7 static char *user_format;
9 void get_commit_format(const char *arg, struct rev_info *rev)
12 static struct cmt_fmt_map {
17 { "raw", 1, CMIT_FMT_RAW },
18 { "medium", 1, CMIT_FMT_MEDIUM },
19 { "short", 1, CMIT_FMT_SHORT },
20 { "email", 1, CMIT_FMT_EMAIL },
21 { "full", 5, CMIT_FMT_FULL },
22 { "fuller", 5, CMIT_FMT_FULLER },
23 { "oneline", 1, CMIT_FMT_ONELINE },
26 rev->use_terminator = 0;
28 rev->commit_format = CMIT_FMT_DEFAULT;
33 if (!prefixcmp(arg, "format:") || !prefixcmp(arg, "tformat:")) {
34 const char *cp = strchr(arg, ':') + 1;
36 user_format = xstrdup(cp);
38 rev->use_terminator = 1;
39 rev->commit_format = CMIT_FMT_USERFORMAT;
42 for (i = 0; i < ARRAY_SIZE(cmt_fmts); i++) {
43 if (!strncmp(arg, cmt_fmts[i].n, cmt_fmts[i].cmp_len) &&
44 !strncmp(arg, cmt_fmts[i].n, strlen(arg))) {
45 if (cmt_fmts[i].v == CMIT_FMT_ONELINE)
46 rev->use_terminator = 1;
47 rev->commit_format = cmt_fmts[i].v;
52 die("invalid --pretty format: %s", arg);
56 * Generic support for pretty-printing the header
58 static int get_one_line(const char *msg)
73 /* High bit set, or ISO-2022-INT */
77 return ((ch & 0x80) || (ch == 0x1b));
80 static int is_rfc2047_special(char ch)
82 return (non_ascii(ch) || (ch == '=') || (ch == '?') || (ch == '_'));
85 static void add_rfc2047(struct strbuf *sb, const char *line, int len,
90 for (i = 0; i < len; i++) {
94 if ((i + 1 < len) && (ch == '=' && line[i+1] == '?'))
97 strbuf_add(sb, line, len);
101 strbuf_grow(sb, len * 3 + strlen(encoding) + 100);
102 strbuf_addf(sb, "=?%s?q?", encoding);
103 for (i = last = 0; i < len; i++) {
104 unsigned ch = line[i] & 0xFF;
106 * We encode ' ' using '=20' even though rfc2047
107 * allows using '_' for readability. Unfortunately,
108 * many programs do not understand this and just
109 * leave the underscore in place.
111 if (is_rfc2047_special(ch) || ch == ' ') {
112 strbuf_add(sb, line + last, i - last);
113 strbuf_addf(sb, "=%02X", ch);
117 strbuf_add(sb, line + last, len - last);
118 strbuf_addstr(sb, "?=");
121 void pp_user_info(const char *what, enum cmit_fmt fmt, struct strbuf *sb,
122 const char *line, enum date_mode dmode,
123 const char *encoding)
129 const char *filler = " ";
131 if (fmt == CMIT_FMT_ONELINE)
133 date = strchr(line, '>');
136 namelen = ++date - line;
137 time = strtoul(date, &date, 10);
138 tz = strtol(date, NULL, 10);
140 if (fmt == CMIT_FMT_EMAIL) {
141 char *name_tail = strchr(line, '<');
142 int display_name_length;
145 while (line < name_tail && isspace(name_tail[-1]))
147 display_name_length = name_tail - line;
149 strbuf_addstr(sb, "From: ");
150 add_rfc2047(sb, line, display_name_length, encoding);
151 strbuf_add(sb, name_tail, namelen - display_name_length);
152 strbuf_addch(sb, '\n');
154 strbuf_addf(sb, "%s: %.*s%.*s\n", what,
155 (fmt == CMIT_FMT_FULLER) ? 4 : 0,
156 filler, namelen, line);
159 case CMIT_FMT_MEDIUM:
160 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, dmode));
163 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, DATE_RFC2822));
165 case CMIT_FMT_FULLER:
166 strbuf_addf(sb, "%sDate: %s\n", what, show_date(time, tz, dmode));
174 static int is_empty_line(const char *line, int *len_p)
177 while (len && isspace(line[len-1]))
183 static void add_merge_info(enum cmit_fmt fmt, struct strbuf *sb,
184 const struct commit *commit, int abbrev)
186 struct commit_list *parent = commit->parents;
188 if ((fmt == CMIT_FMT_ONELINE) || (fmt == CMIT_FMT_EMAIL) ||
189 !parent || !parent->next)
192 strbuf_addstr(sb, "Merge:");
195 struct commit *p = parent->item;
196 const char *hex = NULL;
199 hex = find_unique_abbrev(p->object.sha1, abbrev);
201 hex = sha1_to_hex(p->object.sha1);
202 dots = (abbrev && strlen(hex) != 40) ? "..." : "";
203 parent = parent->next;
205 strbuf_addf(sb, " %s%s", hex, dots);
207 strbuf_addch(sb, '\n');
210 static char *get_header(const struct commit *commit, const char *key)
212 int key_len = strlen(key);
213 const char *line = commit->buffer;
216 const char *eol = strchr(line, '\n'), *next;
221 eol = line + strlen(line);
225 if (eol - line > key_len &&
226 !strncmp(line, key, key_len) &&
227 line[key_len] == ' ') {
228 return xmemdupz(line + key_len + 1, eol - line - key_len - 1);
234 static char *replace_encoding_header(char *buf, const char *encoding)
240 /* guess if there is an encoding header before a \n\n */
241 while (strncmp(cp, "encoding ", strlen("encoding "))) {
242 cp = strchr(cp, '\n');
243 if (!cp || *++cp == '\n')
247 cp = strchr(cp, '\n');
249 return buf; /* should not happen but be defensive */
250 len = cp + 1 - (buf + start);
252 strbuf_init(&tmp, 0);
253 strbuf_attach(&tmp, buf, strlen(buf), strlen(buf) + 1);
254 if (is_encoding_utf8(encoding)) {
255 /* we have re-coded to UTF-8; drop the header */
256 strbuf_remove(&tmp, start, len);
258 /* just replaces XXXX in 'encoding XXXX\n' */
259 strbuf_splice(&tmp, start + strlen("encoding "),
260 len - strlen("encoding \n"),
261 encoding, strlen(encoding));
263 return strbuf_detach(&tmp, NULL);
266 static char *logmsg_reencode(const struct commit *commit,
267 const char *output_encoding)
269 static const char *utf8 = "utf-8";
270 const char *use_encoding;
274 if (!*output_encoding)
276 encoding = get_header(commit, "encoding");
277 use_encoding = encoding ? encoding : utf8;
278 if (!strcmp(use_encoding, output_encoding))
279 if (encoding) /* we'll strip encoding header later */
280 out = xstrdup(commit->buffer);
282 return NULL; /* nothing to do */
284 out = reencode_string(commit->buffer,
285 output_encoding, use_encoding);
287 out = replace_encoding_header(out, output_encoding);
293 static size_t format_person_part(struct strbuf *sb, char part,
294 const char *msg, int len)
296 /* currently all placeholders have same length */
297 const int placeholder_len = 2;
298 int start, end, tz = 0;
299 unsigned long date = 0;
302 /* advance 'end' to point to email start delimiter */
303 for (end = 0; end < len && msg[end] != '<'; end++)
307 * When end points at the '<' that we found, it should have
308 * matching '>' later, which means 'end' must be strictly
314 if (part == 'n') { /* name */
315 while (end > 0 && isspace(msg[end - 1]))
317 strbuf_add(sb, msg, end);
318 return placeholder_len;
320 start = ++end; /* save email start position */
322 /* advance 'end' to point to email end delimiter */
323 for ( ; end < len && msg[end] != '>'; end++)
329 if (part == 'e') { /* email */
330 strbuf_add(sb, msg + start, end - start);
331 return placeholder_len;
334 /* advance 'start' to point to date start delimiter */
335 for (start = end + 1; start < len && isspace(msg[start]); start++)
339 date = strtoul(msg + start, &ep, 10);
340 if (msg + start == ep)
343 if (part == 't') { /* date, UNIX timestamp */
344 strbuf_add(sb, msg + start, ep - (msg + start));
345 return placeholder_len;
349 for (start = ep - msg + 1; start < len && isspace(msg[start]); start++)
351 if (start + 1 < len) {
352 tz = strtoul(msg + start + 1, NULL, 10);
353 if (msg[start] == '-')
359 strbuf_addstr(sb, show_date(date, tz, DATE_NORMAL));
360 return placeholder_len;
361 case 'D': /* date, RFC2822 style */
362 strbuf_addstr(sb, show_date(date, tz, DATE_RFC2822));
363 return placeholder_len;
364 case 'r': /* date, relative */
365 strbuf_addstr(sb, show_date(date, tz, DATE_RELATIVE));
366 return placeholder_len;
367 case 'i': /* date, ISO 8601 */
368 strbuf_addstr(sb, show_date(date, tz, DATE_ISO8601));
369 return placeholder_len;
374 * bogus commit, 'sb' cannot be updated, but we still need to
375 * compute a valid return value.
377 if (part == 'n' || part == 'e' || part == 't' || part == 'd'
378 || part == 'D' || part == 'r' || part == 'i')
379 return placeholder_len;
381 return 0; /* unknown placeholder */
389 struct format_commit_context {
390 const struct commit *commit;
392 /* These offsets are relative to the start of the commit message. */
393 int commit_header_parsed;
394 struct chunk subject;
396 struct chunk committer;
397 struct chunk encoding;
400 /* The following ones are relative to the result struct strbuf. */
401 struct chunk abbrev_commit_hash;
402 struct chunk abbrev_tree_hash;
403 struct chunk abbrev_parent_hashes;
406 static int add_again(struct strbuf *sb, struct chunk *chunk)
409 strbuf_adddup(sb, chunk->off, chunk->len);
414 * We haven't seen this chunk before. Our caller is surely
415 * going to add it the hard way now. Remember the most likely
416 * start of the to-be-added chunk: the current end of the
419 chunk->off = sb->len;
423 static void parse_commit_header(struct format_commit_context *context)
425 const char *msg = context->commit->buffer;
427 enum { HEADER, SUBJECT, BODY } state;
429 for (i = 0, state = HEADER; msg[i] && state < BODY; i++) {
431 for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
434 if (state == SUBJECT) {
435 context->subject.off = i;
436 context->subject.len = eol - i;
441 /* strip empty lines */
442 while (msg[eol] == '\n' && msg[eol + 1] == '\n')
444 } else if (!prefixcmp(msg + i, "author ")) {
445 context->author.off = i + 7;
446 context->author.len = eol - i - 7;
447 } else if (!prefixcmp(msg + i, "committer ")) {
448 context->committer.off = i + 10;
449 context->committer.len = eol - i - 10;
450 } else if (!prefixcmp(msg + i, "encoding ")) {
451 context->encoding.off = i + 9;
452 context->encoding.len = eol - i - 9;
458 context->body_off = i;
459 context->commit_header_parsed = 1;
462 static size_t format_commit_item(struct strbuf *sb, const char *placeholder,
465 struct format_commit_context *c = context;
466 const struct commit *commit = c->commit;
467 const char *msg = commit->buffer;
468 struct commit_list *p;
471 /* these are independent of the commit */
472 switch (placeholder[0]) {
474 if (!prefixcmp(placeholder + 1, "red")) {
475 strbuf_addstr(sb, "\033[31m");
477 } else if (!prefixcmp(placeholder + 1, "green")) {
478 strbuf_addstr(sb, "\033[32m");
480 } else if (!prefixcmp(placeholder + 1, "blue")) {
481 strbuf_addstr(sb, "\033[34m");
483 } else if (!prefixcmp(placeholder + 1, "reset")) {
484 strbuf_addstr(sb, "\033[m");
488 case 'n': /* newline */
489 strbuf_addch(sb, '\n');
492 /* %x00 == NUL, %x0a == LF, etc. */
493 if (0 <= (h1 = hexval_table[0xff & placeholder[1]]) &&
495 0 <= (h2 = hexval_table[0xff & placeholder[2]]) &&
497 strbuf_addch(sb, (h1<<4)|h2);
503 /* these depend on the commit */
504 if (!commit->object.parsed)
505 parse_object(commit->object.sha1);
507 switch (placeholder[0]) {
508 case 'H': /* commit hash */
509 strbuf_addstr(sb, sha1_to_hex(commit->object.sha1));
511 case 'h': /* abbreviated commit hash */
512 if (add_again(sb, &c->abbrev_commit_hash))
514 strbuf_addstr(sb, find_unique_abbrev(commit->object.sha1,
516 c->abbrev_commit_hash.len = sb->len - c->abbrev_commit_hash.off;
518 case 'T': /* tree hash */
519 strbuf_addstr(sb, sha1_to_hex(commit->tree->object.sha1));
521 case 't': /* abbreviated tree hash */
522 if (add_again(sb, &c->abbrev_tree_hash))
524 strbuf_addstr(sb, find_unique_abbrev(commit->tree->object.sha1,
526 c->abbrev_tree_hash.len = sb->len - c->abbrev_tree_hash.off;
528 case 'P': /* parent hashes */
529 for (p = commit->parents; p; p = p->next) {
530 if (p != commit->parents)
531 strbuf_addch(sb, ' ');
532 strbuf_addstr(sb, sha1_to_hex(p->item->object.sha1));
535 case 'p': /* abbreviated parent hashes */
536 if (add_again(sb, &c->abbrev_parent_hashes))
538 for (p = commit->parents; p; p = p->next) {
539 if (p != commit->parents)
540 strbuf_addch(sb, ' ');
541 strbuf_addstr(sb, find_unique_abbrev(
542 p->item->object.sha1, DEFAULT_ABBREV));
544 c->abbrev_parent_hashes.len = sb->len -
545 c->abbrev_parent_hashes.off;
547 case 'm': /* left/right/bottom */
548 strbuf_addch(sb, (commit->object.flags & BOUNDARY)
550 : (commit->object.flags & SYMMETRIC_LEFT)
556 /* For the rest we have to parse the commit header. */
557 if (!c->commit_header_parsed)
558 parse_commit_header(c);
560 switch (placeholder[0]) {
561 case 's': /* subject */
562 strbuf_add(sb, msg + c->subject.off, c->subject.len);
564 case 'a': /* author ... */
565 return format_person_part(sb, placeholder[1],
566 msg + c->author.off, c->author.len);
567 case 'c': /* committer ... */
568 return format_person_part(sb, placeholder[1],
569 msg + c->committer.off, c->committer.len);
570 case 'e': /* encoding */
571 strbuf_add(sb, msg + c->encoding.off, c->encoding.len);
574 strbuf_addstr(sb, msg + c->body_off);
577 return 0; /* unknown placeholder */
580 void format_commit_message(const struct commit *commit,
581 const void *format, struct strbuf *sb)
583 struct format_commit_context context;
585 memset(&context, 0, sizeof(context));
586 context.commit = commit;
587 strbuf_expand(sb, format, format_commit_item, &context);
590 static void pp_header(enum cmit_fmt fmt,
592 enum date_mode dmode,
593 const char *encoding,
594 const struct commit *commit,
598 int parents_shown = 0;
601 const char *line = *msg_p;
602 int linelen = get_one_line(*msg_p);
612 if (fmt == CMIT_FMT_RAW) {
613 strbuf_add(sb, line, linelen);
617 if (!memcmp(line, "parent ", 7)) {
619 die("bad parent line in commit");
623 if (!parents_shown) {
624 struct commit_list *parent;
626 for (parent = commit->parents, num = 0;
628 parent = parent->next, num++)
630 /* with enough slop */
631 strbuf_grow(sb, num * 50 + 20);
632 add_merge_info(fmt, sb, commit, abbrev);
637 * MEDIUM == DEFAULT shows only author with dates.
638 * FULL shows both authors but not dates.
639 * FULLER shows both authors and dates.
641 if (!memcmp(line, "author ", 7)) {
642 strbuf_grow(sb, linelen + 80);
643 pp_user_info("Author", fmt, sb, line + 7, dmode, encoding);
645 if (!memcmp(line, "committer ", 10) &&
646 (fmt == CMIT_FMT_FULL || fmt == CMIT_FMT_FULLER)) {
647 strbuf_grow(sb, linelen + 80);
648 pp_user_info("Commit", fmt, sb, line + 10, dmode, encoding);
653 void pp_title_line(enum cmit_fmt fmt,
657 const char *after_subject,
658 const char *encoding,
663 strbuf_init(&title, 80);
666 const char *line = *msg_p;
667 int linelen = get_one_line(line);
670 if (!linelen || is_empty_line(line, &linelen))
673 strbuf_grow(&title, linelen + 2);
675 if (fmt == CMIT_FMT_EMAIL) {
676 strbuf_addch(&title, '\n');
678 strbuf_addch(&title, ' ');
680 strbuf_add(&title, line, linelen);
683 strbuf_grow(sb, title.len + 1024);
685 strbuf_addstr(sb, subject);
686 add_rfc2047(sb, title.buf, title.len, encoding);
688 strbuf_addbuf(sb, &title);
690 strbuf_addch(sb, '\n');
692 if (need_8bit_cte > 0) {
693 const char *header_fmt =
694 "MIME-Version: 1.0\n"
695 "Content-Type: text/plain; charset=%s\n"
696 "Content-Transfer-Encoding: 8bit\n";
697 strbuf_addf(sb, header_fmt, encoding);
700 strbuf_addstr(sb, after_subject);
702 if (fmt == CMIT_FMT_EMAIL) {
703 strbuf_addch(sb, '\n');
705 strbuf_release(&title);
708 void pp_remainder(enum cmit_fmt fmt,
715 const char *line = *msg_p;
716 int linelen = get_one_line(line);
722 if (is_empty_line(line, &linelen)) {
725 if (fmt == CMIT_FMT_SHORT)
730 strbuf_grow(sb, linelen + indent + 20);
732 memset(sb->buf + sb->len, ' ', indent);
733 strbuf_setlen(sb, sb->len + indent);
735 strbuf_add(sb, line, linelen);
736 strbuf_addch(sb, '\n');
740 void pretty_print_commit(enum cmit_fmt fmt, const struct commit *commit,
741 struct strbuf *sb, int abbrev,
742 const char *subject, const char *after_subject,
743 enum date_mode dmode, int need_8bit_cte)
745 unsigned long beginning_of_body;
747 const char *msg = commit->buffer;
749 const char *encoding;
751 if (fmt == CMIT_FMT_USERFORMAT) {
752 format_commit_message(commit, user_format, sb);
756 encoding = (git_log_output_encoding
757 ? git_log_output_encoding
758 : git_commit_encoding);
761 reencoded = logmsg_reencode(commit, encoding);
766 if (fmt == CMIT_FMT_ONELINE || fmt == CMIT_FMT_EMAIL)
770 * We need to check and emit Content-type: to mark it
771 * as 8-bit if we haven't done so.
773 if (fmt == CMIT_FMT_EMAIL && need_8bit_cte == 0) {
776 for (in_body = i = 0; (ch = msg[i]); i++) {
778 /* author could be non 7-bit ASCII but
779 * the log may be so; skip over the
782 if (ch == '\n' && msg[i+1] == '\n')
785 else if (non_ascii(ch)) {
792 pp_header(fmt, abbrev, dmode, encoding, commit, &msg, sb);
793 if (fmt != CMIT_FMT_ONELINE && !subject) {
794 strbuf_addch(sb, '\n');
797 /* Skip excess blank lines at the beginning of body, if any... */
799 int linelen = get_one_line(msg);
803 if (!is_empty_line(msg, &ll))
808 /* These formats treat the title line specially. */
809 if (fmt == CMIT_FMT_ONELINE || fmt == CMIT_FMT_EMAIL)
810 pp_title_line(fmt, &msg, sb, subject,
811 after_subject, encoding, need_8bit_cte);
813 beginning_of_body = sb->len;
814 if (fmt != CMIT_FMT_ONELINE)
815 pp_remainder(fmt, &msg, sb, indent);
818 /* Make sure there is an EOLN for the non-oneline case */
819 if (fmt != CMIT_FMT_ONELINE)
820 strbuf_addch(sb, '\n');
823 * The caller may append additional body text in e-mail
824 * format. Make sure we did not strip the blank line
825 * between the header and the body.
827 if (fmt == CMIT_FMT_EMAIL && sb->len <= beginning_of_body)
828 strbuf_addch(sb, '\n');