder_arg_value(arg, options[i], &arg,
1244 &argval, &arglen)) {
1245 matched = options[i];
1246 break;
1247 }
1248 }
1249 if (!matched)
1250 break;
1251
1252 if (!arglen)
1253 return 0;
1254 strvec_pushf(args, "--%s=%.*s", matched, (int)arglen, argval);
1255 }
1256 return arg - start;
1257 }
1258
1259 static size_t format_commit_one(struct strbuf *sb, /* in UTF-8 */
1260 const char *placeholder,
1261 void *context)
1262 {
1263 struct format_commit_context *c = context;
1264 const struct commit *commit = c->commit;
1265 const char *msg = c->message;
1266 struct commit_list *p;
1267 const char *arg, *eol;
1268 size_t res;
1269 char **slot;
1270
1271 /* these are independent of the commit */
1272 res = strbuf_expand_literal_cb(sb, placeholder, NULL);
1273 if (res)
1274 return res;
1275
1276 switch (placeholder[0]) {
1277 case 'C':
1278 if (starts_with(placeholder + 1, "(auto)")) {
1279 c->auto_color = want_color(c->pretty_ctx->color);
1280 if (c->auto_color && sb->len)
1281 strbuf_addstr(sb, GIT_COLOR_RESET);
1282 return 7; /* consumed 7 bytes, "C(auto)" */
1283 } else {
1284 int ret = parse_color(sb, placeholder, c);
1285 if (ret)
1286 c->auto_color = 0;
1287 /*
1288 * Otherwise, we decided to treat %C<unknown>
1289 * as a literal string, and the previous
1290 * %C(auto) is still valid.
1291 */
1292 return ret;
1293 }
1294 case 'w':
1295 if (placeholder[1] == '(') {
1296 unsigned long width = 0, indent1 = 0, indent2 = 0;
1297 char *next;
1298 const char *start = placeholder + 2;
1299 const char *end = strchr(start, ')');
1300 if (!end)
1301 return 0;
1302 if (end > start) {
1303 width = strtoul(start, &next, 10);
1304 if (*next == ',') {
1305 indent1 = strtoul(next + 1, &next, 10);
1306 if (*next == ',') {
1307 indent2 = strtoul(next + 1,
1308 &next, 10);
1309 }
1310 }
1311 if (*next != ')')
1312 return 0;
1313 }
1314
1315 /*
1316 * We need to limit the format here as it allows the
1317 * user to prepend arbitrarily many bytes to the buffer
1318 * when rewrapping.
1319 */
1320 if (width > FORMATTING_LIMIT ||
1321 indent1 > FORMATTING_LIMIT ||
1322 indent2 > FORMATTING_LIMIT)
1323 return 0;
1324 rewrap_message_tail(sb, c, width, indent1, indent2);
1325 return end - placeholder + 1;
1326 } else
1327 return 0;
1328
1329 case '<':
1330 case '>':
1331 return parse_padding_placeholder(placeholder, c);
1332 }
1333
1334 if (skip_prefix(placeholder, "(describe", &arg)) {
1335 struct child_process cmd = CHILD_PROCESS_INIT;
1336 struct strbuf out = STRBUF_INIT;
1337 struct strbuf err = STRBUF_INIT;
1338 struct pretty_print_describe_status *describe_status;
1339
1340 describe_status = c->pretty_ctx->describe_status;
1341 if (describe_status) {
1342 if (!describe_status->max_invocations)
1343 return 0;
1344 describe_status->max_invocations--;
1345 }
1346
1347 cmd.git_cmd = 1;
1348 strvec_push(&cmd.args, "describe");
1349
1350 if (*arg == ':') {
1351 arg++;
1352 arg += parse_describe_args(arg, &cmd.args);
1353 }
1354
1355 if (*arg != ')') {
1356 child_process_clear(&cmd);
1357 return 0;
1358 }
1359
1360 strvec_push(&cmd.args, oid_to_hex(&commit->object.oid));
1361 pipe_command(&cmd, NULL, 0, &out, 0, &err, 0);
1362 strbuf_rtrim(&out);
1363 strbuf_addbuf(sb, &out);
1364 strbuf_release(&out);
1365 strbuf_release(&err);
1366 return arg - placeholder + 1;
1367 }
1368
1369 /* these depend on the commit */
1370 if (!commit->object.parsed)
1371 parse_object(the_repository, &commit->object.oid);
1372
1373 switch (placeholder[0]) {
1374 case 'H': /* commit hash */
1375 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_COMMIT));
1376 strbuf_addstr(sb, oid_to_hex(&commit->object.oid));
1377 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_RESET));
1378 return 1;
1379 case 'h': /* abbreviated commit hash */
1380 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_COMMIT));
1381 strbuf_add_unique_abbrev(sb, &commit->object.oid,
1382 c->pretty_ctx->abbrev);
1383 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_RESET));
1384 return 1;
1385 case 'T': /* tree hash */
1386 strbuf_addstr(sb, oid_to_hex(get_commit_tree_oid(commit)));
1387 return 1;
1388 case 't': /* abbreviated tree hash */
1389 strbuf_add_unique_abbrev(sb,
1390 get_commit_tree_oid(commit),
1391 c->pretty_ctx->abbrev);
1392 return 1;
1393 case 'P': /* parent hashes */
1394 for (p = commit->parents; p; p = p->next) {
1395 if (p != commit->parents)
1396 strbuf_addch(sb, ' ');
1397 strbuf_addstr(sb, oid_to_hex(&p->item->object.oid));
1398 }
1399 return 1;
1400 case 'p': /* abbreviated parent hashes */
1401 for (p = commit->parents; p; p = p->next) {
1402 if (p != commit->parents)
1403 strbuf_addch(sb, ' ');
1404 strbuf_add_unique_abbrev(sb, &p->item->object.oid,
1405 c->pretty_ctx->abbrev);
1406 }
1407 return 1;
1408 case 'm': /* left/right/bottom */
1409 strbuf_addstr(sb, get_revision_mark(NULL, commit));
1410 return 1;
1411 case 'd':
1412 format_decorations(sb, commit, c->auto_color);
1413 return 1;
1414 case 'D':
1415 format_decorations_extended(sb, commit, c->auto_color, "", ", ", "");
1416 return 1;
1417 case 'S': /* tag/branch like --source */
1418 if (!(c->pretty_ctx->rev && c->pretty_ctx->rev->sources))
1419 return 0;
1420 slot = revision_sources_at(c->pretty_ctx->rev->sources, commit);
1421 if (!(slot && *slot))
1422 return 0;
1423 strbuf_addstr(sb, *slot);
1424 return 1;
1425 case 'g': /* reflog info */
1426 switch(placeholder[1]) {
1427 case 'd': /* reflog selector */
1428 case 'D':
1429 if (c->pretty_ctx->reflog_info)
1430 get_reflog_selector(sb,
1431 c->pretty_ctx->reflog_info,
1432 &c->pretty_ctx->date_mode,
1433 c->pretty_ctx->date_mode_explicit,
1434 (placeholder[1] == 'd'));
1435 return 2;
1436 case 's': /* reflog message */
1437 if (c->pretty_ctx->reflog_info)
1438 get_reflog_message(sb, c->pretty_ctx->reflog_info);
1439 return 2;
1440 case 'n':
1441 case 'N':
1442 case 'e':
1443 case 'E':
1444 return format_reflog_person(sb,
1445 placeholder[1],
1446 c->pretty_ctx->reflog_info,
1447 &c->pretty_ctx->date_mode);
1448 }
1449 return 0; /* unknown %g placeholder */
1450 case 'N':
1451 if (c->pretty_ctx->notes_message) {
1452 strbuf_addstr(sb, c->pretty_ctx->notes_message);
1453 return 1;
1454 }
1455 return 0;
1456 }
1457
1458 if (placeholder[0] == 'G') {
1459 if (!c->signature_check.result)
1460 check_commit_signature(c->commit, &(c->signature_check));
1461 switch (placeholder[1]) {
1462 case 'G':
1463 if (c->signature_check.gpg_output)
1464 strbuf_addstr(sb, c->signature_check.gpg_output);
1465 break;
1466 case '?':
1467 switch (c->signature_check.result) {
1468 case 'G':
1469 switch (c->signature_check.trust_level) {
1470 case TRUST_UNDEFINED:
1471 case TRUST_NEVER:
1472 strbuf_addch(sb, 'U');
1473 break;
1474 default:
1475 strbuf_addch(sb, 'G');
1476 break;
1477 }
1478 break;
1479 case 'B':
1480 case 'E':
1481 case 'N':
1482 case 'X':
1483 case 'Y':
1484 case 'R':
1485 strbuf_addch(sb, c->signature_check.result);
1486 }
1487 break;
1488 case 'S':
1489 if (c->signature_check.signer)
1490 strbuf_addstr(sb, c->signature_check.signer);
1491 break;
1492 case 'K':
1493 if (c->signature_check.key)
1494 strbuf_addstr(sb, c->signature_check.key);
1495 break;
1496 case 'F':
1497 if (c->signature_check.fingerprint)
1498 strbuf_addstr(sb, c->signature_check.fingerprint);
1499 break;
1500 case 'P':
1501 if (c->signature_check.primary_key_fingerprint)
1502 strbuf_addstr(sb, c->signature_check.primary_key_fingerprint);
1503 break;
1504 case 'T':
1505 switch (c->signature_check.trust_level) {
1506 case TRUST_UNDEFINED:
1507 strbuf_addstr(sb, "undefined");
1508 break;
1509 case TRUST_NEVER:
1510 strbuf_addstr(sb, "never");
1511 break;
1512 case TRUST_MARGINAL:
1513 strbuf_addstr(sb, "marginal");
1514 break;
1515 case TRUST_FULLY:
1516 strbuf_addstr(sb, "fully");
1517 break;
1518 case TRUST_ULTIMATE:
1519 strbuf_addstr(sb, "ultimate");
1520 break;
1521 }
1522 break;
1523 default:
1524 return 0;
1525 }
1526 return 2;
1527 }
1528
1529 /* For the rest we have to parse the commit header. */
1530 if (!c->commit_header_parsed) {
1531 msg = c->message =
1532 repo_logmsg_reencode(c->repository, commit,
1533 &c->commit_encoding, "UTF-8");
1534 parse_commit_header(c);
1535 }
1536
1537 switch (placeholder[0]) {
1538 case 'a': /* author ... */
1539 return format_person_part(sb, placeholder[1],
1540 msg + c->author.off, c->author.len,
1541 &c->pretty_ctx->date_mode);
1542 case 'c': /* committer ... */
1543 return format_person_part(sb, placeholder[1],
1544 msg + c->committer.off, c->committer.len,
1545 &c->pretty_ctx->date_mode);
1546 case 'e': /* encoding */
1547 if (c->commit_encoding)
1548 strbuf_addstr(sb, c->commit_encoding);
1549 return 1;
1550 case 'B': /* raw body */
1551 /* message_off is always left at the initial newline */
1552 strbuf_addstr(sb, msg + c->message_off + 1);
1553 return 1;
1554 }
1555
1556 /* Now we need to parse the commit message. */
1557 if (!c->commit_message_parsed)
1558 parse_commit_message(c);
1559
1560 switch (placeholder[0]) {
1561 case 's': /* subject */
1562 format_subject(sb, msg + c->subject_off, " ");
1563 return 1;
1564 case 'f': /* sanitized subject */
1565 eol = strchrnul(msg + c->subject_off, '\n');
1566 format_sanitized_subject(sb, msg + c->subject_off, eol - (msg + c->subject_off));
1567 return 1;
1568 case 'b': /* body */
1569 strbuf_addstr(sb, msg + c->body_off);
1570 return 1;
1571 }
1572
1573 if (skip_prefix(placeholder, "(trailers", &arg)) {
1574 struct process_trailer_options opts = PROCESS_TRAILER_OPTIONS_INIT;
1575 struct string_list filter_list = STRING_LIST_INIT_NODUP;
1576 struct strbuf sepbuf = STRBUF_INIT;
1577 struct strbuf kvsepbuf = STRBUF_INIT;
1578 size_t ret = 0;
1579
1580 opts.no_divider = 1;
1581
1582 if (*arg == ':') {
1583 arg++;
1584 if (format_set_trailers_options(&opts, &filter_list, &sepbuf, &kvsepbuf, &arg, NULL))
1585 goto trailer_out;
1586 }
1587 if (*arg == ')') {
1588 format_trailers_from_commit(sb, msg + c->subject_off, &opts);
1589 ret = arg - placeholder + 1;
1590 }
1591 trailer_out:
1592 string_list_clear(&filter_list, 0);
1593 strbuf_release(&sepbuf);
1594 return ret;
1595 }
1596
1597 return 0; /* unknown placeholder */
1598 }
1599
1600 static size_t format_and_pad_commit(struct strbuf *sb, /* in UTF-8 */
1601 const char *placeholder,
1602 struct format_commit_context *c)
1603 {
1604 struct strbuf local_sb = STRBUF_INIT;
1605 size_t total_consumed = 0;
1606 int len, padding = c->padding;
1607
1608 if (padding < 0) {
1609 const char *start = strrchr(sb->buf, '\n');
1610 int occupied;
1611 if (!start)
1612 start = sb->buf;
1613 occupied = utf8_strnwidth(start, strlen(start), 1);
1614 occupied += c->pretty_ctx->graph_width;
1615 padding = (-padding) - occupied;
1616 }
1617 while (1) {
1618 int modifier = *placeholder == 'C';
1619 size_t consumed = format_commit_one(&local_sb, placeholder, c);
1620 total_consumed += consumed;
1621
1622 if (!modifier)
1623 break;
1624
1625 placeholder += consumed;
1626 if (*placeholder != '%')
1627 break;
1628 placeholder++;
1629 total_consumed++;
1630 }
1631 len = utf8_strnwidth(local_sb.buf, local_sb.len, 1);
1632
1633 if (c->flush_type == flush_left_and_steal) {
1634 const char *ch = sb->buf + sb->len - 1;
1635 while (len > padding && ch > sb->buf) {
1636 const char *p;
1637 if (*ch == ' ') {
1638 ch--;
1639 padding++;
1640 continue;
1641 }
1642 /* check for trailing ansi sequences */
1643 if (*ch != 'm')
1644 break;
1645 p = ch - 1;
1646 while (p > sb->buf && ch - p < 10 && *p != '\033')
1647 p--;
1648 if (*p != '\033' ||
1649 ch + 1 - p != display_mode_esc_sequence_len(p))
1650 break;
1651 /*
1652 * got a good ansi sequence, put it back to
1653 * local_sb as we're cutting sb
1654 */
1655 strbuf_insert(&local_sb, 0, p, ch + 1 - p);
1656 ch = p - 1;
1657 }
1658 strbuf_setlen(sb, ch + 1 - sb->buf);
1659 c->flush_type = flush_left;
1660 }
1661
1662 if (len > padding) {
1663 switch (c->truncate) {
1664 case trunc_left:
1665 strbuf_utf8_replace(&local_sb,
1666 0, len - (padding - 2),
1667 "..");
1668 break;
1669 case trunc_middle:
1670 strbuf_utf8_replace(&local_sb,
1671 padding / 2 - 1,
1672 len - (padding - 2),
1673 "..");
1674 break;
1675 case trunc_right:
1676 strbuf_utf8_replace(&local_sb,
1677 padding - 2, len - (padding - 2),
1678 "..");
1679 break;
1680 case trunc_none:
1681 break;
1682 }
1683 strbuf_addbuf(sb, &local_sb);
1684 } else {
1685 size_t sb_len = sb->len, offset = 0;
1686 if (c->flush_type == flush_left)
1687 offset = padding - len;
1688 else if (c->flush_type == flush_both)
1689 offset = (padding - len) / 2;
1690 /*
1691 * we calculate padding in columns, now
1692 * convert it back to chars
1693 */
1694 padding = padding - len + local_sb.len;
1695 strbuf_addchars(sb, ' ', padding);
1696 memcpy(sb->buf + sb_len + offset, local_sb.buf,
1697 local_sb.len);
1698 }
1699 strbuf_release(&local_sb);
1700 c->flush_type = no_flush;
1701 return total_consumed;
1702 }
1703
1704 static size_t format_commit_item(struct strbuf *sb, /* in UTF-8 */
1705 const char *placeholder,
1706 void *context)
1707 {
1708 size_t consumed, orig_len;
1709 enum {
1710 NO_MAGIC,
1711 ADD_LF_BEFORE_NON_EMPTY,
1712 DEL_LF_BEFORE_EMPTY,
1713 ADD_SP_BEFORE_NON_EMPTY
1714 } magic = NO_MAGIC;
1715
1716 switch (placeholder[0]) {
1717 case '-':
1718 magic = DEL_LF_BEFORE_EMPTY;
1719 break;
1720 case '+':
1721 magic = ADD_LF_BEFORE_NON_EMPTY;
1722 break;
1723 case ' ':
1724 magic = ADD_SP_BEFORE_NON_EMPTY;
1725 break;
1726 default:
1727 break;
1728 }
1729 if (magic != NO_MAGIC) {
1730 placeholder++;
1731
1732 switch (placeholder[0]) {
1733 case 'w':
1734 /*
1735 * `%+w()` cannot ever expand to a non-empty string,
1736 * and it potentially changes the layout of preceding
1737 * contents. We're thus not able to handle the magic in
1738 * this combination and refuse the pattern.
1739 */
1740 return 0;
1741 };
1742 }
1743
1744 orig_len = sb->len;
1745 if (((struct format_commit_context *)context)->flush_type != no_flush)
1746 consumed = format_and_pad_commit(sb, placeholder, context);
1747 else
1748 consumed = format_commit_one(sb, placeholder, context);
1749 if (magic == NO_MAGIC)
1750 return consumed;
1751
1752 if ((orig_len == sb->len) && magic == DEL_LF_BEFORE_EMPTY) {
1753 while (sb->len && sb->buf[sb->len - 1] == '\n')
1754 strbuf_setlen(sb, sb->len - 1);
1755 } else if (orig_len != sb->len) {
1756 if (magic == ADD_LF_BEFORE_NON_EMPTY)
1757 strbuf_insertstr(sb, orig_len, "\n");
1758 else if (magic == ADD_SP_BEFORE_NON_EMPTY)
1759 strbuf_insertstr(sb, orig_len, " ");
1760 }
1761 return consumed + 1;
1762 }
1763
1764 static size_t userformat_want_item(struct strbuf *sb, const char *placeholder,
1765 void *context)
1766 {
1767 struct userformat_want *w = context;
1768
1769 if (*placeholder == '+' || *placeholder == '-' || *placeholder == ' ')
1770 placeholder++;
1771
1772 switch (*placeholder) {
1773 case 'N':
1774 w->notes = 1;
1775 break;
1776 case 'S':
1777 w->source = 1;
1778 break;
1779 case 'd':
1780 case 'D':
1781 w->decorate = 1;
1782 break;
1783 }
1784 return 0;
1785 }
1786
1787 void userformat_find_requirements(const char *fmt, struct userformat_want *w)
1788 {
1789 struct strbuf dummy = STRBUF_INIT;
1790
1791 if (!fmt) {
1792 if (!user_format)
1793 return;
1794 fmt = user_format;
1795 }
1796 strbuf_expand(&dummy, fmt, userformat_want_item, w);
1797 strbuf_release(&dummy);
1798 }
1799
1800 void repo_format_commit_message(struct repository *r,
1801 const struct commit *commit,
1802 const char *format, struct strbuf *sb,
1803 const struct pretty_print_context *pretty_ctx)
1804 {
1805 struct format_commit_context context = {
1806 .repository = r,
1807 .commit = commit,
1808 .pretty_ctx = pretty_ctx,
1809 .wrap_start = sb->len
1810 };
1811 const char *output_enc = pretty_ctx->output_encoding;
1812 const char *utf8 = "UTF-8";
1813
1814 strbuf_expand(sb, format, format_commit_item, &context);
1815 rewrap_message_tail(sb, &context, 0, 0, 0);
1816
1817 /*
1818 * Convert output to an actual output encoding; note that
1819 * format_commit_item() will always use UTF-8, so we don't
1820 * have to bother if that's what the output wants.
1821 */
1822 if (output_enc) {
1823 if (same_encoding(utf8, output_enc))
1824 output_enc = NULL;
1825 } else {
1826 if (context.commit_encoding &&
1827 !same_encoding(context.commit_encoding, utf8))
1828 output_enc = context.commit_encoding;
1829 }
1830
1831 if (output_enc) {
1832 size_t outsz;
1833 char *out = reencode_string_len(sb->buf, sb->len,
1834 output_enc, utf8, &outsz);
1835 if (out)
1836 strbuf_attach(sb, out, outsz, outsz + 1);
1837 }
1838
1839 free(context.commit_encoding);
1840 repo_unuse_commit_buffer(r, commit, context.message);
1841 }
1842
1843 static void pp_header(struct pretty_print_context *pp,
1844 const char *encoding,
1845 const struct commit *commit,
1846 const char **msg_p,
1847 struct strbuf *sb)
1848 {
1849 int parents_shown = 0;
1850
1851 for (;;) {
1852 const char *name, *line = *msg_p;
1853 int linelen = get_one_line(*msg_p);
1854
1855 if (!linelen)
1856 return;
1857 *msg_p += linelen;
1858
1859 if (linelen == 1)
1860 /* End of header */
1861 return;
1862
1863 if (pp->fmt == CMIT_FMT_RAW) {
1864 strbuf_add(sb, line, linelen);
1865 continue;
1866 }
1867
1868 if (starts_with(line, "parent ")) {
1869 if (linelen != the_hash_algo->hexsz + 8)
1870 die("bad parent line in commit");
1871 continue;
1872 }
1873
1874 if (!parents_shown) {
1875 unsigned num = commit_list_count(commit->parents);
1876 /* with enough slop */
1877 strbuf_grow(sb, num * (GIT_MAX_HEXSZ + 10) + 20);
1878 add_merge_info(pp, sb, commit);
1879 parents_shown = 1;
1880 }
1881
1882 /*
1883 * MEDIUM == DEFAULT shows only author with dates.
1884 * FULL shows both authors but not dates.
1885 * FULLER shows both authors and dates.
1886 */
1887 if (skip_prefix(line, "author ", &name)) {
1888 strbuf_grow(sb, linelen + 80);
1889 pp_user_info(pp, "Author", sb, name, encoding);
1890 }
1891 if (skip_prefix(line, "committer ", &name) &&
1892 (pp->fmt == CMIT_FMT_FULL || pp->fmt == CMIT_FMT_FULLER)) {
1893 strbuf_grow(sb, linelen + 80);
1894 pp_user_info(pp, "Commit", sb, name, encoding);
1895 }
1896 }
1897 }
1898
1899 void pp_title_line(struct pretty_print_context *pp,
1900 const char **msg_p,
1901 struct strbuf *sb,
1902 const char *encoding,
1903 int need_8bit_cte)
1904 {
1905 static const int max_length = 78; /* per rfc2047 */
1906 struct strbuf title;
1907
1908 strbuf_init(&title, 80);
1909 *msg_p = format_subject(&title, *msg_p,
1910 pp->preserve_subject ? "\n" : " ");
1911
1912 strbuf_grow(sb, title.len + 1024);
1913 if (pp->print_email_subject) {
1914 if (pp->rev)
1915 fmt_output_email_subject(sb, pp->rev);
1916 if (pp->encode_email_headers &&
1917 needs_rfc2047_encoding(title.buf, title.len))
1918 add_rfc2047(sb, title.buf, title.len,
1919 encoding, RFC2047_SUBJECT);
1920 else
1921 strbuf_add_wrapped_bytes(sb, title.buf, title.len,
1922 -last_line_length(sb), 1, max_length);
1923 } else {
1924 strbuf_addbuf(sb, &title);
1925 }
1926 strbuf_addch(sb, '\n');
1927
1928 if (need_8bit_cte == 0) {
1929 int i;
1930 for (i = 0; i < pp->in_body_headers.nr; i++) {
1931 if (has_non_ascii(pp->in_body_headers.items[i].string)) {
1932 need_8bit_cte = 1;
1933 break;
1934 }
1935 }
1936 }
1937
1938 if (need_8bit_cte > 0) {
1939 const char *header_fmt =
1940 "MIME-Version: 1.0\n"
1941 "Content-Type: text/plain; charset=%s\n"
1942 "Content-Transfer-Encoding: 8bit\n";
1943 strbuf_addf(sb, header_fmt, encoding);
1944 }
1945 if (pp->after_subject) {
1946 strbuf_addstr(sb, pp->after_subject);
1947 }
1948 if (cmit_fmt_is_mail(pp->fmt)) {
1949 strbuf_addch(sb, '\n');
1950 }
1951
1952 if (pp->in_body_headers.nr) {
1953 int i;
1954 for (i = 0; i < pp->in_body_headers.nr; i++) {
1955 strbuf_addstr(sb, pp->in_body_headers.items[i].string);
1956 free(pp->in_body_headers.items[i].string);
1957 }
1958 string_list_clear(&pp->in_body_headers, 0);
1959 strbuf_addch(sb, '\n');
1960 }
1961
1962 strbuf_release(&title);
1963 }
1964
1965 static int pp_utf8_width(const char *start, const char *end)
1966 {
1967 int width = 0;
1968 size_t remain = end - start;
1969
1970 while (remain) {
1971 int n = utf8_width(&start, &remain);
1972 if (n < 0 || !start)
1973 return -1;
1974 width += n;
1975 }
1976 return width;
1977 }
1978
1979 static void strbuf_add_tabexpand(struct strbuf *sb, int tabwidth,
1980 const char *line, int linelen)
1981 {
1982 const char *tab;
1983
1984 while ((tab = memchr(line, '\t', linelen)) != NULL) {
1985 int width = pp_utf8_width(line, tab);
1986
1987 /*
1988 * If it wasn't well-formed utf8, or it
1989 * had characters with badly defined
1990 * width (control characters etc), just
1991 * give up on trying to align things.
1992 */
1993 if (width < 0)
1994 break;
1995
1996 /* Output the data .. */
1997 strbuf_add(sb, line, tab - line);
1998
1999 /* .. and the de-tabified tab */
2000 strbuf_addchars(sb, ' ', tabwidth - (width % tabwidth));
2001
2002 /* Skip over the printed part .. */
2003 linelen -= tab + 1 - line;
2004 line = tab + 1;
2005 }
2006
2007 /*
2008 * Print out everything after the last tab without
2009 * worrying about width - there's nothing more to
2010 * align.
2011 */
2012 strbuf_add(sb, line, linelen);
2013 }
2014
2015 /*
2016 * pp_handle_indent() prints out the intendation, and
2017 * the whole line (without the final newline), after
2018 * de-tabifying.
2019 */
2020 static void pp_handle_indent(struct pretty_print_context *pp,
2021 struct strbuf *sb, int indent,
2022 const char *line, int linelen)
2023 {
2024 strbuf_addchars(sb, ' ', indent);
2025 if (pp->expand_tabs_in_log)
2026 strbuf_add_tabexpand(sb, pp->expand_tabs_in_log, line, linelen);
2027 else
2028 strbuf_add(sb, line, linelen);
2029 }
2030
2031 static int is_mboxrd_from(const char *line, int len)
2032 {
2033 /*
2034 * a line matching /^From $/ here would only have len == 4
2035 * at this point because is_empty_line would've trimmed all
2036 * trailing space
2037 */
2038 return len > 4 && starts_with(line + strspn(line, ">"), "From ");
2039 }
2040
2041 void pp_remainder(struct pretty_print_context *pp,
2042 const char **msg_p,
2043 struct strbuf *sb,
2044 int indent)
2045 {
2046 int first = 1;
2047 for (;;) {
2048 const char *line = *msg_p;
2049 int linelen = get_one_line(line);
2050 *msg_p += linelen;
2051
2052 if (!linelen)
2053 break;
2054
2055 if (is_blank_line(line, &linelen)) {
2056 if (first)
2057 continue;
2058 if (pp->fmt == CMIT_FMT_SHORT)
2059 break;
2060 }
2061 first = 0;
2062
2063 strbuf_grow(sb, linelen + indent + 20);
2064 if (indent)
2065 pp_handle_indent(pp, sb, indent, line, linelen);
2066 else if (pp->expand_tabs_in_log)
2067 strbuf_add_tabexpand(sb, pp->expand_tabs_in_log,
2068 line, linelen);
2069 else {
2070 if (pp->fmt == CMIT_FMT_MBOXRD &&
2071 is_mboxrd_from(line, linelen))
2072 strbuf_addch(sb, '>');
2073
2074 strbuf_add(sb, line, linelen);
2075 }
2076 strbuf_addch(sb, '\n');
2077 }
2078 }
2079
2080 void pretty_print_commit(struct pretty_print_context *pp,
2081 const struct commit *commit,
2082 struct strbuf *sb)
2083 {
2084 unsigned long beginning_of_body;
2085 int indent = 4;
2086 const char *msg;
2087 const char *reencoded;
2088 const char *encoding;
2089 int need_8bit_cte = pp->need_8bit_cte;
2090
2091 if (pp->fmt == CMIT_FMT_USERFORMAT) {
2092 format_commit_message(commit, user_format, sb, pp);
2093 return;
2094 }
2095
2096 encoding = get_log_output_encoding();
2097 msg = reencoded = logmsg_reencode(commit, NULL, encoding);
2098
2099 if (pp->fmt == CMIT_FMT_ONELINE || cmit_fmt_is_mail(pp->fmt))
2100 indent = 0;
2101
2102 /*
2103 * We need to check and emit Content-type: to mark it
2104 * as 8-bit if we haven't done so.
2105 */
2106 if (cmit_fmt_is_mail(pp->fmt) && need_8bit_cte == 0) {
2107 int i, ch, in_body;
2108
2109 for (in_body = i = 0; (ch = msg[i]); i++) {
2110 if (!in_body) {
2111 /* author could be non 7-bit ASCII but
2112 * the log may be so; skip over the
2113 * header part first.
2114 */
2115 if (ch == '\n' && msg[i+1] == '\n')
2116 in_body = 1;
2117 }
2118 else if (non_ascii(ch)) {
2119 need_8bit_cte = 1;
2120 break;
2121 }
2122 }
2123 }
2124
2125 pp_header(pp, encoding, commit, &msg, sb);
2126 if (pp->fmt != CMIT_FMT_ONELINE && !pp->print_email_subject) {
2127 strbuf_addch(sb, '\n');
2128 }
2129
2130 /* Skip excess blank lines at the beginning of body, if any... */
2131 msg = skip_blank_lines(msg);
2132
2133 /* These formats treat the title line specially. */
2134 if (pp->fmt == CMIT_FMT_ONELINE || cmit_fmt_is_mail(pp->fmt))
2135 pp_title_line(pp, &msg, sb, encoding, need_8bit_cte);
2136
2137 beginning_of_body = sb->len;
2138 if (pp->fmt != CMIT_FMT_ONELINE)
2139 pp_remainder(pp, &msg, sb, indent);
2140 strbuf_rtrim(sb);
2141
2142 /* Make sure there is an EOLN for the non-oneline case */
2143 if (pp->fmt != CMIT_FMT_ONELINE)
2144 strbuf_addch(sb, '\n');
2145
2146 /*
2147 * The caller may append additional body text in e-mail
2148 * format. Make sure we did not strip the blank line
2149 * between the header and the body.
2150 */
2151 if (cmit_fmt_is_mail(pp->fmt) && sb->len <= beginning_of_body)
2152 strbuf_addch(sb, '\n');
2153
2154 unuse_commit_buffer(commit, reencoded);
2155 }
2156
2157 void pp_commit_easy(enum cmit_fmt fmt, const struct commit *commit,
2158 struct strbuf *sb)
2159 {
2160 struct pretty_print_context pp = {0};
2161 pp.fmt = fmt;
2162 pretty_print_commit(&pp, commit, sb);
2163 }
git for Debian