6 #include "string-list.h"
11 static char *user_format;
13 void get_commit_format(const char *arg, struct rev_info *rev)
16 static struct cmt_fmt_map {
21 { "raw", 1, CMIT_FMT_RAW },
22 { "medium", 1, CMIT_FMT_MEDIUM },
23 { "short", 1, CMIT_FMT_SHORT },
24 { "email", 1, CMIT_FMT_EMAIL },
25 { "full", 5, CMIT_FMT_FULL },
26 { "fuller", 5, CMIT_FMT_FULLER },
27 { "oneline", 1, CMIT_FMT_ONELINE },
30 rev->use_terminator = 0;
32 rev->commit_format = CMIT_FMT_DEFAULT;
35 if (!prefixcmp(arg, "format:") || !prefixcmp(arg, "tformat:")) {
36 const char *cp = strchr(arg, ':') + 1;
38 user_format = xstrdup(cp);
40 rev->use_terminator = 1;
41 rev->commit_format = CMIT_FMT_USERFORMAT;
44 for (i = 0; i < ARRAY_SIZE(cmt_fmts); i++) {
45 if (!strncmp(arg, cmt_fmts[i].n, cmt_fmts[i].cmp_len) &&
46 !strncmp(arg, cmt_fmts[i].n, strlen(arg))) {
47 if (cmt_fmts[i].v == CMIT_FMT_ONELINE)
48 rev->use_terminator = 1;
49 rev->commit_format = cmt_fmts[i].v;
54 die("invalid --pretty format: %s", arg);
58 * Generic support for pretty-printing the header
60 static int get_one_line(const char *msg)
75 /* High bit set, or ISO-2022-INT */
79 return ((ch & 0x80) || (ch == 0x1b));
82 static int is_rfc2047_special(char ch)
84 return (non_ascii(ch) || (ch == '=') || (ch == '?') || (ch == '_'));
87 static void add_rfc2047(struct strbuf *sb, const char *line, int len,
92 for (i = 0; i < len; i++) {
96 if ((i + 1 < len) && (ch == '=' && line[i+1] == '?'))
99 strbuf_add(sb, line, len);
103 strbuf_grow(sb, len * 3 + strlen(encoding) + 100);
104 strbuf_addf(sb, "=?%s?q?", encoding);
105 for (i = last = 0; i < len; i++) {
106 unsigned ch = line[i] & 0xFF;
108 * We encode ' ' using '=20' even though rfc2047
109 * allows using '_' for readability. Unfortunately,
110 * many programs do not understand this and just
111 * leave the underscore in place.
113 if (is_rfc2047_special(ch) || ch == ' ') {
114 strbuf_add(sb, line + last, i - last);
115 strbuf_addf(sb, "=%02X", ch);
119 strbuf_add(sb, line + last, len - last);
120 strbuf_addstr(sb, "?=");
123 void pp_user_info(const char *what, enum cmit_fmt fmt, struct strbuf *sb,
124 const char *line, enum date_mode dmode,
125 const char *encoding)
131 const char *filler = " ";
133 if (fmt == CMIT_FMT_ONELINE)
135 date = strchr(line, '>');
138 namelen = ++date - line;
139 time = strtoul(date, &date, 10);
140 tz = strtol(date, NULL, 10);
142 if (fmt == CMIT_FMT_EMAIL) {
143 char *name_tail = strchr(line, '<');
144 int display_name_length;
147 while (line < name_tail && isspace(name_tail[-1]))
149 display_name_length = name_tail - line;
151 strbuf_addstr(sb, "From: ");
152 add_rfc2047(sb, line, display_name_length, encoding);
153 strbuf_add(sb, name_tail, namelen - display_name_length);
154 strbuf_addch(sb, '\n');
156 strbuf_addf(sb, "%s: %.*s%.*s\n", what,
157 (fmt == CMIT_FMT_FULLER) ? 4 : 0,
158 filler, namelen, line);
161 case CMIT_FMT_MEDIUM:
162 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, dmode));
165 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, DATE_RFC2822));
167 case CMIT_FMT_FULLER:
168 strbuf_addf(sb, "%sDate: %s\n", what, show_date(time, tz, dmode));
176 static int is_empty_line(const char *line, int *len_p)
179 while (len && isspace(line[len-1]))
185 static const char *skip_empty_lines(const char *msg)
188 int linelen = get_one_line(msg);
192 if (!is_empty_line(msg, &ll))
199 static void add_merge_info(enum cmit_fmt fmt, struct strbuf *sb,
200 const struct commit *commit, int abbrev)
202 struct commit_list *parent = commit->parents;
204 if ((fmt == CMIT_FMT_ONELINE) || (fmt == CMIT_FMT_EMAIL) ||
205 !parent || !parent->next)
208 strbuf_addstr(sb, "Merge:");
211 struct commit *p = parent->item;
212 const char *hex = NULL;
215 hex = find_unique_abbrev(p->object.sha1, abbrev);
217 hex = sha1_to_hex(p->object.sha1);
218 dots = (abbrev && strlen(hex) != 40) ? "..." : "";
219 parent = parent->next;
221 strbuf_addf(sb, " %s%s", hex, dots);
223 strbuf_addch(sb, '\n');
226 static char *get_header(const struct commit *commit, const char *key)
228 int key_len = strlen(key);
229 const char *line = commit->buffer;
232 const char *eol = strchr(line, '\n'), *next;
237 eol = line + strlen(line);
241 if (eol - line > key_len &&
242 !strncmp(line, key, key_len) &&
243 line[key_len] == ' ') {
244 return xmemdupz(line + key_len + 1, eol - line - key_len - 1);
250 static char *replace_encoding_header(char *buf, const char *encoding)
252 struct strbuf tmp = STRBUF_INIT;
256 /* guess if there is an encoding header before a \n\n */
257 while (strncmp(cp, "encoding ", strlen("encoding "))) {
258 cp = strchr(cp, '\n');
259 if (!cp || *++cp == '\n')
263 cp = strchr(cp, '\n');
265 return buf; /* should not happen but be defensive */
266 len = cp + 1 - (buf + start);
268 strbuf_attach(&tmp, buf, strlen(buf), strlen(buf) + 1);
269 if (is_encoding_utf8(encoding)) {
270 /* we have re-coded to UTF-8; drop the header */
271 strbuf_remove(&tmp, start, len);
273 /* just replaces XXXX in 'encoding XXXX\n' */
274 strbuf_splice(&tmp, start + strlen("encoding "),
275 len - strlen("encoding \n"),
276 encoding, strlen(encoding));
278 return strbuf_detach(&tmp, NULL);
281 static char *logmsg_reencode(const struct commit *commit,
282 const char *output_encoding)
284 static const char *utf8 = "utf-8";
285 const char *use_encoding;
289 if (!*output_encoding)
291 encoding = get_header(commit, "encoding");
292 use_encoding = encoding ? encoding : utf8;
293 if (!strcmp(use_encoding, output_encoding))
294 if (encoding) /* we'll strip encoding header later */
295 out = xstrdup(commit->buffer);
297 return NULL; /* nothing to do */
299 out = reencode_string(commit->buffer,
300 output_encoding, use_encoding);
302 out = replace_encoding_header(out, output_encoding);
308 static int mailmap_name(struct strbuf *sb, const char *email)
310 static struct string_list *mail_map;
314 mail_map = xcalloc(1, sizeof(*mail_map));
315 read_mailmap(mail_map, ".mailmap", NULL);
321 if (!map_email(mail_map, email, buffer, sizeof(buffer)))
323 strbuf_addstr(sb, buffer);
327 static size_t format_person_part(struct strbuf *sb, char part,
328 const char *msg, int len, enum date_mode dmode)
330 /* currently all placeholders have same length */
331 const int placeholder_len = 2;
332 int start, end, tz = 0;
333 unsigned long date = 0;
336 /* advance 'end' to point to email start delimiter */
337 for (end = 0; end < len && msg[end] != '<'; end++)
341 * When end points at the '<' that we found, it should have
342 * matching '>' later, which means 'end' must be strictly
348 if (part == 'n' || part == 'N') { /* name */
349 while (end > 0 && isspace(msg[end - 1]))
351 if (part != 'N' || !msg[end] || !msg[end + 1] ||
352 mailmap_name(sb, msg + end + 2) < 0)
353 strbuf_add(sb, msg, end);
354 return placeholder_len;
356 start = ++end; /* save email start position */
358 /* advance 'end' to point to email end delimiter */
359 for ( ; end < len && msg[end] != '>'; end++)
365 if (part == 'e') { /* email */
366 strbuf_add(sb, msg + start, end - start);
367 return placeholder_len;
370 /* advance 'start' to point to date start delimiter */
371 for (start = end + 1; start < len && isspace(msg[start]); start++)
375 date = strtoul(msg + start, &ep, 10);
376 if (msg + start == ep)
379 if (part == 't') { /* date, UNIX timestamp */
380 strbuf_add(sb, msg + start, ep - (msg + start));
381 return placeholder_len;
385 for (start = ep - msg + 1; start < len && isspace(msg[start]); start++)
387 if (start + 1 < len) {
388 tz = strtoul(msg + start + 1, NULL, 10);
389 if (msg[start] == '-')
395 strbuf_addstr(sb, show_date(date, tz, dmode));
396 return placeholder_len;
397 case 'D': /* date, RFC2822 style */
398 strbuf_addstr(sb, show_date(date, tz, DATE_RFC2822));
399 return placeholder_len;
400 case 'r': /* date, relative */
401 strbuf_addstr(sb, show_date(date, tz, DATE_RELATIVE));
402 return placeholder_len;
403 case 'i': /* date, ISO 8601 */
404 strbuf_addstr(sb, show_date(date, tz, DATE_ISO8601));
405 return placeholder_len;
410 * bogus commit, 'sb' cannot be updated, but we still need to
411 * compute a valid return value.
413 if (part == 'n' || part == 'e' || part == 't' || part == 'd'
414 || part == 'D' || part == 'r' || part == 'i')
415 return placeholder_len;
417 return 0; /* unknown placeholder */
425 struct format_commit_context {
426 const struct commit *commit;
427 enum date_mode dmode;
428 unsigned commit_header_parsed:1;
429 unsigned commit_message_parsed:1;
431 /* These offsets are relative to the start of the commit message. */
433 struct chunk committer;
434 struct chunk encoding;
439 /* The following ones are relative to the result struct strbuf. */
440 struct chunk abbrev_commit_hash;
441 struct chunk abbrev_tree_hash;
442 struct chunk abbrev_parent_hashes;
445 static int add_again(struct strbuf *sb, struct chunk *chunk)
448 strbuf_adddup(sb, chunk->off, chunk->len);
453 * We haven't seen this chunk before. Our caller is surely
454 * going to add it the hard way now. Remember the most likely
455 * start of the to-be-added chunk: the current end of the
458 chunk->off = sb->len;
462 static void parse_commit_header(struct format_commit_context *context)
464 const char *msg = context->commit->buffer;
467 for (i = 0; msg[i]; i++) {
469 for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
474 } else if (!prefixcmp(msg + i, "author ")) {
475 context->author.off = i + 7;
476 context->author.len = eol - i - 7;
477 } else if (!prefixcmp(msg + i, "committer ")) {
478 context->committer.off = i + 10;
479 context->committer.len = eol - i - 10;
480 } else if (!prefixcmp(msg + i, "encoding ")) {
481 context->encoding.off = i + 9;
482 context->encoding.len = eol - i - 9;
486 context->message_off = i;
487 context->commit_header_parsed = 1;
490 const char *format_subject(struct strbuf *sb, const char *msg,
491 const char *line_separator)
496 const char *line = msg;
497 int linelen = get_one_line(line);
500 if (!linelen || is_empty_line(line, &linelen))
505 strbuf_grow(sb, linelen + 2);
507 strbuf_addstr(sb, line_separator);
508 strbuf_add(sb, line, linelen);
514 static void parse_commit_message(struct format_commit_context *c)
516 const char *msg = c->commit->buffer + c->message_off;
517 const char *start = c->commit->buffer;
519 msg = skip_empty_lines(msg);
520 c->subject_off = msg - start;
522 msg = format_subject(NULL, msg, NULL);
523 msg = skip_empty_lines(msg);
524 c->body_off = msg - start;
526 c->commit_message_parsed = 1;
529 static void format_decoration(struct strbuf *sb, const struct commit *commit)
531 struct name_decoration *d;
532 const char *prefix = " (";
534 load_ref_decorations();
535 d = lookup_decoration(&name_decoration, &commit->object);
537 strbuf_addstr(sb, prefix);
539 strbuf_addstr(sb, d->name);
542 if (prefix[0] == ',')
543 strbuf_addch(sb, ')');
546 static size_t format_commit_item(struct strbuf *sb, const char *placeholder,
549 struct format_commit_context *c = context;
550 const struct commit *commit = c->commit;
551 const char *msg = commit->buffer;
552 struct commit_list *p;
555 /* these are independent of the commit */
556 switch (placeholder[0]) {
558 if (placeholder[1] == '(') {
559 const char *end = strchr(placeholder + 2, ')');
560 char color[COLOR_MAXLEN];
563 color_parse_mem(placeholder + 2,
564 end - (placeholder + 2),
565 "--pretty format", color);
566 strbuf_addstr(sb, color);
567 return end - placeholder + 1;
569 if (!prefixcmp(placeholder + 1, "red")) {
570 strbuf_addstr(sb, "\033[31m");
572 } else if (!prefixcmp(placeholder + 1, "green")) {
573 strbuf_addstr(sb, "\033[32m");
575 } else if (!prefixcmp(placeholder + 1, "blue")) {
576 strbuf_addstr(sb, "\033[34m");
578 } else if (!prefixcmp(placeholder + 1, "reset")) {
579 strbuf_addstr(sb, "\033[m");
583 case 'n': /* newline */
584 strbuf_addch(sb, '\n');
587 /* %x00 == NUL, %x0a == LF, etc. */
588 if (0 <= (h1 = hexval_table[0xff & placeholder[1]]) &&
590 0 <= (h2 = hexval_table[0xff & placeholder[2]]) &&
592 strbuf_addch(sb, (h1<<4)|h2);
598 /* these depend on the commit */
599 if (!commit->object.parsed)
600 parse_object(commit->object.sha1);
602 switch (placeholder[0]) {
603 case 'H': /* commit hash */
604 strbuf_addstr(sb, sha1_to_hex(commit->object.sha1));
606 case 'h': /* abbreviated commit hash */
607 if (add_again(sb, &c->abbrev_commit_hash))
609 strbuf_addstr(sb, find_unique_abbrev(commit->object.sha1,
611 c->abbrev_commit_hash.len = sb->len - c->abbrev_commit_hash.off;
613 case 'T': /* tree hash */
614 strbuf_addstr(sb, sha1_to_hex(commit->tree->object.sha1));
616 case 't': /* abbreviated tree hash */
617 if (add_again(sb, &c->abbrev_tree_hash))
619 strbuf_addstr(sb, find_unique_abbrev(commit->tree->object.sha1,
621 c->abbrev_tree_hash.len = sb->len - c->abbrev_tree_hash.off;
623 case 'P': /* parent hashes */
624 for (p = commit->parents; p; p = p->next) {
625 if (p != commit->parents)
626 strbuf_addch(sb, ' ');
627 strbuf_addstr(sb, sha1_to_hex(p->item->object.sha1));
630 case 'p': /* abbreviated parent hashes */
631 if (add_again(sb, &c->abbrev_parent_hashes))
633 for (p = commit->parents; p; p = p->next) {
634 if (p != commit->parents)
635 strbuf_addch(sb, ' ');
636 strbuf_addstr(sb, find_unique_abbrev(
637 p->item->object.sha1, DEFAULT_ABBREV));
639 c->abbrev_parent_hashes.len = sb->len -
640 c->abbrev_parent_hashes.off;
642 case 'm': /* left/right/bottom */
643 strbuf_addch(sb, (commit->object.flags & BOUNDARY)
645 : (commit->object.flags & SYMMETRIC_LEFT)
650 format_decoration(sb, commit);
654 /* For the rest we have to parse the commit header. */
655 if (!c->commit_header_parsed)
656 parse_commit_header(c);
658 switch (placeholder[0]) {
659 case 'a': /* author ... */
660 return format_person_part(sb, placeholder[1],
661 msg + c->author.off, c->author.len,
663 case 'c': /* committer ... */
664 return format_person_part(sb, placeholder[1],
665 msg + c->committer.off, c->committer.len,
667 case 'e': /* encoding */
668 strbuf_add(sb, msg + c->encoding.off, c->encoding.len);
672 /* Now we need to parse the commit message. */
673 if (!c->commit_message_parsed)
674 parse_commit_message(c);
676 switch (placeholder[0]) {
677 case 's': /* subject */
678 format_subject(sb, msg + c->subject_off, " ");
681 strbuf_addstr(sb, msg + c->body_off);
684 return 0; /* unknown placeholder */
687 void format_commit_message(const struct commit *commit,
688 const void *format, struct strbuf *sb,
689 enum date_mode dmode)
691 struct format_commit_context context;
693 memset(&context, 0, sizeof(context));
694 context.commit = commit;
695 context.dmode = dmode;
696 strbuf_expand(sb, format, format_commit_item, &context);
699 static void pp_header(enum cmit_fmt fmt,
701 enum date_mode dmode,
702 const char *encoding,
703 const struct commit *commit,
707 int parents_shown = 0;
710 const char *line = *msg_p;
711 int linelen = get_one_line(*msg_p);
721 if (fmt == CMIT_FMT_RAW) {
722 strbuf_add(sb, line, linelen);
726 if (!memcmp(line, "parent ", 7)) {
728 die("bad parent line in commit");
732 if (!parents_shown) {
733 struct commit_list *parent;
735 for (parent = commit->parents, num = 0;
737 parent = parent->next, num++)
739 /* with enough slop */
740 strbuf_grow(sb, num * 50 + 20);
741 add_merge_info(fmt, sb, commit, abbrev);
746 * MEDIUM == DEFAULT shows only author with dates.
747 * FULL shows both authors but not dates.
748 * FULLER shows both authors and dates.
750 if (!memcmp(line, "author ", 7)) {
751 strbuf_grow(sb, linelen + 80);
752 pp_user_info("Author", fmt, sb, line + 7, dmode, encoding);
754 if (!memcmp(line, "committer ", 10) &&
755 (fmt == CMIT_FMT_FULL || fmt == CMIT_FMT_FULLER)) {
756 strbuf_grow(sb, linelen + 80);
757 pp_user_info("Commit", fmt, sb, line + 10, dmode, encoding);
762 void pp_title_line(enum cmit_fmt fmt,
766 const char *after_subject,
767 const char *encoding,
770 const char *line_separator = (fmt == CMIT_FMT_EMAIL) ? "\n " : " ";
773 strbuf_init(&title, 80);
774 *msg_p = format_subject(&title, *msg_p, line_separator);
776 strbuf_grow(sb, title.len + 1024);
778 strbuf_addstr(sb, subject);
779 add_rfc2047(sb, title.buf, title.len, encoding);
781 strbuf_addbuf(sb, &title);
783 strbuf_addch(sb, '\n');
785 if (need_8bit_cte > 0) {
786 const char *header_fmt =
787 "MIME-Version: 1.0\n"
788 "Content-Type: text/plain; charset=%s\n"
789 "Content-Transfer-Encoding: 8bit\n";
790 strbuf_addf(sb, header_fmt, encoding);
793 strbuf_addstr(sb, after_subject);
795 if (fmt == CMIT_FMT_EMAIL) {
796 strbuf_addch(sb, '\n');
798 strbuf_release(&title);
801 void pp_remainder(enum cmit_fmt fmt,
808 const char *line = *msg_p;
809 int linelen = get_one_line(line);
815 if (is_empty_line(line, &linelen)) {
818 if (fmt == CMIT_FMT_SHORT)
823 strbuf_grow(sb, linelen + indent + 20);
825 memset(sb->buf + sb->len, ' ', indent);
826 strbuf_setlen(sb, sb->len + indent);
828 strbuf_add(sb, line, linelen);
829 strbuf_addch(sb, '\n');
833 char *reencode_commit_message(const struct commit *commit, const char **encoding_p)
835 const char *encoding;
837 encoding = (git_log_output_encoding
838 ? git_log_output_encoding
839 : git_commit_encoding);
843 *encoding_p = encoding;
844 return logmsg_reencode(commit, encoding);
847 void pretty_print_commit(enum cmit_fmt fmt, const struct commit *commit,
848 struct strbuf *sb, int abbrev,
849 const char *subject, const char *after_subject,
850 enum date_mode dmode, int need_8bit_cte)
852 unsigned long beginning_of_body;
854 const char *msg = commit->buffer;
856 const char *encoding;
858 if (fmt == CMIT_FMT_USERFORMAT) {
859 format_commit_message(commit, user_format, sb, dmode);
863 reencoded = reencode_commit_message(commit, &encoding);
868 if (fmt == CMIT_FMT_ONELINE || fmt == CMIT_FMT_EMAIL)
872 * We need to check and emit Content-type: to mark it
873 * as 8-bit if we haven't done so.
875 if (fmt == CMIT_FMT_EMAIL && need_8bit_cte == 0) {
878 for (in_body = i = 0; (ch = msg[i]); i++) {
880 /* author could be non 7-bit ASCII but
881 * the log may be so; skip over the
884 if (ch == '\n' && msg[i+1] == '\n')
887 else if (non_ascii(ch)) {
894 pp_header(fmt, abbrev, dmode, encoding, commit, &msg, sb);
895 if (fmt != CMIT_FMT_ONELINE && !subject) {
896 strbuf_addch(sb, '\n');
899 /* Skip excess blank lines at the beginning of body, if any... */
900 msg = skip_empty_lines(msg);
902 /* These formats treat the title line specially. */
903 if (fmt == CMIT_FMT_ONELINE || fmt == CMIT_FMT_EMAIL)
904 pp_title_line(fmt, &msg, sb, subject,
905 after_subject, encoding, need_8bit_cte);
907 beginning_of_body = sb->len;
908 if (fmt != CMIT_FMT_ONELINE)
909 pp_remainder(fmt, &msg, sb, indent);
912 /* Make sure there is an EOLN for the non-oneline case */
913 if (fmt != CMIT_FMT_ONELINE)
914 strbuf_addch(sb, '\n');
917 * The caller may append additional body text in e-mail
918 * format. Make sure we did not strip the blank line
919 * between the header and the body.
921 if (fmt == CMIT_FMT_EMAIL && sb->len <= beginning_of_body)
922 strbuf_addch(sb, '\n');