Update draft release notes to 1.8.2
[git/jnareb-git.git] / pretty.c
blob07fc0628656c5e37eaf04c969a7273455a391f8d
1 #include "cache.h"
2 #include "commit.h"
3 #include "utf8.h"
4 #include "diff.h"
5 #include "revision.h"
6 #include "string-list.h"
7 #include "mailmap.h"
8 #include "log-tree.h"
9 #include "notes.h"
10 #include "color.h"
11 #include "reflog-walk.h"
12 #include "gpg-interface.h"
14 static char *user_format;
15 static struct cmt_fmt_map {
16 const char *name;
17 enum cmit_fmt format;
18 int is_tformat;
19 int is_alias;
20 const char *user_format;
21 } *commit_formats;
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 static void save_user_format(struct rev_info *rev, const char *cp, int is_tformat)
29 free(user_format);
30 user_format = xstrdup(cp);
31 if (is_tformat)
32 rev->use_terminator = 1;
33 rev->commit_format = CMIT_FMT_USERFORMAT;
36 static int git_pretty_formats_config(const char *var, const char *value, void *cb)
38 struct cmt_fmt_map *commit_format = NULL;
39 const char *name;
40 const char *fmt;
41 int i;
43 if (prefixcmp(var, "pretty."))
44 return 0;
46 name = var + strlen("pretty.");
47 for (i = 0; i < builtin_formats_len; i++) {
48 if (!strcmp(commit_formats[i].name, name))
49 return 0;
52 for (i = builtin_formats_len; i < commit_formats_len; i++) {
53 if (!strcmp(commit_formats[i].name, name)) {
54 commit_format = &commit_formats[i];
55 break;
59 if (!commit_format) {
60 ALLOC_GROW(commit_formats, commit_formats_len+1,
61 commit_formats_alloc);
62 commit_format = &commit_formats[commit_formats_len];
63 memset(commit_format, 0, sizeof(*commit_format));
64 commit_formats_len++;
67 commit_format->name = xstrdup(name);
68 commit_format->format = CMIT_FMT_USERFORMAT;
69 git_config_string(&fmt, var, value);
70 if (!prefixcmp(fmt, "format:") || !prefixcmp(fmt, "tformat:")) {
71 commit_format->is_tformat = fmt[0] == 't';
72 fmt = strchr(fmt, ':') + 1;
73 } else if (strchr(fmt, '%'))
74 commit_format->is_tformat = 1;
75 else
76 commit_format->is_alias = 1;
77 commit_format->user_format = fmt;
79 return 0;
82 static void setup_commit_formats(void)
84 struct cmt_fmt_map builtin_formats[] = {
85 { "raw", CMIT_FMT_RAW, 0 },
86 { "medium", CMIT_FMT_MEDIUM, 0 },
87 { "short", CMIT_FMT_SHORT, 0 },
88 { "email", CMIT_FMT_EMAIL, 0 },
89 { "fuller", CMIT_FMT_FULLER, 0 },
90 { "full", CMIT_FMT_FULL, 0 },
91 { "oneline", CMIT_FMT_ONELINE, 1 }
93 commit_formats_len = ARRAY_SIZE(builtin_formats);
94 builtin_formats_len = commit_formats_len;
95 ALLOC_GROW(commit_formats, commit_formats_len, commit_formats_alloc);
96 memcpy(commit_formats, builtin_formats,
97 sizeof(*builtin_formats)*ARRAY_SIZE(builtin_formats));
99 git_config(git_pretty_formats_config, NULL);
102 static struct cmt_fmt_map *find_commit_format_recursive(const char *sought,
103 const char *original,
104 int num_redirections)
106 struct cmt_fmt_map *found = NULL;
107 size_t found_match_len = 0;
108 int i;
110 if (num_redirections >= commit_formats_len)
111 die("invalid --pretty format: "
112 "'%s' references an alias which points to itself",
113 original);
115 for (i = 0; i < commit_formats_len; i++) {
116 size_t match_len;
118 if (prefixcmp(commit_formats[i].name, sought))
119 continue;
121 match_len = strlen(commit_formats[i].name);
122 if (found == NULL || found_match_len > match_len) {
123 found = &commit_formats[i];
124 found_match_len = match_len;
128 if (found && found->is_alias) {
129 found = find_commit_format_recursive(found->user_format,
130 original,
131 num_redirections+1);
134 return found;
137 static struct cmt_fmt_map *find_commit_format(const char *sought)
139 if (!commit_formats)
140 setup_commit_formats();
142 return find_commit_format_recursive(sought, sought, 0);
145 void get_commit_format(const char *arg, struct rev_info *rev)
147 struct cmt_fmt_map *commit_format;
149 rev->use_terminator = 0;
150 if (!arg || !*arg) {
151 rev->commit_format = CMIT_FMT_DEFAULT;
152 return;
154 if (!prefixcmp(arg, "format:") || !prefixcmp(arg, "tformat:")) {
155 save_user_format(rev, strchr(arg, ':') + 1, arg[0] == 't');
156 return;
159 if (strchr(arg, '%')) {
160 save_user_format(rev, arg, 1);
161 return;
164 commit_format = find_commit_format(arg);
165 if (!commit_format)
166 die("invalid --pretty format: %s", arg);
168 rev->commit_format = commit_format->format;
169 rev->use_terminator = commit_format->is_tformat;
170 if (commit_format->format == CMIT_FMT_USERFORMAT) {
171 save_user_format(rev, commit_format->user_format,
172 commit_format->is_tformat);
177 * Generic support for pretty-printing the header
179 static int get_one_line(const char *msg)
181 int ret = 0;
183 for (;;) {
184 char c = *msg++;
185 if (!c)
186 break;
187 ret++;
188 if (c == '\n')
189 break;
191 return ret;
194 /* High bit set, or ISO-2022-INT */
195 static int non_ascii(int ch)
197 return !isascii(ch) || ch == '\033';
200 int has_non_ascii(const char *s)
202 int ch;
203 if (!s)
204 return 0;
205 while ((ch = *s++) != '\0') {
206 if (non_ascii(ch))
207 return 1;
209 return 0;
212 static int is_rfc822_special(char ch)
214 switch (ch) {
215 case '(':
216 case ')':
217 case '<':
218 case '>':
219 case '[':
220 case ']':
221 case ':':
222 case ';':
223 case '@':
224 case ',':
225 case '.':
226 case '"':
227 case '\\':
228 return 1;
229 default:
230 return 0;
234 static int needs_rfc822_quoting(const char *s, int len)
236 int i;
237 for (i = 0; i < len; i++)
238 if (is_rfc822_special(s[i]))
239 return 1;
240 return 0;
243 static int last_line_length(struct strbuf *sb)
245 int i;
247 /* How many bytes are already used on the last line? */
248 for (i = sb->len - 1; i >= 0; i--)
249 if (sb->buf[i] == '\n')
250 break;
251 return sb->len - (i + 1);
254 static void add_rfc822_quoted(struct strbuf *out, const char *s, int len)
256 int i;
258 /* just a guess, we may have to also backslash-quote */
259 strbuf_grow(out, len + 2);
261 strbuf_addch(out, '"');
262 for (i = 0; i < len; i++) {
263 switch (s[i]) {
264 case '"':
265 case '\\':
266 strbuf_addch(out, '\\');
267 /* fall through */
268 default:
269 strbuf_addch(out, s[i]);
272 strbuf_addch(out, '"');
275 enum rfc2047_type {
276 RFC2047_SUBJECT,
277 RFC2047_ADDRESS,
280 static int is_rfc2047_special(char ch, enum rfc2047_type type)
283 * rfc2047, section 4.2:
285 * 8-bit values which correspond to printable ASCII characters other
286 * than "=", "?", and "_" (underscore), MAY be represented as those
287 * characters. (But see section 5 for restrictions.) In
288 * particular, SPACE and TAB MUST NOT be represented as themselves
289 * within encoded words.
293 * rule out non-ASCII characters and non-printable characters (the
294 * non-ASCII check should be redundant as isprint() is not localized
295 * and only knows about ASCII, but be defensive about that)
297 if (non_ascii(ch) || !isprint(ch))
298 return 1;
301 * rule out special printable characters (' ' should be the only
302 * whitespace character considered printable, but be defensive and use
303 * isspace())
305 if (isspace(ch) || ch == '=' || ch == '?' || ch == '_')
306 return 1;
309 * rfc2047, section 5.3:
311 * As a replacement for a 'word' entity within a 'phrase', for example,
312 * one that precedes an address in a From, To, or Cc header. The ABNF
313 * definition for 'phrase' from RFC 822 thus becomes:
315 * phrase = 1*( encoded-word / word )
317 * In this case the set of characters that may be used in a "Q"-encoded
318 * 'encoded-word' is restricted to: <upper and lower case ASCII
319 * letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
320 * (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
321 * 'phrase' MUST be separated from any adjacent 'word', 'text' or
322 * 'special' by 'linear-white-space'.
325 if (type != RFC2047_ADDRESS)
326 return 0;
328 /* '=' and '_' are special cases and have been checked above */
329 return !(isalnum(ch) || ch == '!' || ch == '*' || ch == '+' || ch == '-' || ch == '/');
332 static int needs_rfc2047_encoding(const char *line, int len,
333 enum rfc2047_type type)
335 int i;
337 for (i = 0; i < len; i++) {
338 int ch = line[i];
339 if (non_ascii(ch) || ch == '\n')
340 return 1;
341 if ((i + 1 < len) && (ch == '=' && line[i+1] == '?'))
342 return 1;
345 return 0;
348 static void add_rfc2047(struct strbuf *sb, const char *line, int len,
349 const char *encoding, enum rfc2047_type type)
351 static const int max_encoded_length = 76; /* per rfc2047 */
352 int i;
353 int line_len = last_line_length(sb);
355 strbuf_grow(sb, len * 3 + strlen(encoding) + 100);
356 strbuf_addf(sb, "=?%s?q?", encoding);
357 line_len += strlen(encoding) + 5; /* 5 for =??q? */
358 for (i = 0; i < len; i++) {
359 unsigned ch = line[i] & 0xFF;
360 int is_special = is_rfc2047_special(ch, type);
363 * According to RFC 2047, we could encode the special character
364 * ' ' (space) with '_' (underscore) for readability. But many
365 * programs do not understand this and just leave the
366 * underscore in place. Thus, we do nothing special here, which
367 * causes ' ' to be encoded as '=20', avoiding this problem.
370 if (line_len + 2 + (is_special ? 3 : 1) > max_encoded_length) {
371 strbuf_addf(sb, "?=\n =?%s?q?", encoding);
372 line_len = strlen(encoding) + 5 + 1; /* =??q? plus SP */
375 if (is_special) {
376 strbuf_addf(sb, "=%02X", ch);
377 line_len += 3;
378 } else {
379 strbuf_addch(sb, ch);
380 line_len++;
383 strbuf_addstr(sb, "?=");
386 void pp_user_info(const struct pretty_print_context *pp,
387 const char *what, struct strbuf *sb,
388 const char *line, const char *encoding)
390 struct strbuf name;
391 struct strbuf mail;
392 struct ident_split ident;
393 int linelen;
394 char *line_end, *date;
395 const char *mailbuf, *namebuf;
396 size_t namelen, maillen;
397 int max_length = 78; /* per rfc2822 */
398 unsigned long time;
399 int tz;
401 if (pp->fmt == CMIT_FMT_ONELINE)
402 return;
404 line_end = strchr(line, '\n');
405 if (!line_end) {
406 line_end = strchr(line, '\0');
407 if (!line_end)
408 return;
411 linelen = ++line_end - line;
412 if (split_ident_line(&ident, line, linelen))
413 return;
416 mailbuf = ident.mail_begin;
417 maillen = ident.mail_end - ident.mail_begin;
418 namebuf = ident.name_begin;
419 namelen = ident.name_end - ident.name_begin;
421 if (pp->mailmap)
422 map_user(pp->mailmap, &mailbuf, &maillen, &namebuf, &namelen);
424 strbuf_init(&mail, 0);
425 strbuf_init(&name, 0);
427 strbuf_add(&mail, mailbuf, maillen);
428 strbuf_add(&name, namebuf, namelen);
430 namelen = name.len + mail.len + 3; /* ' ' + '<' + '>' */
431 time = strtoul(ident.date_begin, &date, 10);
432 tz = strtol(date, NULL, 10);
434 if (pp->fmt == CMIT_FMT_EMAIL) {
435 strbuf_addstr(sb, "From: ");
436 if (needs_rfc2047_encoding(name.buf, name.len, RFC2047_ADDRESS)) {
437 add_rfc2047(sb, name.buf, name.len,
438 encoding, RFC2047_ADDRESS);
439 max_length = 76; /* per rfc2047 */
440 } else if (needs_rfc822_quoting(name.buf, name.len)) {
441 struct strbuf quoted = STRBUF_INIT;
442 add_rfc822_quoted(&quoted, name.buf, name.len);
443 strbuf_add_wrapped_bytes(sb, quoted.buf, quoted.len,
444 -6, 1, max_length);
445 strbuf_release(&quoted);
446 } else {
447 strbuf_add_wrapped_bytes(sb, name.buf, name.len,
448 -6, 1, max_length);
450 if (namelen - name.len + last_line_length(sb) > max_length)
451 strbuf_addch(sb, '\n');
453 strbuf_addf(sb, " <%s>\n", mail.buf);
454 } else {
455 strbuf_addf(sb, "%s: %.*s%s <%s>\n", what,
456 (pp->fmt == CMIT_FMT_FULLER) ? 4 : 0,
457 " ", name.buf, mail.buf);
460 strbuf_release(&mail);
461 strbuf_release(&name);
463 switch (pp->fmt) {
464 case CMIT_FMT_MEDIUM:
465 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, pp->date_mode));
466 break;
467 case CMIT_FMT_EMAIL:
468 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, DATE_RFC2822));
469 break;
470 case CMIT_FMT_FULLER:
471 strbuf_addf(sb, "%sDate: %s\n", what, show_date(time, tz, pp->date_mode));
472 break;
473 default:
474 /* notin' */
475 break;
479 static int is_empty_line(const char *line, int *len_p)
481 int len = *len_p;
482 while (len && isspace(line[len-1]))
483 len--;
484 *len_p = len;
485 return !len;
488 static const char *skip_empty_lines(const char *msg)
490 for (;;) {
491 int linelen = get_one_line(msg);
492 int ll = linelen;
493 if (!linelen)
494 break;
495 if (!is_empty_line(msg, &ll))
496 break;
497 msg += linelen;
499 return msg;
502 static void add_merge_info(const struct pretty_print_context *pp,
503 struct strbuf *sb, const struct commit *commit)
505 struct commit_list *parent = commit->parents;
507 if ((pp->fmt == CMIT_FMT_ONELINE) || (pp->fmt == CMIT_FMT_EMAIL) ||
508 !parent || !parent->next)
509 return;
511 strbuf_addstr(sb, "Merge:");
513 while (parent) {
514 struct commit *p = parent->item;
515 const char *hex = NULL;
516 if (pp->abbrev)
517 hex = find_unique_abbrev(p->object.sha1, pp->abbrev);
518 if (!hex)
519 hex = sha1_to_hex(p->object.sha1);
520 parent = parent->next;
522 strbuf_addf(sb, " %s", hex);
524 strbuf_addch(sb, '\n');
527 static char *get_header(const struct commit *commit, const char *key)
529 int key_len = strlen(key);
530 const char *line = commit->buffer;
532 while (line) {
533 const char *eol = strchr(line, '\n'), *next;
535 if (line == eol)
536 return NULL;
537 if (!eol) {
538 warning("malformed commit (header is missing newline): %s",
539 sha1_to_hex(commit->object.sha1));
540 eol = line + strlen(line);
541 next = NULL;
542 } else
543 next = eol + 1;
544 if (eol - line > key_len &&
545 !strncmp(line, key, key_len) &&
546 line[key_len] == ' ') {
547 return xmemdupz(line + key_len + 1, eol - line - key_len - 1);
549 line = next;
551 return NULL;
554 static char *replace_encoding_header(char *buf, const char *encoding)
556 struct strbuf tmp = STRBUF_INIT;
557 size_t start, len;
558 char *cp = buf;
560 /* guess if there is an encoding header before a \n\n */
561 while (strncmp(cp, "encoding ", strlen("encoding "))) {
562 cp = strchr(cp, '\n');
563 if (!cp || *++cp == '\n')
564 return buf;
566 start = cp - buf;
567 cp = strchr(cp, '\n');
568 if (!cp)
569 return buf; /* should not happen but be defensive */
570 len = cp + 1 - (buf + start);
572 strbuf_attach(&tmp, buf, strlen(buf), strlen(buf) + 1);
573 if (is_encoding_utf8(encoding)) {
574 /* we have re-coded to UTF-8; drop the header */
575 strbuf_remove(&tmp, start, len);
576 } else {
577 /* just replaces XXXX in 'encoding XXXX\n' */
578 strbuf_splice(&tmp, start + strlen("encoding "),
579 len - strlen("encoding \n"),
580 encoding, strlen(encoding));
582 return strbuf_detach(&tmp, NULL);
585 char *logmsg_reencode(const struct commit *commit,
586 const char *output_encoding)
588 static const char *utf8 = "UTF-8";
589 const char *use_encoding;
590 char *encoding;
591 char *out;
593 if (!output_encoding || !*output_encoding)
594 return NULL;
595 encoding = get_header(commit, "encoding");
596 use_encoding = encoding ? encoding : utf8;
597 if (same_encoding(use_encoding, output_encoding))
598 if (encoding) /* we'll strip encoding header later */
599 out = xstrdup(commit->buffer);
600 else
601 return NULL; /* nothing to do */
602 else
603 out = reencode_string(commit->buffer,
604 output_encoding, use_encoding);
605 if (out)
606 out = replace_encoding_header(out, output_encoding);
608 free(encoding);
609 return out;
612 static int mailmap_name(const char **email, size_t *email_len,
613 const char **name, size_t *name_len)
615 static struct string_list *mail_map;
616 if (!mail_map) {
617 mail_map = xcalloc(1, sizeof(*mail_map));
618 read_mailmap(mail_map, NULL);
620 return mail_map->nr && map_user(mail_map, email, email_len, name, name_len);
623 static size_t format_person_part(struct strbuf *sb, char part,
624 const char *msg, int len, enum date_mode dmode)
626 /* currently all placeholders have same length */
627 const int placeholder_len = 2;
628 int tz;
629 unsigned long date = 0;
630 struct ident_split s;
631 const char *name, *mail;
632 size_t maillen, namelen;
634 if (split_ident_line(&s, msg, len) < 0)
635 goto skip;
637 name = s.name_begin;
638 namelen = s.name_end - s.name_begin;
639 mail = s.mail_begin;
640 maillen = s.mail_end - s.mail_begin;
642 if (part == 'N' || part == 'E') /* mailmap lookup */
643 mailmap_name(&mail, &maillen, &name, &namelen);
644 if (part == 'n' || part == 'N') { /* name */
645 strbuf_add(sb, name, namelen);
646 return placeholder_len;
648 if (part == 'e' || part == 'E') { /* email */
649 strbuf_add(sb, mail, maillen);
650 return placeholder_len;
653 if (!s.date_begin)
654 goto skip;
656 date = strtoul(s.date_begin, NULL, 10);
658 if (part == 't') { /* date, UNIX timestamp */
659 strbuf_add(sb, s.date_begin, s.date_end - s.date_begin);
660 return placeholder_len;
663 /* parse tz */
664 tz = strtoul(s.tz_begin + 1, NULL, 10);
665 if (*s.tz_begin == '-')
666 tz = -tz;
668 switch (part) {
669 case 'd': /* date */
670 strbuf_addstr(sb, show_date(date, tz, dmode));
671 return placeholder_len;
672 case 'D': /* date, RFC2822 style */
673 strbuf_addstr(sb, show_date(date, tz, DATE_RFC2822));
674 return placeholder_len;
675 case 'r': /* date, relative */
676 strbuf_addstr(sb, show_date(date, tz, DATE_RELATIVE));
677 return placeholder_len;
678 case 'i': /* date, ISO 8601 */
679 strbuf_addstr(sb, show_date(date, tz, DATE_ISO8601));
680 return placeholder_len;
683 skip:
685 * reading from either a bogus commit, or a reflog entry with
686 * %gn, %ge, etc.; 'sb' cannot be updated, but we still need
687 * to compute a valid return value.
689 if (part == 'n' || part == 'e' || part == 't' || part == 'd'
690 || part == 'D' || part == 'r' || part == 'i')
691 return placeholder_len;
693 return 0; /* unknown placeholder */
696 struct chunk {
697 size_t off;
698 size_t len;
701 struct format_commit_context {
702 const struct commit *commit;
703 const struct pretty_print_context *pretty_ctx;
704 unsigned commit_header_parsed:1;
705 unsigned commit_message_parsed:1;
706 unsigned commit_signature_parsed:1;
707 struct {
708 char *gpg_output;
709 char good_bad;
710 char *signer;
711 } signature;
712 char *message;
713 size_t width, indent1, indent2;
715 /* These offsets are relative to the start of the commit message. */
716 struct chunk author;
717 struct chunk committer;
718 struct chunk encoding;
719 size_t message_off;
720 size_t subject_off;
721 size_t body_off;
723 /* The following ones are relative to the result struct strbuf. */
724 struct chunk abbrev_commit_hash;
725 struct chunk abbrev_tree_hash;
726 struct chunk abbrev_parent_hashes;
727 size_t wrap_start;
730 static int add_again(struct strbuf *sb, struct chunk *chunk)
732 if (chunk->len) {
733 strbuf_adddup(sb, chunk->off, chunk->len);
734 return 1;
738 * We haven't seen this chunk before. Our caller is surely
739 * going to add it the hard way now. Remember the most likely
740 * start of the to-be-added chunk: the current end of the
741 * struct strbuf.
743 chunk->off = sb->len;
744 return 0;
747 static void parse_commit_header(struct format_commit_context *context)
749 const char *msg = context->message;
750 int i;
752 for (i = 0; msg[i]; i++) {
753 int eol;
754 for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
755 ; /* do nothing */
757 if (i == eol) {
758 break;
759 } else if (!prefixcmp(msg + i, "author ")) {
760 context->author.off = i + 7;
761 context->author.len = eol - i - 7;
762 } else if (!prefixcmp(msg + i, "committer ")) {
763 context->committer.off = i + 10;
764 context->committer.len = eol - i - 10;
765 } else if (!prefixcmp(msg + i, "encoding ")) {
766 context->encoding.off = i + 9;
767 context->encoding.len = eol - i - 9;
769 i = eol;
771 context->message_off = i;
772 context->commit_header_parsed = 1;
775 static int istitlechar(char c)
777 return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') ||
778 (c >= '0' && c <= '9') || c == '.' || c == '_';
781 static void format_sanitized_subject(struct strbuf *sb, const char *msg)
783 size_t trimlen;
784 size_t start_len = sb->len;
785 int space = 2;
787 for (; *msg && *msg != '\n'; msg++) {
788 if (istitlechar(*msg)) {
789 if (space == 1)
790 strbuf_addch(sb, '-');
791 space = 0;
792 strbuf_addch(sb, *msg);
793 if (*msg == '.')
794 while (*(msg+1) == '.')
795 msg++;
796 } else
797 space |= 1;
800 /* trim any trailing '.' or '-' characters */
801 trimlen = 0;
802 while (sb->len - trimlen > start_len &&
803 (sb->buf[sb->len - 1 - trimlen] == '.'
804 || sb->buf[sb->len - 1 - trimlen] == '-'))
805 trimlen++;
806 strbuf_remove(sb, sb->len - trimlen, trimlen);
809 const char *format_subject(struct strbuf *sb, const char *msg,
810 const char *line_separator)
812 int first = 1;
814 for (;;) {
815 const char *line = msg;
816 int linelen = get_one_line(line);
818 msg += linelen;
819 if (!linelen || is_empty_line(line, &linelen))
820 break;
822 if (!sb)
823 continue;
824 strbuf_grow(sb, linelen + 2);
825 if (!first)
826 strbuf_addstr(sb, line_separator);
827 strbuf_add(sb, line, linelen);
828 first = 0;
830 return msg;
833 static void parse_commit_message(struct format_commit_context *c)
835 const char *msg = c->message + c->message_off;
836 const char *start = c->message;
838 msg = skip_empty_lines(msg);
839 c->subject_off = msg - start;
841 msg = format_subject(NULL, msg, NULL);
842 msg = skip_empty_lines(msg);
843 c->body_off = msg - start;
845 c->commit_message_parsed = 1;
848 static void format_decoration(struct strbuf *sb, const struct commit *commit)
850 struct name_decoration *d;
851 const char *prefix = " (";
853 load_ref_decorations(DECORATE_SHORT_REFS);
854 d = lookup_decoration(&name_decoration, &commit->object);
855 while (d) {
856 strbuf_addstr(sb, prefix);
857 prefix = ", ";
858 strbuf_addstr(sb, d->name);
859 d = d->next;
861 if (prefix[0] == ',')
862 strbuf_addch(sb, ')');
865 static void strbuf_wrap(struct strbuf *sb, size_t pos,
866 size_t width, size_t indent1, size_t indent2)
868 struct strbuf tmp = STRBUF_INIT;
870 if (pos)
871 strbuf_add(&tmp, sb->buf, pos);
872 strbuf_add_wrapped_text(&tmp, sb->buf + pos,
873 (int) indent1, (int) indent2, (int) width);
874 strbuf_swap(&tmp, sb);
875 strbuf_release(&tmp);
878 static void rewrap_message_tail(struct strbuf *sb,
879 struct format_commit_context *c,
880 size_t new_width, size_t new_indent1,
881 size_t new_indent2)
883 if (c->width == new_width && c->indent1 == new_indent1 &&
884 c->indent2 == new_indent2)
885 return;
886 if (c->wrap_start < sb->len)
887 strbuf_wrap(sb, c->wrap_start, c->width, c->indent1, c->indent2);
888 c->wrap_start = sb->len;
889 c->width = new_width;
890 c->indent1 = new_indent1;
891 c->indent2 = new_indent2;
894 static struct {
895 char result;
896 const char *check;
897 } signature_check[] = {
898 { 'G', ": Good signature from " },
899 { 'B', ": BAD signature from " },
902 static void parse_signature_lines(struct format_commit_context *ctx)
904 const char *buf = ctx->signature.gpg_output;
905 int i;
907 for (i = 0; i < ARRAY_SIZE(signature_check); i++) {
908 const char *found = strstr(buf, signature_check[i].check);
909 const char *next;
910 if (!found)
911 continue;
912 ctx->signature.good_bad = signature_check[i].result;
913 found += strlen(signature_check[i].check);
914 next = strchrnul(found, '\n');
915 ctx->signature.signer = xmemdupz(found, next - found);
916 break;
920 static void parse_commit_signature(struct format_commit_context *ctx)
922 struct strbuf payload = STRBUF_INIT;
923 struct strbuf signature = STRBUF_INIT;
924 struct strbuf gpg_output = STRBUF_INIT;
925 int status;
927 ctx->commit_signature_parsed = 1;
929 if (parse_signed_commit(ctx->commit->object.sha1,
930 &payload, &signature) <= 0)
931 goto out;
932 status = verify_signed_buffer(payload.buf, payload.len,
933 signature.buf, signature.len,
934 &gpg_output);
935 if (status && !gpg_output.len)
936 goto out;
937 ctx->signature.gpg_output = strbuf_detach(&gpg_output, NULL);
938 parse_signature_lines(ctx);
940 out:
941 strbuf_release(&gpg_output);
942 strbuf_release(&payload);
943 strbuf_release(&signature);
947 static int format_reflog_person(struct strbuf *sb,
948 char part,
949 struct reflog_walk_info *log,
950 enum date_mode dmode)
952 const char *ident;
954 if (!log)
955 return 2;
957 ident = get_reflog_ident(log);
958 if (!ident)
959 return 2;
961 return format_person_part(sb, part, ident, strlen(ident), dmode);
964 static size_t format_commit_one(struct strbuf *sb, const char *placeholder,
965 void *context)
967 struct format_commit_context *c = context;
968 const struct commit *commit = c->commit;
969 const char *msg = c->message;
970 struct commit_list *p;
971 int h1, h2;
973 /* these are independent of the commit */
974 switch (placeholder[0]) {
975 case 'C':
976 if (placeholder[1] == '(') {
977 const char *begin = placeholder + 2;
978 const char *end = strchr(begin, ')');
979 char color[COLOR_MAXLEN];
981 if (!end)
982 return 0;
983 if (!prefixcmp(begin, "auto,")) {
984 if (!want_color(c->pretty_ctx->color))
985 return end - placeholder + 1;
986 begin += 5;
988 color_parse_mem(begin,
989 end - begin,
990 "--pretty format", color);
991 strbuf_addstr(sb, color);
992 return end - placeholder + 1;
994 if (!prefixcmp(placeholder + 1, "red")) {
995 strbuf_addstr(sb, GIT_COLOR_RED);
996 return 4;
997 } else if (!prefixcmp(placeholder + 1, "green")) {
998 strbuf_addstr(sb, GIT_COLOR_GREEN);
999 return 6;
1000 } else if (!prefixcmp(placeholder + 1, "blue")) {
1001 strbuf_addstr(sb, GIT_COLOR_BLUE);
1002 return 5;
1003 } else if (!prefixcmp(placeholder + 1, "reset")) {
1004 strbuf_addstr(sb, GIT_COLOR_RESET);
1005 return 6;
1006 } else
1007 return 0;
1008 case 'n': /* newline */
1009 strbuf_addch(sb, '\n');
1010 return 1;
1011 case 'x':
1012 /* %x00 == NUL, %x0a == LF, etc. */
1013 if (0 <= (h1 = hexval_table[0xff & placeholder[1]]) &&
1014 h1 <= 16 &&
1015 0 <= (h2 = hexval_table[0xff & placeholder[2]]) &&
1016 h2 <= 16) {
1017 strbuf_addch(sb, (h1<<4)|h2);
1018 return 3;
1019 } else
1020 return 0;
1021 case 'w':
1022 if (placeholder[1] == '(') {
1023 unsigned long width = 0, indent1 = 0, indent2 = 0;
1024 char *next;
1025 const char *start = placeholder + 2;
1026 const char *end = strchr(start, ')');
1027 if (!end)
1028 return 0;
1029 if (end > start) {
1030 width = strtoul(start, &next, 10);
1031 if (*next == ',') {
1032 indent1 = strtoul(next + 1, &next, 10);
1033 if (*next == ',') {
1034 indent2 = strtoul(next + 1,
1035 &next, 10);
1038 if (*next != ')')
1039 return 0;
1041 rewrap_message_tail(sb, c, width, indent1, indent2);
1042 return end - placeholder + 1;
1043 } else
1044 return 0;
1047 /* these depend on the commit */
1048 if (!commit->object.parsed)
1049 parse_object(commit->object.sha1);
1051 switch (placeholder[0]) {
1052 case 'H': /* commit hash */
1053 strbuf_addstr(sb, sha1_to_hex(commit->object.sha1));
1054 return 1;
1055 case 'h': /* abbreviated commit hash */
1056 if (add_again(sb, &c->abbrev_commit_hash))
1057 return 1;
1058 strbuf_addstr(sb, find_unique_abbrev(commit->object.sha1,
1059 c->pretty_ctx->abbrev));
1060 c->abbrev_commit_hash.len = sb->len - c->abbrev_commit_hash.off;
1061 return 1;
1062 case 'T': /* tree hash */
1063 strbuf_addstr(sb, sha1_to_hex(commit->tree->object.sha1));
1064 return 1;
1065 case 't': /* abbreviated tree hash */
1066 if (add_again(sb, &c->abbrev_tree_hash))
1067 return 1;
1068 strbuf_addstr(sb, find_unique_abbrev(commit->tree->object.sha1,
1069 c->pretty_ctx->abbrev));
1070 c->abbrev_tree_hash.len = sb->len - c->abbrev_tree_hash.off;
1071 return 1;
1072 case 'P': /* parent hashes */
1073 for (p = commit->parents; p; p = p->next) {
1074 if (p != commit->parents)
1075 strbuf_addch(sb, ' ');
1076 strbuf_addstr(sb, sha1_to_hex(p->item->object.sha1));
1078 return 1;
1079 case 'p': /* abbreviated parent hashes */
1080 if (add_again(sb, &c->abbrev_parent_hashes))
1081 return 1;
1082 for (p = commit->parents; p; p = p->next) {
1083 if (p != commit->parents)
1084 strbuf_addch(sb, ' ');
1085 strbuf_addstr(sb, find_unique_abbrev(
1086 p->item->object.sha1,
1087 c->pretty_ctx->abbrev));
1089 c->abbrev_parent_hashes.len = sb->len -
1090 c->abbrev_parent_hashes.off;
1091 return 1;
1092 case 'm': /* left/right/bottom */
1093 strbuf_addstr(sb, get_revision_mark(NULL, commit));
1094 return 1;
1095 case 'd':
1096 format_decoration(sb, commit);
1097 return 1;
1098 case 'g': /* reflog info */
1099 switch(placeholder[1]) {
1100 case 'd': /* reflog selector */
1101 case 'D':
1102 if (c->pretty_ctx->reflog_info)
1103 get_reflog_selector(sb,
1104 c->pretty_ctx->reflog_info,
1105 c->pretty_ctx->date_mode,
1106 c->pretty_ctx->date_mode_explicit,
1107 (placeholder[1] == 'd'));
1108 return 2;
1109 case 's': /* reflog message */
1110 if (c->pretty_ctx->reflog_info)
1111 get_reflog_message(sb, c->pretty_ctx->reflog_info);
1112 return 2;
1113 case 'n':
1114 case 'N':
1115 case 'e':
1116 case 'E':
1117 return format_reflog_person(sb,
1118 placeholder[1],
1119 c->pretty_ctx->reflog_info,
1120 c->pretty_ctx->date_mode);
1122 return 0; /* unknown %g placeholder */
1123 case 'N':
1124 if (c->pretty_ctx->notes_message) {
1125 strbuf_addstr(sb, c->pretty_ctx->notes_message);
1126 return 1;
1128 return 0;
1131 if (placeholder[0] == 'G') {
1132 if (!c->commit_signature_parsed)
1133 parse_commit_signature(c);
1134 switch (placeholder[1]) {
1135 case 'G':
1136 if (c->signature.gpg_output)
1137 strbuf_addstr(sb, c->signature.gpg_output);
1138 break;
1139 case '?':
1140 switch (c->signature.good_bad) {
1141 case 'G':
1142 case 'B':
1143 strbuf_addch(sb, c->signature.good_bad);
1145 break;
1146 case 'S':
1147 if (c->signature.signer)
1148 strbuf_addstr(sb, c->signature.signer);
1149 break;
1151 return 2;
1155 /* For the rest we have to parse the commit header. */
1156 if (!c->commit_header_parsed)
1157 parse_commit_header(c);
1159 switch (placeholder[0]) {
1160 case 'a': /* author ... */
1161 return format_person_part(sb, placeholder[1],
1162 msg + c->author.off, c->author.len,
1163 c->pretty_ctx->date_mode);
1164 case 'c': /* committer ... */
1165 return format_person_part(sb, placeholder[1],
1166 msg + c->committer.off, c->committer.len,
1167 c->pretty_ctx->date_mode);
1168 case 'e': /* encoding */
1169 strbuf_add(sb, msg + c->encoding.off, c->encoding.len);
1170 return 1;
1171 case 'B': /* raw body */
1172 /* message_off is always left at the initial newline */
1173 strbuf_addstr(sb, msg + c->message_off + 1);
1174 return 1;
1177 /* Now we need to parse the commit message. */
1178 if (!c->commit_message_parsed)
1179 parse_commit_message(c);
1181 switch (placeholder[0]) {
1182 case 's': /* subject */
1183 format_subject(sb, msg + c->subject_off, " ");
1184 return 1;
1185 case 'f': /* sanitized subject */
1186 format_sanitized_subject(sb, msg + c->subject_off);
1187 return 1;
1188 case 'b': /* body */
1189 strbuf_addstr(sb, msg + c->body_off);
1190 return 1;
1192 return 0; /* unknown placeholder */
1195 static size_t format_commit_item(struct strbuf *sb, const char *placeholder,
1196 void *context)
1198 int consumed;
1199 size_t orig_len;
1200 enum {
1201 NO_MAGIC,
1202 ADD_LF_BEFORE_NON_EMPTY,
1203 DEL_LF_BEFORE_EMPTY,
1204 ADD_SP_BEFORE_NON_EMPTY
1205 } magic = NO_MAGIC;
1207 switch (placeholder[0]) {
1208 case '-':
1209 magic = DEL_LF_BEFORE_EMPTY;
1210 break;
1211 case '+':
1212 magic = ADD_LF_BEFORE_NON_EMPTY;
1213 break;
1214 case ' ':
1215 magic = ADD_SP_BEFORE_NON_EMPTY;
1216 break;
1217 default:
1218 break;
1220 if (magic != NO_MAGIC)
1221 placeholder++;
1223 orig_len = sb->len;
1224 consumed = format_commit_one(sb, placeholder, context);
1225 if (magic == NO_MAGIC)
1226 return consumed;
1228 if ((orig_len == sb->len) && magic == DEL_LF_BEFORE_EMPTY) {
1229 while (sb->len && sb->buf[sb->len - 1] == '\n')
1230 strbuf_setlen(sb, sb->len - 1);
1231 } else if (orig_len != sb->len) {
1232 if (magic == ADD_LF_BEFORE_NON_EMPTY)
1233 strbuf_insert(sb, orig_len, "\n", 1);
1234 else if (magic == ADD_SP_BEFORE_NON_EMPTY)
1235 strbuf_insert(sb, orig_len, " ", 1);
1237 return consumed + 1;
1240 static size_t userformat_want_item(struct strbuf *sb, const char *placeholder,
1241 void *context)
1243 struct userformat_want *w = context;
1245 if (*placeholder == '+' || *placeholder == '-' || *placeholder == ' ')
1246 placeholder++;
1248 switch (*placeholder) {
1249 case 'N':
1250 w->notes = 1;
1251 break;
1253 return 0;
1256 void userformat_find_requirements(const char *fmt, struct userformat_want *w)
1258 struct strbuf dummy = STRBUF_INIT;
1260 if (!fmt) {
1261 if (!user_format)
1262 return;
1263 fmt = user_format;
1265 strbuf_expand(&dummy, fmt, userformat_want_item, w);
1266 strbuf_release(&dummy);
1269 void format_commit_message(const struct commit *commit,
1270 const char *format, struct strbuf *sb,
1271 const struct pretty_print_context *pretty_ctx)
1273 struct format_commit_context context;
1274 const char *output_enc = pretty_ctx->output_encoding;
1276 memset(&context, 0, sizeof(context));
1277 context.commit = commit;
1278 context.pretty_ctx = pretty_ctx;
1279 context.wrap_start = sb->len;
1280 context.message = logmsg_reencode(commit, output_enc);
1281 if (!context.message)
1282 context.message = commit->buffer;
1284 strbuf_expand(sb, format, format_commit_item, &context);
1285 rewrap_message_tail(sb, &context, 0, 0, 0);
1287 if (context.message != commit->buffer)
1288 free(context.message);
1289 free(context.signature.gpg_output);
1290 free(context.signature.signer);
1293 static void pp_header(const struct pretty_print_context *pp,
1294 const char *encoding,
1295 const struct commit *commit,
1296 const char **msg_p,
1297 struct strbuf *sb)
1299 int parents_shown = 0;
1301 for (;;) {
1302 const char *line = *msg_p;
1303 int linelen = get_one_line(*msg_p);
1305 if (!linelen)
1306 return;
1307 *msg_p += linelen;
1309 if (linelen == 1)
1310 /* End of header */
1311 return;
1313 if (pp->fmt == CMIT_FMT_RAW) {
1314 strbuf_add(sb, line, linelen);
1315 continue;
1318 if (!prefixcmp(line, "parent ")) {
1319 if (linelen != 48)
1320 die("bad parent line in commit");
1321 continue;
1324 if (!parents_shown) {
1325 struct commit_list *parent;
1326 int num;
1327 for (parent = commit->parents, num = 0;
1328 parent;
1329 parent = parent->next, num++)
1331 /* with enough slop */
1332 strbuf_grow(sb, num * 50 + 20);
1333 add_merge_info(pp, sb, commit);
1334 parents_shown = 1;
1338 * MEDIUM == DEFAULT shows only author with dates.
1339 * FULL shows both authors but not dates.
1340 * FULLER shows both authors and dates.
1342 if (!prefixcmp(line, "author ")) {
1343 strbuf_grow(sb, linelen + 80);
1344 pp_user_info(pp, "Author", sb, line + 7, encoding);
1346 if (!prefixcmp(line, "committer ") &&
1347 (pp->fmt == CMIT_FMT_FULL || pp->fmt == CMIT_FMT_FULLER)) {
1348 strbuf_grow(sb, linelen + 80);
1349 pp_user_info(pp, "Commit", sb, line + 10, encoding);
1354 void pp_title_line(const struct pretty_print_context *pp,
1355 const char **msg_p,
1356 struct strbuf *sb,
1357 const char *encoding,
1358 int need_8bit_cte)
1360 static const int max_length = 78; /* per rfc2047 */
1361 struct strbuf title;
1363 strbuf_init(&title, 80);
1364 *msg_p = format_subject(&title, *msg_p,
1365 pp->preserve_subject ? "\n" : " ");
1367 strbuf_grow(sb, title.len + 1024);
1368 if (pp->subject) {
1369 strbuf_addstr(sb, pp->subject);
1370 if (needs_rfc2047_encoding(title.buf, title.len, RFC2047_SUBJECT))
1371 add_rfc2047(sb, title.buf, title.len,
1372 encoding, RFC2047_SUBJECT);
1373 else
1374 strbuf_add_wrapped_bytes(sb, title.buf, title.len,
1375 -last_line_length(sb), 1, max_length);
1376 } else {
1377 strbuf_addbuf(sb, &title);
1379 strbuf_addch(sb, '\n');
1381 if (need_8bit_cte > 0) {
1382 const char *header_fmt =
1383 "MIME-Version: 1.0\n"
1384 "Content-Type: text/plain; charset=%s\n"
1385 "Content-Transfer-Encoding: 8bit\n";
1386 strbuf_addf(sb, header_fmt, encoding);
1388 if (pp->after_subject) {
1389 strbuf_addstr(sb, pp->after_subject);
1391 if (pp->fmt == CMIT_FMT_EMAIL) {
1392 strbuf_addch(sb, '\n');
1394 strbuf_release(&title);
1397 void pp_remainder(const struct pretty_print_context *pp,
1398 const char **msg_p,
1399 struct strbuf *sb,
1400 int indent)
1402 int first = 1;
1403 for (;;) {
1404 const char *line = *msg_p;
1405 int linelen = get_one_line(line);
1406 *msg_p += linelen;
1408 if (!linelen)
1409 break;
1411 if (is_empty_line(line, &linelen)) {
1412 if (first)
1413 continue;
1414 if (pp->fmt == CMIT_FMT_SHORT)
1415 break;
1417 first = 0;
1419 strbuf_grow(sb, linelen + indent + 20);
1420 if (indent) {
1421 memset(sb->buf + sb->len, ' ', indent);
1422 strbuf_setlen(sb, sb->len + indent);
1424 strbuf_add(sb, line, linelen);
1425 strbuf_addch(sb, '\n');
1429 void pretty_print_commit(const struct pretty_print_context *pp,
1430 const struct commit *commit,
1431 struct strbuf *sb)
1433 unsigned long beginning_of_body;
1434 int indent = 4;
1435 const char *msg = commit->buffer;
1436 char *reencoded;
1437 const char *encoding;
1438 int need_8bit_cte = pp->need_8bit_cte;
1440 if (pp->fmt == CMIT_FMT_USERFORMAT) {
1441 format_commit_message(commit, user_format, sb, pp);
1442 return;
1445 encoding = get_log_output_encoding();
1446 reencoded = logmsg_reencode(commit, encoding);
1447 if (reencoded) {
1448 msg = reencoded;
1451 if (pp->fmt == CMIT_FMT_ONELINE || pp->fmt == CMIT_FMT_EMAIL)
1452 indent = 0;
1455 * We need to check and emit Content-type: to mark it
1456 * as 8-bit if we haven't done so.
1458 if (pp->fmt == CMIT_FMT_EMAIL && need_8bit_cte == 0) {
1459 int i, ch, in_body;
1461 for (in_body = i = 0; (ch = msg[i]); i++) {
1462 if (!in_body) {
1463 /* author could be non 7-bit ASCII but
1464 * the log may be so; skip over the
1465 * header part first.
1467 if (ch == '\n' && msg[i+1] == '\n')
1468 in_body = 1;
1470 else if (non_ascii(ch)) {
1471 need_8bit_cte = 1;
1472 break;
1477 pp_header(pp, encoding, commit, &msg, sb);
1478 if (pp->fmt != CMIT_FMT_ONELINE && !pp->subject) {
1479 strbuf_addch(sb, '\n');
1482 /* Skip excess blank lines at the beginning of body, if any... */
1483 msg = skip_empty_lines(msg);
1485 /* These formats treat the title line specially. */
1486 if (pp->fmt == CMIT_FMT_ONELINE || pp->fmt == CMIT_FMT_EMAIL)
1487 pp_title_line(pp, &msg, sb, encoding, need_8bit_cte);
1489 beginning_of_body = sb->len;
1490 if (pp->fmt != CMIT_FMT_ONELINE)
1491 pp_remainder(pp, &msg, sb, indent);
1492 strbuf_rtrim(sb);
1494 /* Make sure there is an EOLN for the non-oneline case */
1495 if (pp->fmt != CMIT_FMT_ONELINE)
1496 strbuf_addch(sb, '\n');
1499 * The caller may append additional body text in e-mail
1500 * format. Make sure we did not strip the blank line
1501 * between the header and the body.
1503 if (pp->fmt == CMIT_FMT_EMAIL && sb->len <= beginning_of_body)
1504 strbuf_addch(sb, '\n');
1506 free(reencoded);
1509 void pp_commit_easy(enum cmit_fmt fmt, const struct commit *commit,
1510 struct strbuf *sb)
1512 struct pretty_print_context pp = {0};
1513 pp.fmt = fmt;
1514 pretty_print_commit(&pp, commit, sb);