?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Merge branch 'bb/date-iso-strict'
[git/mingw.git] / pretty.c
blob5f012a6b8ed6cdf345b350e3fa2f6b821ceebd7d
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"
13
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);
26
27 int commit_format_is_empty(enum cmit_fmt fmt)
28 {
29 return fmt == CMIT_FMT_USERFORMAT && !*user_format;
30 }
31
32 static void save_user_format(struct rev_info *rev, const char *cp, int is_tformat)
33 {
34 free(user_format);
35 user_format = xstrdup(cp);
36 if (is_tformat)
37 rev->use_terminator = 1;
38 rev->commit_format = CMIT_FMT_USERFORMAT;
39 }
40
41 static int git_pretty_formats_config(const char *var, const char *value, void *cb)
42 {
43 struct cmt_fmt_map *commit_format = NULL;
44 const char *name;
45 const char *fmt;
46 int i;
47
48 if (!skip_prefix(var, "pretty.", &name))
49 return 0;
50
51 for (i = 0; i < builtin_formats_len; i++) {
52 if (!strcmp(commit_formats[i].name, name))
53 return 0;
54 }
55
56 for (i = builtin_formats_len; i < commit_formats_len; i++) {
57 if (!strcmp(commit_formats[i].name, name)) {
58 commit_format = &commit_formats[i];
59 break;
60 }
61 }
62
63 if (!commit_format) {
64 ALLOC_GROW(commit_formats, commit_formats_len+1,
65 commit_formats_alloc);
66 commit_format = &commit_formats[commit_formats_len];
67 memset(commit_format, 0, sizeof(*commit_format));
68 commit_formats_len++;
69 }
70
71 commit_format->name = xstrdup(name);
72 commit_format->format = CMIT_FMT_USERFORMAT;
73 if (git_config_string(&fmt, var, value))
74 return -1;
75
76 if (starts_with(fmt, "format:") || starts_with(fmt, "tformat:")) {
77 commit_format->is_tformat = fmt[0] == 't';
78 fmt = strchr(fmt, ':') + 1;
79 } else if (strchr(fmt, '%'))
80 commit_format->is_tformat = 1;
81 else
82 commit_format->is_alias = 1;
83 commit_format->user_format = fmt;
84
85 return 0;
86 }
87
88 static void setup_commit_formats(void)
89 {
90 struct cmt_fmt_map builtin_formats[] = {
91 { "raw", CMIT_FMT_RAW, 0 },
92 { "medium", CMIT_FMT_MEDIUM, 0 },
93 { "short", CMIT_FMT_SHORT, 0 },
94 { "email", CMIT_FMT_EMAIL, 0 },
95 { "fuller", CMIT_FMT_FULLER, 0 },
96 { "full", CMIT_FMT_FULL, 0 },
97 { "oneline", CMIT_FMT_ONELINE, 1 }
98 };
99 commit_formats_len = ARRAY_SIZE(builtin_formats);
100 builtin_formats_len = commit_formats_len;
101 ALLOC_GROW(commit_formats, commit_formats_len, commit_formats_alloc);
102 memcpy(commit_formats, builtin_formats,
103 sizeof(*builtin_formats)*ARRAY_SIZE(builtin_formats));
104
105 git_config(git_pretty_formats_config, NULL);
106 }
107
108 static struct cmt_fmt_map *find_commit_format_recursive(const char *sought,
109 const char *original,
110 int num_redirections)
111 {
112 struct cmt_fmt_map *found = NULL;
113 size_t found_match_len = 0;
114 int i;
115
116 if (num_redirections >= commit_formats_len)
117 die("invalid --pretty format: "
118 "'%s' references an alias which points to itself",
119 original);
120
121 for (i = 0; i < commit_formats_len; i++) {
122 size_t match_len;
123
124 if (!starts_with(commit_formats[i].name, sought))
125 continue;
126
127 match_len = strlen(commit_formats[i].name);
128 if (found == NULL || found_match_len > match_len) {
129 found = &commit_formats[i];
130 found_match_len = match_len;
131 }
132 }
133
134 if (found && found->is_alias) {
135 found = find_commit_format_recursive(found->user_format,
136 original,
137 num_redirections+1);
138 }
139
140 return found;
141 }
142
143 static struct cmt_fmt_map *find_commit_format(const char *sought)
144 {
145 if (!commit_formats)
146 setup_commit_formats();
147
148 return find_commit_format_recursive(sought, sought, 0);
149 }
150
151 void get_commit_format(const char *arg, struct rev_info *rev)
152 {
153 struct cmt_fmt_map *commit_format;
154
155 rev->use_terminator = 0;
156 if (!arg) {
157 rev->commit_format = CMIT_FMT_DEFAULT;
158 return;
159 }
160 if (starts_with(arg, "format:") || starts_with(arg, "tformat:")) {
161 save_user_format(rev, strchr(arg, ':') + 1, arg[0] == 't');
162 return;
163 }
164
165 if (!*arg || strchr(arg, '%')) {
166 save_user_format(rev, arg, 1);
167 return;
168 }
169
170 commit_format = find_commit_format(arg);
171 if (!commit_format)
172 die("invalid --pretty format: %s", arg);
173
174 rev->commit_format = commit_format->format;
175 rev->use_terminator = commit_format->is_tformat;
176 if (commit_format->format == CMIT_FMT_USERFORMAT) {
177 save_user_format(rev, commit_format->user_format,
178 commit_format->is_tformat);
179 }
180 }
181
182 /*
183 * Generic support for pretty-printing the header
184 */
185 static int get_one_line(const char *msg)
186 {
187 int ret = 0;
188
189 for (;;) {
190 char c = *msg++;
191 if (!c)
192 break;
193 ret++;
194 if (c == '\n')
195 break;
196 }
197 return ret;
198 }
199
200 /* High bit set, or ISO-2022-INT */
201 static int non_ascii(int ch)
202 {
203 return !isascii(ch) || ch == '\033';
204 }
205
206 int has_non_ascii(const char *s)
207 {
208 int ch;
209 if (!s)
210 return 0;
211 while ((ch = *s++) != '\0') {
212 if (non_ascii(ch))
213 return 1;
214 }
215 return 0;
216 }
217
218 static int is_rfc822_special(char ch)
219 {
220 switch (ch) {
221 case '(':
222 case ')':
223 case '<':
224 case '>':
225 case '[':
226 case ']':
227 case ':':
228 case ';':
229 case '@':
230 case ',':
231 case '.':
232 case '"':
233 case '\\':
234 return 1;
235 default:
236 return 0;
237 }
238 }
239
240 static int needs_rfc822_quoting(const char *s, int len)
241 {
242 int i;
243 for (i = 0; i < len; i++)
244 if (is_rfc822_special(s[i]))
245 return 1;
246 return 0;
247 }
248
249 static int last_line_length(struct strbuf *sb)
250 {
251 int i;
252
253 /* How many bytes are already used on the last line? */
254 for (i = sb->len - 1; i >= 0; i--)
255 if (sb->buf[i] == '\n')
256 break;
257 return sb->len - (i + 1);
258 }
259
260 static void add_rfc822_quoted(struct strbuf *out, const char *s, int len)
261 {
262 int i;
263
264 /* just a guess, we may have to also backslash-quote */
265 strbuf_grow(out, len + 2);
266
267 strbuf_addch(out, '"');
268 for (i = 0; i < len; i++) {
269 switch (s[i]) {
270 case '"':
271 case '\\':
272 strbuf_addch(out, '\\');
273 /* fall through */
274 default:
275 strbuf_addch(out, s[i]);
276 }
277 }
278 strbuf_addch(out, '"');
279 }
280
281 enum rfc2047_type {
282 RFC2047_SUBJECT,
283 RFC2047_ADDRESS
284 };
285
286 static int is_rfc2047_special(char ch, enum rfc2047_type type)
287 {
288 /*
289 * rfc2047, section 4.2:
290 *
291 * 8-bit values which correspond to printable ASCII characters other
292 * than "=", "?", and "_" (underscore), MAY be represented as those
293 * characters. (But see section 5 for restrictions.) In
294 * particular, SPACE and TAB MUST NOT be represented as themselves
295 * within encoded words.
296 */
297
298 /*
299 * rule out non-ASCII characters and non-printable characters (the
300 * non-ASCII check should be redundant as isprint() is not localized
301 * and only knows about ASCII, but be defensive about that)
302 */
303 if (non_ascii(ch) || !isprint(ch))
304 return 1;
305
306 /*
307 * rule out special printable characters (' ' should be the only
308 * whitespace character considered printable, but be defensive and use
309 * isspace())
310 */
311 if (isspace(ch) || ch == '=' || ch == '?' || ch == '_')
312 return 1;
313
314 /*
315 * rfc2047, section 5.3:
316 *
317 * As a replacement for a 'word' entity within a 'phrase', for example,
318 * one that precedes an address in a From, To, or Cc header. The ABNF
319 * definition for 'phrase' from RFC 822 thus becomes:
320 *
321 * phrase = 1*( encoded-word / word )
322 *
323 * In this case the set of characters that may be used in a "Q"-encoded
324 * 'encoded-word' is restricted to: <upper and lower case ASCII
325 * letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
326 * (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
327 * 'phrase' MUST be separated from any adjacent 'word', 'text' or
328 * 'special' by 'linear-white-space'.
329 */
330
331 if (type != RFC2047_ADDRESS)
332 return 0;
333
334 /* '=' and '_' are special cases and have been checked above */
335 return !(isalnum(ch) || ch == '!' || ch == '*' || ch == '+' || ch == '-' || ch == '/');
336 }
337
338 static int needs_rfc2047_encoding(const char *line, int len,
339 enum rfc2047_type type)
340 {
341 int i;
342
343 for (i = 0; i < len; i++) {
344 int ch = line[i];
345 if (non_ascii(ch) || ch == '\n')
346 return 1;
347 if ((i + 1 < len) && (ch == '=' && line[i+1] == '?'))
348 return 1;
349 }
350
351 return 0;
352 }
353
354 static void add_rfc2047(struct strbuf *sb, const char *line, size_t len,
355 const char *encoding, enum rfc2047_type type)
356 {
357 static const int max_encoded_length = 76; /* per rfc2047 */
358 int i;
359 int line_len = last_line_length(sb);
360
361 strbuf_grow(sb, len * 3 + strlen(encoding) + 100);
362 strbuf_addf(sb, "=?%s?q?", encoding);
363 line_len += strlen(encoding) + 5; /* 5 for =??q? */
364
365 while (len) {
366 /*
367 * RFC 2047, section 5 (3):
368 *
369 * Each 'encoded-word' MUST represent an integral number of
370 * characters. A multi-octet character may not be split across
371 * adjacent 'encoded- word's.
372 */
373 const unsigned char *p = (const unsigned char *)line;
374 int chrlen = mbs_chrlen(&line, &len, encoding);
375 int is_special = (chrlen > 1) || is_rfc2047_special(*p, type);
376
377 /* "=%02X" * chrlen, or the byte itself */
378 const char *encoded_fmt = is_special ? "=%02X" : "%c";
379 int encoded_len = is_special ? 3 * chrlen : 1;
380
381 /*
382 * According to RFC 2047, we could encode the special character
383 * ' ' (space) with '_' (underscore) for readability. But many
384 * programs do not understand this and just leave the
385 * underscore in place. Thus, we do nothing special here, which
386 * causes ' ' to be encoded as '=20', avoiding this problem.
387 */
388
389 if (line_len + encoded_len + 2 > max_encoded_length) {
390 /* It won't fit with trailing "?=" --- break the line */
391 strbuf_addf(sb, "?=\n =?%s?q?", encoding);
392 line_len = strlen(encoding) + 5 + 1; /* =??q? plus SP */
393 }
394
395 for (i = 0; i < chrlen; i++)
396 strbuf_addf(sb, encoded_fmt, p[i]);
397 line_len += encoded_len;
398 }
399 strbuf_addstr(sb, "?=");
400 }
401
402 const char *show_ident_date(const struct ident_split *ident,
403 enum date_mode mode)
404 {
405 unsigned long date = 0;
406 long tz = 0;
407
408 if (ident->date_begin && ident->date_end)
409 date = strtoul(ident->date_begin, NULL, 10);
410 if (date_overflows(date))
411 date = 0;
412 else {
413 if (ident->tz_begin && ident->tz_end)
414 tz = strtol(ident->tz_begin, NULL, 10);
415 if (tz >= INT_MAX || tz <= INT_MIN)
416 tz = 0;
417 }
418 return show_date(date, tz, mode);
419 }
420
421 void pp_user_info(struct pretty_print_context *pp,
422 const char *what, struct strbuf *sb,
423 const char *line, const char *encoding)
424 {
425 struct ident_split ident;
426 char *line_end;
427 const char *mailbuf, *namebuf;
428 size_t namelen, maillen;
429 int max_length = 78; /* per rfc2822 */
430
431 if (pp->fmt == CMIT_FMT_ONELINE)
432 return;
433
434 line_end = strchrnul(line, '\n');
435 if (split_ident_line(&ident, line, line_end - line))
436 return;
437
438 mailbuf = ident.mail_begin;
439 maillen = ident.mail_end - ident.mail_begin;
440 namebuf = ident.name_begin;
441 namelen = ident.name_end - ident.name_begin;
442
443 if (pp->mailmap)
444 map_user(pp->mailmap, &mailbuf, &maillen, &namebuf, &namelen);
445
446 if (pp->fmt == CMIT_FMT_EMAIL) {
447 if (pp->from_ident && ident_cmp(pp->from_ident, &ident)) {
448 struct strbuf buf = STRBUF_INIT;
449
450 strbuf_addstr(&buf, "From: ");
451 strbuf_add(&buf, namebuf, namelen);
452 strbuf_addstr(&buf, " <");
453 strbuf_add(&buf, mailbuf, maillen);
454 strbuf_addstr(&buf, ">\n");
455 string_list_append(&pp->in_body_headers,
456 strbuf_detach(&buf, NULL));
457
458 mailbuf = pp->from_ident->mail_begin;
459 maillen = pp->from_ident->mail_end - mailbuf;
460 namebuf = pp->from_ident->name_begin;
461 namelen = pp->from_ident->name_end - namebuf;
462 }
463
464 strbuf_addstr(sb, "From: ");
465 if (needs_rfc2047_encoding(namebuf, namelen, RFC2047_ADDRESS)) {
466 add_rfc2047(sb, namebuf, namelen,
467 encoding, RFC2047_ADDRESS);
468 max_length = 76; /* per rfc2047 */
469 } else if (needs_rfc822_quoting(namebuf, namelen)) {
470 struct strbuf quoted = STRBUF_INIT;
471 add_rfc822_quoted(&quoted, namebuf, namelen);
472 strbuf_add_wrapped_bytes(sb, quoted.buf, quoted.len,
473 -6, 1, max_length);
474 strbuf_release(&quoted);
475 } else {
476 strbuf_add_wrapped_bytes(sb, namebuf, namelen,
477 -6, 1, max_length);
478 }
479
480 if (max_length <
481 last_line_length(sb) + strlen(" <") + maillen + strlen(">"))
482 strbuf_addch(sb, '\n');
483 strbuf_addf(sb, " <%.*s>\n", (int)maillen, mailbuf);
484 } else {
485 strbuf_addf(sb, "%s: %.*s%.*s <%.*s>\n", what,
486 (pp->fmt == CMIT_FMT_FULLER) ? 4 : 0, " ",
487 (int)namelen, namebuf, (int)maillen, mailbuf);
488 }
489
490 switch (pp->fmt) {
491 case CMIT_FMT_MEDIUM:
492 strbuf_addf(sb, "Date: %s\n",
493 show_ident_date(&ident, pp->date_mode));
494 break;
495 case CMIT_FMT_EMAIL:
496 strbuf_addf(sb, "Date: %s\n",
497 show_ident_date(&ident, DATE_RFC2822));
498 break;
499 case CMIT_FMT_FULLER:
500 strbuf_addf(sb, "%sDate: %s\n", what,
501 show_ident_date(&ident, pp->date_mode));
502 break;
503 default:
504 /* notin' */
505 break;
506 }
507 }
508
509 static int is_empty_line(const char *line, int *len_p)
510 {
511 int len = *len_p;
512 while (len && isspace(line[len - 1]))
513 len--;
514 *len_p = len;
515 return !len;
516 }
517
518 static const char *skip_empty_lines(const char *msg)
519 {
520 for (;;) {
521 int linelen = get_one_line(msg);
522 int ll = linelen;
523 if (!linelen)
524 break;
525 if (!is_empty_line(msg, &ll))
526 break;
527 msg += linelen;
528 }
529 return msg;
530 }
531
532 static void add_merge_info(const struct pretty_print_context *pp,
533 struct strbuf *sb, const struct commit *commit)
534 {
535 struct commit_list *parent = commit->parents;
536
537 if ((pp->fmt == CMIT_FMT_ONELINE) || (pp->fmt == CMIT_FMT_EMAIL) ||
538 !parent || !parent->next)
539 return;
540
541 strbuf_addstr(sb, "Merge:");
542
543 while (parent) {
544 struct commit *p = parent->item;
545 const char *hex = NULL;
546 if (pp->abbrev)
547 hex = find_unique_abbrev(p->object.sha1, pp->abbrev);
548 if (!hex)
549 hex = sha1_to_hex(p->object.sha1);
550 parent = parent->next;
551
552 strbuf_addf(sb, " %s", hex);
553 }
554 strbuf_addch(sb, '\n');
555 }
556
557 static char *get_header(const struct commit *commit, const char *msg,
558 const char *key)
559 {
560 int key_len = strlen(key);
561 const char *line = msg;
562
563 while (line) {
564 const char *eol = strchrnul(line, '\n'), *next;
565
566 if (line == eol)
567 return NULL;
568 if (!*eol) {
569 warning("malformed commit (header is missing newline): %s",
570 sha1_to_hex(commit->object.sha1));
571 next = NULL;
572 } else
573 next = eol + 1;
574 if (eol - line > key_len &&
575 !strncmp(line, key, key_len) &&
576 line[key_len] == ' ') {
577 return xmemdupz(line + key_len + 1, eol - line - key_len - 1);
578 }
579 line = next;
580 }
581 return NULL;
582 }
583
584 static char *replace_encoding_header(char *buf, const char *encoding)
585 {
586 struct strbuf tmp = STRBUF_INIT;
587 size_t start, len;
588 char *cp = buf;
589
590 /* guess if there is an encoding header before a \n\n */
591 while (strncmp(cp, "encoding ", strlen("encoding "))) {
592 cp = strchr(cp, '\n');
593 if (!cp || *++cp == '\n')
594 return buf;
595 }
596 start = cp - buf;
597 cp = strchr(cp, '\n');
598 if (!cp)
599 return buf; /* should not happen but be defensive */
600 len = cp + 1 - (buf + start);
601
602 strbuf_attach(&tmp, buf, strlen(buf), strlen(buf) + 1);
603 if (is_encoding_utf8(encoding)) {
604 /* we have re-coded to UTF-8; drop the header */
605 strbuf_remove(&tmp, start, len);
606 } else {
607 /* just replaces XXXX in 'encoding XXXX\n' */
608 strbuf_splice(&tmp, start + strlen("encoding "),
609 len - strlen("encoding \n"),
610 encoding, strlen(encoding));
611 }
612 return strbuf_detach(&tmp, NULL);
613 }
614
615 const char *logmsg_reencode(const struct commit *commit,
616 char **commit_encoding,
617 const char *output_encoding)
618 {
619 static const char *utf8 = "UTF-8";
620 const char *use_encoding;
621 char *encoding;
622 const char *msg = get_commit_buffer(commit, NULL);
623 char *out;
624
625 if (!output_encoding || !*output_encoding) {
626 if (commit_encoding)
627 *commit_encoding =
628 get_header(commit, msg, "encoding");
629 return msg;
630 }
631 encoding = get_header(commit, msg, "encoding");
632 if (commit_encoding)
633 *commit_encoding = encoding;
634 use_encoding = encoding ? encoding : utf8;
635 if (same_encoding(use_encoding, output_encoding)) {
636 /*
637 * No encoding work to be done. If we have no encoding header
638 * at all, then there's nothing to do, and we can return the
639 * message verbatim (whether newly allocated or not).
640 */
641 if (!encoding)
642 return msg;
643
644 /*
645 * Otherwise, we still want to munge the encoding header in the
646 * result, which will be done by modifying the buffer. If we
647 * are using a fresh copy, we can reuse it. But if we are using
648 * the cached copy from get_commit_buffer, we need to duplicate it
649 * to avoid munging the cached copy.
650 */
651 if (msg == get_cached_commit_buffer(commit, NULL))
652 out = xstrdup(msg);
653 else
654 out = (char *)msg;
655 }
656 else {
657 /*
658 * There's actual encoding work to do. Do the reencoding, which
659 * still leaves the header to be replaced in the next step. At
660 * this point, we are done with msg. If we allocated a fresh
661 * copy, we can free it.
662 */
663 out = reencode_string(msg, output_encoding, use_encoding);
664 if (out)
665 unuse_commit_buffer(commit, msg);
666 }
667
668 /*
669 * This replacement actually consumes the buffer we hand it, so we do
670 * not have to worry about freeing the old "out" here.
671 */
672 if (out)
673 out = replace_encoding_header(out, output_encoding);
674
675 if (!commit_encoding)
676 free(encoding);
677 /*
678 * If the re-encoding failed, out might be NULL here; in that
679 * case we just return the commit message verbatim.
680 */
681 return out ? out : msg;
682 }
683
684 static int mailmap_name(const char **email, size_t *email_len,
685 const char **name, size_t *name_len)
686 {
687 static struct string_list *mail_map;
688 if (!mail_map) {
689 mail_map = xcalloc(1, sizeof(*mail_map));
690 read_mailmap(mail_map, NULL);
691 }
692 return mail_map->nr && map_user(mail_map, email, email_len, name, name_len);
693 }
694
695 static size_t format_person_part(struct strbuf *sb, char part,
696 const char *msg, int len, enum date_mode dmode)
697 {
698 /* currently all placeholders have same length */
699 const int placeholder_len = 2;
700 struct ident_split s;
701 const char *name, *mail;
702 size_t maillen, namelen;
703
704 if (split_ident_line(&s, msg, len) < 0)
705 goto skip;
706
707 name = s.name_begin;
708 namelen = s.name_end - s.name_begin;
709 mail = s.mail_begin;
710 maillen = s.mail_end - s.mail_begin;
711
712 if (part == 'N' || part == 'E') /* mailmap lookup */
713 mailmap_name(&mail, &maillen, &name, &namelen);
714 if (part == 'n' || part == 'N') { /* name */
715 strbuf_add(sb, name, namelen);
716 return placeholder_len;
717 }
718 if (part == 'e' || part == 'E') { /* email */
719 strbuf_add(sb, mail, maillen);
720 return placeholder_len;
721 }
722
723 if (!s.date_begin)
724 goto skip;
725
726 if (part == 't') { /* date, UNIX timestamp */
727 strbuf_add(sb, s.date_begin, s.date_end - s.date_begin);
728 return placeholder_len;
729 }
730
731 switch (part) {
732 case 'd': /* date */
733 strbuf_addstr(sb, show_ident_date(&s, dmode));
734 return placeholder_len;
735 case 'D': /* date, RFC2822 style */
736 strbuf_addstr(sb, show_ident_date(&s, DATE_RFC2822));
737 return placeholder_len;
738 case 'r': /* date, relative */
739 strbuf_addstr(sb, show_ident_date(&s, DATE_RELATIVE));
740 return placeholder_len;
741 case 'i': /* date, ISO 8601-like */
742 strbuf_addstr(sb, show_ident_date(&s, DATE_ISO8601));
743 return placeholder_len;
744 case 'I': /* date, ISO 8601 strict */
745 strbuf_addstr(sb, show_ident_date(&s, DATE_ISO8601_STRICT));
746 return placeholder_len;
747 }
748
749 skip:
750 /*
751 * reading from either a bogus commit, or a reflog entry with
752 * %gn, %ge, etc.; 'sb' cannot be updated, but we still need
753 * to compute a valid return value.
754 */
755 if (part == 'n' || part == 'e' || part == 't' || part == 'd'
756 || part == 'D' || part == 'r' || part == 'i')
757 return placeholder_len;
758
759 return 0; /* unknown placeholder */
760 }
761
762 struct chunk {
763 size_t off;
764 size_t len;
765 };
766
767 enum flush_type {
768 no_flush,
769 flush_right,
770 flush_left,
771 flush_left_and_steal,
772 flush_both
773 };
774
775 enum trunc_type {
776 trunc_none,
777 trunc_left,
778 trunc_middle,
779 trunc_right
780 };
781
782 struct format_commit_context {
783 const struct commit *commit;
784 const struct pretty_print_context *pretty_ctx;
785 unsigned commit_header_parsed:1;
786 unsigned commit_message_parsed:1;
787 struct signature_check signature_check;
788 enum flush_type flush_type;
789 enum trunc_type truncate;
790 const char *message;
791 char *commit_encoding;
792 size_t width, indent1, indent2;
793 int auto_color;
794 int padding;
795
796 /* These offsets are relative to the start of the commit message. */
797 struct chunk author;
798 struct chunk committer;
799 size_t message_off;
800 size_t subject_off;
801 size_t body_off;
802
803 /* The following ones are relative to the result struct strbuf. */
804 struct chunk abbrev_commit_hash;
805 struct chunk abbrev_tree_hash;
806 struct chunk abbrev_parent_hashes;
807 size_t wrap_start;
808 };
809
810 static int add_again(struct strbuf *sb, struct chunk *chunk)
811 {
812 if (chunk->len) {
813 strbuf_adddup(sb, chunk->off, chunk->len);
814 return 1;
815 }
816
817 /*
818 * We haven't seen this chunk before. Our caller is surely
819 * going to add it the hard way now. Remember the most likely
820 * start of the to-be-added chunk: the current end of the
821 * struct strbuf.
822 */
823 chunk->off = sb->len;
824 return 0;
825 }
826
827 static void parse_commit_header(struct format_commit_context *context)
828 {
829 const char *msg = context->message;
830 int i;
831
832 for (i = 0; msg[i]; i++) {
833 int eol;
834 for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
835 ; /* do nothing */
836
837 if (i == eol) {
838 break;
839 } else if (starts_with(msg + i, "author ")) {
840 context->author.off = i + 7;
841 context->author.len = eol - i - 7;
842 } else if (starts_with(msg + i, "committer ")) {
843 context->committer.off = i + 10;
844 context->committer.len = eol - i - 10;
845 }
846 i = eol;
847 }
848 context->message_off = i;
849 context->commit_header_parsed = 1;
850 }
851
852 static int istitlechar(char c)
853 {
854 return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') ||
855 (c >= '0' && c <= '9') || c == '.' || c == '_';
856 }
857
858 static void format_sanitized_subject(struct strbuf *sb, const char *msg)
859 {
860 size_t trimlen;
861 size_t start_len = sb->len;
862 int space = 2;
863
864 for (; *msg && *msg != '\n'; msg++) {
865 if (istitlechar(*msg)) {
866 if (space == 1)
867 strbuf_addch(sb, '-');
868 space = 0;
869 strbuf_addch(sb, *msg);
870 if (*msg == '.')
871 while (*(msg+1) == '.')
872 msg++;
873 } else
874 space |= 1;
875 }
876
877 /* trim any trailing '.' or '-' characters */
878 trimlen = 0;
879 while (sb->len - trimlen > start_len &&
880 (sb->buf[sb->len - 1 - trimlen] == '.'
881 || sb->buf[sb->len - 1 - trimlen] == '-'))
882 trimlen++;
883 strbuf_remove(sb, sb->len - trimlen, trimlen);
884 }
885
886 const char *format_subject(struct strbuf *sb, const char *msg,
887 const char *line_separator)
888 {
889 int first = 1;
890
891 for (;;) {
892 const char *line = msg;
893 int linelen = get_one_line(line);
894
895 msg += linelen;
896 if (!linelen || is_empty_line(line, &linelen))
897 break;
898
899 if (!sb)
900 continue;
901 strbuf_grow(sb, linelen + 2);
902 if (!first)
903 strbuf_addstr(sb, line_separator);
904 strbuf_add(sb, line, linelen);
905 first = 0;
906 }
907 return msg;
908 }
909
910 static void parse_commit_message(struct format_commit_context *c)
911 {
912 const char *msg = c->message + c->message_off;
913 const char *start = c->message;
914
915 msg = skip_empty_lines(msg);
916 c->subject_off = msg - start;
917
918 msg = format_subject(NULL, msg, NULL);
919 msg = skip_empty_lines(msg);
920 c->body_off = msg - start;
921
922 c->commit_message_parsed = 1;
923 }
924
925 static void strbuf_wrap(struct strbuf *sb, size_t pos,
926 size_t width, size_t indent1, size_t indent2)
927 {
928 struct strbuf tmp = STRBUF_INIT;
929
930 if (pos)
931 strbuf_add(&tmp, sb->buf, pos);
932 strbuf_add_wrapped_text(&tmp, sb->buf + pos,
933 (int) indent1, (int) indent2, (int) width);
934 strbuf_swap(&tmp, sb);
935 strbuf_release(&tmp);
936 }
937
938 static void rewrap_message_tail(struct strbuf *sb,
939 struct format_commit_context *c,
940 size_t new_width, size_t new_indent1,
941 size_t new_indent2)
942 {
943 if (c->width == new_width && c->indent1 == new_indent1 &&
944 c->indent2 == new_indent2)
945 return;
946 if (c->wrap_start < sb->len)
947 strbuf_wrap(sb, c->wrap_start, c->width, c->indent1, c->indent2);
948 c->wrap_start = sb->len;
949 c->width = new_width;
950 c->indent1 = new_indent1;
951 c->indent2 = new_indent2;
952 }
953
954 static int format_reflog_person(struct strbuf *sb,
955 char part,
956 struct reflog_walk_info *log,
957 enum date_mode dmode)
958 {
959 const char *ident;
960
961 if (!log)
962 return 2;
963
964 ident = get_reflog_ident(log);
965 if (!ident)
966 return 2;
967
968 return format_person_part(sb, part, ident, strlen(ident), dmode);
969 }
970
971 static size_t parse_color(struct strbuf *sb, /* in UTF-8 */
972 const char *placeholder,
973 struct format_commit_context *c)
974 {
975 if (placeholder[1] == '(') {
976 const char *begin = placeholder + 2;
977 const char *end = strchr(begin, ')');
978 char color[COLOR_MAXLEN];
979
980 if (!end)
981 return 0;
982 if (starts_with(begin, "auto,")) {
983 if (!want_color(c->pretty_ctx->color))
984 return end - placeholder + 1;
985 begin += 5;
986 }
987 color_parse_mem(begin,
988 end - begin,
989 "--pretty format", color);
990 strbuf_addstr(sb, color);
991 return end - placeholder + 1;
992 }
993 if (starts_with(placeholder + 1, "red")) {
994 strbuf_addstr(sb, GIT_COLOR_RED);
995 return 4;
996 } else if (starts_with(placeholder + 1, "green")) {
997 strbuf_addstr(sb, GIT_COLOR_GREEN);
998 return 6;
999 } else if (starts_with(placeholder + 1, "blue")) {
1000 strbuf_addstr(sb, GIT_COLOR_BLUE);
1001 return 5;
1002 } else if (starts_with(placeholder + 1, "reset")) {
1003 strbuf_addstr(sb, GIT_COLOR_RESET);
1004 return 6;
1005 } else
1006 return 0;
1007 }
1008
1009 static size_t parse_padding_placeholder(struct strbuf *sb,
1010 const char *placeholder,
1011 struct format_commit_context *c)
1012 {
1013 const char *ch = placeholder;
1014 enum flush_type flush_type;
1015 int to_column = 0;
1016
1017 switch (*ch++) {
1018 case '<':
1019 flush_type = flush_right;
1020 break;
1021 case '>':
1022 if (*ch == '<') {
1023 flush_type = flush_both;
1024 ch++;
1025 } else if (*ch == '>') {
1026 flush_type = flush_left_and_steal;
1027 ch++;
1028 } else
1029 flush_type = flush_left;
1030 break;
1031 default:
1032 return 0;
1033 }
1034
1035 /* the next value means "wide enough to that column" */
1036 if (*ch == '|') {
1037 to_column = 1;
1038 ch++;
1039 }
1040
1041 if (*ch == '(') {
1042 const char *start = ch + 1;
1043 const char *end = start + strcspn(start, ",)");
1044 char *next;
1045 int width;
1046 if (!end || end == start)
1047 return 0;
1048 width = strtoul(start, &next, 10);
1049 if (next == start || width == 0)
1050 return 0;
1051 c->padding = to_column ? -width : width;
1052 c->flush_type = flush_type;
1053
1054 if (*end == ',') {
1055 start = end + 1;
1056 end = strchr(start, ')');
1057 if (!end || end == start)
1058 return 0;
1059 if (starts_with(start, "trunc)"))
1060 c->truncate = trunc_right;
1061 else if (starts_with(start, "ltrunc)"))
1062 c->truncate = trunc_left;
1063 else if (starts_with(start, "mtrunc)"))
1064 c->truncate = trunc_middle;
1065 else
1066 return 0;
1067 } else
1068 c->truncate = trunc_none;
1069
1070 return end - placeholder + 1;
1071 }
1072 return 0;
1073 }
1074
1075 static size_t format_commit_one(struct strbuf *sb, /* in UTF-8 */
1076 const char *placeholder,
1077 void *context)
1078 {
1079 struct format_commit_context *c = context;
1080 const struct commit *commit = c->commit;
1081 const char *msg = c->message;
1082 struct commit_list *p;
1083 int h1, h2;
1084
1085 /* these are independent of the commit */
1086 switch (placeholder[0]) {
1087 case 'C':
1088 if (starts_with(placeholder + 1, "(auto)")) {
1089 c->auto_color = 1;
1090 return 7; /* consumed 7 bytes, "C(auto)" */
1091 } else {
1092 int ret = parse_color(sb, placeholder, c);
1093 if (ret)
1094 c->auto_color = 0;
1095 /*
1096 * Otherwise, we decided to treat %C<unknown>
1097 * as a literal string, and the previous
1098 * %C(auto) is still valid.
1099 */
1100 return ret;
1101 }
1102 case 'n': /* newline */
1103 strbuf_addch(sb, '\n');
1104 return 1;
1105 case 'x':
1106 /* %x00 == NUL, %x0a == LF, etc. */
1107 if (0 <= (h1 = hexval_table[0xff & placeholder[1]]) &&
1108 h1 <= 16 &&
1109 0 <= (h2 = hexval_table[0xff & placeholder[2]]) &&
1110 h2 <= 16) {
1111 strbuf_addch(sb, (h1<<4)|h2);
1112 return 3;
1113 } else
1114 return 0;
1115 case 'w':
1116 if (placeholder[1] == '(') {
1117 unsigned long width = 0, indent1 = 0, indent2 = 0;
1118 char *next;
1119 const char *start = placeholder + 2;
1120 const char *end = strchr(start, ')');
1121 if (!end)
1122 return 0;
1123 if (end > start) {
1124 width = strtoul(start, &next, 10);
1125 if (*next == ',') {
1126 indent1 = strtoul(next + 1, &next, 10);
1127 if (*next == ',') {
1128 indent2 = strtoul(next + 1,
1129 &next, 10);
1130 }
1131 }
1132 if (*next != ')')
1133 return 0;
1134 }
1135 rewrap_message_tail(sb, c, width, indent1, indent2);
1136 return end - placeholder + 1;
1137 } else
1138 return 0;
1139
1140 case '<':
1141 case '>':
1142 return parse_padding_placeholder(sb, placeholder, c);
1143 }
1144
1145 /* these depend on the commit */
1146 if (!commit->object.parsed)
1147 parse_object(commit->object.sha1);
1148
1149 switch (placeholder[0]) {
1150 case 'H': /* commit hash */
1151 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_COMMIT));
1152 strbuf_addstr(sb, sha1_to_hex(commit->object.sha1));
1153 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_RESET));
1154 return 1;
1155 case 'h': /* abbreviated commit hash */
1156 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_COMMIT));
1157 if (add_again(sb, &c->abbrev_commit_hash)) {
1158 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_RESET));
1159 return 1;
1160 }
1161 strbuf_addstr(sb, find_unique_abbrev(commit->object.sha1,
1162 c->pretty_ctx->abbrev));
1163 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_RESET));
1164 c->abbrev_commit_hash.len = sb->len - c->abbrev_commit_hash.off;
1165 return 1;
1166 case 'T': /* tree hash */
1167 strbuf_addstr(sb, sha1_to_hex(commit->tree->object.sha1));
1168 return 1;
1169 case 't': /* abbreviated tree hash */
1170 if (add_again(sb, &c->abbrev_tree_hash))
1171 return 1;
1172 strbuf_addstr(sb, find_unique_abbrev(commit->tree->object.sha1,
1173 c->pretty_ctx->abbrev));
1174 c->abbrev_tree_hash.len = sb->len - c->abbrev_tree_hash.off;
1175 return 1;
1176 case 'P': /* parent hashes */
1177 for (p = commit->parents; p; p = p->next) {
1178 if (p != commit->parents)
1179 strbuf_addch(sb, ' ');
1180 strbuf_addstr(sb, sha1_to_hex(p->item->object.sha1));
1181 }
1182 return 1;
1183 case 'p': /* abbreviated parent hashes */
1184 if (add_again(sb, &c->abbrev_parent_hashes))
1185 return 1;
1186 for (p = commit->parents; p; p = p->next) {
1187 if (p != commit->parents)
1188 strbuf_addch(sb, ' ');
1189 strbuf_addstr(sb, find_unique_abbrev(
1190 p->item->object.sha1,
1191 c->pretty_ctx->abbrev));
1192 }
1193 c->abbrev_parent_hashes.len = sb->len -
1194 c->abbrev_parent_hashes.off;
1195 return 1;
1196 case 'm': /* left/right/bottom */
1197 strbuf_addstr(sb, get_revision_mark(NULL, commit));
1198 return 1;
1199 case 'd':
1200 load_ref_decorations(DECORATE_SHORT_REFS);
1201 format_decorations(sb, commit, c->auto_color);
1202 return 1;
1203 case 'g': /* reflog info */
1204 switch(placeholder[1]) {
1205 case 'd': /* reflog selector */
1206 case 'D':
1207 if (c->pretty_ctx->reflog_info)
1208 get_reflog_selector(sb,
1209 c->pretty_ctx->reflog_info,
1210 c->pretty_ctx->date_mode,
1211 c->pretty_ctx->date_mode_explicit,
1212 (placeholder[1] == 'd'));
1213 return 2;
1214 case 's': /* reflog message */
1215 if (c->pretty_ctx->reflog_info)
1216 get_reflog_message(sb, c->pretty_ctx->reflog_info);
1217 return 2;
1218 case 'n':
1219 case 'N':
1220 case 'e':
1221 case 'E':
1222 return format_reflog_person(sb,
1223 placeholder[1],
1224 c->pretty_ctx->reflog_info,
1225 c->pretty_ctx->date_mode);
1226 }
1227 return 0; /* unknown %g placeholder */
1228 case 'N':
1229 if (c->pretty_ctx->notes_message) {
1230 strbuf_addstr(sb, c->pretty_ctx->notes_message);
1231 return 1;
1232 }
1233 return 0;
1234 }
1235
1236 if (placeholder[0] == 'G') {
1237 if (!c->signature_check.result)
1238 check_commit_signature(c->commit, &(c->signature_check));
1239 switch (placeholder[1]) {
1240 case 'G':
1241 if (c->signature_check.gpg_output)
1242 strbuf_addstr(sb, c->signature_check.gpg_output);
1243 break;
1244 case '?':
1245 switch (c->signature_check.result) {
1246 case 'G':
1247 case 'B':
1248 case 'U':
1249 case 'N':
1250 strbuf_addch(sb, c->signature_check.result);
1251 }
1252 break;
1253 case 'S':
1254 if (c->signature_check.signer)
1255 strbuf_addstr(sb, c->signature_check.signer);
1256 break;
1257 case 'K':
1258 if (c->signature_check.key)
1259 strbuf_addstr(sb, c->signature_check.key);
1260 break;
1261 default:
1262 return 0;
1263 }
1264 return 2;
1265 }
1266
1267
1268 /* For the rest we have to parse the commit header. */
1269 if (!c->commit_header_parsed)
1270 parse_commit_header(c);
1271
1272 switch (placeholder[0]) {
1273 case 'a': /* author ... */
1274 return format_person_part(sb, placeholder[1],
1275 msg + c->author.off, c->author.len,
1276 c->pretty_ctx->date_mode);
1277 case 'c': /* committer ... */
1278 return format_person_part(sb, placeholder[1],
1279 msg + c->committer.off, c->committer.len,
1280 c->pretty_ctx->date_mode);
1281 case 'e': /* encoding */
1282 if (c->commit_encoding)
1283 strbuf_addstr(sb, c->commit_encoding);
1284 return 1;
1285 case 'B': /* raw body */
1286 /* message_off is always left at the initial newline */
1287 strbuf_addstr(sb, msg + c->message_off + 1);
1288 return 1;
1289 }
1290
1291 /* Now we need to parse the commit message. */
1292 if (!c->commit_message_parsed)
1293 parse_commit_message(c);
1294
1295 switch (placeholder[0]) {
1296 case 's': /* subject */
1297 format_subject(sb, msg + c->subject_off, " ");
1298 return 1;
1299 case 'f': /* sanitized subject */
1300 format_sanitized_subject(sb, msg + c->subject_off);
1301 return 1;
1302 case 'b': /* body */
1303 strbuf_addstr(sb, msg + c->body_off);
1304 return 1;
1305 }
1306 return 0; /* unknown placeholder */
1307 }
1308
1309 static size_t format_and_pad_commit(struct strbuf *sb, /* in UTF-8 */
1310 const char *placeholder,
1311 struct format_commit_context *c)
1312 {
1313 struct strbuf local_sb = STRBUF_INIT;
1314 int total_consumed = 0, len, padding = c->padding;
1315 if (padding < 0) {
1316 const char *start = strrchr(sb->buf, '\n');
1317 int occupied;
1318 if (!start)
1319 start = sb->buf;
1320 occupied = utf8_strnwidth(start, -1, 1);
1321 padding = (-padding) - occupied;
1322 }
1323 while (1) {
1324 int modifier = *placeholder == 'C';
1325 int consumed = format_commit_one(&local_sb, placeholder, c);
1326 total_consumed += consumed;
1327
1328 if (!modifier)
1329 break;
1330
1331 placeholder += consumed;
1332 if (*placeholder != '%')
1333 break;
1334 placeholder++;
1335 total_consumed++;
1336 }
1337 len = utf8_strnwidth(local_sb.buf, -1, 1);
1338
1339 if (c->flush_type == flush_left_and_steal) {
1340 const char *ch = sb->buf + sb->len - 1;
1341 while (len > padding && ch > sb->buf) {
1342 const char *p;
1343 if (*ch == ' ') {
1344 ch--;
1345 padding++;
1346 continue;
1347 }
1348 /* check for trailing ansi sequences */
1349 if (*ch != 'm')
1350 break;
1351 p = ch - 1;
1352 while (ch - p < 10 && *p != '\033')
1353 p--;
1354 if (*p != '\033' ||
1355 ch + 1 - p != display_mode_esc_sequence_len(p))
1356 break;
1357 /*
1358 * got a good ansi sequence, put it back to
1359 * local_sb as we're cutting sb
1360 */
1361 strbuf_insert(&local_sb, 0, p, ch + 1 - p);
1362 ch = p - 1;
1363 }
1364 strbuf_setlen(sb, ch + 1 - sb->buf);
1365 c->flush_type = flush_left;
1366 }
1367
1368 if (len > padding) {
1369 switch (c->truncate) {
1370 case trunc_left:
1371 strbuf_utf8_replace(&local_sb,
1372 0, len - (padding - 2),
1373 "..");
1374 break;
1375 case trunc_middle:
1376 strbuf_utf8_replace(&local_sb,
1377 padding / 2 - 1,
1378 len - (padding - 2),
1379 "..");
1380 break;
1381 case trunc_right:
1382 strbuf_utf8_replace(&local_sb,
1383 padding - 2, len - (padding - 2),
1384 "..");
1385 break;
1386 case trunc_none:
1387 break;
1388 }
1389 strbuf_addbuf(sb, &local_sb);
1390 } else {
1391 int sb_len = sb->len, offset = 0;
1392 if (c->flush_type == flush_left)
1393 offset = padding - len;
1394 else if (c->flush_type == flush_both)
1395 offset = (padding - len) / 2;
1396 /*
1397 * we calculate padding in columns, now
1398 * convert it back to chars
1399 */
1400 padding = padding - len + local_sb.len;
1401 strbuf_grow(sb, padding);
1402 strbuf_setlen(sb, sb_len + padding);
1403 memset(sb->buf + sb_len, ' ', sb->len - sb_len);
1404 memcpy(sb->buf + sb_len + offset, local_sb.buf,
1405 local_sb.len);
1406 }
1407 strbuf_release(&local_sb);
1408 c->flush_type = no_flush;
1409 return total_consumed;
1410 }
1411
1412 static size_t format_commit_item(struct strbuf *sb, /* in UTF-8 */
1413 const char *placeholder,
1414 void *context)
1415 {
1416 int consumed;
1417 size_t orig_len;
1418 enum {
1419 NO_MAGIC,
1420 ADD_LF_BEFORE_NON_EMPTY,
1421 DEL_LF_BEFORE_EMPTY,
1422 ADD_SP_BEFORE_NON_EMPTY
1423 } magic = NO_MAGIC;
1424
1425 switch (placeholder[0]) {
1426 case '-':
1427 magic = DEL_LF_BEFORE_EMPTY;
1428 break;
1429 case '+':
1430 magic = ADD_LF_BEFORE_NON_EMPTY;
1431 break;
1432 case ' ':
1433 magic = ADD_SP_BEFORE_NON_EMPTY;
1434 break;
1435 default:
1436 break;
1437 }
1438 if (magic != NO_MAGIC)
1439 placeholder++;
1440
1441 orig_len = sb->len;
1442 if (((struct format_commit_context *)context)->flush_type != no_flush)
1443 consumed = format_and_pad_commit(sb, placeholder, context);
1444 else
1445 consumed = format_commit_one(sb, placeholder, context);
1446 if (magic == NO_MAGIC)
1447 return consumed;
1448
1449 if ((orig_len == sb->len) && magic == DEL_LF_BEFORE_EMPTY) {
1450 while (sb->len && sb->buf[sb->len - 1] == '\n')
1451 strbuf_setlen(sb, sb->len - 1);
1452 } else if (orig_len != sb->len) {
1453 if (magic == ADD_LF_BEFORE_NON_EMPTY)
1454 strbuf_insert(sb, orig_len, "\n", 1);
1455 else if (magic == ADD_SP_BEFORE_NON_EMPTY)
1456 strbuf_insert(sb, orig_len, " ", 1);
1457 }
1458 return consumed + 1;
1459 }
1460
1461 static size_t userformat_want_item(struct strbuf *sb, const char *placeholder,
1462 void *context)
1463 {
1464 struct userformat_want *w = context;
1465
1466 if (*placeholder == '+' || *placeholder == '-' || *placeholder == ' ')
1467 placeholder++;
1468
1469 switch (*placeholder) {
1470 case 'N':
1471 w->notes = 1;
1472 break;
1473 }
1474 return 0;
1475 }
1476
1477 void userformat_find_requirements(const char *fmt, struct userformat_want *w)
1478 {
1479 struct strbuf dummy = STRBUF_INIT;
1480
1481 if (!fmt) {
1482 if (!user_format)
1483 return;
1484 fmt = user_format;
1485 }
1486 strbuf_expand(&dummy, fmt, userformat_want_item, w);
1487 strbuf_release(&dummy);
1488 }
1489
1490 void format_commit_message(const struct commit *commit,
1491 const char *format, struct strbuf *sb,
1492 const struct pretty_print_context *pretty_ctx)
1493 {
1494 struct format_commit_context context;
1495 const char *output_enc = pretty_ctx->output_encoding;
1496 const char *utf8 = "UTF-8";
1497
1498 memset(&context, 0, sizeof(context));
1499 context.commit = commit;
1500 context.pretty_ctx = pretty_ctx;
1501 context.wrap_start = sb->len;
1502 /*
1503 * convert a commit message to UTF-8 first
1504 * as far as 'format_commit_item' assumes it in UTF-8
1505 */
1506 context.message = logmsg_reencode(commit,
1507 &context.commit_encoding,
1508 utf8);
1509
1510 strbuf_expand(sb, format, format_commit_item, &context);
1511 rewrap_message_tail(sb, &context, 0, 0, 0);
1512
1513 /* then convert a commit message to an actual output encoding */
1514 if (output_enc) {
1515 if (same_encoding(utf8, output_enc))
1516 output_enc = NULL;
1517 } else {
1518 if (context.commit_encoding &&
1519 !same_encoding(context.commit_encoding, utf8))
1520 output_enc = context.commit_encoding;
1521 }
1522
1523 if (output_enc) {
1524 int outsz;
1525 char *out = reencode_string_len(sb->buf, sb->len,
1526 output_enc, utf8, &outsz);
1527 if (out)
1528 strbuf_attach(sb, out, outsz, outsz + 1);
1529 }
1530
1531 free(context.commit_encoding);
1532 unuse_commit_buffer(commit, context.message);
1533 }
1534
1535 static void pp_header(struct pretty_print_context *pp,
1536 const char *encoding,
1537 const struct commit *commit,
1538 const char **msg_p,
1539 struct strbuf *sb)
1540 {
1541 int parents_shown = 0;
1542
1543 for (;;) {
1544 const char *line = *msg_p;
1545 int linelen = get_one_line(*msg_p);
1546
1547 if (!linelen)
1548 return;
1549 *msg_p += linelen;
1550
1551 if (linelen == 1)
1552 /* End of header */
1553 return;
1554
1555 if (pp->fmt == CMIT_FMT_RAW) {
1556 strbuf_add(sb, line, linelen);
1557 continue;
1558 }
1559
1560 if (starts_with(line, "parent ")) {
1561 if (linelen != 48)
1562 die("bad parent line in commit");
1563 continue;
1564 }
1565
1566 if (!parents_shown) {
1567 unsigned num = commit_list_count(commit->parents);
1568 /* with enough slop */
1569 strbuf_grow(sb, num * 50 + 20);
1570 add_merge_info(pp, sb, commit);
1571 parents_shown = 1;
1572 }
1573
1574 /*
1575 * MEDIUM == DEFAULT shows only author with dates.
1576 * FULL shows both authors but not dates.
1577 * FULLER shows both authors and dates.
1578 */
1579 if (starts_with(line, "author ")) {
1580 strbuf_grow(sb, linelen + 80);
1581 pp_user_info(pp, "Author", sb, line + 7, encoding);
1582 }
1583 if (starts_with(line, "committer ") &&
1584 (pp->fmt == CMIT_FMT_FULL || pp->fmt == CMIT_FMT_FULLER)) {
1585 strbuf_grow(sb, linelen + 80);
1586 pp_user_info(pp, "Commit", sb, line + 10, encoding);
1587 }
1588 }
1589 }
1590
1591 void pp_title_line(struct pretty_print_context *pp,
1592 const char **msg_p,
1593 struct strbuf *sb,
1594 const char *encoding,
1595 int need_8bit_cte)
1596 {
1597 static const int max_length = 78; /* per rfc2047 */
1598 struct strbuf title;
1599
1600 strbuf_init(&title, 80);
1601 *msg_p = format_subject(&title, *msg_p,
1602 pp->preserve_subject ? "\n" : " ");
1603
1604 strbuf_grow(sb, title.len + 1024);
1605 if (pp->subject) {
1606 strbuf_addstr(sb, pp->subject);
1607 if (needs_rfc2047_encoding(title.buf, title.len, RFC2047_SUBJECT))
1608 add_rfc2047(sb, title.buf, title.len,
1609 encoding, RFC2047_SUBJECT);
1610 else
1611 strbuf_add_wrapped_bytes(sb, title.buf, title.len,
1612 -last_line_length(sb), 1, max_length);
1613 } else {
1614 strbuf_addbuf(sb, &title);
1615 }
1616 strbuf_addch(sb, '\n');
1617
1618 if (need_8bit_cte == 0) {
1619 int i;
1620 for (i = 0; i < pp->in_body_headers.nr; i++) {
1621 if (has_non_ascii(pp->in_body_headers.items[i].string)) {
1622 need_8bit_cte = 1;
1623 break;
1624 }
1625 }
1626 }
1627
1628 if (need_8bit_cte > 0) {
1629 const char *header_fmt =
1630 "MIME-Version: 1.0\n"
1631 "Content-Type: text/plain; charset=%s\n"
1632 "Content-Transfer-Encoding: 8bit\n";
1633 strbuf_addf(sb, header_fmt, encoding);
1634 }
1635 if (pp->after_subject) {
1636 strbuf_addstr(sb, pp->after_subject);
1637 }
1638 if (pp->fmt == CMIT_FMT_EMAIL) {
1639 strbuf_addch(sb, '\n');
1640 }
1641
1642 if (pp->in_body_headers.nr) {
1643 int i;
1644 for (i = 0; i < pp->in_body_headers.nr; i++) {
1645 strbuf_addstr(sb, pp->in_body_headers.items[i].string);
1646 free(pp->in_body_headers.items[i].string);
1647 }
1648 string_list_clear(&pp->in_body_headers, 0);
1649 strbuf_addch(sb, '\n');
1650 }
1651
1652 strbuf_release(&title);
1653 }
1654
1655 void pp_remainder(struct pretty_print_context *pp,
1656 const char **msg_p,
1657 struct strbuf *sb,
1658 int indent)
1659 {
1660 int first = 1;
1661 for (;;) {
1662 const char *line = *msg_p;
1663 int linelen = get_one_line(line);
1664 *msg_p += linelen;
1665
1666 if (!linelen)
1667 break;
1668
1669 if (is_empty_line(line, &linelen)) {
1670 if (first)
1671 continue;
1672 if (pp->fmt == CMIT_FMT_SHORT)
1673 break;
1674 }
1675 first = 0;
1676
1677 strbuf_grow(sb, linelen + indent + 20);
1678 if (indent) {
1679 memset(sb->buf + sb->len, ' ', indent);
1680 strbuf_setlen(sb, sb->len + indent);
1681 }
1682 strbuf_add(sb, line, linelen);
1683 strbuf_addch(sb, '\n');
1684 }
1685 }
1686
1687 void pretty_print_commit(struct pretty_print_context *pp,
1688 const struct commit *commit,
1689 struct strbuf *sb)
1690 {
1691 unsigned long beginning_of_body;
1692 int indent = 4;
1693 const char *msg;
1694 const char *reencoded;
1695 const char *encoding;
1696 int need_8bit_cte = pp->need_8bit_cte;
1697
1698 if (pp->fmt == CMIT_FMT_USERFORMAT) {
1699 format_commit_message(commit, user_format, sb, pp);
1700 return;
1701 }
1702
1703 encoding = get_log_output_encoding();
1704 msg = reencoded = logmsg_reencode(commit, NULL, encoding);
1705
1706 if (pp->fmt == CMIT_FMT_ONELINE || pp->fmt == CMIT_FMT_EMAIL)
1707 indent = 0;
1708
1709 /*
1710 * We need to check and emit Content-type: to mark it
1711 * as 8-bit if we haven't done so.
1712 */
1713 if (pp->fmt == CMIT_FMT_EMAIL && need_8bit_cte == 0) {
1714 int i, ch, in_body;
1715
1716 for (in_body = i = 0; (ch = msg[i]); i++) {
1717 if (!in_body) {
1718 /* author could be non 7-bit ASCII but
1719 * the log may be so; skip over the
1720 * header part first.
1721 */
1722 if (ch == '\n' && msg[i+1] == '\n')
1723 in_body = 1;
1724 }
1725 else if (non_ascii(ch)) {
1726 need_8bit_cte = 1;
1727 break;
1728 }
1729 }
1730 }
1731
1732 pp_header(pp, encoding, commit, &msg, sb);
1733 if (pp->fmt != CMIT_FMT_ONELINE && !pp->subject) {
1734 strbuf_addch(sb, '\n');
1735 }
1736
1737 /* Skip excess blank lines at the beginning of body, if any... */
1738 msg = skip_empty_lines(msg);
1739
1740 /* These formats treat the title line specially. */
1741 if (pp->fmt == CMIT_FMT_ONELINE || pp->fmt == CMIT_FMT_EMAIL)
1742 pp_title_line(pp, &msg, sb, encoding, need_8bit_cte);
1743
1744 beginning_of_body = sb->len;
1745 if (pp->fmt != CMIT_FMT_ONELINE)
1746 pp_remainder(pp, &msg, sb, indent);
1747 strbuf_rtrim(sb);
1748
1749 /* Make sure there is an EOLN for the non-oneline case */
1750 if (pp->fmt != CMIT_FMT_ONELINE)
1751 strbuf_addch(sb, '\n');
1752
1753 /*
1754 * The caller may append additional body text in e-mail
1755 * format. Make sure we did not strip the blank line
1756 * between the header and the body.
1757 */
1758 if (pp->fmt == CMIT_FMT_EMAIL && sb->len <= beginning_of_body)
1759 strbuf_addch(sb, '\n');
1760
1761 unuse_commit_buffer(commit, reencoded);
1762 }
1763
1764 void pp_commit_easy(enum cmit_fmt fmt, const struct commit *commit,
1765 struct strbuf *sb)
1766 {
1767 struct pretty_print_context pp = {0};
1768 pp.fmt = fmt;
1769 pretty_print_commit(&pp, commit, sb);
1770 }
MinGW port of GIT