rebase -i: Handle the author script all in one place in do_next
[git.git] / pretty.c
blob8f5bd1ab7f119715564fb13cb74de3937d1a3774
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"
13 static char *user_format;
15 static void save_user_format(struct rev_info *rev, const char *cp, int is_tformat)
17 free(user_format);
18 user_format = xstrdup(cp);
19 if (is_tformat)
20 rev->use_terminator = 1;
21 rev->commit_format = CMIT_FMT_USERFORMAT;
24 void get_commit_format(const char *arg, struct rev_info *rev)
26 int i;
27 static struct cmt_fmt_map {
28 const char *n;
29 size_t cmp_len;
30 enum cmit_fmt v;
31 } cmt_fmts[] = {
32 { "raw", 1, CMIT_FMT_RAW },
33 { "medium", 1, CMIT_FMT_MEDIUM },
34 { "short", 1, CMIT_FMT_SHORT },
35 { "email", 1, CMIT_FMT_EMAIL },
36 { "full", 5, CMIT_FMT_FULL },
37 { "fuller", 5, CMIT_FMT_FULLER },
38 { "oneline", 1, CMIT_FMT_ONELINE },
41 rev->use_terminator = 0;
42 if (!arg || !*arg) {
43 rev->commit_format = CMIT_FMT_DEFAULT;
44 return;
46 if (!prefixcmp(arg, "format:") || !prefixcmp(arg, "tformat:")) {
47 save_user_format(rev, strchr(arg, ':') + 1, arg[0] == 't');
48 return;
50 for (i = 0; i < ARRAY_SIZE(cmt_fmts); i++) {
51 if (!strncmp(arg, cmt_fmts[i].n, cmt_fmts[i].cmp_len) &&
52 !strncmp(arg, cmt_fmts[i].n, strlen(arg))) {
53 if (cmt_fmts[i].v == CMIT_FMT_ONELINE)
54 rev->use_terminator = 1;
55 rev->commit_format = cmt_fmts[i].v;
56 return;
59 if (strchr(arg, '%')) {
60 save_user_format(rev, arg, 1);
61 return;
64 die("invalid --pretty format: %s", arg);
68 * Generic support for pretty-printing the header
70 static int get_one_line(const char *msg)
72 int ret = 0;
74 for (;;) {
75 char c = *msg++;
76 if (!c)
77 break;
78 ret++;
79 if (c == '\n')
80 break;
82 return ret;
85 /* High bit set, or ISO-2022-INT */
86 int non_ascii(int ch)
88 return !isascii(ch) || ch == '\033';
91 int has_non_ascii(const char *s)
93 int ch;
94 if (!s)
95 return 0;
96 while ((ch = *s++) != '\0') {
97 if (non_ascii(ch))
98 return 1;
100 return 0;
103 static int is_rfc2047_special(char ch)
105 return (non_ascii(ch) || (ch == '=') || (ch == '?') || (ch == '_'));
108 static void add_rfc2047(struct strbuf *sb, const char *line, int len,
109 const char *encoding)
111 int i, last;
113 for (i = 0; i < len; i++) {
114 int ch = line[i];
115 if (non_ascii(ch))
116 goto needquote;
117 if ((i + 1 < len) && (ch == '=' && line[i+1] == '?'))
118 goto needquote;
120 strbuf_add(sb, line, len);
121 return;
123 needquote:
124 strbuf_grow(sb, len * 3 + strlen(encoding) + 100);
125 strbuf_addf(sb, "=?%s?q?", encoding);
126 for (i = last = 0; i < len; i++) {
127 unsigned ch = line[i] & 0xFF;
129 * We encode ' ' using '=20' even though rfc2047
130 * allows using '_' for readability. Unfortunately,
131 * many programs do not understand this and just
132 * leave the underscore in place.
134 if (is_rfc2047_special(ch) || ch == ' ') {
135 strbuf_add(sb, line + last, i - last);
136 strbuf_addf(sb, "=%02X", ch);
137 last = i + 1;
140 strbuf_add(sb, line + last, len - last);
141 strbuf_addstr(sb, "?=");
144 void pp_user_info(const char *what, enum cmit_fmt fmt, struct strbuf *sb,
145 const char *line, enum date_mode dmode,
146 const char *encoding)
148 char *date;
149 int namelen;
150 unsigned long time;
151 int tz;
153 if (fmt == CMIT_FMT_ONELINE)
154 return;
155 date = strchr(line, '>');
156 if (!date)
157 return;
158 namelen = ++date - line;
159 time = strtoul(date, &date, 10);
160 tz = strtol(date, NULL, 10);
162 if (fmt == CMIT_FMT_EMAIL) {
163 char *name_tail = strchr(line, '<');
164 int display_name_length;
165 if (!name_tail)
166 return;
167 while (line < name_tail && isspace(name_tail[-1]))
168 name_tail--;
169 display_name_length = name_tail - line;
170 strbuf_addstr(sb, "From: ");
171 add_rfc2047(sb, line, display_name_length, encoding);
172 strbuf_add(sb, name_tail, namelen - display_name_length);
173 strbuf_addch(sb, '\n');
174 } else {
175 strbuf_addf(sb, "%s: %.*s%.*s\n", what,
176 (fmt == CMIT_FMT_FULLER) ? 4 : 0,
177 " ", namelen, line);
179 switch (fmt) {
180 case CMIT_FMT_MEDIUM:
181 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, dmode));
182 break;
183 case CMIT_FMT_EMAIL:
184 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, DATE_RFC2822));
185 break;
186 case CMIT_FMT_FULLER:
187 strbuf_addf(sb, "%sDate: %s\n", what, show_date(time, tz, dmode));
188 break;
189 default:
190 /* notin' */
191 break;
195 static int is_empty_line(const char *line, int *len_p)
197 int len = *len_p;
198 while (len && isspace(line[len-1]))
199 len--;
200 *len_p = len;
201 return !len;
204 static const char *skip_empty_lines(const char *msg)
206 for (;;) {
207 int linelen = get_one_line(msg);
208 int ll = linelen;
209 if (!linelen)
210 break;
211 if (!is_empty_line(msg, &ll))
212 break;
213 msg += linelen;
215 return msg;
218 static void add_merge_info(enum cmit_fmt fmt, struct strbuf *sb,
219 const struct commit *commit, int abbrev)
221 struct commit_list *parent = commit->parents;
223 if ((fmt == CMIT_FMT_ONELINE) || (fmt == CMIT_FMT_EMAIL) ||
224 !parent || !parent->next)
225 return;
227 strbuf_addstr(sb, "Merge:");
229 while (parent) {
230 struct commit *p = parent->item;
231 const char *hex = NULL;
232 if (abbrev)
233 hex = find_unique_abbrev(p->object.sha1, abbrev);
234 if (!hex)
235 hex = sha1_to_hex(p->object.sha1);
236 parent = parent->next;
238 strbuf_addf(sb, " %s", hex);
240 strbuf_addch(sb, '\n');
243 static char *get_header(const struct commit *commit, const char *key)
245 int key_len = strlen(key);
246 const char *line = commit->buffer;
248 for (;;) {
249 const char *eol = strchr(line, '\n'), *next;
251 if (line == eol)
252 return NULL;
253 if (!eol) {
254 eol = line + strlen(line);
255 next = NULL;
256 } else
257 next = eol + 1;
258 if (eol - line > key_len &&
259 !strncmp(line, key, key_len) &&
260 line[key_len] == ' ') {
261 return xmemdupz(line + key_len + 1, eol - line - key_len - 1);
263 line = next;
267 static char *replace_encoding_header(char *buf, const char *encoding)
269 struct strbuf tmp = STRBUF_INIT;
270 size_t start, len;
271 char *cp = buf;
273 /* guess if there is an encoding header before a \n\n */
274 while (strncmp(cp, "encoding ", strlen("encoding "))) {
275 cp = strchr(cp, '\n');
276 if (!cp || *++cp == '\n')
277 return buf;
279 start = cp - buf;
280 cp = strchr(cp, '\n');
281 if (!cp)
282 return buf; /* should not happen but be defensive */
283 len = cp + 1 - (buf + start);
285 strbuf_attach(&tmp, buf, strlen(buf), strlen(buf) + 1);
286 if (is_encoding_utf8(encoding)) {
287 /* we have re-coded to UTF-8; drop the header */
288 strbuf_remove(&tmp, start, len);
289 } else {
290 /* just replaces XXXX in 'encoding XXXX\n' */
291 strbuf_splice(&tmp, start + strlen("encoding "),
292 len - strlen("encoding \n"),
293 encoding, strlen(encoding));
295 return strbuf_detach(&tmp, NULL);
298 static char *logmsg_reencode(const struct commit *commit,
299 const char *output_encoding)
301 static const char *utf8 = "UTF-8";
302 const char *use_encoding;
303 char *encoding;
304 char *out;
306 if (!*output_encoding)
307 return NULL;
308 encoding = get_header(commit, "encoding");
309 use_encoding = encoding ? encoding : utf8;
310 if (!strcmp(use_encoding, output_encoding))
311 if (encoding) /* we'll strip encoding header later */
312 out = xstrdup(commit->buffer);
313 else
314 return NULL; /* nothing to do */
315 else
316 out = reencode_string(commit->buffer,
317 output_encoding, use_encoding);
318 if (out)
319 out = replace_encoding_header(out, output_encoding);
321 free(encoding);
322 return out;
325 static int mailmap_name(char *email, int email_len, char *name, int name_len)
327 static struct string_list *mail_map;
328 if (!mail_map) {
329 mail_map = xcalloc(1, sizeof(*mail_map));
330 read_mailmap(mail_map, NULL);
332 return mail_map->nr && map_user(mail_map, email, email_len, name, name_len);
335 static size_t format_person_part(struct strbuf *sb, char part,
336 const char *msg, int len, enum date_mode dmode)
338 /* currently all placeholders have same length */
339 const int placeholder_len = 2;
340 int start, end, tz = 0;
341 unsigned long date = 0;
342 char *ep;
343 const char *name_start, *name_end, *mail_start, *mail_end, *msg_end = msg+len;
344 char person_name[1024];
345 char person_mail[1024];
347 /* advance 'end' to point to email start delimiter */
348 for (end = 0; end < len && msg[end] != '<'; end++)
349 ; /* do nothing */
352 * When end points at the '<' that we found, it should have
353 * matching '>' later, which means 'end' must be strictly
354 * below len - 1.
356 if (end >= len - 2)
357 goto skip;
359 /* Seek for both name and email part */
360 name_start = msg;
361 name_end = msg+end;
362 while (name_end > name_start && isspace(*(name_end-1)))
363 name_end--;
364 mail_start = msg+end+1;
365 mail_end = mail_start;
366 while (mail_end < msg_end && *mail_end != '>')
367 mail_end++;
368 if (mail_end == msg_end)
369 goto skip;
370 end = mail_end-msg;
372 if (part == 'N' || part == 'E') { /* mailmap lookup */
373 strlcpy(person_name, name_start, name_end-name_start+1);
374 strlcpy(person_mail, mail_start, mail_end-mail_start+1);
375 mailmap_name(person_mail, sizeof(person_mail), person_name, sizeof(person_name));
376 name_start = person_name;
377 name_end = name_start + strlen(person_name);
378 mail_start = person_mail;
379 mail_end = mail_start + strlen(person_mail);
381 if (part == 'n' || part == 'N') { /* name */
382 strbuf_add(sb, name_start, name_end-name_start);
383 return placeholder_len;
385 if (part == 'e' || part == 'E') { /* email */
386 strbuf_add(sb, mail_start, mail_end-mail_start);
387 return placeholder_len;
390 /* advance 'start' to point to date start delimiter */
391 for (start = end + 1; start < len && isspace(msg[start]); start++)
392 ; /* do nothing */
393 if (start >= len)
394 goto skip;
395 date = strtoul(msg + start, &ep, 10);
396 if (msg + start == ep)
397 goto skip;
399 if (part == 't') { /* date, UNIX timestamp */
400 strbuf_add(sb, msg + start, ep - (msg + start));
401 return placeholder_len;
404 /* parse tz */
405 for (start = ep - msg + 1; start < len && isspace(msg[start]); start++)
406 ; /* do nothing */
407 if (start + 1 < len) {
408 tz = strtoul(msg + start + 1, NULL, 10);
409 if (msg[start] == '-')
410 tz = -tz;
413 switch (part) {
414 case 'd': /* date */
415 strbuf_addstr(sb, show_date(date, tz, dmode));
416 return placeholder_len;
417 case 'D': /* date, RFC2822 style */
418 strbuf_addstr(sb, show_date(date, tz, DATE_RFC2822));
419 return placeholder_len;
420 case 'r': /* date, relative */
421 strbuf_addstr(sb, show_date(date, tz, DATE_RELATIVE));
422 return placeholder_len;
423 case 'i': /* date, ISO 8601 */
424 strbuf_addstr(sb, show_date(date, tz, DATE_ISO8601));
425 return placeholder_len;
428 skip:
430 * bogus commit, 'sb' cannot be updated, but we still need to
431 * compute a valid return value.
433 if (part == 'n' || part == 'e' || part == 't' || part == 'd'
434 || part == 'D' || part == 'r' || part == 'i')
435 return placeholder_len;
437 return 0; /* unknown placeholder */
440 struct chunk {
441 size_t off;
442 size_t len;
445 struct format_commit_context {
446 const struct commit *commit;
447 const struct pretty_print_context *pretty_ctx;
448 unsigned commit_header_parsed:1;
449 unsigned commit_message_parsed:1;
450 size_t width, indent1, indent2;
452 /* These offsets are relative to the start of the commit message. */
453 struct chunk author;
454 struct chunk committer;
455 struct chunk encoding;
456 size_t message_off;
457 size_t subject_off;
458 size_t body_off;
460 /* The following ones are relative to the result struct strbuf. */
461 struct chunk abbrev_commit_hash;
462 struct chunk abbrev_tree_hash;
463 struct chunk abbrev_parent_hashes;
464 size_t wrap_start;
467 static int add_again(struct strbuf *sb, struct chunk *chunk)
469 if (chunk->len) {
470 strbuf_adddup(sb, chunk->off, chunk->len);
471 return 1;
475 * We haven't seen this chunk before. Our caller is surely
476 * going to add it the hard way now. Remember the most likely
477 * start of the to-be-added chunk: the current end of the
478 * struct strbuf.
480 chunk->off = sb->len;
481 return 0;
484 static void parse_commit_header(struct format_commit_context *context)
486 const char *msg = context->commit->buffer;
487 int i;
489 for (i = 0; msg[i]; i++) {
490 int eol;
491 for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
492 ; /* do nothing */
494 if (i == eol) {
495 break;
496 } else if (!prefixcmp(msg + i, "author ")) {
497 context->author.off = i + 7;
498 context->author.len = eol - i - 7;
499 } else if (!prefixcmp(msg + i, "committer ")) {
500 context->committer.off = i + 10;
501 context->committer.len = eol - i - 10;
502 } else if (!prefixcmp(msg + i, "encoding ")) {
503 context->encoding.off = i + 9;
504 context->encoding.len = eol - i - 9;
506 i = eol;
508 context->message_off = i;
509 context->commit_header_parsed = 1;
512 static int istitlechar(char c)
514 return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') ||
515 (c >= '0' && c <= '9') || c == '.' || c == '_';
518 static void format_sanitized_subject(struct strbuf *sb, const char *msg)
520 size_t trimlen;
521 size_t start_len = sb->len;
522 int space = 2;
524 for (; *msg && *msg != '\n'; msg++) {
525 if (istitlechar(*msg)) {
526 if (space == 1)
527 strbuf_addch(sb, '-');
528 space = 0;
529 strbuf_addch(sb, *msg);
530 if (*msg == '.')
531 while (*(msg+1) == '.')
532 msg++;
533 } else
534 space |= 1;
537 /* trim any trailing '.' or '-' characters */
538 trimlen = 0;
539 while (sb->len - trimlen > start_len &&
540 (sb->buf[sb->len - 1 - trimlen] == '.'
541 || sb->buf[sb->len - 1 - trimlen] == '-'))
542 trimlen++;
543 strbuf_remove(sb, sb->len - trimlen, trimlen);
546 const char *format_subject(struct strbuf *sb, const char *msg,
547 const char *line_separator)
549 int first = 1;
551 for (;;) {
552 const char *line = msg;
553 int linelen = get_one_line(line);
555 msg += linelen;
556 if (!linelen || is_empty_line(line, &linelen))
557 break;
559 if (!sb)
560 continue;
561 strbuf_grow(sb, linelen + 2);
562 if (!first)
563 strbuf_addstr(sb, line_separator);
564 strbuf_add(sb, line, linelen);
565 first = 0;
567 return msg;
570 static void parse_commit_message(struct format_commit_context *c)
572 const char *msg = c->commit->buffer + c->message_off;
573 const char *start = c->commit->buffer;
575 msg = skip_empty_lines(msg);
576 c->subject_off = msg - start;
578 msg = format_subject(NULL, msg, NULL);
579 msg = skip_empty_lines(msg);
580 c->body_off = msg - start;
582 c->commit_message_parsed = 1;
585 static void format_decoration(struct strbuf *sb, const struct commit *commit)
587 struct name_decoration *d;
588 const char *prefix = " (";
590 load_ref_decorations(DECORATE_SHORT_REFS);
591 d = lookup_decoration(&name_decoration, &commit->object);
592 while (d) {
593 strbuf_addstr(sb, prefix);
594 prefix = ", ";
595 strbuf_addstr(sb, d->name);
596 d = d->next;
598 if (prefix[0] == ',')
599 strbuf_addch(sb, ')');
602 static void strbuf_wrap(struct strbuf *sb, size_t pos,
603 size_t width, size_t indent1, size_t indent2)
605 struct strbuf tmp = STRBUF_INIT;
607 if (pos)
608 strbuf_add(&tmp, sb->buf, pos);
609 strbuf_add_wrapped_text(&tmp, sb->buf + pos,
610 (int) indent1, (int) indent2, (int) width);
611 strbuf_swap(&tmp, sb);
612 strbuf_release(&tmp);
615 static void rewrap_message_tail(struct strbuf *sb,
616 struct format_commit_context *c,
617 size_t new_width, size_t new_indent1,
618 size_t new_indent2)
620 if (c->width == new_width && c->indent1 == new_indent1 &&
621 c->indent2 == new_indent2)
622 return;
623 if (c->wrap_start < sb->len)
624 strbuf_wrap(sb, c->wrap_start, c->width, c->indent1, c->indent2);
625 c->wrap_start = sb->len;
626 c->width = new_width;
627 c->indent1 = new_indent1;
628 c->indent2 = new_indent2;
631 static size_t format_commit_one(struct strbuf *sb, const char *placeholder,
632 void *context)
634 struct format_commit_context *c = context;
635 const struct commit *commit = c->commit;
636 const char *msg = commit->buffer;
637 struct commit_list *p;
638 int h1, h2;
640 /* these are independent of the commit */
641 switch (placeholder[0]) {
642 case 'C':
643 if (placeholder[1] == '(') {
644 const char *end = strchr(placeholder + 2, ')');
645 char color[COLOR_MAXLEN];
646 if (!end)
647 return 0;
648 color_parse_mem(placeholder + 2,
649 end - (placeholder + 2),
650 "--pretty format", color);
651 strbuf_addstr(sb, color);
652 return end - placeholder + 1;
654 if (!prefixcmp(placeholder + 1, "red")) {
655 strbuf_addstr(sb, GIT_COLOR_RED);
656 return 4;
657 } else if (!prefixcmp(placeholder + 1, "green")) {
658 strbuf_addstr(sb, GIT_COLOR_GREEN);
659 return 6;
660 } else if (!prefixcmp(placeholder + 1, "blue")) {
661 strbuf_addstr(sb, GIT_COLOR_BLUE);
662 return 5;
663 } else if (!prefixcmp(placeholder + 1, "reset")) {
664 strbuf_addstr(sb, GIT_COLOR_RESET);
665 return 6;
666 } else
667 return 0;
668 case 'n': /* newline */
669 strbuf_addch(sb, '\n');
670 return 1;
671 case 'x':
672 /* %x00 == NUL, %x0a == LF, etc. */
673 if (0 <= (h1 = hexval_table[0xff & placeholder[1]]) &&
674 h1 <= 16 &&
675 0 <= (h2 = hexval_table[0xff & placeholder[2]]) &&
676 h2 <= 16) {
677 strbuf_addch(sb, (h1<<4)|h2);
678 return 3;
679 } else
680 return 0;
681 case 'w':
682 if (placeholder[1] == '(') {
683 unsigned long width = 0, indent1 = 0, indent2 = 0;
684 char *next;
685 const char *start = placeholder + 2;
686 const char *end = strchr(start, ')');
687 if (!end)
688 return 0;
689 if (end > start) {
690 width = strtoul(start, &next, 10);
691 if (*next == ',') {
692 indent1 = strtoul(next + 1, &next, 10);
693 if (*next == ',') {
694 indent2 = strtoul(next + 1,
695 &next, 10);
698 if (*next != ')')
699 return 0;
701 rewrap_message_tail(sb, c, width, indent1, indent2);
702 return end - placeholder + 1;
703 } else
704 return 0;
707 /* these depend on the commit */
708 if (!commit->object.parsed)
709 parse_object(commit->object.sha1);
711 switch (placeholder[0]) {
712 case 'H': /* commit hash */
713 strbuf_addstr(sb, sha1_to_hex(commit->object.sha1));
714 return 1;
715 case 'h': /* abbreviated commit hash */
716 if (add_again(sb, &c->abbrev_commit_hash))
717 return 1;
718 strbuf_addstr(sb, find_unique_abbrev(commit->object.sha1,
719 DEFAULT_ABBREV));
720 c->abbrev_commit_hash.len = sb->len - c->abbrev_commit_hash.off;
721 return 1;
722 case 'T': /* tree hash */
723 strbuf_addstr(sb, sha1_to_hex(commit->tree->object.sha1));
724 return 1;
725 case 't': /* abbreviated tree hash */
726 if (add_again(sb, &c->abbrev_tree_hash))
727 return 1;
728 strbuf_addstr(sb, find_unique_abbrev(commit->tree->object.sha1,
729 DEFAULT_ABBREV));
730 c->abbrev_tree_hash.len = sb->len - c->abbrev_tree_hash.off;
731 return 1;
732 case 'P': /* parent hashes */
733 for (p = commit->parents; p; p = p->next) {
734 if (p != commit->parents)
735 strbuf_addch(sb, ' ');
736 strbuf_addstr(sb, sha1_to_hex(p->item->object.sha1));
738 return 1;
739 case 'p': /* abbreviated parent hashes */
740 if (add_again(sb, &c->abbrev_parent_hashes))
741 return 1;
742 for (p = commit->parents; p; p = p->next) {
743 if (p != commit->parents)
744 strbuf_addch(sb, ' ');
745 strbuf_addstr(sb, find_unique_abbrev(
746 p->item->object.sha1, DEFAULT_ABBREV));
748 c->abbrev_parent_hashes.len = sb->len -
749 c->abbrev_parent_hashes.off;
750 return 1;
751 case 'm': /* left/right/bottom */
752 strbuf_addch(sb, (commit->object.flags & BOUNDARY)
753 ? '-'
754 : (commit->object.flags & SYMMETRIC_LEFT)
755 ? '<'
756 : '>');
757 return 1;
758 case 'd':
759 format_decoration(sb, commit);
760 return 1;
761 case 'g': /* reflog info */
762 switch(placeholder[1]) {
763 case 'd': /* reflog selector */
764 case 'D':
765 if (c->pretty_ctx->reflog_info)
766 get_reflog_selector(sb,
767 c->pretty_ctx->reflog_info,
768 c->pretty_ctx->date_mode,
769 (placeholder[1] == 'd'));
770 return 2;
771 case 's': /* reflog message */
772 if (c->pretty_ctx->reflog_info)
773 get_reflog_message(sb, c->pretty_ctx->reflog_info);
774 return 2;
776 return 0; /* unknown %g placeholder */
777 case 'N':
778 get_commit_notes(commit, sb, git_log_output_encoding ?
779 git_log_output_encoding : git_commit_encoding, 0);
780 return 1;
783 /* For the rest we have to parse the commit header. */
784 if (!c->commit_header_parsed)
785 parse_commit_header(c);
787 switch (placeholder[0]) {
788 case 'a': /* author ... */
789 return format_person_part(sb, placeholder[1],
790 msg + c->author.off, c->author.len,
791 c->pretty_ctx->date_mode);
792 case 'c': /* committer ... */
793 return format_person_part(sb, placeholder[1],
794 msg + c->committer.off, c->committer.len,
795 c->pretty_ctx->date_mode);
796 case 'e': /* encoding */
797 strbuf_add(sb, msg + c->encoding.off, c->encoding.len);
798 return 1;
801 /* Now we need to parse the commit message. */
802 if (!c->commit_message_parsed)
803 parse_commit_message(c);
805 switch (placeholder[0]) {
806 case 's': /* subject */
807 format_subject(sb, msg + c->subject_off, " ");
808 return 1;
809 case 'f': /* sanitized subject */
810 format_sanitized_subject(sb, msg + c->subject_off);
811 return 1;
812 case 'b': /* body */
813 strbuf_addstr(sb, msg + c->body_off);
814 return 1;
816 return 0; /* unknown placeholder */
819 static size_t format_commit_item(struct strbuf *sb, const char *placeholder,
820 void *context)
822 int consumed;
823 size_t orig_len;
824 enum {
825 NO_MAGIC,
826 ADD_LF_BEFORE_NON_EMPTY,
827 DEL_LF_BEFORE_EMPTY,
828 } magic = NO_MAGIC;
830 switch (placeholder[0]) {
831 case '-':
832 magic = DEL_LF_BEFORE_EMPTY;
833 break;
834 case '+':
835 magic = ADD_LF_BEFORE_NON_EMPTY;
836 break;
837 default:
838 break;
840 if (magic != NO_MAGIC)
841 placeholder++;
843 orig_len = sb->len;
844 consumed = format_commit_one(sb, placeholder, context);
845 if (magic == NO_MAGIC)
846 return consumed;
848 if ((orig_len == sb->len) && magic == DEL_LF_BEFORE_EMPTY) {
849 while (sb->len && sb->buf[sb->len - 1] == '\n')
850 strbuf_setlen(sb, sb->len - 1);
851 } else if ((orig_len != sb->len) && magic == ADD_LF_BEFORE_NON_EMPTY) {
852 strbuf_insert(sb, orig_len, "\n", 1);
854 return consumed + 1;
857 void format_commit_message(const struct commit *commit,
858 const char *format, struct strbuf *sb,
859 const struct pretty_print_context *pretty_ctx)
861 struct format_commit_context context;
863 memset(&context, 0, sizeof(context));
864 context.commit = commit;
865 context.pretty_ctx = pretty_ctx;
866 context.wrap_start = sb->len;
867 strbuf_expand(sb, format, format_commit_item, &context);
868 rewrap_message_tail(sb, &context, 0, 0, 0);
871 static void pp_header(enum cmit_fmt fmt,
872 int abbrev,
873 enum date_mode dmode,
874 const char *encoding,
875 const struct commit *commit,
876 const char **msg_p,
877 struct strbuf *sb)
879 int parents_shown = 0;
881 for (;;) {
882 const char *line = *msg_p;
883 int linelen = get_one_line(*msg_p);
885 if (!linelen)
886 return;
887 *msg_p += linelen;
889 if (linelen == 1)
890 /* End of header */
891 return;
893 if (fmt == CMIT_FMT_RAW) {
894 strbuf_add(sb, line, linelen);
895 continue;
898 if (!memcmp(line, "parent ", 7)) {
899 if (linelen != 48)
900 die("bad parent line in commit");
901 continue;
904 if (!parents_shown) {
905 struct commit_list *parent;
906 int num;
907 for (parent = commit->parents, num = 0;
908 parent;
909 parent = parent->next, num++)
911 /* with enough slop */
912 strbuf_grow(sb, num * 50 + 20);
913 add_merge_info(fmt, sb, commit, abbrev);
914 parents_shown = 1;
918 * MEDIUM == DEFAULT shows only author with dates.
919 * FULL shows both authors but not dates.
920 * FULLER shows both authors and dates.
922 if (!memcmp(line, "author ", 7)) {
923 strbuf_grow(sb, linelen + 80);
924 pp_user_info("Author", fmt, sb, line + 7, dmode, encoding);
926 if (!memcmp(line, "committer ", 10) &&
927 (fmt == CMIT_FMT_FULL || fmt == CMIT_FMT_FULLER)) {
928 strbuf_grow(sb, linelen + 80);
929 pp_user_info("Commit", fmt, sb, line + 10, dmode, encoding);
934 void pp_title_line(enum cmit_fmt fmt,
935 const char **msg_p,
936 struct strbuf *sb,
937 const char *subject,
938 const char *after_subject,
939 const char *encoding,
940 int need_8bit_cte)
942 const char *line_separator = (fmt == CMIT_FMT_EMAIL) ? "\n " : " ";
943 struct strbuf title;
945 strbuf_init(&title, 80);
946 *msg_p = format_subject(&title, *msg_p, line_separator);
948 strbuf_grow(sb, title.len + 1024);
949 if (subject) {
950 strbuf_addstr(sb, subject);
951 add_rfc2047(sb, title.buf, title.len, encoding);
952 } else {
953 strbuf_addbuf(sb, &title);
955 strbuf_addch(sb, '\n');
957 if (need_8bit_cte > 0) {
958 const char *header_fmt =
959 "MIME-Version: 1.0\n"
960 "Content-Type: text/plain; charset=%s\n"
961 "Content-Transfer-Encoding: 8bit\n";
962 strbuf_addf(sb, header_fmt, encoding);
964 if (after_subject) {
965 strbuf_addstr(sb, after_subject);
967 if (fmt == CMIT_FMT_EMAIL) {
968 strbuf_addch(sb, '\n');
970 strbuf_release(&title);
973 void pp_remainder(enum cmit_fmt fmt,
974 const char **msg_p,
975 struct strbuf *sb,
976 int indent)
978 int first = 1;
979 for (;;) {
980 const char *line = *msg_p;
981 int linelen = get_one_line(line);
982 *msg_p += linelen;
984 if (!linelen)
985 break;
987 if (is_empty_line(line, &linelen)) {
988 if (first)
989 continue;
990 if (fmt == CMIT_FMT_SHORT)
991 break;
993 first = 0;
995 strbuf_grow(sb, linelen + indent + 20);
996 if (indent) {
997 memset(sb->buf + sb->len, ' ', indent);
998 strbuf_setlen(sb, sb->len + indent);
1000 strbuf_add(sb, line, linelen);
1001 strbuf_addch(sb, '\n');
1005 char *reencode_commit_message(const struct commit *commit, const char **encoding_p)
1007 const char *encoding;
1009 encoding = (git_log_output_encoding
1010 ? git_log_output_encoding
1011 : git_commit_encoding);
1012 if (!encoding)
1013 encoding = "UTF-8";
1014 if (encoding_p)
1015 *encoding_p = encoding;
1016 return logmsg_reencode(commit, encoding);
1019 void pretty_print_commit(enum cmit_fmt fmt, const struct commit *commit,
1020 struct strbuf *sb,
1021 const struct pretty_print_context *context)
1023 unsigned long beginning_of_body;
1024 int indent = 4;
1025 const char *msg = commit->buffer;
1026 char *reencoded;
1027 const char *encoding;
1028 int need_8bit_cte = context->need_8bit_cte;
1030 if (fmt == CMIT_FMT_USERFORMAT) {
1031 format_commit_message(commit, user_format, sb, context);
1032 return;
1035 reencoded = reencode_commit_message(commit, &encoding);
1036 if (reencoded) {
1037 msg = reencoded;
1040 if (fmt == CMIT_FMT_ONELINE || fmt == CMIT_FMT_EMAIL)
1041 indent = 0;
1044 * We need to check and emit Content-type: to mark it
1045 * as 8-bit if we haven't done so.
1047 if (fmt == CMIT_FMT_EMAIL && need_8bit_cte == 0) {
1048 int i, ch, in_body;
1050 for (in_body = i = 0; (ch = msg[i]); i++) {
1051 if (!in_body) {
1052 /* author could be non 7-bit ASCII but
1053 * the log may be so; skip over the
1054 * header part first.
1056 if (ch == '\n' && msg[i+1] == '\n')
1057 in_body = 1;
1059 else if (non_ascii(ch)) {
1060 need_8bit_cte = 1;
1061 break;
1066 pp_header(fmt, context->abbrev, context->date_mode, encoding,
1067 commit, &msg, sb);
1068 if (fmt != CMIT_FMT_ONELINE && !context->subject) {
1069 strbuf_addch(sb, '\n');
1072 /* Skip excess blank lines at the beginning of body, if any... */
1073 msg = skip_empty_lines(msg);
1075 /* These formats treat the title line specially. */
1076 if (fmt == CMIT_FMT_ONELINE || fmt == CMIT_FMT_EMAIL)
1077 pp_title_line(fmt, &msg, sb, context->subject,
1078 context->after_subject, encoding, need_8bit_cte);
1080 beginning_of_body = sb->len;
1081 if (fmt != CMIT_FMT_ONELINE)
1082 pp_remainder(fmt, &msg, sb, indent);
1083 strbuf_rtrim(sb);
1085 /* Make sure there is an EOLN for the non-oneline case */
1086 if (fmt != CMIT_FMT_ONELINE)
1087 strbuf_addch(sb, '\n');
1090 * The caller may append additional body text in e-mail
1091 * format. Make sure we did not strip the blank line
1092 * between the header and the body.
1094 if (fmt == CMIT_FMT_EMAIL && sb->len <= beginning_of_body)
1095 strbuf_addch(sb, '\n');
1097 if (fmt != CMIT_FMT_ONELINE)
1098 get_commit_notes(commit, sb, encoding,
1099 NOTES_SHOW_HEADER | NOTES_INDENT);
1101 free(reencoded);