7 static struct cmt_fmt_map {
12 { "raw", 1, CMIT_FMT_RAW },
13 { "medium", 1, CMIT_FMT_MEDIUM },
14 { "short", 1, CMIT_FMT_SHORT },
15 { "email", 1, CMIT_FMT_EMAIL },
16 { "full", 5, CMIT_FMT_FULL },
17 { "fuller", 5, CMIT_FMT_FULLER },
18 { "oneline", 1, CMIT_FMT_ONELINE },
19 { "format:", 7, CMIT_FMT_USERFORMAT},
22 static char *user_format;
24 enum cmit_fmt get_commit_format(const char *arg)
29 return CMIT_FMT_DEFAULT;
32 if (!prefixcmp(arg, "format:")) {
35 user_format = xstrdup(arg + 7);
36 return CMIT_FMT_USERFORMAT;
38 for (i = 0; i < ARRAY_SIZE(cmt_fmts); i++) {
39 if (!strncmp(arg, cmt_fmts[i].n, cmt_fmts[i].cmp_len) &&
40 !strncmp(arg, cmt_fmts[i].n, strlen(arg)))
44 die("invalid --pretty format: %s", arg);
48 * Generic support for pretty-printing the header
50 static int get_one_line(const char *msg)
65 /* High bit set, or ISO-2022-INT */
69 return ((ch & 0x80) || (ch == 0x1b));
72 static int is_rfc2047_special(char ch)
74 return (non_ascii(ch) || (ch == '=') || (ch == '?') || (ch == '_'));
77 static void add_rfc2047(struct strbuf *sb, const char *line, int len,
82 for (i = 0; i < len; i++) {
86 if ((i + 1 < len) && (ch == '=' && line[i+1] == '?'))
89 strbuf_add(sb, line, len);
93 strbuf_grow(sb, len * 3 + strlen(encoding) + 100);
94 strbuf_addf(sb, "=?%s?q?", encoding);
95 for (i = last = 0; i < len; i++) {
96 unsigned ch = line[i] & 0xFF;
98 * We encode ' ' using '=20' even though rfc2047
99 * allows using '_' for readability. Unfortunately,
100 * many programs do not understand this and just
101 * leave the underscore in place.
103 if (is_rfc2047_special(ch) || ch == ' ') {
104 strbuf_add(sb, line + last, i - last);
105 strbuf_addf(sb, "=%02X", ch);
109 strbuf_add(sb, line + last, len - last);
110 strbuf_addstr(sb, "?=");
113 static void add_user_info(const char *what, enum cmit_fmt fmt, struct strbuf *sb,
114 const char *line, enum date_mode dmode,
115 const char *encoding)
121 const char *filler = " ";
123 if (fmt == CMIT_FMT_ONELINE)
125 date = strchr(line, '>');
128 namelen = ++date - line;
129 time = strtoul(date, &date, 10);
130 tz = strtol(date, NULL, 10);
132 if (fmt == CMIT_FMT_EMAIL) {
133 char *name_tail = strchr(line, '<');
134 int display_name_length;
137 while (line < name_tail && isspace(name_tail[-1]))
139 display_name_length = name_tail - line;
141 strbuf_addstr(sb, "From: ");
142 add_rfc2047(sb, line, display_name_length, encoding);
143 strbuf_add(sb, name_tail, namelen - display_name_length);
144 strbuf_addch(sb, '\n');
146 strbuf_addf(sb, "%s: %.*s%.*s\n", what,
147 (fmt == CMIT_FMT_FULLER) ? 4 : 0,
148 filler, namelen, line);
151 case CMIT_FMT_MEDIUM:
152 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, dmode));
155 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, DATE_RFC2822));
157 case CMIT_FMT_FULLER:
158 strbuf_addf(sb, "%sDate: %s\n", what, show_date(time, tz, dmode));
166 static int is_empty_line(const char *line, int *len_p)
169 while (len && isspace(line[len-1]))
175 static void add_merge_info(enum cmit_fmt fmt, struct strbuf *sb,
176 const struct commit *commit, int abbrev)
178 struct commit_list *parent = commit->parents;
180 if ((fmt == CMIT_FMT_ONELINE) || (fmt == CMIT_FMT_EMAIL) ||
181 !parent || !parent->next)
184 strbuf_addstr(sb, "Merge:");
187 struct commit *p = parent->item;
188 const char *hex = NULL;
191 hex = find_unique_abbrev(p->object.sha1, abbrev);
193 hex = sha1_to_hex(p->object.sha1);
194 dots = (abbrev && strlen(hex) != 40) ? "..." : "";
195 parent = parent->next;
197 strbuf_addf(sb, " %s%s", hex, dots);
199 strbuf_addch(sb, '\n');
202 static char *get_header(const struct commit *commit, const char *key)
204 int key_len = strlen(key);
205 const char *line = commit->buffer;
208 const char *eol = strchr(line, '\n'), *next;
213 eol = line + strlen(line);
217 if (eol - line > key_len &&
218 !strncmp(line, key, key_len) &&
219 line[key_len] == ' ') {
220 return xmemdupz(line + key_len + 1, eol - line - key_len - 1);
226 static char *replace_encoding_header(char *buf, const char *encoding)
232 /* guess if there is an encoding header before a \n\n */
233 while (strncmp(cp, "encoding ", strlen("encoding "))) {
234 cp = strchr(cp, '\n');
235 if (!cp || *++cp == '\n')
239 cp = strchr(cp, '\n');
241 return buf; /* should not happen but be defensive */
242 len = cp + 1 - (buf + start);
244 strbuf_init(&tmp, 0);
245 strbuf_attach(&tmp, buf, strlen(buf), strlen(buf) + 1);
246 if (is_encoding_utf8(encoding)) {
247 /* we have re-coded to UTF-8; drop the header */
248 strbuf_remove(&tmp, start, len);
250 /* just replaces XXXX in 'encoding XXXX\n' */
251 strbuf_splice(&tmp, start + strlen("encoding "),
252 len - strlen("encoding \n"),
253 encoding, strlen(encoding));
255 return strbuf_detach(&tmp, NULL);
258 static char *logmsg_reencode(const struct commit *commit,
259 const char *output_encoding)
261 static const char *utf8 = "utf-8";
262 const char *use_encoding;
266 if (!*output_encoding)
268 encoding = get_header(commit, "encoding");
269 use_encoding = encoding ? encoding : utf8;
270 if (!strcmp(use_encoding, output_encoding))
271 if (encoding) /* we'll strip encoding header later */
272 out = xstrdup(commit->buffer);
274 return NULL; /* nothing to do */
276 out = reencode_string(commit->buffer,
277 output_encoding, use_encoding);
279 out = replace_encoding_header(out, output_encoding);
285 static void format_person_part(struct strbuf *sb, char part,
286 const char *msg, int len)
288 int start, end, tz = 0;
293 for (end = 0; end < len && msg[end] != '<'; end++)
296 while (end > 0 && isspace(msg[end - 1]))
298 if (part == 'n') { /* name */
299 strbuf_add(sb, msg, end);
307 for (end = start + 1; end < len && msg[end] != '>'; end++)
313 if (part == 'e') { /* email */
314 strbuf_add(sb, msg + start, end - start);
319 for (start = end + 1; start < len && isspace(msg[start]); start++)
323 date = strtoul(msg + start, &ep, 10);
324 if (msg + start == ep)
327 if (part == 't') { /* date, UNIX timestamp */
328 strbuf_add(sb, msg + start, ep - (msg + start));
333 for (start = ep - msg + 1; start < len && isspace(msg[start]); start++)
335 if (start + 1 < len) {
336 tz = strtoul(msg + start + 1, NULL, 10);
337 if (msg[start] == '-')
343 strbuf_addstr(sb, show_date(date, tz, DATE_NORMAL));
345 case 'D': /* date, RFC2822 style */
346 strbuf_addstr(sb, show_date(date, tz, DATE_RFC2822));
348 case 'r': /* date, relative */
349 strbuf_addstr(sb, show_date(date, tz, DATE_RELATIVE));
351 case 'i': /* date, ISO 8601 */
352 strbuf_addstr(sb, show_date(date, tz, DATE_ISO8601));
362 struct format_commit_context {
363 const struct commit *commit;
365 /* These offsets are relative to the start of the commit message. */
366 int commit_header_parsed;
367 struct chunk subject;
369 struct chunk committer;
370 struct chunk encoding;
373 /* The following ones are relative to the result struct strbuf. */
374 struct chunk abbrev_commit_hash;
375 struct chunk abbrev_tree_hash;
376 struct chunk abbrev_parent_hashes;
379 static int add_again(struct strbuf *sb, struct chunk *chunk)
382 strbuf_adddup(sb, chunk->off, chunk->len);
387 * We haven't seen this chunk before. Our caller is surely
388 * going to add it the hard way now. Remember the most likely
389 * start of the to-be-added chunk: the current end of the
392 chunk->off = sb->len;
396 static void parse_commit_header(struct format_commit_context *context)
398 const char *msg = context->commit->buffer;
400 enum { HEADER, SUBJECT, BODY } state;
402 for (i = 0, state = HEADER; msg[i] && state < BODY; i++) {
404 for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
407 if (state == SUBJECT) {
408 context->subject.off = i;
409 context->subject.len = eol - i;
414 /* strip empty lines */
415 while (msg[eol] == '\n' && msg[eol + 1] == '\n')
417 } else if (!prefixcmp(msg + i, "author ")) {
418 context->author.off = i + 7;
419 context->author.len = eol - i - 7;
420 } else if (!prefixcmp(msg + i, "committer ")) {
421 context->committer.off = i + 10;
422 context->committer.len = eol - i - 10;
423 } else if (!prefixcmp(msg + i, "encoding ")) {
424 context->encoding.off = i + 9;
425 context->encoding.len = eol - i - 9;
431 context->body_off = i;
432 context->commit_header_parsed = 1;
435 static void format_commit_item(struct strbuf *sb, const char *placeholder,
438 struct format_commit_context *c = context;
439 const struct commit *commit = c->commit;
440 const char *msg = commit->buffer;
441 struct commit_list *p;
443 /* these are independent of the commit */
444 switch (placeholder[0]) {
446 switch (placeholder[3]) {
448 strbuf_addstr(sb, "\033[31m");
450 case 'e': /* green */
451 strbuf_addstr(sb, "\033[32m");
454 strbuf_addstr(sb, "\033[34m");
456 case 's': /* reset color */
457 strbuf_addstr(sb, "\033[m");
460 case 'n': /* newline */
461 strbuf_addch(sb, '\n');
465 /* these depend on the commit */
466 if (!commit->object.parsed)
467 parse_object(commit->object.sha1);
469 switch (placeholder[0]) {
470 case 'H': /* commit hash */
471 strbuf_addstr(sb, sha1_to_hex(commit->object.sha1));
473 case 'h': /* abbreviated commit hash */
474 if (add_again(sb, &c->abbrev_commit_hash))
476 strbuf_addstr(sb, find_unique_abbrev(commit->object.sha1,
478 c->abbrev_commit_hash.len = sb->len - c->abbrev_commit_hash.off;
480 case 'T': /* tree hash */
481 strbuf_addstr(sb, sha1_to_hex(commit->tree->object.sha1));
483 case 't': /* abbreviated tree hash */
484 if (add_again(sb, &c->abbrev_tree_hash))
486 strbuf_addstr(sb, find_unique_abbrev(commit->tree->object.sha1,
488 c->abbrev_tree_hash.len = sb->len - c->abbrev_tree_hash.off;
490 case 'P': /* parent hashes */
491 for (p = commit->parents; p; p = p->next) {
492 if (p != commit->parents)
493 strbuf_addch(sb, ' ');
494 strbuf_addstr(sb, sha1_to_hex(p->item->object.sha1));
497 case 'p': /* abbreviated parent hashes */
498 if (add_again(sb, &c->abbrev_parent_hashes))
500 for (p = commit->parents; p; p = p->next) {
501 if (p != commit->parents)
502 strbuf_addch(sb, ' ');
503 strbuf_addstr(sb, find_unique_abbrev(
504 p->item->object.sha1, DEFAULT_ABBREV));
506 c->abbrev_parent_hashes.len = sb->len -
507 c->abbrev_parent_hashes.off;
509 case 'm': /* left/right/bottom */
510 strbuf_addch(sb, (commit->object.flags & BOUNDARY)
512 : (commit->object.flags & SYMMETRIC_LEFT)
518 /* For the rest we have to parse the commit header. */
519 if (!c->commit_header_parsed)
520 parse_commit_header(c);
522 switch (placeholder[0]) {
524 strbuf_add(sb, msg + c->subject.off, c->subject.len);
527 format_person_part(sb, placeholder[1],
528 msg + c->author.off, c->author.len);
531 format_person_part(sb, placeholder[1],
532 msg + c->committer.off, c->committer.len);
535 strbuf_add(sb, msg + c->encoding.off, c->encoding.len);
538 strbuf_addstr(sb, msg + c->body_off);
543 void format_commit_message(const struct commit *commit,
544 const void *format, struct strbuf *sb)
546 const char *placeholders[] = {
547 "H", /* commit hash */
548 "h", /* abbreviated commit hash */
550 "t", /* abbreviated tree hash */
551 "P", /* parent hashes */
552 "p", /* abbreviated parent hashes */
553 "an", /* author name */
554 "ae", /* author email */
555 "ad", /* author date */
556 "aD", /* author date, RFC2822 style */
557 "ar", /* author date, relative */
558 "at", /* author date, UNIX timestamp */
559 "ai", /* author date, ISO 8601 */
560 "cn", /* committer name */
561 "ce", /* committer email */
562 "cd", /* committer date */
563 "cD", /* committer date, RFC2822 style */
564 "cr", /* committer date, relative */
565 "ct", /* committer date, UNIX timestamp */
566 "ci", /* committer date, ISO 8601 */
571 "Cgreen", /* green */
573 "Creset", /* reset color */
575 "m", /* left/right/bottom */
578 struct format_commit_context context;
580 memset(&context, 0, sizeof(context));
581 context.commit = commit;
582 strbuf_expand(sb, format, placeholders, format_commit_item, &context);
585 static void pp_header(enum cmit_fmt fmt,
587 enum date_mode dmode,
588 const char *encoding,
589 const struct commit *commit,
593 int parents_shown = 0;
596 const char *line = *msg_p;
597 int linelen = get_one_line(*msg_p);
607 if (fmt == CMIT_FMT_RAW) {
608 strbuf_add(sb, line, linelen);
612 if (!memcmp(line, "parent ", 7)) {
614 die("bad parent line in commit");
618 if (!parents_shown) {
619 struct commit_list *parent;
621 for (parent = commit->parents, num = 0;
623 parent = parent->next, num++)
625 /* with enough slop */
626 strbuf_grow(sb, num * 50 + 20);
627 add_merge_info(fmt, sb, commit, abbrev);
632 * MEDIUM == DEFAULT shows only author with dates.
633 * FULL shows both authors but not dates.
634 * FULLER shows both authors and dates.
636 if (!memcmp(line, "author ", 7)) {
637 strbuf_grow(sb, linelen + 80);
638 add_user_info("Author", fmt, sb, line + 7, dmode, encoding);
640 if (!memcmp(line, "committer ", 10) &&
641 (fmt == CMIT_FMT_FULL || fmt == CMIT_FMT_FULLER)) {
642 strbuf_grow(sb, linelen + 80);
643 add_user_info("Commit", fmt, sb, line + 10, dmode, encoding);
648 static void pp_title_line(enum cmit_fmt fmt,
652 const char *after_subject,
653 const char *encoding,
658 strbuf_init(&title, 80);
661 const char *line = *msg_p;
662 int linelen = get_one_line(line);
665 if (!linelen || is_empty_line(line, &linelen))
668 strbuf_grow(&title, linelen + 2);
670 if (fmt == CMIT_FMT_EMAIL) {
671 strbuf_addch(&title, '\n');
673 strbuf_addch(&title, ' ');
675 strbuf_add(&title, line, linelen);
678 strbuf_grow(sb, title.len + 1024);
680 strbuf_addstr(sb, subject);
681 add_rfc2047(sb, title.buf, title.len, encoding);
683 strbuf_addbuf(sb, &title);
685 strbuf_addch(sb, '\n');
687 if (plain_non_ascii) {
688 const char *header_fmt =
689 "MIME-Version: 1.0\n"
690 "Content-Type: text/plain; charset=%s\n"
691 "Content-Transfer-Encoding: 8bit\n";
692 strbuf_addf(sb, header_fmt, encoding);
695 strbuf_addstr(sb, after_subject);
697 if (fmt == CMIT_FMT_EMAIL) {
698 strbuf_addch(sb, '\n');
700 strbuf_release(&title);
703 static void pp_remainder(enum cmit_fmt fmt,
710 const char *line = *msg_p;
711 int linelen = get_one_line(line);
717 if (is_empty_line(line, &linelen)) {
720 if (fmt == CMIT_FMT_SHORT)
725 strbuf_grow(sb, linelen + indent + 20);
727 memset(sb->buf + sb->len, ' ', indent);
728 strbuf_setlen(sb, sb->len + indent);
730 strbuf_add(sb, line, linelen);
731 strbuf_addch(sb, '\n');
735 void pretty_print_commit(enum cmit_fmt fmt, const struct commit *commit,
736 struct strbuf *sb, int abbrev,
737 const char *subject, const char *after_subject,
738 enum date_mode dmode, int plain_non_ascii)
740 unsigned long beginning_of_body;
742 const char *msg = commit->buffer;
744 const char *encoding;
746 if (fmt == CMIT_FMT_USERFORMAT) {
747 format_commit_message(commit, user_format, sb);
751 encoding = (git_log_output_encoding
752 ? git_log_output_encoding
753 : git_commit_encoding);
756 reencoded = logmsg_reencode(commit, encoding);
761 if (fmt == CMIT_FMT_ONELINE || fmt == CMIT_FMT_EMAIL)
764 /* After-subject is used to pass in Content-Type: multipart
765 * MIME header; in that case we do not have to do the
766 * plaintext content type even if the commit message has
767 * non 7-bit ASCII character. Otherwise, check if we need
768 * to say this is not a 7-bit ASCII.
770 if (fmt == CMIT_FMT_EMAIL && !after_subject) {
773 for (in_body = i = 0; (ch = msg[i]); i++) {
775 /* author could be non 7-bit ASCII but
776 * the log may be so; skip over the
779 if (ch == '\n' && msg[i+1] == '\n')
782 else if (non_ascii(ch)) {
789 pp_header(fmt, abbrev, dmode, encoding, commit, &msg, sb);
790 if (fmt != CMIT_FMT_ONELINE && !subject) {
791 strbuf_addch(sb, '\n');
794 /* Skip excess blank lines at the beginning of body, if any... */
796 int linelen = get_one_line(msg);
800 if (!is_empty_line(msg, &ll))
805 /* These formats treat the title line specially. */
806 if (fmt == CMIT_FMT_ONELINE || fmt == CMIT_FMT_EMAIL)
807 pp_title_line(fmt, &msg, sb, subject,
808 after_subject, encoding, plain_non_ascii);
810 beginning_of_body = sb->len;
811 if (fmt != CMIT_FMT_ONELINE)
812 pp_remainder(fmt, &msg, sb, indent);
815 /* Make sure there is an EOLN for the non-oneline case */
816 if (fmt != CMIT_FMT_ONELINE)
817 strbuf_addch(sb, '\n');
820 * The caller may append additional body text in e-mail
821 * format. Make sure we did not strip the blank line
822 * between the header and the body.
824 if (fmt == CMIT_FMT_EMAIL && sb->len <= beginning_of_body)
825 strbuf_addch(sb, '\n');