fsck: mark unused parameters in various fsck callbacks
[git.git] / pretty.c
blob7862be105d937d64749bda07b7f3f147e6621a98
1 #include "git-compat-util.h"
2 #include "alloc.h"
3 #include "config.h"
4 #include "commit.h"
5 #include "environment.h"
6 #include "gettext.h"
7 #include "hash.h"
8 #include "hex.h"
9 #include "utf8.h"
10 #include "diff.h"
11 #include "pager.h"
12 #include "revision.h"
13 #include "string-list.h"
14 #include "mailmap.h"
15 #include "log-tree.h"
16 #include "notes.h"
17 #include "color.h"
18 #include "reflog-walk.h"
19 #include "gpg-interface.h"
20 #include "trailer.h"
21 #include "run-command.h"
22 #include "object-name.h"
25 * The limit for formatting directives, which enable the caller to append
26 * arbitrarily many bytes to the formatted buffer. This includes padding
27 * and wrapping formatters.
29 #define FORMATTING_LIMIT (16 * 1024)
31 static char *user_format;
32 static struct cmt_fmt_map {
33 const char *name;
34 enum cmit_fmt format;
35 int is_tformat;
36 int expand_tabs_in_log;
37 int is_alias;
38 enum date_mode_type default_date_mode_type;
39 const char *user_format;
40 } *commit_formats;
41 static size_t builtin_formats_len;
42 static size_t commit_formats_len;
43 static size_t commit_formats_alloc;
44 static struct cmt_fmt_map *find_commit_format(const char *sought);
46 int commit_format_is_empty(enum cmit_fmt fmt)
48 return fmt == CMIT_FMT_USERFORMAT && !*user_format;
51 static void save_user_format(struct rev_info *rev, const char *cp, int is_tformat)
53 free(user_format);
54 user_format = xstrdup(cp);
55 if (is_tformat)
56 rev->use_terminator = 1;
57 rev->commit_format = CMIT_FMT_USERFORMAT;
60 static int git_pretty_formats_config(const char *var, const char *value,
61 const struct config_context *ctx UNUSED,
62 void *cb UNUSED)
64 struct cmt_fmt_map *commit_format = NULL;
65 const char *name;
66 const char *fmt;
67 int i;
69 if (!skip_prefix(var, "pretty.", &name))
70 return 0;
72 for (i = 0; i < builtin_formats_len; i++) {
73 if (!strcmp(commit_formats[i].name, name))
74 return 0;
77 for (i = builtin_formats_len; i < commit_formats_len; i++) {
78 if (!strcmp(commit_formats[i].name, name)) {
79 commit_format = &commit_formats[i];
80 break;
84 if (!commit_format) {
85 ALLOC_GROW(commit_formats, commit_formats_len+1,
86 commit_formats_alloc);
87 commit_format = &commit_formats[commit_formats_len];
88 memset(commit_format, 0, sizeof(*commit_format));
89 commit_formats_len++;
92 commit_format->name = xstrdup(name);
93 commit_format->format = CMIT_FMT_USERFORMAT;
94 if (git_config_string(&fmt, var, value))
95 return -1;
97 if (skip_prefix(fmt, "format:", &fmt))
98 commit_format->is_tformat = 0;
99 else if (skip_prefix(fmt, "tformat:", &fmt) || strchr(fmt, '%'))
100 commit_format->is_tformat = 1;
101 else
102 commit_format->is_alias = 1;
103 commit_format->user_format = fmt;
105 return 0;
108 static void setup_commit_formats(void)
110 struct cmt_fmt_map builtin_formats[] = {
111 { "raw", CMIT_FMT_RAW, 0, 0 },
112 { "medium", CMIT_FMT_MEDIUM, 0, 8 },
113 { "short", CMIT_FMT_SHORT, 0, 0 },
114 { "email", CMIT_FMT_EMAIL, 0, 0 },
115 { "mboxrd", CMIT_FMT_MBOXRD, 0, 0 },
116 { "fuller", CMIT_FMT_FULLER, 0, 8 },
117 { "full", CMIT_FMT_FULL, 0, 8 },
118 { "oneline", CMIT_FMT_ONELINE, 1, 0 },
119 { "reference", CMIT_FMT_USERFORMAT, 1, 0,
120 0, DATE_SHORT, "%C(auto)%h (%s, %ad)" },
122 * Please update $__git_log_pretty_formats in
123 * git-completion.bash when you add new formats.
126 commit_formats_len = ARRAY_SIZE(builtin_formats);
127 builtin_formats_len = commit_formats_len;
128 ALLOC_GROW(commit_formats, commit_formats_len, commit_formats_alloc);
129 COPY_ARRAY(commit_formats, builtin_formats,
130 ARRAY_SIZE(builtin_formats));
132 git_config(git_pretty_formats_config, NULL);
135 static struct cmt_fmt_map *find_commit_format_recursive(const char *sought,
136 const char *original,
137 int num_redirections)
139 struct cmt_fmt_map *found = NULL;
140 size_t found_match_len = 0;
141 int i;
143 if (num_redirections >= commit_formats_len)
144 die("invalid --pretty format: "
145 "'%s' references an alias which points to itself",
146 original);
148 for (i = 0; i < commit_formats_len; i++) {
149 size_t match_len;
151 if (!starts_with(commit_formats[i].name, sought))
152 continue;
154 match_len = strlen(commit_formats[i].name);
155 if (found == NULL || found_match_len > match_len) {
156 found = &commit_formats[i];
157 found_match_len = match_len;
161 if (found && found->is_alias) {
162 found = find_commit_format_recursive(found->user_format,
163 original,
164 num_redirections+1);
167 return found;
170 static struct cmt_fmt_map *find_commit_format(const char *sought)
172 if (!commit_formats)
173 setup_commit_formats();
175 return find_commit_format_recursive(sought, sought, 0);
178 void get_commit_format(const char *arg, struct rev_info *rev)
180 struct cmt_fmt_map *commit_format;
182 rev->use_terminator = 0;
183 if (!arg) {
184 rev->commit_format = CMIT_FMT_DEFAULT;
185 return;
187 if (skip_prefix(arg, "format:", &arg)) {
188 save_user_format(rev, arg, 0);
189 return;
192 if (!*arg || skip_prefix(arg, "tformat:", &arg) || strchr(arg, '%')) {
193 save_user_format(rev, arg, 1);
194 return;
197 commit_format = find_commit_format(arg);
198 if (!commit_format)
199 die("invalid --pretty format: %s", arg);
201 rev->commit_format = commit_format->format;
202 rev->use_terminator = commit_format->is_tformat;
203 rev->expand_tabs_in_log_default = commit_format->expand_tabs_in_log;
204 if (!rev->date_mode_explicit && commit_format->default_date_mode_type)
205 rev->date_mode.type = commit_format->default_date_mode_type;
206 if (commit_format->format == CMIT_FMT_USERFORMAT) {
207 save_user_format(rev, commit_format->user_format,
208 commit_format->is_tformat);
213 * Generic support for pretty-printing the header
215 static int get_one_line(const char *msg)
217 int ret = 0;
219 for (;;) {
220 char c = *msg++;
221 if (!c)
222 break;
223 ret++;
224 if (c == '\n')
225 break;
227 return ret;
230 /* High bit set, or ISO-2022-INT */
231 static int non_ascii(int ch)
233 return !isascii(ch) || ch == '\033';
236 int has_non_ascii(const char *s)
238 int ch;
239 if (!s)
240 return 0;
241 while ((ch = *s++) != '\0') {
242 if (non_ascii(ch))
243 return 1;
245 return 0;
248 static int is_rfc822_special(char ch)
250 switch (ch) {
251 case '(':
252 case ')':
253 case '<':
254 case '>':
255 case '[':
256 case ']':
257 case ':':
258 case ';':
259 case '@':
260 case ',':
261 case '.':
262 case '"':
263 case '\\':
264 return 1;
265 default:
266 return 0;
270 static int needs_rfc822_quoting(const char *s, int len)
272 int i;
273 for (i = 0; i < len; i++)
274 if (is_rfc822_special(s[i]))
275 return 1;
276 return 0;
279 static int last_line_length(struct strbuf *sb)
281 int i;
283 /* How many bytes are already used on the last line? */
284 for (i = sb->len - 1; i >= 0; i--)
285 if (sb->buf[i] == '\n')
286 break;
287 return sb->len - (i + 1);
290 static void add_rfc822_quoted(struct strbuf *out, const char *s, int len)
292 int i;
294 /* just a guess, we may have to also backslash-quote */
295 strbuf_grow(out, len + 2);
297 strbuf_addch(out, '"');
298 for (i = 0; i < len; i++) {
299 switch (s[i]) {
300 case '"':
301 case '\\':
302 strbuf_addch(out, '\\');
303 /* fall through */
304 default:
305 strbuf_addch(out, s[i]);
308 strbuf_addch(out, '"');
311 enum rfc2047_type {
312 RFC2047_SUBJECT,
313 RFC2047_ADDRESS
316 static int is_rfc2047_special(char ch, enum rfc2047_type type)
319 * rfc2047, section 4.2:
321 * 8-bit values which correspond to printable ASCII characters other
322 * than "=", "?", and "_" (underscore), MAY be represented as those
323 * characters. (But see section 5 for restrictions.) In
324 * particular, SPACE and TAB MUST NOT be represented as themselves
325 * within encoded words.
329 * rule out non-ASCII characters and non-printable characters (the
330 * non-ASCII check should be redundant as isprint() is not localized
331 * and only knows about ASCII, but be defensive about that)
333 if (non_ascii(ch) || !isprint(ch))
334 return 1;
337 * rule out special printable characters (' ' should be the only
338 * whitespace character considered printable, but be defensive and use
339 * isspace())
341 if (isspace(ch) || ch == '=' || ch == '?' || ch == '_')
342 return 1;
345 * rfc2047, section 5.3:
347 * As a replacement for a 'word' entity within a 'phrase', for example,
348 * one that precedes an address in a From, To, or Cc header. The ABNF
349 * definition for 'phrase' from RFC 822 thus becomes:
351 * phrase = 1*( encoded-word / word )
353 * In this case the set of characters that may be used in a "Q"-encoded
354 * 'encoded-word' is restricted to: <upper and lower case ASCII
355 * letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
356 * (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
357 * 'phrase' MUST be separated from any adjacent 'word', 'text' or
358 * 'special' by 'linear-white-space'.
361 if (type != RFC2047_ADDRESS)
362 return 0;
364 /* '=' and '_' are special cases and have been checked above */
365 return !(isalnum(ch) || ch == '!' || ch == '*' || ch == '+' || ch == '-' || ch == '/');
368 static int needs_rfc2047_encoding(const char *line, int len)
370 int i;
372 for (i = 0; i < len; i++) {
373 int ch = line[i];
374 if (non_ascii(ch) || ch == '\n')
375 return 1;
376 if ((i + 1 < len) && (ch == '=' && line[i+1] == '?'))
377 return 1;
380 return 0;
383 static void add_rfc2047(struct strbuf *sb, const char *line, size_t len,
384 const char *encoding, enum rfc2047_type type)
386 static const int max_encoded_length = 76; /* per rfc2047 */
387 int i;
388 int line_len = last_line_length(sb);
390 strbuf_grow(sb, len * 3 + strlen(encoding) + 100);
391 strbuf_addf(sb, "=?%s?q?", encoding);
392 line_len += strlen(encoding) + 5; /* 5 for =??q? */
394 while (len) {
396 * RFC 2047, section 5 (3):
398 * Each 'encoded-word' MUST represent an integral number of
399 * characters. A multi-octet character may not be split across
400 * adjacent 'encoded- word's.
402 const unsigned char *p = (const unsigned char *)line;
403 int chrlen = mbs_chrlen(&line, &len, encoding);
404 int is_special = (chrlen > 1) || is_rfc2047_special(*p, type);
406 /* "=%02X" * chrlen, or the byte itself */
407 const char *encoded_fmt = is_special ? "=%02X" : "%c";
408 int encoded_len = is_special ? 3 * chrlen : 1;
411 * According to RFC 2047, we could encode the special character
412 * ' ' (space) with '_' (underscore) for readability. But many
413 * programs do not understand this and just leave the
414 * underscore in place. Thus, we do nothing special here, which
415 * causes ' ' to be encoded as '=20', avoiding this problem.
418 if (line_len + encoded_len + 2 > max_encoded_length) {
419 /* It won't fit with trailing "?=" --- break the line */
420 strbuf_addf(sb, "?=\n =?%s?q?", encoding);
421 line_len = strlen(encoding) + 5 + 1; /* =??q? plus SP */
424 for (i = 0; i < chrlen; i++)
425 strbuf_addf(sb, encoded_fmt, p[i]);
426 line_len += encoded_len;
428 strbuf_addstr(sb, "?=");
431 const char *show_ident_date(const struct ident_split *ident,
432 const struct date_mode *mode)
434 timestamp_t date = 0;
435 long tz = 0;
437 if (ident->date_begin && ident->date_end)
438 date = parse_timestamp(ident->date_begin, NULL, 10);
439 if (date_overflows(date))
440 date = 0;
441 else {
442 if (ident->tz_begin && ident->tz_end)
443 tz = strtol(ident->tz_begin, NULL, 10);
444 if (tz >= INT_MAX || tz <= INT_MIN)
445 tz = 0;
447 return show_date(date, tz, mode);
450 static inline void strbuf_add_with_color(struct strbuf *sb, const char *color,
451 const char *buf, size_t buflen)
453 strbuf_addstr(sb, color);
454 strbuf_add(sb, buf, buflen);
455 if (*color)
456 strbuf_addstr(sb, GIT_COLOR_RESET);
459 static void append_line_with_color(struct strbuf *sb, struct grep_opt *opt,
460 const char *line, size_t linelen,
461 int color, enum grep_context ctx,
462 enum grep_header_field field)
464 const char *buf, *eol, *line_color, *match_color;
465 regmatch_t match;
466 int eflags = 0;
468 buf = line;
469 eol = buf + linelen;
471 if (!opt || !want_color(color) || opt->invert)
472 goto end;
474 line_color = opt->colors[GREP_COLOR_SELECTED];
475 match_color = opt->colors[GREP_COLOR_MATCH_SELECTED];
477 while (grep_next_match(opt, buf, eol, ctx, &match, field, eflags)) {
478 if (match.rm_so == match.rm_eo)
479 break;
481 strbuf_add_with_color(sb, line_color, buf, match.rm_so);
482 strbuf_add_with_color(sb, match_color, buf + match.rm_so,
483 match.rm_eo - match.rm_so);
484 buf += match.rm_eo;
485 eflags = REG_NOTBOL;
488 if (eflags)
489 strbuf_add_with_color(sb, line_color, buf, eol - buf);
490 else {
491 end:
492 strbuf_add(sb, buf, eol - buf);
496 static int use_in_body_from(const struct pretty_print_context *pp,
497 const struct ident_split *ident)
499 if (pp->rev && pp->rev->force_in_body_from)
500 return 1;
501 if (ident_cmp(pp->from_ident, ident))
502 return 1;
503 return 0;
506 void pp_user_info(struct pretty_print_context *pp,
507 const char *what, struct strbuf *sb,
508 const char *line, const char *encoding)
510 struct ident_split ident;
511 char *line_end;
512 const char *mailbuf, *namebuf;
513 size_t namelen, maillen;
514 int max_length = 78; /* per rfc2822 */
516 if (pp->fmt == CMIT_FMT_ONELINE)
517 return;
519 line_end = strchrnul(line, '\n');
520 if (split_ident_line(&ident, line, line_end - line))
521 return;
523 mailbuf = ident.mail_begin;
524 maillen = ident.mail_end - ident.mail_begin;
525 namebuf = ident.name_begin;
526 namelen = ident.name_end - ident.name_begin;
528 if (pp->mailmap)
529 map_user(pp->mailmap, &mailbuf, &maillen, &namebuf, &namelen);
531 if (cmit_fmt_is_mail(pp->fmt)) {
532 if (pp->from_ident && use_in_body_from(pp, &ident)) {
533 struct strbuf buf = STRBUF_INIT;
535 strbuf_addstr(&buf, "From: ");
536 strbuf_add(&buf, namebuf, namelen);
537 strbuf_addstr(&buf, " <");
538 strbuf_add(&buf, mailbuf, maillen);
539 strbuf_addstr(&buf, ">\n");
540 string_list_append(&pp->in_body_headers,
541 strbuf_detach(&buf, NULL));
543 mailbuf = pp->from_ident->mail_begin;
544 maillen = pp->from_ident->mail_end - mailbuf;
545 namebuf = pp->from_ident->name_begin;
546 namelen = pp->from_ident->name_end - namebuf;
549 strbuf_addstr(sb, "From: ");
550 if (pp->encode_email_headers &&
551 needs_rfc2047_encoding(namebuf, namelen)) {
552 add_rfc2047(sb, namebuf, namelen,
553 encoding, RFC2047_ADDRESS);
554 max_length = 76; /* per rfc2047 */
555 } else if (needs_rfc822_quoting(namebuf, namelen)) {
556 struct strbuf quoted = STRBUF_INIT;
557 add_rfc822_quoted(&quoted, namebuf, namelen);
558 strbuf_add_wrapped_bytes(sb, quoted.buf, quoted.len,
559 -6, 1, max_length);
560 strbuf_release(&quoted);
561 } else {
562 strbuf_add_wrapped_bytes(sb, namebuf, namelen,
563 -6, 1, max_length);
566 if (max_length <
567 last_line_length(sb) + strlen(" <") + maillen + strlen(">"))
568 strbuf_addch(sb, '\n');
569 strbuf_addf(sb, " <%.*s>\n", (int)maillen, mailbuf);
570 } else {
571 struct strbuf id = STRBUF_INIT;
572 enum grep_header_field field = GREP_HEADER_FIELD_MAX;
573 struct grep_opt *opt = pp->rev ? &pp->rev->grep_filter : NULL;
575 if (!strcmp(what, "Author"))
576 field = GREP_HEADER_AUTHOR;
577 else if (!strcmp(what, "Commit"))
578 field = GREP_HEADER_COMMITTER;
580 strbuf_addf(sb, "%s: ", what);
581 if (pp->fmt == CMIT_FMT_FULLER)
582 strbuf_addchars(sb, ' ', 4);
584 strbuf_addf(&id, "%.*s <%.*s>", (int)namelen, namebuf,
585 (int)maillen, mailbuf);
587 append_line_with_color(sb, opt, id.buf, id.len, pp->color,
588 GREP_CONTEXT_HEAD, field);
589 strbuf_addch(sb, '\n');
590 strbuf_release(&id);
593 switch (pp->fmt) {
594 case CMIT_FMT_MEDIUM:
595 strbuf_addf(sb, "Date: %s\n",
596 show_ident_date(&ident, &pp->date_mode));
597 break;
598 case CMIT_FMT_EMAIL:
599 case CMIT_FMT_MBOXRD:
600 strbuf_addf(sb, "Date: %s\n",
601 show_ident_date(&ident, DATE_MODE(RFC2822)));
602 break;
603 case CMIT_FMT_FULLER:
604 strbuf_addf(sb, "%sDate: %s\n", what,
605 show_ident_date(&ident, &pp->date_mode));
606 break;
607 default:
608 /* notin' */
609 break;
613 static int is_blank_line(const char *line, int *len_p)
615 int len = *len_p;
616 while (len && isspace(line[len - 1]))
617 len--;
618 *len_p = len;
619 return !len;
622 const char *skip_blank_lines(const char *msg)
624 for (;;) {
625 int linelen = get_one_line(msg);
626 int ll = linelen;
627 if (!linelen)
628 break;
629 if (!is_blank_line(msg, &ll))
630 break;
631 msg += linelen;
633 return msg;
636 static void add_merge_info(const struct pretty_print_context *pp,
637 struct strbuf *sb, const struct commit *commit)
639 struct commit_list *parent = commit->parents;
641 if ((pp->fmt == CMIT_FMT_ONELINE) || (cmit_fmt_is_mail(pp->fmt)) ||
642 !parent || !parent->next)
643 return;
645 strbuf_addstr(sb, "Merge:");
647 while (parent) {
648 struct object_id *oidp = &parent->item->object.oid;
649 strbuf_addch(sb, ' ');
650 if (pp->abbrev)
651 strbuf_add_unique_abbrev(sb, oidp, pp->abbrev);
652 else
653 strbuf_addstr(sb, oid_to_hex(oidp));
654 parent = parent->next;
656 strbuf_addch(sb, '\n');
659 static char *get_header(const char *msg, const char *key)
661 size_t len;
662 const char *v = find_commit_header(msg, key, &len);
663 return v ? xmemdupz(v, len) : NULL;
666 static char *replace_encoding_header(char *buf, const char *encoding)
668 struct strbuf tmp = STRBUF_INIT;
669 size_t start, len;
670 char *cp = buf;
672 /* guess if there is an encoding header before a \n\n */
673 while (!starts_with(cp, "encoding ")) {
674 cp = strchr(cp, '\n');
675 if (!cp || *++cp == '\n')
676 return buf;
678 start = cp - buf;
679 cp = strchr(cp, '\n');
680 if (!cp)
681 return buf; /* should not happen but be defensive */
682 len = cp + 1 - (buf + start);
684 strbuf_attach(&tmp, buf, strlen(buf), strlen(buf) + 1);
685 if (is_encoding_utf8(encoding)) {
686 /* we have re-coded to UTF-8; drop the header */
687 strbuf_remove(&tmp, start, len);
688 } else {
689 /* just replaces XXXX in 'encoding XXXX\n' */
690 strbuf_splice(&tmp, start + strlen("encoding "),
691 len - strlen("encoding \n"),
692 encoding, strlen(encoding));
694 return strbuf_detach(&tmp, NULL);
697 const char *repo_logmsg_reencode(struct repository *r,
698 const struct commit *commit,
699 char **commit_encoding,
700 const char *output_encoding)
702 static const char *utf8 = "UTF-8";
703 const char *use_encoding;
704 char *encoding;
705 const char *msg = repo_get_commit_buffer(r, commit, NULL);
706 char *out;
708 if (!output_encoding || !*output_encoding) {
709 if (commit_encoding)
710 *commit_encoding = get_header(msg, "encoding");
711 return msg;
713 encoding = get_header(msg, "encoding");
714 if (commit_encoding)
715 *commit_encoding = encoding;
716 use_encoding = encoding ? encoding : utf8;
717 if (same_encoding(use_encoding, output_encoding)) {
719 * No encoding work to be done. If we have no encoding header
720 * at all, then there's nothing to do, and we can return the
721 * message verbatim (whether newly allocated or not).
723 if (!encoding)
724 return msg;
727 * Otherwise, we still want to munge the encoding header in the
728 * result, which will be done by modifying the buffer. If we
729 * are using a fresh copy, we can reuse it. But if we are using
730 * the cached copy from repo_get_commit_buffer, we need to duplicate it
731 * to avoid munging the cached copy.
733 if (msg == get_cached_commit_buffer(r, commit, NULL))
734 out = xstrdup(msg);
735 else
736 out = (char *)msg;
738 else {
740 * There's actual encoding work to do. Do the reencoding, which
741 * still leaves the header to be replaced in the next step. At
742 * this point, we are done with msg. If we allocated a fresh
743 * copy, we can free it.
745 out = reencode_string(msg, output_encoding, use_encoding);
746 if (out)
747 repo_unuse_commit_buffer(r, commit, msg);
751 * This replacement actually consumes the buffer we hand it, so we do
752 * not have to worry about freeing the old "out" here.
754 if (out)
755 out = replace_encoding_header(out, output_encoding);
757 if (!commit_encoding)
758 free(encoding);
760 * If the re-encoding failed, out might be NULL here; in that
761 * case we just return the commit message verbatim.
763 return out ? out : msg;
766 static int mailmap_name(const char **email, size_t *email_len,
767 const char **name, size_t *name_len)
769 static struct string_list *mail_map;
770 if (!mail_map) {
771 CALLOC_ARRAY(mail_map, 1);
772 read_mailmap(mail_map);
774 return mail_map->nr && map_user(mail_map, email, email_len, name, name_len);
777 static size_t format_person_part(struct strbuf *sb, char part,
778 const char *msg, int len,
779 const struct date_mode *dmode)
781 /* currently all placeholders have same length */
782 const int placeholder_len = 2;
783 struct ident_split s;
784 const char *name, *mail;
785 size_t maillen, namelen;
787 if (split_ident_line(&s, msg, len) < 0)
788 goto skip;
790 name = s.name_begin;
791 namelen = s.name_end - s.name_begin;
792 mail = s.mail_begin;
793 maillen = s.mail_end - s.mail_begin;
795 if (part == 'N' || part == 'E' || part == 'L') /* mailmap lookup */
796 mailmap_name(&mail, &maillen, &name, &namelen);
797 if (part == 'n' || part == 'N') { /* name */
798 strbuf_add(sb, name, namelen);
799 return placeholder_len;
801 if (part == 'e' || part == 'E') { /* email */
802 strbuf_add(sb, mail, maillen);
803 return placeholder_len;
805 if (part == 'l' || part == 'L') { /* local-part */
806 const char *at = memchr(mail, '@', maillen);
807 if (at)
808 maillen = at - mail;
809 strbuf_add(sb, mail, maillen);
810 return placeholder_len;
813 if (!s.date_begin)
814 goto skip;
816 if (part == 't') { /* date, UNIX timestamp */
817 strbuf_add(sb, s.date_begin, s.date_end - s.date_begin);
818 return placeholder_len;
821 switch (part) {
822 case 'd': /* date */
823 strbuf_addstr(sb, show_ident_date(&s, dmode));
824 return placeholder_len;
825 case 'D': /* date, RFC2822 style */
826 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(RFC2822)));
827 return placeholder_len;
828 case 'r': /* date, relative */
829 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(RELATIVE)));
830 return placeholder_len;
831 case 'i': /* date, ISO 8601-like */
832 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(ISO8601)));
833 return placeholder_len;
834 case 'I': /* date, ISO 8601 strict */
835 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(ISO8601_STRICT)));
836 return placeholder_len;
837 case 'h': /* date, human */
838 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(HUMAN)));
839 return placeholder_len;
840 case 's':
841 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(SHORT)));
842 return placeholder_len;
845 skip:
847 * reading from either a bogus commit, or a reflog entry with
848 * %gn, %ge, etc.; 'sb' cannot be updated, but we still need
849 * to compute a valid return value.
851 if (part == 'n' || part == 'e' || part == 't' || part == 'd'
852 || part == 'D' || part == 'r' || part == 'i')
853 return placeholder_len;
855 return 0; /* unknown placeholder */
858 struct chunk {
859 size_t off;
860 size_t len;
863 enum flush_type {
864 no_flush,
865 flush_right,
866 flush_left,
867 flush_left_and_steal,
868 flush_both
871 enum trunc_type {
872 trunc_none,
873 trunc_left,
874 trunc_middle,
875 trunc_right
878 struct format_commit_context {
879 struct repository *repository;
880 const struct commit *commit;
881 const struct pretty_print_context *pretty_ctx;
882 unsigned commit_header_parsed:1;
883 unsigned commit_message_parsed:1;
884 struct signature_check signature_check;
885 enum flush_type flush_type;
886 enum trunc_type truncate;
887 const char *message;
888 char *commit_encoding;
889 size_t width, indent1, indent2;
890 int auto_color;
891 int padding;
893 /* These offsets are relative to the start of the commit message. */
894 struct chunk author;
895 struct chunk committer;
896 size_t message_off;
897 size_t subject_off;
898 size_t body_off;
900 /* The following ones are relative to the result struct strbuf. */
901 size_t wrap_start;
904 static void parse_commit_header(struct format_commit_context *context)
906 const char *msg = context->message;
907 int i;
909 for (i = 0; msg[i]; i++) {
910 const char *name;
911 int eol;
912 for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
913 ; /* do nothing */
915 if (i == eol) {
916 break;
917 } else if (skip_prefix(msg + i, "author ", &name)) {
918 context->author.off = name - msg;
919 context->author.len = msg + eol - name;
920 } else if (skip_prefix(msg + i, "committer ", &name)) {
921 context->committer.off = name - msg;
922 context->committer.len = msg + eol - name;
924 i = eol;
926 context->message_off = i;
927 context->commit_header_parsed = 1;
930 static int istitlechar(char c)
932 return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') ||
933 (c >= '0' && c <= '9') || c == '.' || c == '_';
936 void format_sanitized_subject(struct strbuf *sb, const char *msg, size_t len)
938 size_t trimlen;
939 size_t start_len = sb->len;
940 int space = 2;
941 int i;
943 for (i = 0; i < len; i++) {
944 if (istitlechar(msg[i])) {
945 if (space == 1)
946 strbuf_addch(sb, '-');
947 space = 0;
948 strbuf_addch(sb, msg[i]);
949 if (msg[i] == '.')
950 while (msg[i+1] == '.')
951 i++;
952 } else
953 space |= 1;
956 /* trim any trailing '.' or '-' characters */
957 trimlen = 0;
958 while (sb->len - trimlen > start_len &&
959 (sb->buf[sb->len - 1 - trimlen] == '.'
960 || sb->buf[sb->len - 1 - trimlen] == '-'))
961 trimlen++;
962 strbuf_remove(sb, sb->len - trimlen, trimlen);
965 const char *format_subject(struct strbuf *sb, const char *msg,
966 const char *line_separator)
968 int first = 1;
970 for (;;) {
971 const char *line = msg;
972 int linelen = get_one_line(line);
974 msg += linelen;
975 if (!linelen || is_blank_line(line, &linelen))
976 break;
978 if (!sb)
979 continue;
980 strbuf_grow(sb, linelen + 2);
981 if (!first)
982 strbuf_addstr(sb, line_separator);
983 strbuf_add(sb, line, linelen);
984 first = 0;
986 return msg;
989 static void parse_commit_message(struct format_commit_context *c)
991 const char *msg = c->message + c->message_off;
992 const char *start = c->message;
994 msg = skip_blank_lines(msg);
995 c->subject_off = msg - start;
997 msg = format_subject(NULL, msg, NULL);
998 msg = skip_blank_lines(msg);
999 c->body_off = msg - start;
1001 c->commit_message_parsed = 1;
1004 static void strbuf_wrap(struct strbuf *sb, size_t pos,
1005 size_t width, size_t indent1, size_t indent2)
1007 struct strbuf tmp = STRBUF_INIT;
1009 if (pos)
1010 strbuf_add(&tmp, sb->buf, pos);
1011 strbuf_add_wrapped_text(&tmp, sb->buf + pos,
1012 cast_size_t_to_int(indent1),
1013 cast_size_t_to_int(indent2),
1014 cast_size_t_to_int(width));
1015 strbuf_swap(&tmp, sb);
1016 strbuf_release(&tmp);
1019 static void rewrap_message_tail(struct strbuf *sb,
1020 struct format_commit_context *c,
1021 size_t new_width, size_t new_indent1,
1022 size_t new_indent2)
1024 if (c->width == new_width && c->indent1 == new_indent1 &&
1025 c->indent2 == new_indent2)
1026 return;
1027 if (c->wrap_start < sb->len)
1028 strbuf_wrap(sb, c->wrap_start, c->width, c->indent1, c->indent2);
1029 c->wrap_start = sb->len;
1030 c->width = new_width;
1031 c->indent1 = new_indent1;
1032 c->indent2 = new_indent2;
1035 static int format_reflog_person(struct strbuf *sb,
1036 char part,
1037 struct reflog_walk_info *log,
1038 const struct date_mode *dmode)
1040 const char *ident;
1042 if (!log)
1043 return 2;
1045 ident = get_reflog_ident(log);
1046 if (!ident)
1047 return 2;
1049 return format_person_part(sb, part, ident, strlen(ident), dmode);
1052 static size_t parse_color(struct strbuf *sb, /* in UTF-8 */
1053 const char *placeholder,
1054 struct format_commit_context *c)
1056 const char *rest = placeholder;
1057 const char *basic_color = NULL;
1059 if (placeholder[1] == '(') {
1060 const char *begin = placeholder + 2;
1061 const char *end = strchr(begin, ')');
1062 char color[COLOR_MAXLEN];
1064 if (!end)
1065 return 0;
1067 if (skip_prefix(begin, "auto,", &begin)) {
1068 if (!want_color(c->pretty_ctx->color))
1069 return end - placeholder + 1;
1070 } else if (skip_prefix(begin, "always,", &begin)) {
1071 /* nothing to do; we do not respect want_color at all */
1072 } else {
1073 /* the default is the same as "auto" */
1074 if (!want_color(c->pretty_ctx->color))
1075 return end - placeholder + 1;
1078 if (color_parse_mem(begin, end - begin, color) < 0)
1079 die(_("unable to parse --pretty format"));
1080 strbuf_addstr(sb, color);
1081 return end - placeholder + 1;
1085 * We handle things like "%C(red)" above; for historical reasons, there
1086 * are a few colors that can be specified without parentheses (and
1087 * they cannot support things like "auto" or "always" at all).
1089 if (skip_prefix(placeholder + 1, "red", &rest))
1090 basic_color = GIT_COLOR_RED;
1091 else if (skip_prefix(placeholder + 1, "green", &rest))
1092 basic_color = GIT_COLOR_GREEN;
1093 else if (skip_prefix(placeholder + 1, "blue", &rest))
1094 basic_color = GIT_COLOR_BLUE;
1095 else if (skip_prefix(placeholder + 1, "reset", &rest))
1096 basic_color = GIT_COLOR_RESET;
1098 if (basic_color && want_color(c->pretty_ctx->color))
1099 strbuf_addstr(sb, basic_color);
1101 return rest - placeholder;
1104 static size_t parse_padding_placeholder(const char *placeholder,
1105 struct format_commit_context *c)
1107 const char *ch = placeholder;
1108 enum flush_type flush_type;
1109 int to_column = 0;
1111 switch (*ch++) {
1112 case '<':
1113 flush_type = flush_right;
1114 break;
1115 case '>':
1116 if (*ch == '<') {
1117 flush_type = flush_both;
1118 ch++;
1119 } else if (*ch == '>') {
1120 flush_type = flush_left_and_steal;
1121 ch++;
1122 } else
1123 flush_type = flush_left;
1124 break;
1125 default:
1126 return 0;
1129 /* the next value means "wide enough to that column" */
1130 if (*ch == '|') {
1131 to_column = 1;
1132 ch++;
1135 if (*ch == '(') {
1136 const char *start = ch + 1;
1137 const char *end = start + strcspn(start, ",)");
1138 char *next;
1139 int width;
1140 if (!*end || end == start)
1141 return 0;
1142 width = strtol(start, &next, 10);
1145 * We need to limit the amount of padding, or otherwise this
1146 * would allow the user to pad the buffer by arbitrarily many
1147 * bytes and thus cause resource exhaustion.
1149 if (width < -FORMATTING_LIMIT || width > FORMATTING_LIMIT)
1150 return 0;
1152 if (next == start || width == 0)
1153 return 0;
1154 if (width < 0) {
1155 if (to_column)
1156 width += term_columns();
1157 if (width < 0)
1158 return 0;
1160 c->padding = to_column ? -width : width;
1161 c->flush_type = flush_type;
1163 if (*end == ',') {
1164 start = end + 1;
1165 end = strchr(start, ')');
1166 if (!end || end == start)
1167 return 0;
1168 if (starts_with(start, "trunc)"))
1169 c->truncate = trunc_right;
1170 else if (starts_with(start, "ltrunc)"))
1171 c->truncate = trunc_left;
1172 else if (starts_with(start, "mtrunc)"))
1173 c->truncate = trunc_middle;
1174 else
1175 return 0;
1176 } else
1177 c->truncate = trunc_none;
1179 return end - placeholder + 1;
1181 return 0;
1184 static int match_placeholder_arg_value(const char *to_parse, const char *candidate,
1185 const char **end, const char **valuestart,
1186 size_t *valuelen)
1188 const char *p;
1190 if (!(skip_prefix(to_parse, candidate, &p)))
1191 return 0;
1192 if (valuestart) {
1193 if (*p == '=') {
1194 *valuestart = p + 1;
1195 *valuelen = strcspn(*valuestart, ",)");
1196 p = *valuestart + *valuelen;
1197 } else {
1198 if (*p != ',' && *p != ')')
1199 return 0;
1200 *valuestart = NULL;
1201 *valuelen = 0;
1204 if (*p == ',') {
1205 *end = p + 1;
1206 return 1;
1208 if (*p == ')') {
1209 *end = p;
1210 return 1;
1212 return 0;
1215 static int match_placeholder_bool_arg(const char *to_parse, const char *candidate,
1216 const char **end, int *val)
1218 const char *argval;
1219 char *strval;
1220 size_t arglen;
1221 int v;
1223 if (!match_placeholder_arg_value(to_parse, candidate, end, &argval, &arglen))
1224 return 0;
1226 if (!argval) {
1227 *val = 1;
1228 return 1;
1231 strval = xstrndup(argval, arglen);
1232 v = git_parse_maybe_bool(strval);
1233 free(strval);
1235 if (v == -1)
1236 return 0;
1238 *val = v;
1240 return 1;
1243 static int format_trailer_match_cb(const struct strbuf *key, void *ud)
1245 const struct string_list *list = ud;
1246 const struct string_list_item *item;
1248 for_each_string_list_item (item, list) {
1249 if (key->len == (uintptr_t)item->util &&
1250 !strncasecmp(item->string, key->buf, key->len))
1251 return 1;
1253 return 0;
1256 static struct strbuf *expand_separator(struct strbuf *sb,
1257 const char *argval, size_t arglen)
1259 char *fmt = xstrndup(argval, arglen);
1260 const char *format = fmt;
1262 strbuf_reset(sb);
1263 while (strbuf_expand_step(sb, &format)) {
1264 size_t len;
1266 if (skip_prefix(format, "%", &format))
1267 strbuf_addch(sb, '%');
1268 else if ((len = strbuf_expand_literal(sb, format)))
1269 format += len;
1270 else
1271 strbuf_addch(sb, '%');
1273 free(fmt);
1274 return sb;
1277 int format_set_trailers_options(struct process_trailer_options *opts,
1278 struct string_list *filter_list,
1279 struct strbuf *sepbuf,
1280 struct strbuf *kvsepbuf,
1281 const char **arg,
1282 char **invalid_arg)
1284 for (;;) {
1285 const char *argval;
1286 size_t arglen;
1288 if (**arg == ')')
1289 break;
1291 if (match_placeholder_arg_value(*arg, "key", arg, &argval, &arglen)) {
1292 uintptr_t len = arglen;
1294 if (!argval)
1295 return -1;
1297 if (len && argval[len - 1] == ':')
1298 len--;
1299 string_list_append(filter_list, argval)->util = (char *)len;
1301 opts->filter = format_trailer_match_cb;
1302 opts->filter_data = filter_list;
1303 opts->only_trailers = 1;
1304 } else if (match_placeholder_arg_value(*arg, "separator", arg, &argval, &arglen)) {
1305 opts->separator = expand_separator(sepbuf, argval, arglen);
1306 } else if (match_placeholder_arg_value(*arg, "key_value_separator", arg, &argval, &arglen)) {
1307 opts->key_value_separator = expand_separator(kvsepbuf, argval, arglen);
1308 } else if (!match_placeholder_bool_arg(*arg, "only", arg, &opts->only_trailers) &&
1309 !match_placeholder_bool_arg(*arg, "unfold", arg, &opts->unfold) &&
1310 !match_placeholder_bool_arg(*arg, "keyonly", arg, &opts->key_only) &&
1311 !match_placeholder_bool_arg(*arg, "valueonly", arg, &opts->value_only)) {
1312 if (invalid_arg) {
1313 size_t len = strcspn(*arg, ",)");
1314 *invalid_arg = xstrndup(*arg, len);
1316 return -1;
1319 return 0;
1322 static size_t parse_describe_args(const char *start, struct strvec *args)
1324 struct {
1325 char *name;
1326 enum {
1327 DESCRIBE_ARG_BOOL,
1328 DESCRIBE_ARG_INTEGER,
1329 DESCRIBE_ARG_STRING,
1330 } type;
1331 } option[] = {
1332 { "tags", DESCRIBE_ARG_BOOL},
1333 { "abbrev", DESCRIBE_ARG_INTEGER },
1334 { "exclude", DESCRIBE_ARG_STRING },
1335 { "match", DESCRIBE_ARG_STRING },
1337 const char *arg = start;
1339 for (;;) {
1340 int found = 0;
1341 const char *argval;
1342 size_t arglen = 0;
1343 int optval = 0;
1344 int i;
1346 for (i = 0; !found && i < ARRAY_SIZE(option); i++) {
1347 switch (option[i].type) {
1348 case DESCRIBE_ARG_BOOL:
1349 if (match_placeholder_bool_arg(arg, option[i].name, &arg, &optval)) {
1350 if (optval)
1351 strvec_pushf(args, "--%s", option[i].name);
1352 else
1353 strvec_pushf(args, "--no-%s", option[i].name);
1354 found = 1;
1356 break;
1357 case DESCRIBE_ARG_INTEGER:
1358 if (match_placeholder_arg_value(arg, option[i].name, &arg,
1359 &argval, &arglen)) {
1360 char *endptr;
1361 if (!arglen)
1362 return 0;
1363 strtol(argval, &endptr, 10);
1364 if (endptr - argval != arglen)
1365 return 0;
1366 strvec_pushf(args, "--%s=%.*s", option[i].name, (int)arglen, argval);
1367 found = 1;
1369 break;
1370 case DESCRIBE_ARG_STRING:
1371 if (match_placeholder_arg_value(arg, option[i].name, &arg,
1372 &argval, &arglen)) {
1373 if (!arglen)
1374 return 0;
1375 strvec_pushf(args, "--%s=%.*s", option[i].name, (int)arglen, argval);
1376 found = 1;
1378 break;
1381 if (!found)
1382 break;
1385 return arg - start;
1388 static size_t format_commit_one(struct strbuf *sb, /* in UTF-8 */
1389 const char *placeholder,
1390 void *context)
1392 struct format_commit_context *c = context;
1393 const struct commit *commit = c->commit;
1394 const char *msg = c->message;
1395 struct commit_list *p;
1396 const char *arg, *eol;
1397 size_t res;
1398 char **slot;
1400 /* these are independent of the commit */
1401 res = strbuf_expand_literal(sb, placeholder);
1402 if (res)
1403 return res;
1405 switch (placeholder[0]) {
1406 case 'C':
1407 if (starts_with(placeholder + 1, "(auto)")) {
1408 c->auto_color = want_color(c->pretty_ctx->color);
1409 if (c->auto_color && sb->len)
1410 strbuf_addstr(sb, GIT_COLOR_RESET);
1411 return 7; /* consumed 7 bytes, "C(auto)" */
1412 } else {
1413 int ret = parse_color(sb, placeholder, c);
1414 if (ret)
1415 c->auto_color = 0;
1417 * Otherwise, we decided to treat %C<unknown>
1418 * as a literal string, and the previous
1419 * %C(auto) is still valid.
1421 return ret;
1423 case 'w':
1424 if (placeholder[1] == '(') {
1425 unsigned long width = 0, indent1 = 0, indent2 = 0;
1426 char *next;
1427 const char *start = placeholder + 2;
1428 const char *end = strchr(start, ')');
1429 if (!end)
1430 return 0;
1431 if (end > start) {
1432 width = strtoul(start, &next, 10);
1433 if (*next == ',') {
1434 indent1 = strtoul(next + 1, &next, 10);
1435 if (*next == ',') {
1436 indent2 = strtoul(next + 1,
1437 &next, 10);
1440 if (*next != ')')
1441 return 0;
1445 * We need to limit the format here as it allows the
1446 * user to prepend arbitrarily many bytes to the buffer
1447 * when rewrapping.
1449 if (width > FORMATTING_LIMIT ||
1450 indent1 > FORMATTING_LIMIT ||
1451 indent2 > FORMATTING_LIMIT)
1452 return 0;
1453 rewrap_message_tail(sb, c, width, indent1, indent2);
1454 return end - placeholder + 1;
1455 } else
1456 return 0;
1458 case '<':
1459 case '>':
1460 return parse_padding_placeholder(placeholder, c);
1463 if (skip_prefix(placeholder, "(describe", &arg)) {
1464 struct child_process cmd = CHILD_PROCESS_INIT;
1465 struct strbuf out = STRBUF_INIT;
1466 struct strbuf err = STRBUF_INIT;
1467 struct pretty_print_describe_status *describe_status;
1469 describe_status = c->pretty_ctx->describe_status;
1470 if (describe_status) {
1471 if (!describe_status->max_invocations)
1472 return 0;
1473 describe_status->max_invocations--;
1476 cmd.git_cmd = 1;
1477 strvec_push(&cmd.args, "describe");
1479 if (*arg == ':') {
1480 arg++;
1481 arg += parse_describe_args(arg, &cmd.args);
1484 if (*arg != ')') {
1485 child_process_clear(&cmd);
1486 return 0;
1489 strvec_push(&cmd.args, oid_to_hex(&commit->object.oid));
1490 pipe_command(&cmd, NULL, 0, &out, 0, &err, 0);
1491 strbuf_rtrim(&out);
1492 strbuf_addbuf(sb, &out);
1493 strbuf_release(&out);
1494 strbuf_release(&err);
1495 return arg - placeholder + 1;
1498 /* these depend on the commit */
1499 if (!commit->object.parsed)
1500 parse_object(the_repository, &commit->object.oid);
1502 switch (placeholder[0]) {
1503 case 'H': /* commit hash */
1504 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_COMMIT));
1505 strbuf_addstr(sb, oid_to_hex(&commit->object.oid));
1506 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_RESET));
1507 return 1;
1508 case 'h': /* abbreviated commit hash */
1509 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_COMMIT));
1510 strbuf_add_unique_abbrev(sb, &commit->object.oid,
1511 c->pretty_ctx->abbrev);
1512 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_RESET));
1513 return 1;
1514 case 'T': /* tree hash */
1515 strbuf_addstr(sb, oid_to_hex(get_commit_tree_oid(commit)));
1516 return 1;
1517 case 't': /* abbreviated tree hash */
1518 strbuf_add_unique_abbrev(sb,
1519 get_commit_tree_oid(commit),
1520 c->pretty_ctx->abbrev);
1521 return 1;
1522 case 'P': /* parent hashes */
1523 for (p = commit->parents; p; p = p->next) {
1524 if (p != commit->parents)
1525 strbuf_addch(sb, ' ');
1526 strbuf_addstr(sb, oid_to_hex(&p->item->object.oid));
1528 return 1;
1529 case 'p': /* abbreviated parent hashes */
1530 for (p = commit->parents; p; p = p->next) {
1531 if (p != commit->parents)
1532 strbuf_addch(sb, ' ');
1533 strbuf_add_unique_abbrev(sb, &p->item->object.oid,
1534 c->pretty_ctx->abbrev);
1536 return 1;
1537 case 'm': /* left/right/bottom */
1538 strbuf_addstr(sb, get_revision_mark(NULL, commit));
1539 return 1;
1540 case 'd':
1541 format_decorations(sb, commit, c->auto_color);
1542 return 1;
1543 case 'D':
1544 format_decorations_extended(sb, commit, c->auto_color, "", ", ", "");
1545 return 1;
1546 case 'S': /* tag/branch like --source */
1547 if (!(c->pretty_ctx->rev && c->pretty_ctx->rev->sources))
1548 return 0;
1549 slot = revision_sources_at(c->pretty_ctx->rev->sources, commit);
1550 if (!(slot && *slot))
1551 return 0;
1552 strbuf_addstr(sb, *slot);
1553 return 1;
1554 case 'g': /* reflog info */
1555 switch(placeholder[1]) {
1556 case 'd': /* reflog selector */
1557 case 'D':
1558 if (c->pretty_ctx->reflog_info)
1559 get_reflog_selector(sb,
1560 c->pretty_ctx->reflog_info,
1561 &c->pretty_ctx->date_mode,
1562 c->pretty_ctx->date_mode_explicit,
1563 (placeholder[1] == 'd'));
1564 return 2;
1565 case 's': /* reflog message */
1566 if (c->pretty_ctx->reflog_info)
1567 get_reflog_message(sb, c->pretty_ctx->reflog_info);
1568 return 2;
1569 case 'n':
1570 case 'N':
1571 case 'e':
1572 case 'E':
1573 return format_reflog_person(sb,
1574 placeholder[1],
1575 c->pretty_ctx->reflog_info,
1576 &c->pretty_ctx->date_mode);
1578 return 0; /* unknown %g placeholder */
1579 case 'N':
1580 if (c->pretty_ctx->notes_message) {
1581 strbuf_addstr(sb, c->pretty_ctx->notes_message);
1582 return 1;
1584 return 0;
1587 if (placeholder[0] == 'G') {
1588 if (!c->signature_check.result)
1589 check_commit_signature(c->commit, &(c->signature_check));
1590 switch (placeholder[1]) {
1591 case 'G':
1592 if (c->signature_check.output)
1593 strbuf_addstr(sb, c->signature_check.output);
1594 break;
1595 case '?':
1596 switch (c->signature_check.result) {
1597 case 'G':
1598 switch (c->signature_check.trust_level) {
1599 case TRUST_UNDEFINED:
1600 case TRUST_NEVER:
1601 strbuf_addch(sb, 'U');
1602 break;
1603 default:
1604 strbuf_addch(sb, 'G');
1605 break;
1607 break;
1608 case 'B':
1609 case 'E':
1610 case 'N':
1611 case 'X':
1612 case 'Y':
1613 case 'R':
1614 strbuf_addch(sb, c->signature_check.result);
1616 break;
1617 case 'S':
1618 if (c->signature_check.signer)
1619 strbuf_addstr(sb, c->signature_check.signer);
1620 break;
1621 case 'K':
1622 if (c->signature_check.key)
1623 strbuf_addstr(sb, c->signature_check.key);
1624 break;
1625 case 'F':
1626 if (c->signature_check.fingerprint)
1627 strbuf_addstr(sb, c->signature_check.fingerprint);
1628 break;
1629 case 'P':
1630 if (c->signature_check.primary_key_fingerprint)
1631 strbuf_addstr(sb, c->signature_check.primary_key_fingerprint);
1632 break;
1633 case 'T':
1634 strbuf_addstr(sb, gpg_trust_level_to_str(c->signature_check.trust_level));
1635 break;
1636 default:
1637 return 0;
1639 return 2;
1642 /* For the rest we have to parse the commit header. */
1643 if (!c->commit_header_parsed) {
1644 msg = c->message =
1645 repo_logmsg_reencode(c->repository, commit,
1646 &c->commit_encoding, "UTF-8");
1647 parse_commit_header(c);
1650 switch (placeholder[0]) {
1651 case 'a': /* author ... */
1652 return format_person_part(sb, placeholder[1],
1653 msg + c->author.off, c->author.len,
1654 &c->pretty_ctx->date_mode);
1655 case 'c': /* committer ... */
1656 return format_person_part(sb, placeholder[1],
1657 msg + c->committer.off, c->committer.len,
1658 &c->pretty_ctx->date_mode);
1659 case 'e': /* encoding */
1660 if (c->commit_encoding)
1661 strbuf_addstr(sb, c->commit_encoding);
1662 return 1;
1663 case 'B': /* raw body */
1664 /* message_off is always left at the initial newline */
1665 strbuf_addstr(sb, msg + c->message_off + 1);
1666 return 1;
1669 /* Now we need to parse the commit message. */
1670 if (!c->commit_message_parsed)
1671 parse_commit_message(c);
1673 switch (placeholder[0]) {
1674 case 's': /* subject */
1675 format_subject(sb, msg + c->subject_off, " ");
1676 return 1;
1677 case 'f': /* sanitized subject */
1678 eol = strchrnul(msg + c->subject_off, '\n');
1679 format_sanitized_subject(sb, msg + c->subject_off, eol - (msg + c->subject_off));
1680 return 1;
1681 case 'b': /* body */
1682 strbuf_addstr(sb, msg + c->body_off);
1683 return 1;
1686 if (skip_prefix(placeholder, "(trailers", &arg)) {
1687 struct process_trailer_options opts = PROCESS_TRAILER_OPTIONS_INIT;
1688 struct string_list filter_list = STRING_LIST_INIT_NODUP;
1689 struct strbuf sepbuf = STRBUF_INIT;
1690 struct strbuf kvsepbuf = STRBUF_INIT;
1691 size_t ret = 0;
1693 opts.no_divider = 1;
1695 if (*arg == ':') {
1696 arg++;
1697 if (format_set_trailers_options(&opts, &filter_list, &sepbuf, &kvsepbuf, &arg, NULL))
1698 goto trailer_out;
1700 if (*arg == ')') {
1701 format_trailers_from_commit(sb, msg + c->subject_off, &opts);
1702 ret = arg - placeholder + 1;
1704 trailer_out:
1705 string_list_clear(&filter_list, 0);
1706 strbuf_release(&sepbuf);
1707 return ret;
1710 return 0; /* unknown placeholder */
1713 static size_t format_and_pad_commit(struct strbuf *sb, /* in UTF-8 */
1714 const char *placeholder,
1715 struct format_commit_context *c)
1717 struct strbuf local_sb = STRBUF_INIT;
1718 size_t total_consumed = 0;
1719 int len, padding = c->padding;
1721 if (padding < 0) {
1722 const char *start = strrchr(sb->buf, '\n');
1723 int occupied;
1724 if (!start)
1725 start = sb->buf;
1726 occupied = utf8_strnwidth(start, strlen(start), 1);
1727 occupied += c->pretty_ctx->graph_width;
1728 padding = (-padding) - occupied;
1730 while (1) {
1731 int modifier = *placeholder == 'C';
1732 size_t consumed = format_commit_one(&local_sb, placeholder, c);
1733 total_consumed += consumed;
1735 if (!modifier)
1736 break;
1738 placeholder += consumed;
1739 if (*placeholder != '%')
1740 break;
1741 placeholder++;
1742 total_consumed++;
1744 len = utf8_strnwidth(local_sb.buf, local_sb.len, 1);
1746 if (c->flush_type == flush_left_and_steal) {
1747 const char *ch = sb->buf + sb->len - 1;
1748 while (len > padding && ch > sb->buf) {
1749 const char *p;
1750 if (*ch == ' ') {
1751 ch--;
1752 padding++;
1753 continue;
1755 /* check for trailing ansi sequences */
1756 if (*ch != 'm')
1757 break;
1758 p = ch - 1;
1759 while (p > sb->buf && ch - p < 10 && *p != '\033')
1760 p--;
1761 if (*p != '\033' ||
1762 ch + 1 - p != display_mode_esc_sequence_len(p))
1763 break;
1765 * got a good ansi sequence, put it back to
1766 * local_sb as we're cutting sb
1768 strbuf_insert(&local_sb, 0, p, ch + 1 - p);
1769 ch = p - 1;
1771 strbuf_setlen(sb, ch + 1 - sb->buf);
1772 c->flush_type = flush_left;
1775 if (len > padding) {
1776 switch (c->truncate) {
1777 case trunc_left:
1778 strbuf_utf8_replace(&local_sb,
1779 0, len - (padding - 2),
1780 "..");
1781 break;
1782 case trunc_middle:
1783 strbuf_utf8_replace(&local_sb,
1784 padding / 2 - 1,
1785 len - (padding - 2),
1786 "..");
1787 break;
1788 case trunc_right:
1789 strbuf_utf8_replace(&local_sb,
1790 padding - 2, len - (padding - 2),
1791 "..");
1792 break;
1793 case trunc_none:
1794 break;
1796 strbuf_addbuf(sb, &local_sb);
1797 } else {
1798 size_t sb_len = sb->len, offset = 0;
1799 if (c->flush_type == flush_left)
1800 offset = padding - len;
1801 else if (c->flush_type == flush_both)
1802 offset = (padding - len) / 2;
1804 * we calculate padding in columns, now
1805 * convert it back to chars
1807 padding = padding - len + local_sb.len;
1808 strbuf_addchars(sb, ' ', padding);
1809 memcpy(sb->buf + sb_len + offset, local_sb.buf,
1810 local_sb.len);
1812 strbuf_release(&local_sb);
1813 c->flush_type = no_flush;
1814 return total_consumed;
1817 static size_t format_commit_item(struct strbuf *sb, /* in UTF-8 */
1818 const char *placeholder,
1819 struct format_commit_context *context)
1821 size_t consumed, orig_len;
1822 enum {
1823 NO_MAGIC,
1824 ADD_LF_BEFORE_NON_EMPTY,
1825 DEL_LF_BEFORE_EMPTY,
1826 ADD_SP_BEFORE_NON_EMPTY
1827 } magic = NO_MAGIC;
1829 switch (placeholder[0]) {
1830 case '-':
1831 magic = DEL_LF_BEFORE_EMPTY;
1832 break;
1833 case '+':
1834 magic = ADD_LF_BEFORE_NON_EMPTY;
1835 break;
1836 case ' ':
1837 magic = ADD_SP_BEFORE_NON_EMPTY;
1838 break;
1839 default:
1840 break;
1842 if (magic != NO_MAGIC) {
1843 placeholder++;
1845 switch (placeholder[0]) {
1846 case 'w':
1848 * `%+w()` cannot ever expand to a non-empty string,
1849 * and it potentially changes the layout of preceding
1850 * contents. We're thus not able to handle the magic in
1851 * this combination and refuse the pattern.
1853 return 0;
1857 orig_len = sb->len;
1858 if ((context)->flush_type != no_flush)
1859 consumed = format_and_pad_commit(sb, placeholder, context);
1860 else
1861 consumed = format_commit_one(sb, placeholder, context);
1862 if (magic == NO_MAGIC)
1863 return consumed;
1865 if ((orig_len == sb->len) && magic == DEL_LF_BEFORE_EMPTY) {
1866 while (sb->len && sb->buf[sb->len - 1] == '\n')
1867 strbuf_setlen(sb, sb->len - 1);
1868 } else if (orig_len != sb->len) {
1869 if (magic == ADD_LF_BEFORE_NON_EMPTY)
1870 strbuf_insertstr(sb, orig_len, "\n");
1871 else if (magic == ADD_SP_BEFORE_NON_EMPTY)
1872 strbuf_insertstr(sb, orig_len, " ");
1874 return consumed + 1;
1877 void userformat_find_requirements(const char *fmt, struct userformat_want *w)
1879 struct strbuf dummy = STRBUF_INIT;
1881 if (!fmt) {
1882 if (!user_format)
1883 return;
1884 fmt = user_format;
1886 while (strbuf_expand_step(&dummy, &fmt)) {
1887 if (skip_prefix(fmt, "%", &fmt))
1888 continue;
1890 if (*fmt == '+' || *fmt == '-' || *fmt == ' ')
1891 fmt++;
1893 switch (*fmt) {
1894 case 'N':
1895 w->notes = 1;
1896 break;
1897 case 'S':
1898 w->source = 1;
1899 break;
1900 case 'd':
1901 case 'D':
1902 w->decorate = 1;
1903 break;
1906 strbuf_release(&dummy);
1909 void repo_format_commit_message(struct repository *r,
1910 const struct commit *commit,
1911 const char *format, struct strbuf *sb,
1912 const struct pretty_print_context *pretty_ctx)
1914 struct format_commit_context context = {
1915 .repository = r,
1916 .commit = commit,
1917 .pretty_ctx = pretty_ctx,
1918 .wrap_start = sb->len
1920 const char *output_enc = pretty_ctx->output_encoding;
1921 const char *utf8 = "UTF-8";
1923 while (strbuf_expand_step(sb, &format)) {
1924 size_t len;
1926 if (skip_prefix(format, "%", &format))
1927 strbuf_addch(sb, '%');
1928 else if ((len = format_commit_item(sb, format, &context)))
1929 format += len;
1930 else
1931 strbuf_addch(sb, '%');
1933 rewrap_message_tail(sb, &context, 0, 0, 0);
1936 * Convert output to an actual output encoding; note that
1937 * format_commit_item() will always use UTF-8, so we don't
1938 * have to bother if that's what the output wants.
1940 if (output_enc) {
1941 if (same_encoding(utf8, output_enc))
1942 output_enc = NULL;
1943 } else {
1944 if (context.commit_encoding &&
1945 !same_encoding(context.commit_encoding, utf8))
1946 output_enc = context.commit_encoding;
1949 if (output_enc) {
1950 size_t outsz;
1951 char *out = reencode_string_len(sb->buf, sb->len,
1952 output_enc, utf8, &outsz);
1953 if (out)
1954 strbuf_attach(sb, out, outsz, outsz + 1);
1957 free(context.commit_encoding);
1958 repo_unuse_commit_buffer(r, commit, context.message);
1961 static void pp_header(struct pretty_print_context *pp,
1962 const char *encoding,
1963 const struct commit *commit,
1964 const char **msg_p,
1965 struct strbuf *sb)
1967 int parents_shown = 0;
1969 for (;;) {
1970 const char *name, *line = *msg_p;
1971 int linelen = get_one_line(*msg_p);
1973 if (!linelen)
1974 return;
1975 *msg_p += linelen;
1977 if (linelen == 1)
1978 /* End of header */
1979 return;
1981 if (pp->fmt == CMIT_FMT_RAW) {
1982 strbuf_add(sb, line, linelen);
1983 continue;
1986 if (starts_with(line, "parent ")) {
1987 if (linelen != the_hash_algo->hexsz + 8)
1988 die("bad parent line in commit");
1989 continue;
1992 if (!parents_shown) {
1993 unsigned num = commit_list_count(commit->parents);
1994 /* with enough slop */
1995 strbuf_grow(sb, num * (GIT_MAX_HEXSZ + 10) + 20);
1996 add_merge_info(pp, sb, commit);
1997 parents_shown = 1;
2001 * MEDIUM == DEFAULT shows only author with dates.
2002 * FULL shows both authors but not dates.
2003 * FULLER shows both authors and dates.
2005 if (skip_prefix(line, "author ", &name)) {
2006 strbuf_grow(sb, linelen + 80);
2007 pp_user_info(pp, "Author", sb, name, encoding);
2009 if (skip_prefix(line, "committer ", &name) &&
2010 (pp->fmt == CMIT_FMT_FULL || pp->fmt == CMIT_FMT_FULLER)) {
2011 strbuf_grow(sb, linelen + 80);
2012 pp_user_info(pp, "Commit", sb, name, encoding);
2017 void pp_title_line(struct pretty_print_context *pp,
2018 const char **msg_p,
2019 struct strbuf *sb,
2020 const char *encoding,
2021 int need_8bit_cte)
2023 static const int max_length = 78; /* per rfc2047 */
2024 struct strbuf title;
2026 strbuf_init(&title, 80);
2027 *msg_p = format_subject(&title, *msg_p,
2028 pp->preserve_subject ? "\n" : " ");
2030 strbuf_grow(sb, title.len + 1024);
2031 if (pp->print_email_subject) {
2032 if (pp->rev)
2033 fmt_output_email_subject(sb, pp->rev);
2034 if (pp->encode_email_headers &&
2035 needs_rfc2047_encoding(title.buf, title.len))
2036 add_rfc2047(sb, title.buf, title.len,
2037 encoding, RFC2047_SUBJECT);
2038 else
2039 strbuf_add_wrapped_bytes(sb, title.buf, title.len,
2040 -last_line_length(sb), 1, max_length);
2041 } else {
2042 strbuf_addbuf(sb, &title);
2044 strbuf_addch(sb, '\n');
2046 if (need_8bit_cte == 0) {
2047 int i;
2048 for (i = 0; i < pp->in_body_headers.nr; i++) {
2049 if (has_non_ascii(pp->in_body_headers.items[i].string)) {
2050 need_8bit_cte = 1;
2051 break;
2056 if (need_8bit_cte > 0) {
2057 const char *header_fmt =
2058 "MIME-Version: 1.0\n"
2059 "Content-Type: text/plain; charset=%s\n"
2060 "Content-Transfer-Encoding: 8bit\n";
2061 strbuf_addf(sb, header_fmt, encoding);
2063 if (pp->after_subject) {
2064 strbuf_addstr(sb, pp->after_subject);
2066 if (cmit_fmt_is_mail(pp->fmt)) {
2067 strbuf_addch(sb, '\n');
2070 if (pp->in_body_headers.nr) {
2071 int i;
2072 for (i = 0; i < pp->in_body_headers.nr; i++) {
2073 strbuf_addstr(sb, pp->in_body_headers.items[i].string);
2074 free(pp->in_body_headers.items[i].string);
2076 string_list_clear(&pp->in_body_headers, 0);
2077 strbuf_addch(sb, '\n');
2080 strbuf_release(&title);
2083 static int pp_utf8_width(const char *start, const char *end)
2085 int width = 0;
2086 size_t remain = end - start;
2088 while (remain) {
2089 int n = utf8_width(&start, &remain);
2090 if (n < 0 || !start)
2091 return -1;
2092 width += n;
2094 return width;
2097 static void strbuf_add_tabexpand(struct strbuf *sb, struct grep_opt *opt,
2098 int color, int tabwidth, const char *line,
2099 int linelen)
2101 const char *tab;
2103 while ((tab = memchr(line, '\t', linelen)) != NULL) {
2104 int width = pp_utf8_width(line, tab);
2107 * If it wasn't well-formed utf8, or it
2108 * had characters with badly defined
2109 * width (control characters etc), just
2110 * give up on trying to align things.
2112 if (width < 0)
2113 break;
2115 /* Output the data .. */
2116 append_line_with_color(sb, opt, line, tab - line, color,
2117 GREP_CONTEXT_BODY,
2118 GREP_HEADER_FIELD_MAX);
2120 /* .. and the de-tabified tab */
2121 strbuf_addchars(sb, ' ', tabwidth - (width % tabwidth));
2123 /* Skip over the printed part .. */
2124 linelen -= tab + 1 - line;
2125 line = tab + 1;
2129 * Print out everything after the last tab without
2130 * worrying about width - there's nothing more to
2131 * align.
2133 append_line_with_color(sb, opt, line, linelen, color, GREP_CONTEXT_BODY,
2134 GREP_HEADER_FIELD_MAX);
2138 * pp_handle_indent() prints out the intendation, and
2139 * the whole line (without the final newline), after
2140 * de-tabifying.
2142 static void pp_handle_indent(struct pretty_print_context *pp,
2143 struct strbuf *sb, int indent,
2144 const char *line, int linelen)
2146 struct grep_opt *opt = pp->rev ? &pp->rev->grep_filter : NULL;
2148 strbuf_addchars(sb, ' ', indent);
2149 if (pp->expand_tabs_in_log)
2150 strbuf_add_tabexpand(sb, opt, pp->color, pp->expand_tabs_in_log,
2151 line, linelen);
2152 else
2153 append_line_with_color(sb, opt, line, linelen, pp->color,
2154 GREP_CONTEXT_BODY,
2155 GREP_HEADER_FIELD_MAX);
2158 static int is_mboxrd_from(const char *line, int len)
2161 * a line matching /^From $/ here would only have len == 4
2162 * at this point because is_empty_line would've trimmed all
2163 * trailing space
2165 return len > 4 && starts_with(line + strspn(line, ">"), "From ");
2168 void pp_remainder(struct pretty_print_context *pp,
2169 const char **msg_p,
2170 struct strbuf *sb,
2171 int indent)
2173 struct grep_opt *opt = pp->rev ? &pp->rev->grep_filter : NULL;
2174 int first = 1;
2176 for (;;) {
2177 const char *line = *msg_p;
2178 int linelen = get_one_line(line);
2179 *msg_p += linelen;
2181 if (!linelen)
2182 break;
2184 if (is_blank_line(line, &linelen)) {
2185 if (first)
2186 continue;
2187 if (pp->fmt == CMIT_FMT_SHORT)
2188 break;
2190 first = 0;
2192 strbuf_grow(sb, linelen + indent + 20);
2193 if (indent)
2194 pp_handle_indent(pp, sb, indent, line, linelen);
2195 else if (pp->expand_tabs_in_log)
2196 strbuf_add_tabexpand(sb, opt, pp->color,
2197 pp->expand_tabs_in_log, line,
2198 linelen);
2199 else {
2200 if (pp->fmt == CMIT_FMT_MBOXRD &&
2201 is_mboxrd_from(line, linelen))
2202 strbuf_addch(sb, '>');
2204 append_line_with_color(sb, opt, line, linelen,
2205 pp->color, GREP_CONTEXT_BODY,
2206 GREP_HEADER_FIELD_MAX);
2208 strbuf_addch(sb, '\n');
2212 void pretty_print_commit(struct pretty_print_context *pp,
2213 const struct commit *commit,
2214 struct strbuf *sb)
2216 unsigned long beginning_of_body;
2217 int indent = 4;
2218 const char *msg;
2219 const char *reencoded;
2220 const char *encoding;
2221 int need_8bit_cte = pp->need_8bit_cte;
2223 if (pp->fmt == CMIT_FMT_USERFORMAT) {
2224 repo_format_commit_message(the_repository, commit,
2225 user_format, sb, pp);
2226 return;
2229 encoding = get_log_output_encoding();
2230 msg = reencoded = repo_logmsg_reencode(the_repository, commit, NULL,
2231 encoding);
2233 if (pp->fmt == CMIT_FMT_ONELINE || cmit_fmt_is_mail(pp->fmt))
2234 indent = 0;
2237 * We need to check and emit Content-type: to mark it
2238 * as 8-bit if we haven't done so.
2240 if (cmit_fmt_is_mail(pp->fmt) && need_8bit_cte == 0) {
2241 int i, ch, in_body;
2243 for (in_body = i = 0; (ch = msg[i]); i++) {
2244 if (!in_body) {
2245 /* author could be non 7-bit ASCII but
2246 * the log may be so; skip over the
2247 * header part first.
2249 if (ch == '\n' && msg[i+1] == '\n')
2250 in_body = 1;
2252 else if (non_ascii(ch)) {
2253 need_8bit_cte = 1;
2254 break;
2259 pp_header(pp, encoding, commit, &msg, sb);
2260 if (pp->fmt != CMIT_FMT_ONELINE && !pp->print_email_subject) {
2261 strbuf_addch(sb, '\n');
2264 /* Skip excess blank lines at the beginning of body, if any... */
2265 msg = skip_blank_lines(msg);
2267 /* These formats treat the title line specially. */
2268 if (pp->fmt == CMIT_FMT_ONELINE || cmit_fmt_is_mail(pp->fmt))
2269 pp_title_line(pp, &msg, sb, encoding, need_8bit_cte);
2271 beginning_of_body = sb->len;
2272 if (pp->fmt != CMIT_FMT_ONELINE)
2273 pp_remainder(pp, &msg, sb, indent);
2274 strbuf_rtrim(sb);
2276 /* Make sure there is an EOLN for the non-oneline case */
2277 if (pp->fmt != CMIT_FMT_ONELINE)
2278 strbuf_addch(sb, '\n');
2281 * The caller may append additional body text in e-mail
2282 * format. Make sure we did not strip the blank line
2283 * between the header and the body.
2285 if (cmit_fmt_is_mail(pp->fmt) && sb->len <= beginning_of_body)
2286 strbuf_addch(sb, '\n');
2288 repo_unuse_commit_buffer(the_repository, commit, reencoded);
2291 void pp_commit_easy(enum cmit_fmt fmt, const struct commit *commit,
2292 struct strbuf *sb)
2294 struct pretty_print_context pp = {0};
2295 pp.fmt = fmt;
2296 pretty_print_commit(&pp, commit, sb);