6 #include "string-list.h"
11 #include "reflog-walk.h"
12 #include "gpg-interface.h"
14 static char *user_format;
15 static struct cmt_fmt_map {
20 const char *user_format;
22 static size_t builtin_formats_len;
23 static size_t commit_formats_len;
24 static size_t commit_formats_alloc;
25 static struct cmt_fmt_map *find_commit_format(const char *sought);
27 int commit_format_is_empty(enum cmit_fmt fmt)
29 return fmt == CMIT_FMT_USERFORMAT && !*user_format;
32 static void save_user_format(struct rev_info *rev, const char *cp, int is_tformat)
35 user_format = xstrdup(cp);
37 rev->use_terminator = 1;
38 rev->commit_format = CMIT_FMT_USERFORMAT;
41 static int git_pretty_formats_config(const char *var, const char *value, void *cb)
43 struct cmt_fmt_map *commit_format = NULL;
48 if (!skip_prefix(var, "pretty.", &name))
51 for (i = 0; i < builtin_formats_len; i++) {
52 if (!strcmp(commit_formats[i].name, name))
56 for (i = builtin_formats_len; i < commit_formats_len; i++) {
57 if (!strcmp(commit_formats[i].name, name)) {
58 commit_format = &commit_formats[i];
64 ALLOC_GROW(commit_formats, commit_formats_len+1,
65 commit_formats_alloc);
66 commit_format = &commit_formats[commit_formats_len];
67 memset(commit_format, 0, sizeof(*commit_format));
71 commit_format->name = xstrdup(name);
72 commit_format->format = CMIT_FMT_USERFORMAT;
73 git_config_string(&fmt, var, value);
74 if (starts_with(fmt, "format:") || starts_with(fmt, "tformat:")) {
75 commit_format->is_tformat = fmt[0] == 't';
76 fmt = strchr(fmt, ':') + 1;
77 } else if (strchr(fmt, '%'))
78 commit_format->is_tformat = 1;
80 commit_format->is_alias = 1;
81 commit_format->user_format = fmt;
86 static void setup_commit_formats(void)
88 struct cmt_fmt_map builtin_formats[] = {
89 { "raw", CMIT_FMT_RAW, 0 },
90 { "medium", CMIT_FMT_MEDIUM, 0 },
91 { "short", CMIT_FMT_SHORT, 0 },
92 { "email", CMIT_FMT_EMAIL, 0 },
93 { "fuller", CMIT_FMT_FULLER, 0 },
94 { "full", CMIT_FMT_FULL, 0 },
95 { "oneline", CMIT_FMT_ONELINE, 1 }
97 commit_formats_len = ARRAY_SIZE(builtin_formats);
98 builtin_formats_len = commit_formats_len;
99 ALLOC_GROW(commit_formats, commit_formats_len, commit_formats_alloc);
100 memcpy(commit_formats, builtin_formats,
101 sizeof(*builtin_formats)*ARRAY_SIZE(builtin_formats));
103 git_config(git_pretty_formats_config, NULL);
106 static struct cmt_fmt_map *find_commit_format_recursive(const char *sought,
107 const char *original,
108 int num_redirections)
110 struct cmt_fmt_map *found = NULL;
111 size_t found_match_len = 0;
114 if (num_redirections >= commit_formats_len)
115 die("invalid --pretty format: "
116 "'%s' references an alias which points to itself",
119 for (i = 0; i < commit_formats_len; i++) {
122 if (!starts_with(commit_formats[i].name, sought))
125 match_len = strlen(commit_formats[i].name);
126 if (found == NULL || found_match_len > match_len) {
127 found = &commit_formats[i];
128 found_match_len = match_len;
132 if (found && found->is_alias) {
133 found = find_commit_format_recursive(found->user_format,
141 static struct cmt_fmt_map *find_commit_format(const char *sought)
144 setup_commit_formats();
146 return find_commit_format_recursive(sought, sought, 0);
149 void get_commit_format(const char *arg, struct rev_info *rev)
151 struct cmt_fmt_map *commit_format;
153 rev->use_terminator = 0;
155 rev->commit_format = CMIT_FMT_DEFAULT;
158 if (starts_with(arg, "format:") || starts_with(arg, "tformat:")) {
159 save_user_format(rev, strchr(arg, ':') + 1, arg[0] == 't');
163 if (!*arg || strchr(arg, '%')) {
164 save_user_format(rev, arg, 1);
168 commit_format = find_commit_format(arg);
170 die("invalid --pretty format: %s", arg);
172 rev->commit_format = commit_format->format;
173 rev->use_terminator = commit_format->is_tformat;
174 if (commit_format->format == CMIT_FMT_USERFORMAT) {
175 save_user_format(rev, commit_format->user_format,
176 commit_format->is_tformat);
181 * Generic support for pretty-printing the header
183 static int get_one_line(const char *msg)
198 /* High bit set, or ISO-2022-INT */
199 static int non_ascii(int ch)
201 return !isascii(ch) || ch == '\033';
204 int has_non_ascii(const char *s)
209 while ((ch = *s++) != '\0') {
216 static int is_rfc822_special(char ch)
238 static int needs_rfc822_quoting(const char *s, int len)
241 for (i = 0; i < len; i++)
242 if (is_rfc822_special(s[i]))
247 static int last_line_length(struct strbuf *sb)
251 /* How many bytes are already used on the last line? */
252 for (i = sb->len - 1; i >= 0; i--)
253 if (sb->buf[i] == '\n')
255 return sb->len - (i + 1);
258 static void add_rfc822_quoted(struct strbuf *out, const char *s, int len)
262 /* just a guess, we may have to also backslash-quote */
263 strbuf_grow(out, len + 2);
265 strbuf_addch(out, '"');
266 for (i = 0; i < len; i++) {
270 strbuf_addch(out, '\\');
273 strbuf_addch(out, s[i]);
276 strbuf_addch(out, '"');
284 static int is_rfc2047_special(char ch, enum rfc2047_type type)
287 * rfc2047, section 4.2:
289 * 8-bit values which correspond to printable ASCII characters other
290 * than "=", "?", and "_" (underscore), MAY be represented as those
291 * characters. (But see section 5 for restrictions.) In
292 * particular, SPACE and TAB MUST NOT be represented as themselves
293 * within encoded words.
297 * rule out non-ASCII characters and non-printable characters (the
298 * non-ASCII check should be redundant as isprint() is not localized
299 * and only knows about ASCII, but be defensive about that)
301 if (non_ascii(ch) || !isprint(ch))
305 * rule out special printable characters (' ' should be the only
306 * whitespace character considered printable, but be defensive and use
309 if (isspace(ch) || ch == '=' || ch == '?' || ch == '_')
313 * rfc2047, section 5.3:
315 * As a replacement for a 'word' entity within a 'phrase', for example,
316 * one that precedes an address in a From, To, or Cc header. The ABNF
317 * definition for 'phrase' from RFC 822 thus becomes:
319 * phrase = 1*( encoded-word / word )
321 * In this case the set of characters that may be used in a "Q"-encoded
322 * 'encoded-word' is restricted to: <upper and lower case ASCII
323 * letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
324 * (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
325 * 'phrase' MUST be separated from any adjacent 'word', 'text' or
326 * 'special' by 'linear-white-space'.
329 if (type != RFC2047_ADDRESS)
332 /* '=' and '_' are special cases and have been checked above */
333 return !(isalnum(ch) || ch == '!' || ch == '*' || ch == '+' || ch == '-' || ch == '/');
336 static int needs_rfc2047_encoding(const char *line, int len,
337 enum rfc2047_type type)
341 for (i = 0; i < len; i++) {
343 if (non_ascii(ch) || ch == '\n')
345 if ((i + 1 < len) && (ch == '=' && line[i+1] == '?'))
352 static void add_rfc2047(struct strbuf *sb, const char *line, size_t len,
353 const char *encoding, enum rfc2047_type type)
355 static const int max_encoded_length = 76; /* per rfc2047 */
357 int line_len = last_line_length(sb);
359 strbuf_grow(sb, len * 3 + strlen(encoding) + 100);
360 strbuf_addf(sb, "=?%s?q?", encoding);
361 line_len += strlen(encoding) + 5; /* 5 for =??q? */
365 * RFC 2047, section 5 (3):
367 * Each 'encoded-word' MUST represent an integral number of
368 * characters. A multi-octet character may not be split across
369 * adjacent 'encoded- word's.
371 const unsigned char *p = (const unsigned char *)line;
372 int chrlen = mbs_chrlen(&line, &len, encoding);
373 int is_special = (chrlen > 1) || is_rfc2047_special(*p, type);
375 /* "=%02X" * chrlen, or the byte itself */
376 const char *encoded_fmt = is_special ? "=%02X" : "%c";
377 int encoded_len = is_special ? 3 * chrlen : 1;
380 * According to RFC 2047, we could encode the special character
381 * ' ' (space) with '_' (underscore) for readability. But many
382 * programs do not understand this and just leave the
383 * underscore in place. Thus, we do nothing special here, which
384 * causes ' ' to be encoded as '=20', avoiding this problem.
387 if (line_len + encoded_len + 2 > max_encoded_length) {
388 /* It won't fit with trailing "?=" --- break the line */
389 strbuf_addf(sb, "?=\n =?%s?q?", encoding);
390 line_len = strlen(encoding) + 5 + 1; /* =??q? plus SP */
393 for (i = 0; i < chrlen; i++)
394 strbuf_addf(sb, encoded_fmt, p[i]);
395 line_len += encoded_len;
397 strbuf_addstr(sb, "?=");
400 const char *show_ident_date(const struct ident_split *ident,
403 unsigned long date = 0;
406 if (ident->date_begin && ident->date_end)
407 date = strtoul(ident->date_begin, NULL, 10);
408 if (date_overflows(date))
411 if (ident->tz_begin && ident->tz_end)
412 tz = strtol(ident->tz_begin, NULL, 10);
413 if (tz >= INT_MAX || tz <= INT_MIN)
416 return show_date(date, tz, mode);
419 void pp_user_info(struct pretty_print_context *pp,
420 const char *what, struct strbuf *sb,
421 const char *line, const char *encoding)
423 struct ident_split ident;
425 const char *mailbuf, *namebuf;
426 size_t namelen, maillen;
427 int max_length = 78; /* per rfc2822 */
429 if (pp->fmt == CMIT_FMT_ONELINE)
432 line_end = strchrnul(line, '\n');
433 if (split_ident_line(&ident, line, line_end - line))
436 mailbuf = ident.mail_begin;
437 maillen = ident.mail_end - ident.mail_begin;
438 namebuf = ident.name_begin;
439 namelen = ident.name_end - ident.name_begin;
442 map_user(pp->mailmap, &mailbuf, &maillen, &namebuf, &namelen);
444 if (pp->fmt == CMIT_FMT_EMAIL) {
445 if (pp->from_ident && ident_cmp(pp->from_ident, &ident)) {
446 struct strbuf buf = STRBUF_INIT;
448 strbuf_addstr(&buf, "From: ");
449 strbuf_add(&buf, namebuf, namelen);
450 strbuf_addstr(&buf, " <");
451 strbuf_add(&buf, mailbuf, maillen);
452 strbuf_addstr(&buf, ">\n");
453 string_list_append(&pp->in_body_headers,
454 strbuf_detach(&buf, NULL));
456 mailbuf = pp->from_ident->mail_begin;
457 maillen = pp->from_ident->mail_end - mailbuf;
458 namebuf = pp->from_ident->name_begin;
459 namelen = pp->from_ident->name_end - namebuf;
462 strbuf_addstr(sb, "From: ");
463 if (needs_rfc2047_encoding(namebuf, namelen, RFC2047_ADDRESS)) {
464 add_rfc2047(sb, namebuf, namelen,
465 encoding, RFC2047_ADDRESS);
466 max_length = 76; /* per rfc2047 */
467 } else if (needs_rfc822_quoting(namebuf, namelen)) {
468 struct strbuf quoted = STRBUF_INIT;
469 add_rfc822_quoted("ed, namebuf, namelen);
470 strbuf_add_wrapped_bytes(sb, quoted.buf, quoted.len,
472 strbuf_release("ed);
474 strbuf_add_wrapped_bytes(sb, namebuf, namelen,
479 last_line_length(sb) + strlen(" <") + maillen + strlen(">"))
480 strbuf_addch(sb, '\n');
481 strbuf_addf(sb, " <%.*s>\n", (int)maillen, mailbuf);
483 strbuf_addf(sb, "%s: %.*s%.*s <%.*s>\n", what,
484 (pp->fmt == CMIT_FMT_FULLER) ? 4 : 0, " ",
485 (int)namelen, namebuf, (int)maillen, mailbuf);
489 case CMIT_FMT_MEDIUM:
490 strbuf_addf(sb, "Date: %s\n",
491 show_ident_date(&ident, pp->date_mode));
494 strbuf_addf(sb, "Date: %s\n",
495 show_ident_date(&ident, DATE_RFC2822));
497 case CMIT_FMT_FULLER:
498 strbuf_addf(sb, "%sDate: %s\n", what,
499 show_ident_date(&ident, pp->date_mode));
507 static int is_empty_line(const char *line, int *len_p)
510 while (len && isspace(line[len - 1]))
516 static const char *skip_empty_lines(const char *msg)
519 int linelen = get_one_line(msg);
523 if (!is_empty_line(msg, &ll))
530 static void add_merge_info(const struct pretty_print_context *pp,
531 struct strbuf *sb, const struct commit *commit)
533 struct commit_list *parent = commit->parents;
535 if ((pp->fmt == CMIT_FMT_ONELINE) || (pp->fmt == CMIT_FMT_EMAIL) ||
536 !parent || !parent->next)
539 strbuf_addstr(sb, "Merge:");
542 struct commit *p = parent->item;
543 const char *hex = NULL;
545 hex = find_unique_abbrev(p->object.sha1, pp->abbrev);
547 hex = sha1_to_hex(p->object.sha1);
548 parent = parent->next;
550 strbuf_addf(sb, " %s", hex);
552 strbuf_addch(sb, '\n');
555 static char *get_header(const struct commit *commit, const char *msg,
558 int key_len = strlen(key);
559 const char *line = msg;
562 const char *eol = strchrnul(line, '\n'), *next;
567 warning("malformed commit (header is missing newline): %s",
568 sha1_to_hex(commit->object.sha1));
572 if (eol - line > key_len &&
573 !strncmp(line, key, key_len) &&
574 line[key_len] == ' ') {
575 return xmemdupz(line + key_len + 1, eol - line - key_len - 1);
582 static char *replace_encoding_header(char *buf, const char *encoding)
584 struct strbuf tmp = STRBUF_INIT;
588 /* guess if there is an encoding header before a \n\n */
589 while (strncmp(cp, "encoding ", strlen("encoding "))) {
590 cp = strchr(cp, '\n');
591 if (!cp || *++cp == '\n')
595 cp = strchr(cp, '\n');
597 return buf; /* should not happen but be defensive */
598 len = cp + 1 - (buf + start);
600 strbuf_attach(&tmp, buf, strlen(buf), strlen(buf) + 1);
601 if (is_encoding_utf8(encoding)) {
602 /* we have re-coded to UTF-8; drop the header */
603 strbuf_remove(&tmp, start, len);
605 /* just replaces XXXX in 'encoding XXXX\n' */
606 strbuf_splice(&tmp, start + strlen("encoding "),
607 len - strlen("encoding \n"),
608 encoding, strlen(encoding));
610 return strbuf_detach(&tmp, NULL);
613 const char *logmsg_reencode(const struct commit *commit,
614 char **commit_encoding,
615 const char *output_encoding)
617 static const char *utf8 = "UTF-8";
618 const char *use_encoding;
620 const char *msg = get_commit_buffer(commit, NULL);
623 if (!output_encoding || !*output_encoding) {
626 get_header(commit, msg, "encoding");
629 encoding = get_header(commit, msg, "encoding");
631 *commit_encoding = encoding;
632 use_encoding = encoding ? encoding : utf8;
633 if (same_encoding(use_encoding, output_encoding)) {
635 * No encoding work to be done. If we have no encoding header
636 * at all, then there's nothing to do, and we can return the
637 * message verbatim (whether newly allocated or not).
643 * Otherwise, we still want to munge the encoding header in the
644 * result, which will be done by modifying the buffer. If we
645 * are using a fresh copy, we can reuse it. But if we are using
646 * the cached copy from get_commit_buffer, we need to duplicate it
647 * to avoid munging the cached copy.
649 if (msg == get_cached_commit_buffer(commit, NULL))
656 * There's actual encoding work to do. Do the reencoding, which
657 * still leaves the header to be replaced in the next step. At
658 * this point, we are done with msg. If we allocated a fresh
659 * copy, we can free it.
661 out = reencode_string(msg, output_encoding, use_encoding);
663 unuse_commit_buffer(commit, msg);
667 * This replacement actually consumes the buffer we hand it, so we do
668 * not have to worry about freeing the old "out" here.
671 out = replace_encoding_header(out, output_encoding);
673 if (!commit_encoding)
676 * If the re-encoding failed, out might be NULL here; in that
677 * case we just return the commit message verbatim.
679 return out ? out : msg;
682 static int mailmap_name(const char **email, size_t *email_len,
683 const char **name, size_t *name_len)
685 static struct string_list *mail_map;
687 mail_map = xcalloc(1, sizeof(*mail_map));
688 read_mailmap(mail_map, NULL);
690 return mail_map->nr && map_user(mail_map, email, email_len, name, name_len);
693 static size_t format_person_part(struct strbuf *sb, char part,
694 const char *msg, int len, enum date_mode dmode)
696 /* currently all placeholders have same length */
697 const int placeholder_len = 2;
698 struct ident_split s;
699 const char *name, *mail;
700 size_t maillen, namelen;
702 if (split_ident_line(&s, msg, len) < 0)
706 namelen = s.name_end - s.name_begin;
708 maillen = s.mail_end - s.mail_begin;
710 if (part == 'N' || part == 'E') /* mailmap lookup */
711 mailmap_name(&mail, &maillen, &name, &namelen);
712 if (part == 'n' || part == 'N') { /* name */
713 strbuf_add(sb, name, namelen);
714 return placeholder_len;
716 if (part == 'e' || part == 'E') { /* email */
717 strbuf_add(sb, mail, maillen);
718 return placeholder_len;
724 if (part == 't') { /* date, UNIX timestamp */
725 strbuf_add(sb, s.date_begin, s.date_end - s.date_begin);
726 return placeholder_len;
731 strbuf_addstr(sb, show_ident_date(&s, dmode));
732 return placeholder_len;
733 case 'D': /* date, RFC2822 style */
734 strbuf_addstr(sb, show_ident_date(&s, DATE_RFC2822));
735 return placeholder_len;
736 case 'r': /* date, relative */
737 strbuf_addstr(sb, show_ident_date(&s, DATE_RELATIVE));
738 return placeholder_len;
739 case 'i': /* date, ISO 8601 */
740 strbuf_addstr(sb, show_ident_date(&s, DATE_ISO8601));
741 return placeholder_len;
746 * reading from either a bogus commit, or a reflog entry with
747 * %gn, %ge, etc.; 'sb' cannot be updated, but we still need
748 * to compute a valid return value.
750 if (part == 'n' || part == 'e' || part == 't' || part == 'd'
751 || part == 'D' || part == 'r' || part == 'i')
752 return placeholder_len;
754 return 0; /* unknown placeholder */
766 flush_left_and_steal,
777 struct format_commit_context {
778 const struct commit *commit;
779 const struct pretty_print_context *pretty_ctx;
780 unsigned commit_header_parsed:1;
781 unsigned commit_message_parsed:1;
782 struct signature_check signature_check;
783 enum flush_type flush_type;
784 enum trunc_type truncate;
786 char *commit_encoding;
787 size_t width, indent1, indent2;
791 /* These offsets are relative to the start of the commit message. */
793 struct chunk committer;
798 /* The following ones are relative to the result struct strbuf. */
799 struct chunk abbrev_commit_hash;
800 struct chunk abbrev_tree_hash;
801 struct chunk abbrev_parent_hashes;
805 static int add_again(struct strbuf *sb, struct chunk *chunk)
808 strbuf_adddup(sb, chunk->off, chunk->len);
813 * We haven't seen this chunk before. Our caller is surely
814 * going to add it the hard way now. Remember the most likely
815 * start of the to-be-added chunk: the current end of the
818 chunk->off = sb->len;
822 static void parse_commit_header(struct format_commit_context *context)
824 const char *msg = context->message;
827 for (i = 0; msg[i]; i++) {
829 for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
834 } else if (starts_with(msg + i, "author ")) {
835 context->author.off = i + 7;
836 context->author.len = eol - i - 7;
837 } else if (starts_with(msg + i, "committer ")) {
838 context->committer.off = i + 10;
839 context->committer.len = eol - i - 10;
843 context->message_off = i;
844 context->commit_header_parsed = 1;
847 static int istitlechar(char c)
849 return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') ||
850 (c >= '0' && c <= '9') || c == '.' || c == '_';
853 static void format_sanitized_subject(struct strbuf *sb, const char *msg)
856 size_t start_len = sb->len;
859 for (; *msg && *msg != '\n'; msg++) {
860 if (istitlechar(*msg)) {
862 strbuf_addch(sb, '-');
864 strbuf_addch(sb, *msg);
866 while (*(msg+1) == '.')
872 /* trim any trailing '.' or '-' characters */
874 while (sb->len - trimlen > start_len &&
875 (sb->buf[sb->len - 1 - trimlen] == '.'
876 || sb->buf[sb->len - 1 - trimlen] == '-'))
878 strbuf_remove(sb, sb->len - trimlen, trimlen);
881 const char *format_subject(struct strbuf *sb, const char *msg,
882 const char *line_separator)
887 const char *line = msg;
888 int linelen = get_one_line(line);
891 if (!linelen || is_empty_line(line, &linelen))
896 strbuf_grow(sb, linelen + 2);
898 strbuf_addstr(sb, line_separator);
899 strbuf_add(sb, line, linelen);
905 static void parse_commit_message(struct format_commit_context *c)
907 const char *msg = c->message + c->message_off;
908 const char *start = c->message;
910 msg = skip_empty_lines(msg);
911 c->subject_off = msg - start;
913 msg = format_subject(NULL, msg, NULL);
914 msg = skip_empty_lines(msg);
915 c->body_off = msg - start;
917 c->commit_message_parsed = 1;
920 static void strbuf_wrap(struct strbuf *sb, size_t pos,
921 size_t width, size_t indent1, size_t indent2)
923 struct strbuf tmp = STRBUF_INIT;
926 strbuf_add(&tmp, sb->buf, pos);
927 strbuf_add_wrapped_text(&tmp, sb->buf + pos,
928 (int) indent1, (int) indent2, (int) width);
929 strbuf_swap(&tmp, sb);
930 strbuf_release(&tmp);
933 static void rewrap_message_tail(struct strbuf *sb,
934 struct format_commit_context *c,
935 size_t new_width, size_t new_indent1,
938 if (c->width == new_width && c->indent1 == new_indent1 &&
939 c->indent2 == new_indent2)
941 if (c->wrap_start < sb->len)
942 strbuf_wrap(sb, c->wrap_start, c->width, c->indent1, c->indent2);
943 c->wrap_start = sb->len;
944 c->width = new_width;
945 c->indent1 = new_indent1;
946 c->indent2 = new_indent2;
949 static int format_reflog_person(struct strbuf *sb,
951 struct reflog_walk_info *log,
952 enum date_mode dmode)
959 ident = get_reflog_ident(log);
963 return format_person_part(sb, part, ident, strlen(ident), dmode);
966 static size_t parse_color(struct strbuf *sb, /* in UTF-8 */
967 const char *placeholder,
968 struct format_commit_context *c)
970 if (placeholder[1] == '(') {
971 const char *begin = placeholder + 2;
972 const char *end = strchr(begin, ')');
973 char color[COLOR_MAXLEN];
977 if (starts_with(begin, "auto,")) {
978 if (!want_color(c->pretty_ctx->color))
979 return end - placeholder + 1;
982 color_parse_mem(begin,
984 "--pretty format", color);
985 strbuf_addstr(sb, color);
986 return end - placeholder + 1;
988 if (starts_with(placeholder + 1, "red")) {
989 strbuf_addstr(sb, GIT_COLOR_RED);
991 } else if (starts_with(placeholder + 1, "green")) {
992 strbuf_addstr(sb, GIT_COLOR_GREEN);
994 } else if (starts_with(placeholder + 1, "blue")) {
995 strbuf_addstr(sb, GIT_COLOR_BLUE);
997 } else if (starts_with(placeholder + 1, "reset")) {
998 strbuf_addstr(sb, GIT_COLOR_RESET);
1004 static size_t parse_padding_placeholder(struct strbuf *sb,
1005 const char *placeholder,
1006 struct format_commit_context *c)
1008 const char *ch = placeholder;
1009 enum flush_type flush_type;
1014 flush_type = flush_right;
1018 flush_type = flush_both;
1020 } else if (*ch == '>') {
1021 flush_type = flush_left_and_steal;
1024 flush_type = flush_left;
1030 /* the next value means "wide enough to that column" */
1037 const char *start = ch + 1;
1038 const char *end = start + strcspn(start, ",)");
1041 if (!end || end == start)
1043 width = strtoul(start, &next, 10);
1044 if (next == start || width == 0)
1046 c->padding = to_column ? -width : width;
1047 c->flush_type = flush_type;
1051 end = strchr(start, ')');
1052 if (!end || end == start)
1054 if (starts_with(start, "trunc)"))
1055 c->truncate = trunc_right;
1056 else if (starts_with(start, "ltrunc)"))
1057 c->truncate = trunc_left;
1058 else if (starts_with(start, "mtrunc)"))
1059 c->truncate = trunc_middle;
1063 c->truncate = trunc_none;
1065 return end - placeholder + 1;
1070 static size_t format_commit_one(struct strbuf *sb, /* in UTF-8 */
1071 const char *placeholder,
1074 struct format_commit_context *c = context;
1075 const struct commit *commit = c->commit;
1076 const char *msg = c->message;
1077 struct commit_list *p;
1080 /* these are independent of the commit */
1081 switch (placeholder[0]) {
1083 if (starts_with(placeholder + 1, "(auto)")) {
1085 return 7; /* consumed 7 bytes, "C(auto)" */
1087 int ret = parse_color(sb, placeholder, c);
1091 * Otherwise, we decided to treat %C<unknown>
1092 * as a literal string, and the previous
1093 * %C(auto) is still valid.
1097 case 'n': /* newline */
1098 strbuf_addch(sb, '\n');
1101 /* %x00 == NUL, %x0a == LF, etc. */
1102 if (0 <= (h1 = hexval_table[0xff & placeholder[1]]) &&
1104 0 <= (h2 = hexval_table[0xff & placeholder[2]]) &&
1106 strbuf_addch(sb, (h1<<4)|h2);
1111 if (placeholder[1] == '(') {
1112 unsigned long width = 0, indent1 = 0, indent2 = 0;
1114 const char *start = placeholder + 2;
1115 const char *end = strchr(start, ')');
1119 width = strtoul(start, &next, 10);
1121 indent1 = strtoul(next + 1, &next, 10);
1123 indent2 = strtoul(next + 1,
1130 rewrap_message_tail(sb, c, width, indent1, indent2);
1131 return end - placeholder + 1;
1137 return parse_padding_placeholder(sb, placeholder, c);
1140 /* these depend on the commit */
1141 if (!commit->object.parsed)
1142 parse_object(commit->object.sha1);
1144 switch (placeholder[0]) {
1145 case 'H': /* commit hash */
1146 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_COMMIT));
1147 strbuf_addstr(sb, sha1_to_hex(commit->object.sha1));
1148 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_RESET));
1150 case 'h': /* abbreviated commit hash */
1151 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_COMMIT));
1152 if (add_again(sb, &c->abbrev_commit_hash)) {
1153 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_RESET));
1156 strbuf_addstr(sb, find_unique_abbrev(commit->object.sha1,
1157 c->pretty_ctx->abbrev));
1158 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_RESET));
1159 c->abbrev_commit_hash.len = sb->len - c->abbrev_commit_hash.off;
1161 case 'T': /* tree hash */
1162 strbuf_addstr(sb, sha1_to_hex(commit->tree->object.sha1));
1164 case 't': /* abbreviated tree hash */
1165 if (add_again(sb, &c->abbrev_tree_hash))
1167 strbuf_addstr(sb, find_unique_abbrev(commit->tree->object.sha1,
1168 c->pretty_ctx->abbrev));
1169 c->abbrev_tree_hash.len = sb->len - c->abbrev_tree_hash.off;
1171 case 'P': /* parent hashes */
1172 for (p = commit->parents; p; p = p->next) {
1173 if (p != commit->parents)
1174 strbuf_addch(sb, ' ');
1175 strbuf_addstr(sb, sha1_to_hex(p->item->object.sha1));
1178 case 'p': /* abbreviated parent hashes */
1179 if (add_again(sb, &c->abbrev_parent_hashes))
1181 for (p = commit->parents; p; p = p->next) {
1182 if (p != commit->parents)
1183 strbuf_addch(sb, ' ');
1184 strbuf_addstr(sb, find_unique_abbrev(
1185 p->item->object.sha1,
1186 c->pretty_ctx->abbrev));
1188 c->abbrev_parent_hashes.len = sb->len -
1189 c->abbrev_parent_hashes.off;
1191 case 'm': /* left/right/bottom */
1192 strbuf_addstr(sb, get_revision_mark(NULL, commit));
1195 load_ref_decorations(DECORATE_SHORT_REFS);
1196 format_decorations(sb, commit, c->auto_color);
1198 case 'g': /* reflog info */
1199 switch(placeholder[1]) {
1200 case 'd': /* reflog selector */
1202 if (c->pretty_ctx->reflog_info)
1203 get_reflog_selector(sb,
1204 c->pretty_ctx->reflog_info,
1205 c->pretty_ctx->date_mode,
1206 c->pretty_ctx->date_mode_explicit,
1207 (placeholder[1] == 'd'));
1209 case 's': /* reflog message */
1210 if (c->pretty_ctx->reflog_info)
1211 get_reflog_message(sb, c->pretty_ctx->reflog_info);
1217 return format_reflog_person(sb,
1219 c->pretty_ctx->reflog_info,
1220 c->pretty_ctx->date_mode);
1222 return 0; /* unknown %g placeholder */
1224 if (c->pretty_ctx->notes_message) {
1225 strbuf_addstr(sb, c->pretty_ctx->notes_message);
1231 if (placeholder[0] == 'G') {
1232 if (!c->signature_check.result)
1233 check_commit_signature(c->commit, &(c->signature_check));
1234 switch (placeholder[1]) {
1236 if (c->signature_check.gpg_output)
1237 strbuf_addstr(sb, c->signature_check.gpg_output);
1240 switch (c->signature_check.result) {
1245 strbuf_addch(sb, c->signature_check.result);
1249 if (c->signature_check.signer)
1250 strbuf_addstr(sb, c->signature_check.signer);
1253 if (c->signature_check.key)
1254 strbuf_addstr(sb, c->signature_check.key);
1263 /* For the rest we have to parse the commit header. */
1264 if (!c->commit_header_parsed)
1265 parse_commit_header(c);
1267 switch (placeholder[0]) {
1268 case 'a': /* author ... */
1269 return format_person_part(sb, placeholder[1],
1270 msg + c->author.off, c->author.len,
1271 c->pretty_ctx->date_mode);
1272 case 'c': /* committer ... */
1273 return format_person_part(sb, placeholder[1],
1274 msg + c->committer.off, c->committer.len,
1275 c->pretty_ctx->date_mode);
1276 case 'e': /* encoding */
1277 if (c->commit_encoding)
1278 strbuf_addstr(sb, c->commit_encoding);
1280 case 'B': /* raw body */
1281 /* message_off is always left at the initial newline */
1282 strbuf_addstr(sb, msg + c->message_off + 1);
1286 /* Now we need to parse the commit message. */
1287 if (!c->commit_message_parsed)
1288 parse_commit_message(c);
1290 switch (placeholder[0]) {
1291 case 's': /* subject */
1292 format_subject(sb, msg + c->subject_off, " ");
1294 case 'f': /* sanitized subject */
1295 format_sanitized_subject(sb, msg + c->subject_off);
1297 case 'b': /* body */
1298 strbuf_addstr(sb, msg + c->body_off);
1301 return 0; /* unknown placeholder */
1304 static size_t format_and_pad_commit(struct strbuf *sb, /* in UTF-8 */
1305 const char *placeholder,
1306 struct format_commit_context *c)
1308 struct strbuf local_sb = STRBUF_INIT;
1309 int total_consumed = 0, len, padding = c->padding;
1311 const char *start = strrchr(sb->buf, '\n');
1315 occupied = utf8_strnwidth(start, -1, 1);
1316 padding = (-padding) - occupied;
1319 int modifier = *placeholder == 'C';
1320 int consumed = format_commit_one(&local_sb, placeholder, c);
1321 total_consumed += consumed;
1326 placeholder += consumed;
1327 if (*placeholder != '%')
1332 len = utf8_strnwidth(local_sb.buf, -1, 1);
1334 if (c->flush_type == flush_left_and_steal) {
1335 const char *ch = sb->buf + sb->len - 1;
1336 while (len > padding && ch > sb->buf) {
1343 /* check for trailing ansi sequences */
1347 while (ch - p < 10 && *p != '\033')
1350 ch + 1 - p != display_mode_esc_sequence_len(p))
1353 * got a good ansi sequence, put it back to
1354 * local_sb as we're cutting sb
1356 strbuf_insert(&local_sb, 0, p, ch + 1 - p);
1359 strbuf_setlen(sb, ch + 1 - sb->buf);
1360 c->flush_type = flush_left;
1363 if (len > padding) {
1364 switch (c->truncate) {
1366 strbuf_utf8_replace(&local_sb,
1367 0, len - (padding - 2),
1371 strbuf_utf8_replace(&local_sb,
1373 len - (padding - 2),
1377 strbuf_utf8_replace(&local_sb,
1378 padding - 2, len - (padding - 2),
1384 strbuf_addbuf(sb, &local_sb);
1386 int sb_len = sb->len, offset = 0;
1387 if (c->flush_type == flush_left)
1388 offset = padding - len;
1389 else if (c->flush_type == flush_both)
1390 offset = (padding - len) / 2;
1392 * we calculate padding in columns, now
1393 * convert it back to chars
1395 padding = padding - len + local_sb.len;
1396 strbuf_grow(sb, padding);
1397 strbuf_setlen(sb, sb_len + padding);
1398 memset(sb->buf + sb_len, ' ', sb->len - sb_len);
1399 memcpy(sb->buf + sb_len + offset, local_sb.buf,
1402 strbuf_release(&local_sb);
1403 c->flush_type = no_flush;
1404 return total_consumed;
1407 static size_t format_commit_item(struct strbuf *sb, /* in UTF-8 */
1408 const char *placeholder,
1415 ADD_LF_BEFORE_NON_EMPTY,
1416 DEL_LF_BEFORE_EMPTY,
1417 ADD_SP_BEFORE_NON_EMPTY
1420 switch (placeholder[0]) {
1422 magic = DEL_LF_BEFORE_EMPTY;
1425 magic = ADD_LF_BEFORE_NON_EMPTY;
1428 magic = ADD_SP_BEFORE_NON_EMPTY;
1433 if (magic != NO_MAGIC)
1437 if (((struct format_commit_context *)context)->flush_type != no_flush)
1438 consumed = format_and_pad_commit(sb, placeholder, context);
1440 consumed = format_commit_one(sb, placeholder, context);
1441 if (magic == NO_MAGIC)
1444 if ((orig_len == sb->len) && magic == DEL_LF_BEFORE_EMPTY) {
1445 while (sb->len && sb->buf[sb->len - 1] == '\n')
1446 strbuf_setlen(sb, sb->len - 1);
1447 } else if (orig_len != sb->len) {
1448 if (magic == ADD_LF_BEFORE_NON_EMPTY)
1449 strbuf_insert(sb, orig_len, "\n", 1);
1450 else if (magic == ADD_SP_BEFORE_NON_EMPTY)
1451 strbuf_insert(sb, orig_len, " ", 1);
1453 return consumed + 1;
1456 static size_t userformat_want_item(struct strbuf *sb, const char *placeholder,
1459 struct userformat_want *w = context;
1461 if (*placeholder == '+' || *placeholder == '-' || *placeholder == ' ')
1464 switch (*placeholder) {
1472 void userformat_find_requirements(const char *fmt, struct userformat_want *w)
1474 struct strbuf dummy = STRBUF_INIT;
1481 strbuf_expand(&dummy, fmt, userformat_want_item, w);
1482 strbuf_release(&dummy);
1485 void format_commit_message(const struct commit *commit,
1486 const char *format, struct strbuf *sb,
1487 const struct pretty_print_context *pretty_ctx)
1489 struct format_commit_context context;
1490 const char *output_enc = pretty_ctx->output_encoding;
1491 const char *utf8 = "UTF-8";
1493 memset(&context, 0, sizeof(context));
1494 context.commit = commit;
1495 context.pretty_ctx = pretty_ctx;
1496 context.wrap_start = sb->len;
1498 * convert a commit message to UTF-8 first
1499 * as far as 'format_commit_item' assumes it in UTF-8
1501 context.message = logmsg_reencode(commit,
1502 &context.commit_encoding,
1505 strbuf_expand(sb, format, format_commit_item, &context);
1506 rewrap_message_tail(sb, &context, 0, 0, 0);
1508 /* then convert a commit message to an actual output encoding */
1510 if (same_encoding(utf8, output_enc))
1513 if (context.commit_encoding &&
1514 !same_encoding(context.commit_encoding, utf8))
1515 output_enc = context.commit_encoding;
1520 char *out = reencode_string_len(sb->buf, sb->len,
1521 output_enc, utf8, &outsz);
1523 strbuf_attach(sb, out, outsz, outsz + 1);
1526 free(context.commit_encoding);
1527 unuse_commit_buffer(commit, context.message);
1530 static void pp_header(struct pretty_print_context *pp,
1531 const char *encoding,
1532 const struct commit *commit,
1536 int parents_shown = 0;
1539 const char *line = *msg_p;
1540 int linelen = get_one_line(*msg_p);
1550 if (pp->fmt == CMIT_FMT_RAW) {
1551 strbuf_add(sb, line, linelen);
1555 if (starts_with(line, "parent ")) {
1557 die("bad parent line in commit");
1561 if (!parents_shown) {
1562 unsigned num = commit_list_count(commit->parents);
1563 /* with enough slop */
1564 strbuf_grow(sb, num * 50 + 20);
1565 add_merge_info(pp, sb, commit);
1570 * MEDIUM == DEFAULT shows only author with dates.
1571 * FULL shows both authors but not dates.
1572 * FULLER shows both authors and dates.
1574 if (starts_with(line, "author ")) {
1575 strbuf_grow(sb, linelen + 80);
1576 pp_user_info(pp, "Author", sb, line + 7, encoding);
1578 if (starts_with(line, "committer ") &&
1579 (pp->fmt == CMIT_FMT_FULL || pp->fmt == CMIT_FMT_FULLER)) {
1580 strbuf_grow(sb, linelen + 80);
1581 pp_user_info(pp, "Commit", sb, line + 10, encoding);
1586 void pp_title_line(struct pretty_print_context *pp,
1589 const char *encoding,
1592 static const int max_length = 78; /* per rfc2047 */
1593 struct strbuf title;
1595 strbuf_init(&title, 80);
1596 *msg_p = format_subject(&title, *msg_p,
1597 pp->preserve_subject ? "\n" : " ");
1599 strbuf_grow(sb, title.len + 1024);
1601 strbuf_addstr(sb, pp->subject);
1602 if (needs_rfc2047_encoding(title.buf, title.len, RFC2047_SUBJECT))
1603 add_rfc2047(sb, title.buf, title.len,
1604 encoding, RFC2047_SUBJECT);
1606 strbuf_add_wrapped_bytes(sb, title.buf, title.len,
1607 -last_line_length(sb), 1, max_length);
1609 strbuf_addbuf(sb, &title);
1611 strbuf_addch(sb, '\n');
1613 if (need_8bit_cte == 0) {
1615 for (i = 0; i < pp->in_body_headers.nr; i++) {
1616 if (has_non_ascii(pp->in_body_headers.items[i].string)) {
1623 if (need_8bit_cte > 0) {
1624 const char *header_fmt =
1625 "MIME-Version: 1.0\n"
1626 "Content-Type: text/plain; charset=%s\n"
1627 "Content-Transfer-Encoding: 8bit\n";
1628 strbuf_addf(sb, header_fmt, encoding);
1630 if (pp->after_subject) {
1631 strbuf_addstr(sb, pp->after_subject);
1633 if (pp->fmt == CMIT_FMT_EMAIL) {
1634 strbuf_addch(sb, '\n');
1637 if (pp->in_body_headers.nr) {
1639 for (i = 0; i < pp->in_body_headers.nr; i++) {
1640 strbuf_addstr(sb, pp->in_body_headers.items[i].string);
1641 free(pp->in_body_headers.items[i].string);
1643 string_list_clear(&pp->in_body_headers, 0);
1644 strbuf_addch(sb, '\n');
1647 strbuf_release(&title);
1650 void pp_remainder(struct pretty_print_context *pp,
1657 const char *line = *msg_p;
1658 int linelen = get_one_line(line);
1664 if (is_empty_line(line, &linelen)) {
1667 if (pp->fmt == CMIT_FMT_SHORT)
1672 strbuf_grow(sb, linelen + indent + 20);
1674 memset(sb->buf + sb->len, ' ', indent);
1675 strbuf_setlen(sb, sb->len + indent);
1677 strbuf_add(sb, line, linelen);
1678 strbuf_addch(sb, '\n');
1682 void pretty_print_commit(struct pretty_print_context *pp,
1683 const struct commit *commit,
1686 unsigned long beginning_of_body;
1689 const char *reencoded;
1690 const char *encoding;
1691 int need_8bit_cte = pp->need_8bit_cte;
1693 if (pp->fmt == CMIT_FMT_USERFORMAT) {
1694 format_commit_message(commit, user_format, sb, pp);
1698 encoding = get_log_output_encoding();
1699 msg = reencoded = logmsg_reencode(commit, NULL, encoding);
1701 if (pp->fmt == CMIT_FMT_ONELINE || pp->fmt == CMIT_FMT_EMAIL)
1705 * We need to check and emit Content-type: to mark it
1706 * as 8-bit if we haven't done so.
1708 if (pp->fmt == CMIT_FMT_EMAIL && need_8bit_cte == 0) {
1711 for (in_body = i = 0; (ch = msg[i]); i++) {
1713 /* author could be non 7-bit ASCII but
1714 * the log may be so; skip over the
1715 * header part first.
1717 if (ch == '\n' && msg[i+1] == '\n')
1720 else if (non_ascii(ch)) {
1727 pp_header(pp, encoding, commit, &msg, sb);
1728 if (pp->fmt != CMIT_FMT_ONELINE && !pp->subject) {
1729 strbuf_addch(sb, '\n');
1732 /* Skip excess blank lines at the beginning of body, if any... */
1733 msg = skip_empty_lines(msg);
1735 /* These formats treat the title line specially. */
1736 if (pp->fmt == CMIT_FMT_ONELINE || pp->fmt == CMIT_FMT_EMAIL)
1737 pp_title_line(pp, &msg, sb, encoding, need_8bit_cte);
1739 beginning_of_body = sb->len;
1740 if (pp->fmt != CMIT_FMT_ONELINE)
1741 pp_remainder(pp, &msg, sb, indent);
1744 /* Make sure there is an EOLN for the non-oneline case */
1745 if (pp->fmt != CMIT_FMT_ONELINE)
1746 strbuf_addch(sb, '\n');
1749 * The caller may append additional body text in e-mail
1750 * format. Make sure we did not strip the blank line
1751 * between the header and the body.
1753 if (pp->fmt == CMIT_FMT_EMAIL && sb->len <= beginning_of_body)
1754 strbuf_addch(sb, '\n');
1756 unuse_commit_buffer(commit, reencoded);
1759 void pp_commit_easy(enum cmit_fmt fmt, const struct commit *commit,
1762 struct pretty_print_context pp = {0};
1764 pretty_print_commit(&pp, commit, sb);