t8005: use more portable character encoding names
[debian-git.git] / pretty.c
blobe5328dab5b2978d90d20f47ed799782e7b2fb63b
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 "color.h"
11 static char *user_format;
13 static void save_user_format(struct rev_info *rev, const char *cp, int is_tformat)
15 free(user_format);
16 user_format = xstrdup(cp);
17 if (is_tformat)
18 rev->use_terminator = 1;
19 rev->commit_format = CMIT_FMT_USERFORMAT;
22 void get_commit_format(const char *arg, struct rev_info *rev)
24 int i;
25 static struct cmt_fmt_map {
26 const char *n;
27 size_t cmp_len;
28 enum cmit_fmt v;
29 } cmt_fmts[] = {
30 { "raw", 1, CMIT_FMT_RAW },
31 { "medium", 1, CMIT_FMT_MEDIUM },
32 { "short", 1, CMIT_FMT_SHORT },
33 { "email", 1, CMIT_FMT_EMAIL },
34 { "full", 5, CMIT_FMT_FULL },
35 { "fuller", 5, CMIT_FMT_FULLER },
36 { "oneline", 1, CMIT_FMT_ONELINE },
39 rev->use_terminator = 0;
40 if (!arg || !*arg) {
41 rev->commit_format = CMIT_FMT_DEFAULT;
42 return;
44 if (!prefixcmp(arg, "format:") || !prefixcmp(arg, "tformat:")) {
45 save_user_format(rev, strchr(arg, ':') + 1, arg[0] == 't');
46 return;
48 for (i = 0; i < ARRAY_SIZE(cmt_fmts); i++) {
49 if (!strncmp(arg, cmt_fmts[i].n, cmt_fmts[i].cmp_len) &&
50 !strncmp(arg, cmt_fmts[i].n, strlen(arg))) {
51 if (cmt_fmts[i].v == CMIT_FMT_ONELINE)
52 rev->use_terminator = 1;
53 rev->commit_format = cmt_fmts[i].v;
54 return;
57 if (strchr(arg, '%')) {
58 save_user_format(rev, arg, 1);
59 return;
62 die("invalid --pretty format: %s", arg);
66 * Generic support for pretty-printing the header
68 static int get_one_line(const char *msg)
70 int ret = 0;
72 for (;;) {
73 char c = *msg++;
74 if (!c)
75 break;
76 ret++;
77 if (c == '\n')
78 break;
80 return ret;
83 /* High bit set, or ISO-2022-INT */
84 int non_ascii(int ch)
86 return !isascii(ch) || ch == '\033';
89 static int is_rfc2047_special(char ch)
91 return (non_ascii(ch) || (ch == '=') || (ch == '?') || (ch == '_'));
94 static void add_rfc2047(struct strbuf *sb, const char *line, int len,
95 const char *encoding)
97 int i, last;
99 for (i = 0; i < len; i++) {
100 int ch = line[i];
101 if (non_ascii(ch))
102 goto needquote;
103 if ((i + 1 < len) && (ch == '=' && line[i+1] == '?'))
104 goto needquote;
106 strbuf_add(sb, line, len);
107 return;
109 needquote:
110 strbuf_grow(sb, len * 3 + strlen(encoding) + 100);
111 strbuf_addf(sb, "=?%s?q?", encoding);
112 for (i = last = 0; i < len; i++) {
113 unsigned ch = line[i] & 0xFF;
115 * We encode ' ' using '=20' even though rfc2047
116 * allows using '_' for readability. Unfortunately,
117 * many programs do not understand this and just
118 * leave the underscore in place.
120 if (is_rfc2047_special(ch) || ch == ' ') {
121 strbuf_add(sb, line + last, i - last);
122 strbuf_addf(sb, "=%02X", ch);
123 last = i + 1;
126 strbuf_add(sb, line + last, len - last);
127 strbuf_addstr(sb, "?=");
130 void pp_user_info(const char *what, enum cmit_fmt fmt, struct strbuf *sb,
131 const char *line, enum date_mode dmode,
132 const char *encoding)
134 char *date;
135 int namelen;
136 unsigned long time;
137 int tz;
139 if (fmt == CMIT_FMT_ONELINE)
140 return;
141 date = strchr(line, '>');
142 if (!date)
143 return;
144 namelen = ++date - line;
145 time = strtoul(date, &date, 10);
146 tz = strtol(date, NULL, 10);
148 if (fmt == CMIT_FMT_EMAIL) {
149 char *name_tail = strchr(line, '<');
150 int display_name_length;
151 if (!name_tail)
152 return;
153 while (line < name_tail && isspace(name_tail[-1]))
154 name_tail--;
155 display_name_length = name_tail - line;
156 strbuf_addstr(sb, "From: ");
157 add_rfc2047(sb, line, display_name_length, encoding);
158 strbuf_add(sb, name_tail, namelen - display_name_length);
159 strbuf_addch(sb, '\n');
160 } else {
161 strbuf_addf(sb, "%s: %.*s%.*s\n", what,
162 (fmt == CMIT_FMT_FULLER) ? 4 : 0,
163 " ", namelen, line);
165 switch (fmt) {
166 case CMIT_FMT_MEDIUM:
167 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, dmode));
168 break;
169 case CMIT_FMT_EMAIL:
170 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, DATE_RFC2822));
171 break;
172 case CMIT_FMT_FULLER:
173 strbuf_addf(sb, "%sDate: %s\n", what, show_date(time, tz, dmode));
174 break;
175 default:
176 /* notin' */
177 break;
181 static int is_empty_line(const char *line, int *len_p)
183 int len = *len_p;
184 while (len && isspace(line[len-1]))
185 len--;
186 *len_p = len;
187 return !len;
190 static const char *skip_empty_lines(const char *msg)
192 for (;;) {
193 int linelen = get_one_line(msg);
194 int ll = linelen;
195 if (!linelen)
196 break;
197 if (!is_empty_line(msg, &ll))
198 break;
199 msg += linelen;
201 return msg;
204 static void add_merge_info(enum cmit_fmt fmt, struct strbuf *sb,
205 const struct commit *commit, int abbrev)
207 struct commit_list *parent = commit->parents;
209 if ((fmt == CMIT_FMT_ONELINE) || (fmt == CMIT_FMT_EMAIL) ||
210 !parent || !parent->next)
211 return;
213 strbuf_addstr(sb, "Merge:");
215 while (parent) {
216 struct commit *p = parent->item;
217 const char *hex = NULL;
218 if (abbrev)
219 hex = find_unique_abbrev(p->object.sha1, abbrev);
220 if (!hex)
221 hex = sha1_to_hex(p->object.sha1);
222 parent = parent->next;
224 strbuf_addf(sb, " %s", hex);
226 strbuf_addch(sb, '\n');
229 static char *get_header(const struct commit *commit, const char *key)
231 int key_len = strlen(key);
232 const char *line = commit->buffer;
234 for (;;) {
235 const char *eol = strchr(line, '\n'), *next;
237 if (line == eol)
238 return NULL;
239 if (!eol) {
240 eol = line + strlen(line);
241 next = NULL;
242 } else
243 next = eol + 1;
244 if (eol - line > key_len &&
245 !strncmp(line, key, key_len) &&
246 line[key_len] == ' ') {
247 return xmemdupz(line + key_len + 1, eol - line - key_len - 1);
249 line = next;
253 static char *replace_encoding_header(char *buf, const char *encoding)
255 struct strbuf tmp = STRBUF_INIT;
256 size_t start, len;
257 char *cp = buf;
259 /* guess if there is an encoding header before a \n\n */
260 while (strncmp(cp, "encoding ", strlen("encoding "))) {
261 cp = strchr(cp, '\n');
262 if (!cp || *++cp == '\n')
263 return buf;
265 start = cp - buf;
266 cp = strchr(cp, '\n');
267 if (!cp)
268 return buf; /* should not happen but be defensive */
269 len = cp + 1 - (buf + start);
271 strbuf_attach(&tmp, buf, strlen(buf), strlen(buf) + 1);
272 if (is_encoding_utf8(encoding)) {
273 /* we have re-coded to UTF-8; drop the header */
274 strbuf_remove(&tmp, start, len);
275 } else {
276 /* just replaces XXXX in 'encoding XXXX\n' */
277 strbuf_splice(&tmp, start + strlen("encoding "),
278 len - strlen("encoding \n"),
279 encoding, strlen(encoding));
281 return strbuf_detach(&tmp, NULL);
284 static char *logmsg_reencode(const struct commit *commit,
285 const char *output_encoding)
287 static const char *utf8 = "UTF-8";
288 const char *use_encoding;
289 char *encoding;
290 char *out;
292 if (!*output_encoding)
293 return NULL;
294 encoding = get_header(commit, "encoding");
295 use_encoding = encoding ? encoding : utf8;
296 if (!strcmp(use_encoding, output_encoding))
297 if (encoding) /* we'll strip encoding header later */
298 out = xstrdup(commit->buffer);
299 else
300 return NULL; /* nothing to do */
301 else
302 out = reencode_string(commit->buffer,
303 output_encoding, use_encoding);
304 if (out)
305 out = replace_encoding_header(out, output_encoding);
307 free(encoding);
308 return out;
311 static int mailmap_name(char *email, int email_len, char *name, int name_len)
313 static struct string_list *mail_map;
314 if (!mail_map) {
315 mail_map = xcalloc(1, sizeof(*mail_map));
316 read_mailmap(mail_map, NULL);
318 return mail_map->nr && map_user(mail_map, email, email_len, name, name_len);
321 static size_t format_person_part(struct strbuf *sb, char part,
322 const char *msg, int len, enum date_mode dmode)
324 /* currently all placeholders have same length */
325 const int placeholder_len = 2;
326 int start, end, tz = 0;
327 unsigned long date = 0;
328 char *ep;
329 const char *name_start, *name_end, *mail_start, *mail_end, *msg_end = msg+len;
330 char person_name[1024];
331 char person_mail[1024];
333 /* advance 'end' to point to email start delimiter */
334 for (end = 0; end < len && msg[end] != '<'; end++)
335 ; /* do nothing */
338 * When end points at the '<' that we found, it should have
339 * matching '>' later, which means 'end' must be strictly
340 * below len - 1.
342 if (end >= len - 2)
343 goto skip;
345 /* Seek for both name and email part */
346 name_start = msg;
347 name_end = msg+end;
348 while (name_end > name_start && isspace(*(name_end-1)))
349 name_end--;
350 mail_start = msg+end+1;
351 mail_end = mail_start;
352 while (mail_end < msg_end && *mail_end != '>')
353 mail_end++;
354 if (mail_end == msg_end)
355 goto skip;
356 end = mail_end-msg;
358 if (part == 'N' || part == 'E') { /* mailmap lookup */
359 strlcpy(person_name, name_start, name_end-name_start+1);
360 strlcpy(person_mail, mail_start, mail_end-mail_start+1);
361 mailmap_name(person_mail, sizeof(person_mail), person_name, sizeof(person_name));
362 name_start = person_name;
363 name_end = name_start + strlen(person_name);
364 mail_start = person_mail;
365 mail_end = mail_start + strlen(person_mail);
367 if (part == 'n' || part == 'N') { /* name */
368 strbuf_add(sb, name_start, name_end-name_start);
369 return placeholder_len;
371 if (part == 'e' || part == 'E') { /* email */
372 strbuf_add(sb, mail_start, mail_end-mail_start);
373 return placeholder_len;
376 /* advance 'start' to point to date start delimiter */
377 for (start = end + 1; start < len && isspace(msg[start]); start++)
378 ; /* do nothing */
379 if (start >= len)
380 goto skip;
381 date = strtoul(msg + start, &ep, 10);
382 if (msg + start == ep)
383 goto skip;
385 if (part == 't') { /* date, UNIX timestamp */
386 strbuf_add(sb, msg + start, ep - (msg + start));
387 return placeholder_len;
390 /* parse tz */
391 for (start = ep - msg + 1; start < len && isspace(msg[start]); start++)
392 ; /* do nothing */
393 if (start + 1 < len) {
394 tz = strtoul(msg + start + 1, NULL, 10);
395 if (msg[start] == '-')
396 tz = -tz;
399 switch (part) {
400 case 'd': /* date */
401 strbuf_addstr(sb, show_date(date, tz, dmode));
402 return placeholder_len;
403 case 'D': /* date, RFC2822 style */
404 strbuf_addstr(sb, show_date(date, tz, DATE_RFC2822));
405 return placeholder_len;
406 case 'r': /* date, relative */
407 strbuf_addstr(sb, show_date(date, tz, DATE_RELATIVE));
408 return placeholder_len;
409 case 'i': /* date, ISO 8601 */
410 strbuf_addstr(sb, show_date(date, tz, DATE_ISO8601));
411 return placeholder_len;
414 skip:
416 * bogus commit, 'sb' cannot be updated, but we still need to
417 * compute a valid return value.
419 if (part == 'n' || part == 'e' || part == 't' || part == 'd'
420 || part == 'D' || part == 'r' || part == 'i')
421 return placeholder_len;
423 return 0; /* unknown placeholder */
426 struct chunk {
427 size_t off;
428 size_t len;
431 struct format_commit_context {
432 const struct commit *commit;
433 enum date_mode dmode;
434 unsigned commit_header_parsed:1;
435 unsigned commit_message_parsed:1;
437 /* These offsets are relative to the start of the commit message. */
438 struct chunk author;
439 struct chunk committer;
440 struct chunk encoding;
441 size_t message_off;
442 size_t subject_off;
443 size_t body_off;
445 /* The following ones are relative to the result struct strbuf. */
446 struct chunk abbrev_commit_hash;
447 struct chunk abbrev_tree_hash;
448 struct chunk abbrev_parent_hashes;
451 static int add_again(struct strbuf *sb, struct chunk *chunk)
453 if (chunk->len) {
454 strbuf_adddup(sb, chunk->off, chunk->len);
455 return 1;
459 * We haven't seen this chunk before. Our caller is surely
460 * going to add it the hard way now. Remember the most likely
461 * start of the to-be-added chunk: the current end of the
462 * struct strbuf.
464 chunk->off = sb->len;
465 return 0;
468 static void parse_commit_header(struct format_commit_context *context)
470 const char *msg = context->commit->buffer;
471 int i;
473 for (i = 0; msg[i]; i++) {
474 int eol;
475 for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
476 ; /* do nothing */
478 if (i == eol) {
479 break;
480 } else if (!prefixcmp(msg + i, "author ")) {
481 context->author.off = i + 7;
482 context->author.len = eol - i - 7;
483 } else if (!prefixcmp(msg + i, "committer ")) {
484 context->committer.off = i + 10;
485 context->committer.len = eol - i - 10;
486 } else if (!prefixcmp(msg + i, "encoding ")) {
487 context->encoding.off = i + 9;
488 context->encoding.len = eol - i - 9;
490 i = eol;
492 context->message_off = i;
493 context->commit_header_parsed = 1;
496 static int istitlechar(char c)
498 return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') ||
499 (c >= '0' && c <= '9') || c == '.' || c == '_';
502 static void format_sanitized_subject(struct strbuf *sb, const char *msg)
504 size_t trimlen;
505 size_t start_len = sb->len;
506 int space = 2;
508 for (; *msg && *msg != '\n'; msg++) {
509 if (istitlechar(*msg)) {
510 if (space == 1)
511 strbuf_addch(sb, '-');
512 space = 0;
513 strbuf_addch(sb, *msg);
514 if (*msg == '.')
515 while (*(msg+1) == '.')
516 msg++;
517 } else
518 space |= 1;
521 /* trim any trailing '.' or '-' characters */
522 trimlen = 0;
523 while (sb->len - trimlen > start_len &&
524 (sb->buf[sb->len - 1 - trimlen] == '.'
525 || sb->buf[sb->len - 1 - trimlen] == '-'))
526 trimlen++;
527 strbuf_remove(sb, sb->len - trimlen, trimlen);
530 const char *format_subject(struct strbuf *sb, const char *msg,
531 const char *line_separator)
533 int first = 1;
535 for (;;) {
536 const char *line = msg;
537 int linelen = get_one_line(line);
539 msg += linelen;
540 if (!linelen || is_empty_line(line, &linelen))
541 break;
543 if (!sb)
544 continue;
545 strbuf_grow(sb, linelen + 2);
546 if (!first)
547 strbuf_addstr(sb, line_separator);
548 strbuf_add(sb, line, linelen);
549 first = 0;
551 return msg;
554 static void parse_commit_message(struct format_commit_context *c)
556 const char *msg = c->commit->buffer + c->message_off;
557 const char *start = c->commit->buffer;
559 msg = skip_empty_lines(msg);
560 c->subject_off = msg - start;
562 msg = format_subject(NULL, msg, NULL);
563 msg = skip_empty_lines(msg);
564 c->body_off = msg - start;
566 c->commit_message_parsed = 1;
569 static void format_decoration(struct strbuf *sb, const struct commit *commit)
571 struct name_decoration *d;
572 const char *prefix = " (";
574 load_ref_decorations();
575 d = lookup_decoration(&name_decoration, &commit->object);
576 while (d) {
577 strbuf_addstr(sb, prefix);
578 prefix = ", ";
579 strbuf_addstr(sb, d->name);
580 d = d->next;
582 if (prefix[0] == ',')
583 strbuf_addch(sb, ')');
586 static size_t format_commit_item(struct strbuf *sb, const char *placeholder,
587 void *context)
589 struct format_commit_context *c = context;
590 const struct commit *commit = c->commit;
591 const char *msg = commit->buffer;
592 struct commit_list *p;
593 int h1, h2;
595 /* these are independent of the commit */
596 switch (placeholder[0]) {
597 case 'C':
598 if (placeholder[1] == '(') {
599 const char *end = strchr(placeholder + 2, ')');
600 char color[COLOR_MAXLEN];
601 if (!end)
602 return 0;
603 color_parse_mem(placeholder + 2,
604 end - (placeholder + 2),
605 "--pretty format", color);
606 strbuf_addstr(sb, color);
607 return end - placeholder + 1;
609 if (!prefixcmp(placeholder + 1, "red")) {
610 strbuf_addstr(sb, GIT_COLOR_RED);
611 return 4;
612 } else if (!prefixcmp(placeholder + 1, "green")) {
613 strbuf_addstr(sb, GIT_COLOR_GREEN);
614 return 6;
615 } else if (!prefixcmp(placeholder + 1, "blue")) {
616 strbuf_addstr(sb, GIT_COLOR_BLUE);
617 return 5;
618 } else if (!prefixcmp(placeholder + 1, "reset")) {
619 strbuf_addstr(sb, GIT_COLOR_RESET);
620 return 6;
621 } else
622 return 0;
623 case 'n': /* newline */
624 strbuf_addch(sb, '\n');
625 return 1;
626 case 'x':
627 /* %x00 == NUL, %x0a == LF, etc. */
628 if (0 <= (h1 = hexval_table[0xff & placeholder[1]]) &&
629 h1 <= 16 &&
630 0 <= (h2 = hexval_table[0xff & placeholder[2]]) &&
631 h2 <= 16) {
632 strbuf_addch(sb, (h1<<4)|h2);
633 return 3;
634 } else
635 return 0;
638 /* these depend on the commit */
639 if (!commit->object.parsed)
640 parse_object(commit->object.sha1);
642 switch (placeholder[0]) {
643 case 'H': /* commit hash */
644 strbuf_addstr(sb, sha1_to_hex(commit->object.sha1));
645 return 1;
646 case 'h': /* abbreviated commit hash */
647 if (add_again(sb, &c->abbrev_commit_hash))
648 return 1;
649 strbuf_addstr(sb, find_unique_abbrev(commit->object.sha1,
650 DEFAULT_ABBREV));
651 c->abbrev_commit_hash.len = sb->len - c->abbrev_commit_hash.off;
652 return 1;
653 case 'T': /* tree hash */
654 strbuf_addstr(sb, sha1_to_hex(commit->tree->object.sha1));
655 return 1;
656 case 't': /* abbreviated tree hash */
657 if (add_again(sb, &c->abbrev_tree_hash))
658 return 1;
659 strbuf_addstr(sb, find_unique_abbrev(commit->tree->object.sha1,
660 DEFAULT_ABBREV));
661 c->abbrev_tree_hash.len = sb->len - c->abbrev_tree_hash.off;
662 return 1;
663 case 'P': /* parent hashes */
664 for (p = commit->parents; p; p = p->next) {
665 if (p != commit->parents)
666 strbuf_addch(sb, ' ');
667 strbuf_addstr(sb, sha1_to_hex(p->item->object.sha1));
669 return 1;
670 case 'p': /* abbreviated parent hashes */
671 if (add_again(sb, &c->abbrev_parent_hashes))
672 return 1;
673 for (p = commit->parents; p; p = p->next) {
674 if (p != commit->parents)
675 strbuf_addch(sb, ' ');
676 strbuf_addstr(sb, find_unique_abbrev(
677 p->item->object.sha1, DEFAULT_ABBREV));
679 c->abbrev_parent_hashes.len = sb->len -
680 c->abbrev_parent_hashes.off;
681 return 1;
682 case 'm': /* left/right/bottom */
683 strbuf_addch(sb, (commit->object.flags & BOUNDARY)
684 ? '-'
685 : (commit->object.flags & SYMMETRIC_LEFT)
686 ? '<'
687 : '>');
688 return 1;
689 case 'd':
690 format_decoration(sb, commit);
691 return 1;
694 /* For the rest we have to parse the commit header. */
695 if (!c->commit_header_parsed)
696 parse_commit_header(c);
698 switch (placeholder[0]) {
699 case 'a': /* author ... */
700 return format_person_part(sb, placeholder[1],
701 msg + c->author.off, c->author.len,
702 c->dmode);
703 case 'c': /* committer ... */
704 return format_person_part(sb, placeholder[1],
705 msg + c->committer.off, c->committer.len,
706 c->dmode);
707 case 'e': /* encoding */
708 strbuf_add(sb, msg + c->encoding.off, c->encoding.len);
709 return 1;
712 /* Now we need to parse the commit message. */
713 if (!c->commit_message_parsed)
714 parse_commit_message(c);
716 switch (placeholder[0]) {
717 case 's': /* subject */
718 format_subject(sb, msg + c->subject_off, " ");
719 return 1;
720 case 'f': /* sanitized subject */
721 format_sanitized_subject(sb, msg + c->subject_off);
722 return 1;
723 case 'b': /* body */
724 strbuf_addstr(sb, msg + c->body_off);
725 return 1;
727 return 0; /* unknown placeholder */
730 void format_commit_message(const struct commit *commit,
731 const void *format, struct strbuf *sb,
732 enum date_mode dmode)
734 struct format_commit_context context;
736 memset(&context, 0, sizeof(context));
737 context.commit = commit;
738 context.dmode = dmode;
739 strbuf_expand(sb, format, format_commit_item, &context);
742 static void pp_header(enum cmit_fmt fmt,
743 int abbrev,
744 enum date_mode dmode,
745 const char *encoding,
746 const struct commit *commit,
747 const char **msg_p,
748 struct strbuf *sb)
750 int parents_shown = 0;
752 for (;;) {
753 const char *line = *msg_p;
754 int linelen = get_one_line(*msg_p);
756 if (!linelen)
757 return;
758 *msg_p += linelen;
760 if (linelen == 1)
761 /* End of header */
762 return;
764 if (fmt == CMIT_FMT_RAW) {
765 strbuf_add(sb, line, linelen);
766 continue;
769 if (!memcmp(line, "parent ", 7)) {
770 if (linelen != 48)
771 die("bad parent line in commit");
772 continue;
775 if (!parents_shown) {
776 struct commit_list *parent;
777 int num;
778 for (parent = commit->parents, num = 0;
779 parent;
780 parent = parent->next, num++)
782 /* with enough slop */
783 strbuf_grow(sb, num * 50 + 20);
784 add_merge_info(fmt, sb, commit, abbrev);
785 parents_shown = 1;
789 * MEDIUM == DEFAULT shows only author with dates.
790 * FULL shows both authors but not dates.
791 * FULLER shows both authors and dates.
793 if (!memcmp(line, "author ", 7)) {
794 strbuf_grow(sb, linelen + 80);
795 pp_user_info("Author", fmt, sb, line + 7, dmode, encoding);
797 if (!memcmp(line, "committer ", 10) &&
798 (fmt == CMIT_FMT_FULL || fmt == CMIT_FMT_FULLER)) {
799 strbuf_grow(sb, linelen + 80);
800 pp_user_info("Commit", fmt, sb, line + 10, dmode, encoding);
805 void pp_title_line(enum cmit_fmt fmt,
806 const char **msg_p,
807 struct strbuf *sb,
808 const char *subject,
809 const char *after_subject,
810 const char *encoding,
811 int need_8bit_cte)
813 const char *line_separator = (fmt == CMIT_FMT_EMAIL) ? "\n " : " ";
814 struct strbuf title;
816 strbuf_init(&title, 80);
817 *msg_p = format_subject(&title, *msg_p, line_separator);
819 strbuf_grow(sb, title.len + 1024);
820 if (subject) {
821 strbuf_addstr(sb, subject);
822 add_rfc2047(sb, title.buf, title.len, encoding);
823 } else {
824 strbuf_addbuf(sb, &title);
826 strbuf_addch(sb, '\n');
828 if (need_8bit_cte > 0) {
829 const char *header_fmt =
830 "MIME-Version: 1.0\n"
831 "Content-Type: text/plain; charset=%s\n"
832 "Content-Transfer-Encoding: 8bit\n";
833 strbuf_addf(sb, header_fmt, encoding);
835 if (after_subject) {
836 strbuf_addstr(sb, after_subject);
838 if (fmt == CMIT_FMT_EMAIL) {
839 strbuf_addch(sb, '\n');
841 strbuf_release(&title);
844 void pp_remainder(enum cmit_fmt fmt,
845 const char **msg_p,
846 struct strbuf *sb,
847 int indent)
849 int first = 1;
850 for (;;) {
851 const char *line = *msg_p;
852 int linelen = get_one_line(line);
853 *msg_p += linelen;
855 if (!linelen)
856 break;
858 if (is_empty_line(line, &linelen)) {
859 if (first)
860 continue;
861 if (fmt == CMIT_FMT_SHORT)
862 break;
864 first = 0;
866 strbuf_grow(sb, linelen + indent + 20);
867 if (indent) {
868 memset(sb->buf + sb->len, ' ', indent);
869 strbuf_setlen(sb, sb->len + indent);
871 strbuf_add(sb, line, linelen);
872 strbuf_addch(sb, '\n');
876 char *reencode_commit_message(const struct commit *commit, const char **encoding_p)
878 const char *encoding;
880 encoding = (git_log_output_encoding
881 ? git_log_output_encoding
882 : git_commit_encoding);
883 if (!encoding)
884 encoding = "UTF-8";
885 if (encoding_p)
886 *encoding_p = encoding;
887 return logmsg_reencode(commit, encoding);
890 void pretty_print_commit(enum cmit_fmt fmt, const struct commit *commit,
891 struct strbuf *sb, int abbrev,
892 const char *subject, const char *after_subject,
893 enum date_mode dmode, int need_8bit_cte)
895 unsigned long beginning_of_body;
896 int indent = 4;
897 const char *msg = commit->buffer;
898 char *reencoded;
899 const char *encoding;
901 if (fmt == CMIT_FMT_USERFORMAT) {
902 format_commit_message(commit, user_format, sb, dmode);
903 return;
906 reencoded = reencode_commit_message(commit, &encoding);
907 if (reencoded) {
908 msg = reencoded;
911 if (fmt == CMIT_FMT_ONELINE || fmt == CMIT_FMT_EMAIL)
912 indent = 0;
915 * We need to check and emit Content-type: to mark it
916 * as 8-bit if we haven't done so.
918 if (fmt == CMIT_FMT_EMAIL && need_8bit_cte == 0) {
919 int i, ch, in_body;
921 for (in_body = i = 0; (ch = msg[i]); i++) {
922 if (!in_body) {
923 /* author could be non 7-bit ASCII but
924 * the log may be so; skip over the
925 * header part first.
927 if (ch == '\n' && msg[i+1] == '\n')
928 in_body = 1;
930 else if (non_ascii(ch)) {
931 need_8bit_cte = 1;
932 break;
937 pp_header(fmt, abbrev, dmode, encoding, commit, &msg, sb);
938 if (fmt != CMIT_FMT_ONELINE && !subject) {
939 strbuf_addch(sb, '\n');
942 /* Skip excess blank lines at the beginning of body, if any... */
943 msg = skip_empty_lines(msg);
945 /* These formats treat the title line specially. */
946 if (fmt == CMIT_FMT_ONELINE || fmt == CMIT_FMT_EMAIL)
947 pp_title_line(fmt, &msg, sb, subject,
948 after_subject, encoding, need_8bit_cte);
950 beginning_of_body = sb->len;
951 if (fmt != CMIT_FMT_ONELINE)
952 pp_remainder(fmt, &msg, sb, indent);
953 strbuf_rtrim(sb);
955 /* Make sure there is an EOLN for the non-oneline case */
956 if (fmt != CMIT_FMT_ONELINE)
957 strbuf_addch(sb, '\n');
960 * The caller may append additional body text in e-mail
961 * format. Make sure we did not strip the blank line
962 * between the header and the body.
964 if (fmt == CMIT_FMT_EMAIL && sb->len <= beginning_of_body)
965 strbuf_addch(sb, '\n');
966 free(reencoded);