@leroysheep / CrazyWebTemps
Code Commits Pulls CI/CD 82,037
test
bisect jch maint master next seen test todo
c 2,386 lines 58.4 KB
Raw
1 #define USE_THE_REPOSITORY_VARIABLE
2 #define DISABLE_SIGN_COMPARE_WARNINGS
3
4 #include "git-compat-util.h"
5 #include "config.h"
6 #include "commit.h"
7 #include "environment.h"
8 #include "gettext.h"
9 #include "hash.h"
10 #include "hex.h"
11 #include "utf8.h"
12 #include "diff.h"
13 #include "pager.h"
14 #include "revision.h"
15 #include "string-list.h"
16 #include "mailmap.h"
17 #include "log-tree.h"
18 #include "notes.h"
19 #include "color.h"
20 #include "reflog-walk.h"
21 #include "gpg-interface.h"
22 #include "trailer.h"
23 #include "run-command.h"
24 #include "object-name.h"
25
26 /*
27 * The limit for formatting directives, which enable the caller to append
28 * arbitrarily many bytes to the formatted buffer. This includes padding
29 * and wrapping formatters.
30 */
31 #define FORMATTING_LIMIT (16 * 1024)
32
33 static char *user_format;
34 static struct cmt_fmt_map {
35 const char *name;
36 enum cmit_fmt format;
37 int is_tformat;
38 int expand_tabs_in_log;
39 int is_alias;
40 enum date_mode_type default_date_mode_type;
41 const char *user_format;
42 } *commit_formats;
43 static size_t builtin_formats_len;
44 static size_t commit_formats_len;
45 static size_t commit_formats_alloc;
46 static struct cmt_fmt_map *find_commit_format(const char *sought);
47
48 int commit_format_is_empty(enum cmit_fmt fmt)
49 {
50 return fmt == CMIT_FMT_USERFORMAT && !*user_format;
51 }
52
53 static void save_user_format(struct rev_info *rev, const char *cp, int is_tformat)
54 {
55 free(user_format);
56 user_format = xstrdup(cp);
57 if (is_tformat)
58 rev->use_terminator = 1;
59 rev->commit_format = CMIT_FMT_USERFORMAT;
60 }
61
62 static int git_pretty_formats_config(const char *var, const char *value,
63 const struct config_context *ctx UNUSED,
64 void *cb UNUSED)
65 {
66 struct cmt_fmt_map *commit_format = NULL;
67 const char *name, *stripped;
68 char *fmt;
69 int i;
70
71 if (!skip_prefix(var, "pretty.", &name))
72 return 0;
73
74 for (i = 0; i < builtin_formats_len; i++) {
75 if (!strcmp(commit_formats[i].name, name))
76 return 0;
77 }
78
79 for (i = builtin_formats_len; i < commit_formats_len; i++) {
80 if (!strcmp(commit_formats[i].name, name)) {
81 commit_format = &commit_formats[i];
82 break;
83 }
84 }
85
86 if (!commit_format) {
87 ALLOC_GROW(commit_formats, commit_formats_len+1,
88 commit_formats_alloc);
89 commit_format = &commit_formats[commit_formats_len];
90 memset(commit_format, 0, sizeof(*commit_format));
91 commit_formats_len++;
92 }
93
94 free((char *)commit_format->name);
95 commit_format->name = xstrdup(name);
96 commit_format->format = CMIT_FMT_USERFORMAT;
97 if (git_config_string(&fmt, var, value))
98 return -1;
99
100 free((char *)commit_format->user_format);
101 if (skip_prefix(fmt, "format:", &stripped)) {
102 commit_format->is_tformat = 0;
103 commit_format->user_format = xstrdup(stripped);
104 free(fmt);
105 } else if (skip_prefix(fmt, "tformat:", &stripped)) {
106 commit_format->is_tformat = 1;
107 commit_format->user_format = xstrdup(stripped);
108 free(fmt);
109 } else if (strchr(fmt, '%')) {
110 commit_format->is_tformat = 1;
111 commit_format->user_format = fmt;
112 } else {
113 commit_format->is_alias = 1;
114 commit_format->user_format = fmt;
115 }
116
117 return 0;
118 }
119
120 static void setup_commit_formats(void)
121 {
122 struct cmt_fmt_map builtin_formats[] = {
123 { "raw", CMIT_FMT_RAW, 0, 0 },
124 { "medium", CMIT_FMT_MEDIUM, 0, 8 },
125 { "short", CMIT_FMT_SHORT, 0, 0 },
126 { "email", CMIT_FMT_EMAIL, 0, 0 },
127 { "mboxrd", CMIT_FMT_MBOXRD, 0, 0 },
128 { "fuller", CMIT_FMT_FULLER, 0, 8 },
129 { "full", CMIT_FMT_FULL, 0, 8 },
130 { "oneline", CMIT_FMT_ONELINE, 1, 0 },
131 { "reference", CMIT_FMT_USERFORMAT, 1, 0,
132 0, DATE_SHORT, "%C(auto)%h (%s, %ad)" },
133 /*
134 * Please update $__git_log_pretty_formats in
135 * git-completion.bash when you add new formats.
136 */
137 };
138 commit_formats_len = ARRAY_SIZE(builtin_formats);
139 builtin_formats_len = commit_formats_len;
140 ALLOC_GROW(commit_formats, commit_formats_len, commit_formats_alloc);
141 COPY_ARRAY(commit_formats, builtin_formats,
142 ARRAY_SIZE(builtin_formats));
143
144 repo_config(the_repository, git_pretty_formats_config, NULL);
145 }
146
147 static struct cmt_fmt_map *find_commit_format_recursive(const char *sought,
148 const char *original,
149 int num_redirections)
150 {
151 struct cmt_fmt_map *found = NULL;
152 size_t found_match_len = 0;
153 int i;
154
155 if (num_redirections >= commit_formats_len)
156 die("invalid --pretty format: "
157 "'%s' references an alias which points to itself",
158 original);
159
160 for (i = 0; i < commit_formats_len; i++) {
161 size_t match_len;
162
163 if (!istarts_with(commit_formats[i].name, sought))
164 continue;
165
166 match_len = strlen(commit_formats[i].name);
167 if (found == NULL || found_match_len > match_len) {
168 found = &commit_formats[i];
169 found_match_len = match_len;
170 }
171 }
172
173 if (found && found->is_alias) {
174 found = find_commit_format_recursive(found->user_format,
175 original,
176 num_redirections+1);
177 }
178
179 return found;
180 }
181
182 static struct cmt_fmt_map *find_commit_format(const char *sought)
183 {
184 if (!commit_formats)
185 setup_commit_formats();
186
187 return find_commit_format_recursive(sought, sought, 0);
188 }
189
190 void get_commit_format(const char *arg, struct rev_info *rev)
191 {
192 struct cmt_fmt_map *commit_format;
193
194 rev->use_terminator = 0;
195 if (!arg) {
196 rev->commit_format = CMIT_FMT_DEFAULT;
197 return;
198 }
199 if (skip_prefix(arg, "format:", &arg)) {
200 save_user_format(rev, arg, 0);
201 return;
202 }
203
204 if (!*arg || skip_prefix(arg, "tformat:", &arg) || strchr(arg, '%')) {
205 save_user_format(rev, arg, 1);
206 return;
207 }
208
209 commit_format = find_commit_format(arg);
210 if (!commit_format)
211 die("invalid --pretty format: %s", arg);
212
213 rev->commit_format = commit_format->format;
214 rev->use_terminator = commit_format->is_tformat;
215 rev->expand_tabs_in_log_default = commit_format->expand_tabs_in_log;
216 if (!rev->date_mode_explicit && commit_format->default_date_mode_type)
217 rev->date_mode.type = commit_format->default_date_mode_type;
218 if (commit_format->format == CMIT_FMT_USERFORMAT) {
219 save_user_format(rev, commit_format->user_format,
220 commit_format->is_tformat);
221 }
222 }
223
224 /*
225 * Generic support for pretty-printing the header
226 */
227 static int get_one_line(const char *msg)
228 {
229 int ret = 0;
230
231 for (;;) {
232 char c = *msg++;
233 if (!c)
234 break;
235 ret++;
236 if (c == '\n')
237 break;
238 }
239 return ret;
240 }
241
242 /* High bit set, or ISO-2022-INT */
243 static int non_ascii(int ch)
244 {
245 return !isascii(ch) || ch == '\033';
246 }
247
248 int has_non_ascii(const char *s)
249 {
250 int ch;
251 if (!s)
252 return 0;
253 while ((ch = *s++) != '\0') {
254 if (non_ascii(ch))
255 return 1;
256 }
257 return 0;
258 }
259
260 static int is_rfc822_special(char ch)
261 {
262 switch (ch) {
263 case '(':
264 case ')':
265 case '<':
266 case '>':
267 case '[':
268 case ']':
269 case ':':
270 case ';':
271 case '@':
272 case ',':
273 case '.':
274 case '"':
275 case '\\':
276 return 1;
277 default:
278 return 0;
279 }
280 }
281
282 static int needs_rfc822_quoting(const char *s, int len)
283 {
284 int i;
285 for (i = 0; i < len; i++)
286 if (is_rfc822_special(s[i]))
287 return 1;
288 return 0;
289 }
290
291 static int last_line_length(struct strbuf *sb)
292 {
293 int i;
294
295 /* How many bytes are already used on the last line? */
296 for (i = sb->len - 1; i >= 0; i--)
297 if (sb->buf[i] == '\n')
298 break;
299 return sb->len - (i + 1);
300 }
301
302 static void add_rfc822_quoted(struct strbuf *out, const char *s, int len)
303 {
304 int i;
305
306 /* just a guess, we may have to also backslash-quote */
307 strbuf_grow(out, len + 2);
308
309 strbuf_addch(out, '"');
310 for (i = 0; i < len; i++) {
311 switch (s[i]) {
312 case '"':
313 case '\\':
314 strbuf_addch(out, '\\');
315 /* fall through */
316 default:
317 strbuf_addch(out, s[i]);
318 }
319 }
320 strbuf_addch(out, '"');
321 }
322
323 enum rfc2047_type {
324 RFC2047_SUBJECT,
325 RFC2047_ADDRESS
326 };
327
328 static int is_rfc2047_special(char ch, enum rfc2047_type type)
329 {
330 /*
331 * rfc2047, section 4.2:
332 *
333 * 8-bit values which correspond to printable ASCII characters other
334 * than "=", "?", and "_" (underscore), MAY be represented as those
335 * characters. (But see section 5 for restrictions.) In
336 * particular, SPACE and TAB MUST NOT be represented as themselves
337 * within encoded words.
338 */
339
340 /*
341 * rule out non-ASCII characters and non-printable characters (the
342 * non-ASCII check should be redundant as isprint() is not localized
343 * and only knows about ASCII, but be defensive about that)
344 */
345 if (non_ascii(ch) || !isprint(ch))
346 return 1;
347
348 /*
349 * rule out special printable characters (' ' should be the only
350 * whitespace character considered printable, but be defensive and use
351 * isspace())
352 */
353 if (isspace(ch) || ch == '=' || ch == '?' || ch == '_')
354 return 1;
355
356 /*
357 * rfc2047, section 5.3:
358 *
359 * As a replacement for a 'word' entity within a 'phrase', for example,
360 * one that precedes an address in a From, To, or Cc header. The ABNF
361 * definition for 'phrase' from RFC 822 thus becomes:
362 *
363 * phrase = 1*( encoded-word / word )
364 *
365 * In this case the set of characters that may be used in a "Q"-encoded
366 * 'encoded-word' is restricted to: <upper and lower case ASCII
367 * letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
368 * (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
369 * 'phrase' MUST be separated from any adjacent 'word', 'text' or
370 * 'special' by 'linear-white-space'.
371 */
372
373 if (type != RFC2047_ADDRESS)
374 return 0;
375
376 /* '=' and '_' are special cases and have been checked above */
377 return !(isalnum(ch) || ch == '!' || ch == '*' || ch == '+' || ch == '-' || ch == '/');
378 }
379
380 static int needs_rfc2047_encoding(const char *line, int len)
381 {
382 int i;
383
384 for (i = 0; i < len; i++) {
385 int ch = line[i];
386 if (non_ascii(ch) || ch == '\n')
387 return 1;
388 if ((i + 1 < len) && (ch == '=' && line[i+1] == '?'))
389 return 1;
390 }
391
392 return 0;
393 }
394
395 static void add_rfc2047(struct strbuf *sb, const char *line, size_t len,
396 const char *encoding, enum rfc2047_type type)
397 {
398 static const int max_encoded_length = 76; /* per rfc2047 */
399 int i;
400 int line_len = last_line_length(sb);
401
402 strbuf_addf(sb, "=?%s?q?", encoding);
403 line_len += strlen(encoding) + 5; /* 5 for =??q? */
404
405 while (len) {
406 /*
407 * RFC 2047, section 5 (3):
408 *
409 * Each 'encoded-word' MUST represent an integral number of
410 * characters. A multi-octet character may not be split across
411 * adjacent 'encoded- word's.
412 */
413 const unsigned char *p = (const unsigned char *)line;
414 int chrlen = mbs_chrlen(&line, &len, encoding);
415 int is_special = (chrlen > 1) || is_rfc2047_special(*p, type);
416
417 /* "=%02X" * chrlen, or the byte itself */
418 const char *encoded_fmt = is_special ? "=%02X" : "%c";
419 int encoded_len = is_special ? 3 * chrlen : 1;
420
421 /*
422 * According to RFC 2047, we could encode the special character
423 * ' ' (space) with '_' (underscore) for readability. But many
424 * programs do not understand this and just leave the
425 * underscore in place. Thus, we do nothing special here, which
426 * causes ' ' to be encoded as '=20', avoiding this problem.
427 */
428
429 if (line_len + encoded_len + 2 > max_encoded_length) {
430 /* It won't fit with trailing "?=" --- break the line */
431 strbuf_addf(sb, "?=\n =?%s?q?", encoding);
432 line_len = strlen(encoding) + 5 + 1; /* =??q? plus SP */
433 }
434
435 for (i = 0; i < chrlen; i++)
436 strbuf_addf(sb, encoded_fmt, p[i]);
437 line_len += encoded_len;
438 }
439 strbuf_addstr(sb, "?=");
440 }
441
442 const char *show_ident_date(const struct ident_split *ident,
443 struct date_mode mode)
444 {
445 timestamp_t date = 0;
446 long tz = 0;
447
448 if (ident->date_begin && ident->date_end)
449 date = parse_timestamp(ident->date_begin, NULL, 10);
450 if (date_overflows(date))
451 date = 0;
452 else {
453 if (ident->tz_begin && ident->tz_end)
454 tz = strtol(ident->tz_begin, NULL, 10);
455 if (tz >= INT_MAX || tz <= INT_MIN)
456 tz = 0;
457 }
458 return show_date(date, tz, mode);
459 }
460
461 static inline void strbuf_add_with_color(struct strbuf *sb, const char *color,
462 const char *buf, size_t buflen)
463 {
464 strbuf_addstr(sb, color);
465 strbuf_add(sb, buf, buflen);
466 if (*color)
467 strbuf_addstr(sb, GIT_COLOR_RESET);
468 }
469
470 static void append_line_with_color(struct strbuf *sb, struct grep_opt *opt,
471 const char *line, size_t linelen,
472 enum git_colorbool color, enum grep_context ctx,
473 enum grep_header_field field)
474 {
475 const char *buf, *eol, *line_color, *match_color;
476 regmatch_t match;
477 int eflags = 0;
478
479 buf = line;
480 eol = buf + linelen;
481
482 if (!opt || !want_color(color) || opt->invert)
483 goto end;
484
485 line_color = opt->colors[GREP_COLOR_SELECTED];
486 match_color = opt->colors[GREP_COLOR_MATCH_SELECTED];
487
488 while (grep_next_match(opt, buf, eol, ctx, &match, field, eflags)) {
489 if (match.rm_so == match.rm_eo)
490 break;
491
492 strbuf_add_with_color(sb, line_color, buf, match.rm_so);
493 strbuf_add_with_color(sb, match_color, buf + match.rm_so,
494 match.rm_eo - match.rm_so);
495 buf += match.rm_eo;
496 eflags = REG_NOTBOL;
497 }
498
499 if (eflags)
500 strbuf_add_with_color(sb, line_color, buf, eol - buf);
501 else {
502 end:
503 strbuf_add(sb, buf, eol - buf);
504 }
505 }
506
507 static int use_in_body_from(const struct pretty_print_context *pp,
508 const struct ident_split *ident)
509 {
510 if (pp->rev && pp->rev->force_in_body_from)
511 return 1;
512 if (ident_cmp(pp->from_ident, ident))
513 return 1;
514 return 0;
515 }
516
517 void pp_user_info(struct pretty_print_context *pp,
518 const char *what, struct strbuf *sb,
519 const char *line, const char *encoding)
520 {
521 struct ident_split ident;
522 char *line_end;
523 const char *mailbuf, *namebuf;
524 size_t namelen, maillen;
525 int max_length = 78; /* per rfc2822 */
526
527 if (pp->fmt == CMIT_FMT_ONELINE)
528 return;
529
530 line_end = strchrnul(line, '\n');
531 if (split_ident_line(&ident, line, line_end - line))
532 return;
533
534 mailbuf = ident.mail_begin;
535 maillen = ident.mail_end - ident.mail_begin;
536 namebuf = ident.name_begin;
537 namelen = ident.name_end - ident.name_begin;
538
539 if (pp->mailmap)
540 map_user(pp->mailmap, &mailbuf, &maillen, &namebuf, &namelen);
541
542 if (cmit_fmt_is_mail(pp->fmt)) {
543 if (pp->from_ident && use_in_body_from(pp, &ident)) {
544 struct strbuf buf = STRBUF_INIT;
545
546 strbuf_addstr(&buf, "From: ");
547 strbuf_add(&buf, namebuf, namelen);
548 strbuf_addstr(&buf, " <");
549 strbuf_add(&buf, mailbuf, maillen);
550 strbuf_addstr(&buf, ">\n");
551 string_list_append(&pp->in_body_headers,
552 strbuf_detach(&buf, NULL));
553
554 mailbuf = pp->from_ident->mail_begin;
555 maillen = pp->from_ident->mail_end - mailbuf;
556 namebuf = pp->from_ident->name_begin;
557 namelen = pp->from_ident->name_end - namebuf;
558 }
559
560 strbuf_addstr(sb, "From: ");
561 if (pp->encode_email_headers &&
562 needs_rfc2047_encoding(namebuf, namelen)) {
563 add_rfc2047(sb, namebuf, namelen,
564 encoding, RFC2047_ADDRESS);
565 max_length = 76; /* per rfc2047 */
566 } else if (needs_rfc822_quoting(namebuf, namelen)) {
567 struct strbuf quoted = STRBUF_INIT;
568 add_rfc822_quoted(&quoted, namebuf, namelen);
569 strbuf_add_wrapped_bytes(sb, quoted.buf, quoted.len,
570 -6, 1, max_length);
571 strbuf_release(&quoted);
572 } else {
573 strbuf_add_wrapped_bytes(sb, namebuf, namelen,
574 -6, 1, max_length);
575 }
576
577 if (max_length <
578 last_line_length(sb) + strlen(" <") + maillen + strlen(">"))
579 strbuf_addch(sb, '\n');
580 strbuf_addf(sb, " <%.*s>\n", (int)maillen, mailbuf);
581 } else {
582 struct strbuf id = STRBUF_INIT;
583 enum grep_header_field field = GREP_HEADER_FIELD_MAX;
584 struct grep_opt *opt = pp->rev ? &pp->rev->grep_filter : NULL;
585
586 if (!strcmp(what, "Author"))
587 field = GREP_HEADER_AUTHOR;
588 else if (!strcmp(what, "Commit"))
589 field = GREP_HEADER_COMMITTER;
590
591 strbuf_addf(sb, "%s: ", what);
592 if (pp->fmt == CMIT_FMT_FULLER)
593 strbuf_addchars(sb, ' ', 4);
594
595 strbuf_addf(&id, "%.*s <%.*s>", (int)namelen, namebuf,
596 (int)maillen, mailbuf);
597
598 append_line_with_color(sb, opt, id.buf, id.len, pp->color,
599 GREP_CONTEXT_HEAD, field);
600 strbuf_addch(sb, '\n');
601 strbuf_release(&id);
602 }
603
604 switch (pp->fmt) {
605 case CMIT_FMT_MEDIUM:
606 strbuf_addf(sb, "Date: %s\n",
607 show_ident_date(&ident, pp->date_mode));
608 break;
609 case CMIT_FMT_EMAIL:
610 case CMIT_FMT_MBOXRD:
611 strbuf_addf(sb, "Date: %s\n",
612 show_ident_date(&ident, DATE_MODE(RFC2822)));
613 break;
614 case CMIT_FMT_FULLER:
615 strbuf_addf(sb, "%sDate: %s\n", what,
616 show_ident_date(&ident, pp->date_mode));
617 break;
618 default:
619 /* notin' */
620 break;
621 }
622 }
623
624 static int is_blank_line(const char *line, int *len_p)
625 {
626 int len = *len_p;
627 while (len && isspace(line[len - 1]))
628 len--;
629 *len_p = len;
630 return !len;
631 }
632
633 const char *skip_blank_lines(const char *msg)
634 {
635 for (;;) {
636 int linelen = get_one_line(msg);
637 int ll = linelen;
638 if (!linelen)
639 break;
640 if (!is_blank_line(msg, &ll))
641 break;
642 msg += linelen;
643 }
644 return msg;
645 }
646
647 static void add_merge_info(const struct pretty_print_context *pp,
648 struct strbuf *sb, const struct commit *commit)
649 {
650 struct commit_list *parent = commit->parents;
651
652 if ((pp->fmt == CMIT_FMT_ONELINE) || (cmit_fmt_is_mail(pp->fmt)) ||
653 !parent || !parent->next)
654 return;
655
656 strbuf_addstr(sb, "Merge:");
657
658 while (parent) {
659 struct object_id *oidp = &parent->item->object.oid;
660 strbuf_addch(sb, ' ');
661 if (pp->abbrev)
662 strbuf_add_unique_abbrev(sb, oidp, pp->abbrev);
663 else
664 strbuf_add_oid_hex(sb, oidp);
665 parent = parent->next;
666 }
667 strbuf_addch(sb, '\n');
668 }
669
670 static char *get_header(const char *msg, const char *key)
671 {
672 size_t len;
673 const char *v = find_commit_header(msg, key, &len);
674 return v ? xmemdupz(v, len) : NULL;
675 }
676
677 static char *replace_encoding_header(char *buf, const char *encoding)
678 {
679 struct strbuf tmp = STRBUF_INIT;
680 size_t start, len;
681 char *cp = buf;
682
683 /* guess if there is an encoding header before a \n\n */
684 while (!starts_with(cp, "encoding ")) {
685 cp = strchr(cp, '\n');
686 if (!cp || *++cp == '\n')
687 return buf;
688 }
689 start = cp - buf;
690 cp = strchr(cp, '\n');
691 if (!cp)
692 return buf; /* should not happen but be defensive */
693 len = cp + 1 - (buf + start);
694
695 strbuf_attach(&tmp, buf, strlen(buf), strlen(buf) + 1);
696 if (is_encoding_utf8(encoding)) {
697 /* we have re-coded to UTF-8; drop the header */
698 strbuf_remove(&tmp, start, len);
699 } else {
700 /* just replaces XXXX in 'encoding XXXX\n' */
701 strbuf_splice(&tmp, start + strlen("encoding "),
702 len - strlen("encoding \n"),
703 encoding, strlen(encoding));
704 }
705 return strbuf_detach(&tmp, NULL);
706 }
707
708 const char *repo_logmsg_reencode(struct repository *r,
709 const struct commit *commit,
710 char **commit_encoding,
711 const char *output_encoding)
712 {
713 static const char *utf8 = "UTF-8";
714 const char *use_encoding;
715 char *encoding;
716 const char *msg = repo_get_commit_buffer(r, commit, NULL);
717 char *out;
718
719 if (!output_encoding || !*output_encoding) {
720 if (commit_encoding)
721 *commit_encoding = get_header(msg, "encoding");
722 return msg;
723 }
724 encoding = get_header(msg, "encoding");
725 if (commit_encoding)
726 *commit_encoding = encoding;
727 use_encoding = encoding ? encoding : utf8;
728 if (same_encoding(use_encoding, output_encoding)) {
729 /*
730 * No encoding work to be done. If we have no encoding header
731 * at all, then there's nothing to do, and we can return the
732 * message verbatim (whether newly allocated or not).
733 */
734 if (!encoding)
735 return msg;
736
737 /*
738 * Otherwise, we still want to munge the encoding header in the
739 * result, which will be done by modifying the buffer. If we
740 * are using a fresh copy, we can reuse it. But if we are using
741 * the cached copy from repo_get_commit_buffer, we need to duplicate it
742 * to avoid munging the cached copy.
743 */
744 if (msg == get_cached_commit_buffer(r, commit, NULL))
745 out = xstrdup(msg);
746 else
747 out = (char *)msg;
748 }
749 else {
750 /*
751 * There's actual encoding work to do. Do the reencoding, which
752 * still leaves the header to be replaced in the next step. At
753 * this point, we are done with msg. If we allocated a fresh
754 * copy, we can free it.
755 */
756 out = reencode_string(msg, output_encoding, use_encoding);
757 if (out)
758 repo_unuse_commit_buffer(r, commit, msg);
759 }
760
761 /*
762 * This replacement actually consumes the buffer we hand it, so we do
763 * not have to worry about freeing the old "out" here.
764 */
765 if (out)
766 out = replace_encoding_header(out, output_encoding);
767
768 if (!commit_encoding)
769 free(encoding);
770 /*
771 * If the re-encoding failed, out might be NULL here; in that
772 * case we just return the commit message verbatim.
773 */
774 return out ? out : msg;
775 }
776
777 static int mailmap_name(const char **email, size_t *email_len,
778 const char **name, size_t *name_len)
779 {
780 static struct string_list *mail_map;
781 if (!mail_map) {
782 CALLOC_ARRAY(mail_map, 1);
783 read_mailmap(the_repository, mail_map);
784 }
785 return mail_map->nr && map_user(mail_map, email, email_len, name, name_len);
786 }
787
788 static size_t format_person_part(struct strbuf *sb, char part,
789 const char *msg, int len,
790 struct date_mode dmode)
791 {
792 /* currently all placeholders have same length */
793 const int placeholder_len = 2;
794 struct ident_split s;
795 const char *name, *mail;
796 size_t maillen, namelen;
797
798 if (split_ident_line(&s, msg, len) < 0)
799 goto skip;
800
801 name = s.name_begin;
802 namelen = s.name_end - s.name_begin;
803 mail = s.mail_begin;
804 maillen = s.mail_end - s.mail_begin;
805
806 if (part == 'N' || part == 'E' || part == 'L') /* mailmap lookup */
807 mailmap_name(&mail, &maillen, &name, &namelen);
808 if (part == 'n' || part == 'N') { /* name */
809 strbuf_add(sb, name, namelen);
810 return placeholder_len;
811 }
812 if (part == 'e' || part == 'E') { /* email */
813 strbuf_add(sb, mail, maillen);
814 return placeholder_len;
815 }
816 if (part == 'l' || part == 'L') { /* local-part */
817 const char *at = memchr(mail, '@', maillen);
818 if (at)
819 maillen = at - mail;
820 strbuf_add(sb, mail, maillen);
821 return placeholder_len;
822 }
823
824 if (!s.date_begin)
825 goto skip;
826
827 if (part == 't') { /* date, UNIX timestamp */
828 strbuf_add(sb, s.date_begin, s.date_end - s.date_begin);
829 return placeholder_len;
830 }
831
832 switch (part) {
833 case 'd': /* date */
834 strbuf_addstr(sb, show_ident_date(&s, dmode));
835 return placeholder_len;
836 case 'D': /* date, RFC2822 style */
837 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(RFC2822)));
838 return placeholder_len;
839 case 'r': /* date, relative */
840 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(RELATIVE)));
841 return placeholder_len;
842 case 'i': /* date, ISO 8601-like */
843 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(ISO8601)));
844 return placeholder_len;
845 case 'I': /* date, ISO 8601 strict */
846 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(ISO8601_STRICT)));
847 return placeholder_len;
848 case 'h': /* date, human */
849 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(HUMAN)));
850 return placeholder_len;
851 case 's':
852 strbuf_addstr(sb, show_ident_date(&s, DATE_MODE(SHORT)));
853 return placeholder_len;
854 }
855
856 skip:
857 /*
858 * reading from either a bogus commit, or a reflog entry with
859 * %gn, %ge, etc.; 'sb' cannot be updated, but we still need
860 * to compute a valid return value.
861 */
862 if (part == 'n' || part == 'e' || part == 't' || part == 'd'
863 || part == 'D' || part == 'r' || part == 'i')
864 return placeholder_len;
865
866 return 0; /* unknown placeholder */
867 }
868
869 struct chunk {
870 size_t off;
871 size_t len;
872 };
873
874 enum flush_type {
875 no_flush,
876 flush_right,
877 flush_left,
878 flush_left_and_steal,
879 flush_both
880 };
881
882 enum trunc_type {
883 trunc_none,
884 trunc_left,
885 trunc_middle,
886 trunc_right
887 };
888
889 struct format_commit_context {
890 struct repository *repository;
891 const struct commit *commit;
892 const struct pretty_print_context *pretty_ctx;
893 unsigned commit_header_parsed:1;
894 unsigned commit_message_parsed:1;
895 struct signature_check signature_check;
896 enum flush_type flush_type;
897 enum trunc_type truncate;
898 const char *message;
899 char *commit_encoding;
900 size_t width, indent1, indent2;
901 enum git_colorbool auto_color;
902 int padding;
903
904 /* These offsets are relative to the start of the commit message. */
905 struct chunk author;
906 struct chunk committer;
907 size_t message_off;
908 size_t subject_off;
909 size_t body_off;
910
911 /* The following ones are relative to the result struct strbuf. */
912 size_t wrap_start;
913 };
914
915 static void parse_commit_header(struct format_commit_context *context)
916 {
917 const char *msg = context->message;
918 int i;
919
920 for (i = 0; msg[i]; i++) {
921 const char *name;
922 int eol;
923 for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
924 ; /* do nothing */
925
926 if (i == eol) {
927 break;
928 } else if (skip_prefix(msg + i, "author ", &name)) {
929 context->author.off = name - msg;
930 context->author.len = msg + eol - name;
931 } else if (skip_prefix(msg + i, "committer ", &name)) {
932 context->committer.off = name - msg;
933 context->committer.len = msg + eol - name;
934 }
935 i = eol;
936 }
937 context->message_off = i;
938 context->commit_header_parsed = 1;
939 }
940
941 static int istitlechar(char c)
942 {
943 return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') ||
944 (c >= '0' && c <= '9') || c == '.' || c == '_';
945 }
946
947 void format_sanitized_subject(struct strbuf *sb, const char *msg, size_t len)
948 {
949 size_t trimlen;
950 size_t start_len = sb->len;
951 int space = 2;
952 int i;
953
954 for (i = 0; i < len; i++) {
955 if (istitlechar(msg[i])) {
956 if (space == 1)
957 strbuf_addch(sb, '-');
958 space = 0;
959 strbuf_addch(sb, msg[i]);
960 if (msg[i] == '.')
961 while (msg[i+1] == '.')
962 i++;
963 } else
964 space |= 1;
965 }
966
967 /* trim any trailing '.' or '-' characters */
968 trimlen = 0;
969 while (sb->len - trimlen > start_len &&
970 (sb->buf[sb->len - 1 - trimlen] == '.'
971 || sb->buf[sb->len - 1 - trimlen] == '-'))
972 trimlen++;
973 strbuf_remove(sb, sb->len - trimlen, trimlen);
974 }
975
976 const char *format_subject(struct strbuf *sb, const char *msg,
977 const char *line_separator)
978 {
979 int first = 1;
980
981 for (;;) {
982 const char *line = msg;
983 int linelen = get_one_line(line);
984
985 msg += linelen;
986 if (!linelen || is_blank_line(line, &linelen))
987 break;
988
989 if (!sb)
990 continue;
991 strbuf_grow(sb, linelen + 2);
992 if (!first)
993 strbuf_addstr(sb, line_separator);
994 strbuf_add(sb, line, linelen);
995 first = 0;
996 }
997 return msg;
998 }
999
1000 static void parse_commit_message(struct format_commit_context *c)
1001 {
1002 const char *msg = c->message + c->message_off;
1003 const char *start = c->message;
1004
1005 msg = skip_blank_lines(msg);
1006 c->subject_off = msg - start;
1007
1008 msg = format_subject(NULL, msg, NULL);
1009 msg = skip_blank_lines(msg);
1010 c->body_off = msg - start;
1011
1012 c->commit_message_parsed = 1;
1013 }
1014
1015 static void strbuf_wrap(struct strbuf *sb, size_t pos,
1016 size_t width, size_t indent1, size_t indent2)
1017 {
1018 struct strbuf tmp = STRBUF_INIT;
1019
1020 if (pos)
1021 strbuf_add(&tmp, sb->buf, pos);
1022 strbuf_add_wrapped_text(&tmp, sb->buf + pos,
1023 cast_size_t_to_int(indent1),
1024 cast_size_t_to_int(indent2),
1025 cast_size_t_to_int(width));
1026 strbuf_swap(&tmp, sb);
1027 strbuf_release(&tmp);
1028 }
1029
1030 static void rewrap_message_tail(struct strbuf *sb,
1031 struct format_commit_context *c,
1032 size_t new_width, size_t new_indent1,
1033 size_t new_indent2)
1034 {
1035 if (c->width == new_width && c->indent1 == new_indent1 &&
1036 c->indent2 == new_indent2)
1037 return;
1038 if (c->wrap_start < sb->len)
1039 strbuf_wrap(sb, c->wrap_start, c->width, c->indent1, c->indent2);
1040 c->wrap_start = sb->len;
1041 c->width = new_width;
1042 c->indent1 = new_indent1;
1043 c->indent2 = new_indent2;
1044 }
1045
1046 static int format_reflog_person(struct strbuf *sb,
1047 char part,
1048 struct reflog_walk_info *log,
1049 struct date_mode dmode)
1050 {
1051 const char *ident;
1052
1053 if (!log)
1054 return 2;
1055
1056 ident = get_reflog_ident(log);
1057 if (!ident)
1058 return 2;
1059
1060 return format_person_part(sb, part, ident, strlen(ident), dmode);
1061 }
1062
1063 static size_t parse_color(struct strbuf *sb, /* in UTF-8 */
1064 const char *placeholder,
1065 struct format_commit_context *c)
1066 {
1067 const char *rest = placeholder;
1068 const char *basic_color = NULL;
1069
1070 if (placeholder[1] == '(') {
1071 const char *begin = placeholder + 2;
1072 const char *end = strchr(begin, ')');
1073 char color[COLOR_MAXLEN];
1074
1075 if (!end)
1076 return 0;
1077
1078 if (skip_prefix(begin, "auto,", &begin)) {
1079 if (!want_color(c->pretty_ctx->color))
1080 return end - placeholder + 1;
1081 } else if (skip_prefix(begin, "always,", &begin)) {
1082 /* nothing to do; we do not respect want_color at all */
1083 } else {
1084 /* the default is the same as "auto" */
1085 if (!want_color(c->pretty_ctx->color))
1086 return end - placeholder + 1;
1087 }
1088
1089 if (color_parse_mem(begin, end - begin, color) < 0)
1090 die(_("unable to parse --pretty format"));
1091 strbuf_addstr(sb, color);
1092 return end - placeholder + 1;
1093 }
1094
1095 /*
1096 * We handle things like "%C(red)" above; for historical reasons, there
1097 * are a few colors that can be specified without parentheses (and
1098 * they cannot support things like "auto" or "always" at all).
1099 */
1100 if (skip_prefix(placeholder + 1, "red", &rest))
1101 basic_color = GIT_COLOR_RED;
1102 else if (skip_prefix(placeholder + 1, "green", &rest))
1103 basic_color = GIT_COLOR_GREEN;
1104 else if (skip_prefix(placeholder + 1, "blue", &rest))
1105 basic_color = GIT_COLOR_BLUE;
1106 else if (skip_prefix(placeholder + 1, "reset", &rest))
1107 basic_color = GIT_COLOR_RESET;
1108
1109 if (basic_color && want_color(c->pretty_ctx->color))
1110 strbuf_addstr(sb, basic_color);
1111
1112 return rest - placeholder;
1113 }
1114
1115 static size_t parse_padding_placeholder(const char *placeholder,
1116 struct format_commit_context *c)
1117 {
1118 const char *ch = placeholder;
1119 enum flush_type flush_type;
1120 int to_column = 0;
1121
1122 switch (*ch++) {
1123 case '<':
1124 flush_type = flush_right;
1125 break;
1126 case '>':
1127 if (*ch == '<') {
1128 flush_type = flush_both;
1129 ch++;
1130 } else if (*ch == '>') {
1131 flush_type = flush_left_and_steal;
1132 ch++;
1133 } else
1134 flush_type = flush_left;
1135 break;
1136 default:
1137 return 0;
1138 }
1139
1140 /* the next value means "wide enough to that column" */
1141 if (*ch == '|') {
1142 to_column = 1;
1143 ch++;
1144 }
1145
1146 if (*ch == '(') {
1147 const char *start = ch + 1;
1148 const char *end = start + strcspn(start, ",)");
1149 char *next;
1150 int width;
1151 if (!*end || end == start)
1152 return 0;
1153 width = strtol(start, &next, 10);
1154
1155 /*
1156 * We need to limit the amount of padding, or otherwise this
1157 * would allow the user to pad the buffer by arbitrarily many
1158 * bytes and thus cause resource exhaustion.
1159 */
1160 if (width < -FORMATTING_LIMIT || width > FORMATTING_LIMIT)
1161 return 0;
1162
1163 if (next == start || width == 0)
1164 return 0;
1165 if (width < 0) {
1166 if (to_column)
1167 width += term_columns();
1168 if (width < 0)
1169 return 0;
1170 }
1171 c->padding = to_column ? -width : width;
1172 c->flush_type = flush_type;
1173
1174 if (*end == ',') {
1175 start = end + 1;
1176 end = strchr(start, ')');
1177 if (!end || end == start)
1178 return 0;
1179 if (starts_with(start, "trunc)"))
1180 c->truncate = trunc_right;
1181 else if (starts_with(start, "ltrunc)"))
1182 c->truncate = trunc_left;
1183 else if (starts_with(start, "mtrunc)"))
1184 c->truncate = trunc_middle;
1185 else
1186 return 0;
1187 } else
1188 c->truncate = trunc_none;
1189
1190 return end - placeholder + 1;
1191 }
1192 return 0;
1193 }
1194
1195 static int match_placeholder_arg_value(const char *to_parse, const char *candidate,
1196 const char **end, const char **valuestart,
1197 size_t *valuelen)
1198 {
1199 const char *p;
1200
1201 if (!(skip_prefix(to_parse, candidate, &p)))
1202 return 0;
1203 if (valuestart) {
1204 if (*p == '=') {
1205 *valuestart = p + 1;
1206 *valuelen = strcspn(*valuestart, ",)");
1207 p = *valuestart + *valuelen;
1208 } else {
1209 if (*p != ',' && *p != ')')
1210 return 0;
1211 *valuestart = NULL;
1212 *valuelen = 0;
1213 }
1214 }
1215 if (*p == ',') {
1216 *end = p + 1;
1217 return 1;
1218 }
1219 if (*p == ')') {
1220 *end = p;
1221 return 1;
1222 }
1223 return 0;
1224 }
1225
1226 static int match_placeholder_bool_arg(const char *to_parse, const char *candidate,
1227 const char **end, int *val)
1228 {
1229 const char *argval;
1230 char *strval;
1231 size_t arglen;
1232 int v;
1233
1234 if (!match_placeholder_arg_value(to_parse, candidate, end, &argval, &arglen))
1235 return 0;
1236
1237 if (!argval) {
1238 *val = 1;
1239 return 1;
1240 }
1241
1242 strval = xstrndup(argval, arglen);
1243 v = git_parse_maybe_bool(strval);
1244 free(strval);
1245
1246 if (v == -1)
1247 return 0;
1248
1249 *val = v;
1250
1251 return 1;
1252 }
1253
1254 static int format_trailer_match_cb(const struct strbuf *key, void *ud)
1255 {
1256 const struct string_list *list = ud;
1257 const struct string_list_item *item;
1258
1259 for_each_string_list_item (item, list) {
1260 if (key->len == (uintptr_t)item->util &&
1261 !strncasecmp(item->string, key->buf, key->len))
1262 return 1;
1263 }
1264 return 0;
1265 }
1266
1267 static struct strbuf *expand_string_arg(struct strbuf *sb,
1268 const char *argval, size_t arglen)
1269 {
1270 char *fmt = xstrndup(argval, arglen);
1271 const char *format = fmt;
1272
1273 strbuf_reset(sb);
1274 while (strbuf_expand_step(sb, &format)) {
1275 size_t len;
1276
1277 if (skip_prefix(format, "%", &format))
1278 strbuf_addch(sb, '%');
1279 else if ((len = strbuf_expand_literal(sb, format)))
1280 format += len;
1281 else
1282 strbuf_addch(sb, '%');
1283 }
1284 free(fmt);
1285 return sb;
1286 }
1287
1288 int format_set_trailers_options(struct process_trailer_options *opts,
1289 struct string_list *filter_list,
1290 struct strbuf *sepbuf,
1291 struct strbuf *kvsepbuf,
1292 const char **arg,
1293 char **invalid_arg)
1294 {
1295 for (;;) {
1296 const char *argval;
1297 size_t arglen;
1298
1299 if (**arg == ')')
1300 break;
1301
1302 if (match_placeholder_arg_value(*arg, "key", arg, &argval, &arglen)) {
1303 uintptr_t len = arglen;
1304
1305 if (!argval)
1306 return -1;
1307
1308 if (len && argval[len - 1] == ':')
1309 len--;
1310 string_list_append(filter_list, argval)->util = (char *)len;
1311
1312 opts->filter = format_trailer_match_cb;
1313 opts->filter_data = filter_list;
1314 opts->only_trailers = 1;
1315 } else if (match_placeholder_arg_value(*arg, "separator", arg, &argval, &arglen)) {
1316 opts->separator = expand_string_arg(sepbuf, argval, arglen);
1317 } else if (match_placeholder_arg_value(*arg, "key_value_separator", arg, &argval, &arglen)) {
1318 opts->key_value_separator = expand_string_arg(kvsepbuf, argval, arglen);
1319 } else if (!match_placeholder_bool_arg(*arg, "only", arg, &opts->only_trailers) &&
1320 !match_placeholder_bool_arg(*arg, "unfold", arg, &opts->unfold) &&
1321 !match_placeholder_bool_arg(*arg, "keyonly", arg, &opts->key_only) &&
1322 !match_placeholder_bool_arg(*arg, "valueonly", arg, &opts->value_only)) {
1323 if (invalid_arg) {
1324 size_t len = strcspn(*arg, ",)");
1325 *invalid_arg = xstrndup(*arg, len);
1326 }
1327 return -1;
1328 }
1329 }
1330 return 0;
1331 }
1332
1333 static size_t parse_describe_args(const char *start, struct strvec *args)
1334 {
1335 struct {
1336 const char *name;
1337 enum {
1338 DESCRIBE_ARG_BOOL,
1339 DESCRIBE_ARG_INTEGER,
1340 DESCRIBE_ARG_STRING,
1341 } type;
1342 } option[] = {
1343 { "tags", DESCRIBE_ARG_BOOL},
1344 { "abbrev", DESCRIBE_ARG_INTEGER },
1345 { "exclude", DESCRIBE_ARG_STRING },
1346 { "match", DESCRIBE_ARG_STRING },
1347 };
1348 const char *arg = start;
1349
1350 for (;;) {
1351 int found = 0;
1352 const char *argval;
1353 size_t arglen = 0;
1354 int optval = 0;
1355 int i;
1356
1357 for (i = 0; !found && i < ARRAY_SIZE(option); i++) {
1358 switch (option[i].type) {
1359 case DESCRIBE_ARG_BOOL:
1360 if (match_placeholder_bool_arg(arg, option[i].name, &arg, &optval)) {
1361 if (optval)
1362 strvec_pushf(args, "--%s", option[i].name);
1363 else
1364 strvec_pushf(args, "--no-%s", option[i].name);
1365 found = 1;
1366 }
1367 break;
1368 case DESCRIBE_ARG_INTEGER:
1369 if (match_placeholder_arg_value(arg, option[i].name, &arg,
1370 &argval, &arglen)) {
1371 char *endptr;
1372 if (!arglen)
1373 return 0;
1374 strtol(argval, &endptr, 10);
1375 if (endptr - argval != arglen)
1376 return 0;
1377 strvec_pushf(args, "--%s=%.*s", option[i].name, (int)arglen, argval);
1378 found = 1;
1379 }
1380 break;
1381 case DESCRIBE_ARG_STRING:
1382 if (match_placeholder_arg_value(arg, option[i].name, &arg,
1383 &argval, &arglen)) {
1384 if (!arglen)
1385 return 0;
1386 strvec_pushf(args, "--%s=%.*s", option[i].name, (int)arglen, argval);
1387 found = 1;
1388 }
1389 break;
1390 }
1391 }
1392 if (!found)
1393 break;
1394
1395 }
1396 return arg - start;
1397 }
1398
1399
1400 static int parse_decoration_option(const char **arg,
1401 const char *name,
1402 char **opt)
1403 {
1404 const char *argval;
1405 size_t arglen;
1406
1407 if (match_placeholder_arg_value(*arg, name, arg, &argval, &arglen)) {
1408 struct strbuf sb = STRBUF_INIT;
1409
1410 expand_string_arg(&sb, argval, arglen);
1411 *opt = strbuf_detach(&sb, NULL);
1412 return 1;
1413 }
1414 return 0;
1415 }
1416
1417 static void parse_decoration_options(const char **arg,
1418 struct decoration_options *opts)
1419 {
1420 while (parse_decoration_option(arg, "prefix", &opts->prefix) ||
1421 parse_decoration_option(arg, "suffix", &opts->suffix) ||
1422 parse_decoration_option(arg, "separator", &opts->separator) ||
1423 parse_decoration_option(arg, "pointer", &opts->pointer) ||
1424 parse_decoration_option(arg, "tag", &opts->tag))
1425 ;
1426 }
1427
1428 static void free_decoration_options(const struct decoration_options *opts)
1429 {
1430 free(opts->prefix);
1431 free(opts->suffix);
1432 free(opts->separator);
1433 free(opts->pointer);
1434 free(opts->tag);
1435 }
1436
1437 static size_t format_commit_one(struct strbuf *sb, /* in UTF-8 */
1438 const char *placeholder,
1439 void *context)
1440 {
1441 struct format_commit_context *c = context;
1442 const struct commit *commit = c->commit;
1443 const char *msg = c->message;
1444 struct commit_list *p;
1445 const char *arg, *eol;
1446 size_t res;
1447 char **slot;
1448
1449 /* these are independent of the commit */
1450 res = strbuf_expand_literal(sb, placeholder);
1451 if (res)
1452 return res;
1453
1454 switch (placeholder[0]) {
1455 case 'C':
1456 if (starts_with(placeholder + 1, "(auto)")) {
1457 c->auto_color = c->pretty_ctx->color;
1458 if (want_color(c->auto_color) && sb->len)
1459 strbuf_addstr(sb, GIT_COLOR_RESET);
1460 return 7; /* consumed 7 bytes, "C(auto)" */
1461 } else {
1462 int ret = parse_color(sb, placeholder, c);
1463 if (ret)
1464 c->auto_color = GIT_COLOR_NEVER;
1465 /*
1466 * Otherwise, we decided to treat %C<unknown>
1467 * as a literal string, and the previous
1468 * %C(auto) is still valid.
1469 */
1470 return ret;
1471 }
1472 case 'w':
1473 if (placeholder[1] == '(') {
1474 unsigned long width = 0, indent1 = 0, indent2 = 0;
1475 char *next;
1476 const char *start = placeholder + 2;
1477 const char *end = strchr(start, ')');
1478 if (!end)
1479 return 0;
1480 if (end > start) {
1481 width = strtoul(start, &next, 10);
1482 if (*next == ',') {
1483 indent1 = strtoul(next + 1, &next, 10);
1484 if (*next == ',') {
1485 indent2 = strtoul(next + 1,
1486 &next, 10);
1487 }
1488 }
1489 if (*next != ')')
1490 return 0;
1491 }
1492
1493 /*
1494 * We need to limit the format here as it allows the
1495 * user to prepend arbitrarily many bytes to the buffer
1496 * when rewrapping.
1497 */
1498 if (width > FORMATTING_LIMIT ||
1499 indent1 > FORMATTING_LIMIT ||
1500 indent2 > FORMATTING_LIMIT)
1501 return 0;
1502 rewrap_message_tail(sb, c, width, indent1, indent2);
1503 return end - placeholder + 1;
1504 } else
1505 return 0;
1506
1507 case '<':
1508 case '>':
1509 return parse_padding_placeholder(placeholder, c);
1510 }
1511
1512 if (skip_prefix(placeholder, "(describe", &arg)) {
1513 struct child_process cmd = CHILD_PROCESS_INIT;
1514 struct strbuf out = STRBUF_INIT;
1515 struct strbuf err = STRBUF_INIT;
1516 struct pretty_print_describe_status *describe_status;
1517
1518 describe_status = c->pretty_ctx->describe_status;
1519 if (describe_status) {
1520 if (!describe_status->max_invocations)
1521 return 0;
1522 describe_status->max_invocations--;
1523 }
1524
1525 cmd.git_cmd = 1;
1526 strvec_push(&cmd.args, "describe");
1527
1528 if (*arg == ':') {
1529 arg++;
1530 arg += parse_describe_args(arg, &cmd.args);
1531 }
1532
1533 if (*arg != ')') {
1534 child_process_clear(&cmd);
1535 return 0;
1536 }
1537
1538 strvec_push(&cmd.args, oid_to_hex(&commit->object.oid));
1539 pipe_command(&cmd, NULL, 0, &out, 0, &err, 0);
1540 strbuf_rtrim(&out);
1541 strbuf_addbuf(sb, &out);
1542 strbuf_release(&out);
1543 strbuf_release(&err);
1544 return arg - placeholder + 1;
1545 }
1546
1547 /* these depend on the commit */
1548 if (!commit->object.parsed)
1549 parse_object(the_repository, &commit->object.oid);
1550
1551 if (starts_with(placeholder, "(count)")) {
1552 if (!c->pretty_ctx->rev)
1553 die(_("%s is not supported by this command"), "%(count)");
1554 strbuf_addf(sb, "%0*d", decimal_width(c->pretty_ctx->rev->total),
1555 c->pretty_ctx->rev->nr);
1556 return 7;
1557 }
1558
1559 if (starts_with(placeholder, "(total)")) {
1560 if (!c->pretty_ctx->rev)
1561 die(_("%s is not supported by this command"), "%(total)");
1562 strbuf_addf(sb, "%d", c->pretty_ctx->rev->total);
1563 return 7;
1564 }
1565
1566 switch (placeholder[0]) {
1567 case 'H': /* commit hash */
1568 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_COMMIT));
1569 strbuf_add_oid_hex(sb, &commit->object.oid);
1570 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_RESET));
1571 return 1;
1572 case 'h': /* abbreviated commit hash */
1573 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_COMMIT));
1574 strbuf_add_unique_abbrev(sb, &commit->object.oid,
1575 c->pretty_ctx->abbrev);
1576 strbuf_addstr(sb, diff_get_color(c->auto_color, DIFF_RESET));
1577 return 1;
1578 case 'T': /* tree hash */
1579 strbuf_add_oid_hex(sb, get_commit_tree_oid(commit));
1580 return 1;
1581 case 't': /* abbreviated tree hash */
1582 strbuf_add_unique_abbrev(sb,
1583 get_commit_tree_oid(commit),
1584 c->pretty_ctx->abbrev);
1585 return 1;
1586 case 'P': /* parent hashes */
1587 for (p = commit->parents; p; p = p->next) {
1588 if (p != commit->parents)
1589 strbuf_addch(sb, ' ');
1590 strbuf_add_oid_hex(sb, &p->item->object.oid);
1591 }
1592 return 1;
1593 case 'p': /* abbreviated parent hashes */
1594 for (p = commit->parents; p; p = p->next) {
1595 if (p != commit->parents)
1596 strbuf_addch(sb, ' ');
1597 strbuf_add_unique_abbrev(sb, &p->item->object.oid,
1598 c->pretty_ctx->abbrev);
1599 }
1600 return 1;
1601 case 'm': /* left/right/bottom */
1602 strbuf_addstr(sb, get_revision_mark(NULL, commit));
1603 return 1;
1604 case 'd':
1605 format_decorations(sb, commit, c->auto_color, NULL);
1606 return 1;
1607 case 'D':
1608 {
1609 const struct decoration_options opts = {
1610 .prefix = (char *) "",
1611 .suffix = (char *) "",
1612 };
1613
1614 format_decorations(sb, commit, c->auto_color, &opts);
1615 return 1;
1616 }
1617 case 'S': /* tag/branch like --source */
1618 if (!(c->pretty_ctx->rev && c->pretty_ctx->rev->sources))
1619 return 0;
1620 slot = revision_sources_at(c->pretty_ctx->rev->sources, commit);
1621 if (!(slot && *slot))
1622 return 0;
1623 strbuf_addstr(sb, *slot);
1624 return 1;
1625 case 'g': /* reflog info */
1626 switch(placeholder[1]) {
1627 case 'd': /* reflog selector */
1628 case 'D':
1629 if (c->pretty_ctx->reflog_info)
1630 get_reflog_selector(sb,
1631 c->pretty_ctx->reflog_info,
1632 c->pretty_ctx->date_mode,
1633 c->pretty_ctx->date_mode_explicit,
1634 (placeholder[1] == 'd'));
1635 return 2;
1636 case 's': /* reflog message */
1637 if (c->pretty_ctx->reflog_info)
1638 get_reflog_message(sb, c->pretty_ctx->reflog_info);
1639 return 2;
1640 case 'n':
1641 case 'N':
1642 case 'e':
1643 case 'E':
1644 return format_reflog_person(sb,
1645 placeholder[1],
1646 c->pretty_ctx->reflog_info,
1647 c->pretty_ctx->date_mode);
1648 }
1649 return 0; /* unknown %g placeholder */
1650 case 'N':
1651 if (c->pretty_ctx->notes_message) {
1652 strbuf_addstr(sb, c->pretty_ctx->notes_message);
1653 return 1;
1654 }
1655 return 0;
1656 }
1657
1658 if (placeholder[0] == 'G') {
1659 if (!c->signature_check.result)
1660 check_commit_signature(c->commit, &(c->signature_check));
1661 switch (placeholder[1]) {
1662 case 'G':
1663 if (c->signature_check.output)
1664 strbuf_addstr(sb, c->signature_check.output);
1665 break;
1666 case '?':
1667 switch (c->signature_check.result) {
1668 case 'G':
1669 switch (c->signature_check.trust_level) {
1670 case TRUST_UNDEFINED:
1671 case TRUST_NEVER:
1672 strbuf_addch(sb, 'U');
1673 break;
1674 default:
1675 strbuf_addch(sb, 'G');
1676 break;
1677 }
1678 break;
1679 case 'B':
1680 case 'E':
1681 case 'N':
1682 case 'X':
1683 case 'Y':
1684 case 'R':
1685 strbuf_addch(sb, c->signature_check.result);
1686 }
1687 break;
1688 case 'S':
1689 if (c->signature_check.signer)
1690 strbuf_addstr(sb, c->signature_check.signer);
1691 break;
1692 case 'K':
1693 if (c->signature_check.key)
1694 strbuf_addstr(sb, c->signature_check.key);
1695 break;
1696 case 'F':
1697 if (c->signature_check.fingerprint)
1698 strbuf_addstr(sb, c->signature_check.fingerprint);
1699 break;
1700 case 'P':
1701 if (c->signature_check.primary_key_fingerprint)
1702 strbuf_addstr(sb, c->signature_check.primary_key_fingerprint);
1703 break;
1704 case 'T':
1705 strbuf_addstr(sb, gpg_trust_level_to_str(c->signature_check.trust_level));
1706 break;
1707 default:
1708 return 0;
1709 }
1710 return 2;
1711 }
1712
1713 if (skip_prefix(placeholder, "(decorate", &arg)) {
1714 struct decoration_options opts = { NULL };
1715 size_t ret = 0;
1716
1717 if (*arg == ':') {
1718 arg++;
1719 parse_decoration_options(&arg, &opts);
1720 }
1721 if (*arg == ')') {
1722 format_decorations(sb, commit, c->auto_color, &opts);
1723 ret = arg - placeholder + 1;
1724 }
1725
1726 free_decoration_options(&opts);
1727 return ret;
1728 }
1729
1730 /* For the rest we have to parse the commit header. */
1731 if (!c->commit_header_parsed) {
1732 msg = c->message =
1733 repo_logmsg_reencode(c->repository, commit,
1734 &c->commit_encoding, "UTF-8");
1735 parse_commit_header(c);
1736 }
1737
1738 switch (placeholder[0]) {
1739 case 'a': /* author ... */
1740 return format_person_part(sb, placeholder[1],
1741 msg + c->author.off, c->author.len,
1742 c->pretty_ctx->date_mode);
1743 case 'c': /* committer ... */
1744 return format_person_part(sb, placeholder[1],
1745 msg + c->committer.off, c->committer.len,
1746 c->pretty_ctx->date_mode);
1747 case 'e': /* encoding */
1748 if (c->commit_encoding)
1749 strbuf_addstr(sb, c->commit_encoding);
1750 return 1;
1751 case 'B': /* raw body */
1752 /* message_off is always left at the initial newline */
1753 strbuf_addstr(sb, msg + c->message_off + 1);
1754 return 1;
1755 }
1756
1757 /* Now we need to parse the commit message. */
1758 if (!c->commit_message_parsed)
1759 parse_commit_message(c);
1760
1761 switch (placeholder[0]) {
1762 case 's': /* subject */
1763 format_subject(sb, msg + c->subject_off, " ");
1764 return 1;
1765 case 'f': /* sanitized subject */
1766 eol = strchrnul(msg + c->subject_off, '\n');
1767 format_sanitized_subject(sb, msg + c->subject_off, eol - (msg + c->subject_off));
1768 return 1;
1769 case 'b': /* body */
1770 strbuf_addstr(sb, msg + c->body_off);
1771 return 1;
1772 }
1773
1774 if (skip_prefix(placeholder, "(trailers", &arg)) {
1775 struct process_trailer_options opts = PROCESS_TRAILER_OPTIONS_INIT;
1776 struct string_list filter_list = STRING_LIST_INIT_NODUP;
1777 struct strbuf sepbuf = STRBUF_INIT;
1778 struct strbuf kvsepbuf = STRBUF_INIT;
1779 size_t ret = 0;
1780
1781 opts.no_divider = 1;
1782
1783 if (*arg == ':') {
1784 arg++;
1785 if (format_set_trailers_options(&opts, &filter_list, &sepbuf, &kvsepbuf, &arg, NULL))
1786 goto trailer_out;
1787 }
1788 if (*arg == ')') {
1789 format_trailers_from_commit(&opts, msg + c->subject_off, sb);
1790 ret = arg - placeholder + 1;
1791 }
1792 trailer_out:
1793 string_list_clear(&filter_list, 0);
1794 strbuf_release(&kvsepbuf);
1795 strbuf_release(&sepbuf);
1796 return ret;
1797 }
1798
1799 return 0; /* unknown placeholder */
1800 }
1801
1802 static size_t format_and_pad_commit(struct strbuf *sb, /* in UTF-8 */
1803 const char *placeholder,
1804 struct format_commit_context *c)
1805 {
1806 struct strbuf local_sb = STRBUF_INIT;
1807 size_t total_consumed = 0;
1808 int len, padding = c->padding;
1809
1810 if (padding < 0) {
1811 const char *start = strrchr(sb->buf, '\n');
1812 int occupied;
1813 if (!start)
1814 start = sb->buf;
1815 occupied = utf8_strnwidth(start, strlen(start), 1);
1816 occupied += c->pretty_ctx->graph_width;
1817 padding = (-padding) - occupied;
1818 }
1819 while (1) {
1820 int modifier = *placeholder == 'C';
1821 size_t consumed = format_commit_one(&local_sb, placeholder, c);
1822 total_consumed += consumed;
1823
1824 if (!modifier)
1825 break;
1826
1827 placeholder += consumed;
1828 if (*placeholder != '%')
1829 break;
1830 placeholder++;
1831 total_consumed++;
1832 }
1833 len = utf8_strnwidth(local_sb.buf, local_sb.len, 1);
1834
1835 if (c->flush_type == flush_left_and_steal) {
1836 const char *ch = sb->buf + sb->len - 1;
1837 while (len > padding && ch > sb->buf) {
1838 const char *p;
1839 if (*ch == ' ') {
1840 ch--;
1841 padding++;
1842 continue;
1843 }
1844 /* check for trailing ansi sequences */
1845 if (*ch != 'm')
1846 break;
1847 p = ch - 1;
1848 while (p > sb->buf && ch - p < 10 && *p != '\033')
1849 p--;
1850 if (*p != '\033' ||
1851 ch + 1 - p != display_mode_esc_sequence_len(p))
1852 break;
1853 /*
1854 * got a good ansi sequence, put it back to
1855 * local_sb as we're cutting sb
1856 */
1857 strbuf_insert(&local_sb, 0, p, ch + 1 - p);
1858 ch = p - 1;
1859 }
1860 strbuf_setlen(sb, ch + 1 - sb->buf);
1861 c->flush_type = flush_left;
1862 }
1863
1864 if (len > padding) {
1865 switch (c->truncate) {
1866 case trunc_left:
1867 strbuf_utf8_replace(&local_sb,
1868 0, len - (padding - 2),
1869 "..");
1870 break;
1871 case trunc_middle:
1872 strbuf_utf8_replace(&local_sb,
1873 padding / 2 - 1,
1874 len - (padding - 2),
1875 "..");
1876 break;
1877 case trunc_right:
1878 strbuf_utf8_replace(&local_sb,
1879 padding - 2, len - (padding - 2),
1880 "..");
1881 break;
1882 case trunc_none:
1883 break;
1884 }
1885 strbuf_addbuf(sb, &local_sb);
1886 } else {
1887 size_t sb_len = sb->len, offset = 0;
1888 if (c->flush_type == flush_left)
1889 offset = padding - len;
1890 else if (c->flush_type == flush_both)
1891 offset = (padding - len) / 2;
1892 /*
1893 * we calculate padding in columns, now
1894 * convert it back to chars
1895 */
1896 padding = padding - len + local_sb.len;
1897 strbuf_addchars(sb, ' ', padding);
1898 memcpy(sb->buf + sb_len + offset, local_sb.buf,
1899 local_sb.len);
1900 }
1901 strbuf_release(&local_sb);
1902 c->flush_type = no_flush;
1903 return total_consumed;
1904 }
1905
1906 static size_t format_commit_item(struct strbuf *sb, /* in UTF-8 */
1907 const char *placeholder,
1908 struct format_commit_context *context)
1909 {
1910 size_t consumed, orig_len;
1911 enum {
1912 NO_MAGIC,
1913 ADD_LF_BEFORE_NON_EMPTY,
1914 DEL_LF_BEFORE_EMPTY,
1915 ADD_SP_BEFORE_NON_EMPTY
1916 } magic = NO_MAGIC;
1917
1918 switch (placeholder[0]) {
1919 case '-':
1920 magic = DEL_LF_BEFORE_EMPTY;
1921 break;
1922 case '+':
1923 magic = ADD_LF_BEFORE_NON_EMPTY;
1924 break;
1925 case ' ':
1926 magic = ADD_SP_BEFORE_NON_EMPTY;
1927 break;
1928 default:
1929 break;
1930 }
1931 if (magic != NO_MAGIC) {
1932 placeholder++;
1933
1934 switch (placeholder[0]) {
1935 case 'w':
1936 /*
1937 * `%+w()` cannot ever expand to a non-empty string,
1938 * and it potentially changes the layout of preceding
1939 * contents. We're thus not able to handle the magic in
1940 * this combination and refuse the pattern.
1941 */
1942 return 0;
1943 };
1944 }
1945
1946 orig_len = sb->len;
1947 if (context->flush_type == no_flush)
1948 consumed = format_commit_one(sb, placeholder, context);
1949 else
1950 consumed = format_and_pad_commit(sb, placeholder, context);
1951 if (magic == NO_MAGIC)
1952 return consumed;
1953
1954 if ((orig_len == sb->len) && magic == DEL_LF_BEFORE_EMPTY) {
1955 while (sb->len && sb->buf[sb->len - 1] == '\n')
1956 strbuf_setlen(sb, sb->len - 1);
1957 } else if (orig_len != sb->len) {
1958 if (magic == ADD_LF_BEFORE_NON_EMPTY)
1959 strbuf_insertstr(sb, orig_len, "\n");
1960 else if (magic == ADD_SP_BEFORE_NON_EMPTY)
1961 strbuf_insertstr(sb, orig_len, " ");
1962 }
1963 return consumed + 1;
1964 }
1965
1966 void userformat_find_requirements(const char *fmt, struct userformat_want *w)
1967 {
1968 if (!fmt) {
1969 if (!user_format)
1970 return;
1971 fmt = user_format;
1972 }
1973 while ((fmt = strchr(fmt, '%'))) {
1974 fmt++;
1975 if (skip_prefix(fmt, "%", &fmt))
1976 continue;
1977
1978 if (*fmt == '+' || *fmt == '-' || *fmt == ' ')
1979 fmt++;
1980
1981 switch (*fmt) {
1982 case 'N':
1983 w->notes = 1;
1984 break;
1985 case 'S':
1986 w->source = 1;
1987 break;
1988 case 'd':
1989 case 'D':
1990 w->decorate = 1;
1991 break;
1992 case '(':
1993 if (starts_with(fmt + 1, "decorate"))
1994 w->decorate = 1;
1995 break;
1996 }
1997 }
1998 }
1999
2000 void repo_format_commit_message(struct repository *r,
2001 const struct commit *commit,
2002 const char *format, struct strbuf *sb,
2003 const struct pretty_print_context *pretty_ctx)
2004 {
2005 struct format_commit_context context = {
2006 .repository = r,
2007 .commit = commit,
2008 .pretty_ctx = pretty_ctx,
2009 .wrap_start = sb->len
2010 };
2011 const char *output_enc = pretty_ctx->output_encoding;
2012 const char *utf8 = "UTF-8";
2013
2014 while (strbuf_expand_step(sb, &format)) {
2015 size_t len;
2016
2017 if (skip_prefix(format, "%", &format))
2018 strbuf_addch(sb, '%');
2019 else if ((len = format_commit_item(sb, format, &context)))
2020 format += len;
2021 else
2022 strbuf_addch(sb, '%');
2023 }
2024 rewrap_message_tail(sb, &context, 0, 0, 0);
2025
2026 /*
2027 * Convert output to an actual output encoding; note that
2028 * format_commit_item() will always use UTF-8, so we don't
2029 * have to bother if that's what the output wants.
2030 */
2031 if (output_enc) {
2032 if (same_encoding(utf8, output_enc))
2033 output_enc = NULL;
2034 } else {
2035 if (context.commit_encoding &&
2036 !same_encoding(context.commit_encoding, utf8))
2037 output_enc = context.commit_encoding;
2038 }
2039
2040 if (output_enc) {
2041 size_t outsz;
2042 char *out = reencode_string_len(sb->buf, sb->len,
2043 output_enc, utf8, &outsz);
2044 if (out)
2045 strbuf_attach(sb, out, outsz, outsz + 1);
2046 }
2047
2048 free(context.commit_encoding);
2049 repo_unuse_commit_buffer(r, commit, context.message);
2050 signature_check_clear(&context.signature_check);
2051 }
2052
2053 static void pp_header(struct pretty_print_context *pp,
2054 const char *encoding,
2055 const struct commit *commit,
2056 const char **msg_p,
2057 struct strbuf *sb)
2058 {
2059 int parents_shown = 0;
2060
2061 for (;;) {
2062 const char *name, *line = *msg_p;
2063 int linelen = get_one_line(*msg_p);
2064
2065 if (!linelen)
2066 return;
2067 *msg_p += linelen;
2068
2069 if (linelen == 1)
2070 /* End of header */
2071 return;
2072
2073 if (pp->fmt == CMIT_FMT_RAW) {
2074 strbuf_add(sb, line, linelen);
2075 continue;
2076 }
2077
2078 if (starts_with(line, "parent ")) {
2079 if (linelen != the_hash_algo->hexsz + 8)
2080 die("bad parent line in commit");
2081 continue;
2082 }
2083
2084 if (!parents_shown) {
2085 unsigned num = commit_list_count(commit->parents);
2086 /* with enough slop */
2087 strbuf_grow(sb, num * (GIT_MAX_HEXSZ + 10) + 20);
2088 add_merge_info(pp, sb, commit);
2089 parents_shown = 1;
2090 }
2091
2092 /*
2093 * MEDIUM == DEFAULT shows only author with dates.
2094 * FULL shows both authors but not dates.
2095 * FULLER shows both authors and dates.
2096 */
2097 if (skip_prefix(line, "author ", &name)) {
2098 strbuf_grow(sb, linelen + 80);
2099 pp_user_info(pp, "Author", sb, name, encoding);
2100 }
2101 if (skip_prefix(line, "committer ", &name) &&
2102 (pp->fmt == CMIT_FMT_FULL || pp->fmt == CMIT_FMT_FULLER)) {
2103 strbuf_grow(sb, linelen + 80);
2104 pp_user_info(pp, "Commit", sb, name, encoding);
2105 }
2106 }
2107 }
2108
2109 void pp_email_subject(struct pretty_print_context *pp,
2110 const char **msg_p,
2111 struct strbuf *sb,
2112 const char *encoding,
2113 int need_8bit_cte)
2114 {
2115 static const int max_length = 78; /* per rfc2047 */
2116 struct strbuf title;
2117
2118 strbuf_init(&title, 80);
2119 *msg_p = format_subject(&title, *msg_p,
2120 pp->preserve_subject ? "\n" : " ");
2121
2122 strbuf_grow(sb, title.len + 1024);
2123 fmt_output_email_subject(sb, pp->rev);
2124 if (pp->encode_email_headers &&
2125 needs_rfc2047_encoding(title.buf, title.len))
2126 add_rfc2047(sb, title.buf, title.len,
2127 encoding, RFC2047_SUBJECT);
2128 else
2129 strbuf_add_wrapped_bytes(sb, title.buf, title.len,
2130 -last_line_length(sb), 1, max_length);
2131 strbuf_addch(sb, '\n');
2132
2133 if (need_8bit_cte == 0) {
2134 int i;
2135 for (i = 0; i < pp->in_body_headers.nr; i++) {
2136 if (has_non_ascii(pp->in_body_headers.items[i].string)) {
2137 need_8bit_cte = 1;
2138 break;
2139 }
2140 }
2141 }
2142
2143 if (need_8bit_cte > 0) {
2144 const char *header_fmt =
2145 "MIME-Version: 1.0\n"
2146 "Content-Type: text/plain; charset=%s\n"
2147 "Content-Transfer-Encoding: 8bit\n";
2148 strbuf_addf(sb, header_fmt, encoding);
2149 }
2150 if (pp->after_subject) {
2151 strbuf_addstr(sb, pp->after_subject);
2152 }
2153
2154 strbuf_addch(sb, '\n');
2155
2156 if (pp->in_body_headers.nr) {
2157 int i;
2158 for (i = 0; i < pp->in_body_headers.nr; i++) {
2159 strbuf_addstr(sb, pp->in_body_headers.items[i].string);
2160 free(pp->in_body_headers.items[i].string);
2161 }
2162 string_list_clear(&pp->in_body_headers, 0);
2163 strbuf_addch(sb, '\n');
2164 }
2165
2166 strbuf_release(&title);
2167 }
2168
2169 static int pp_utf8_width(const char *start, const char *end)
2170 {
2171 int width = 0;
2172 size_t remain = end - start;
2173
2174 while (remain) {
2175 int n = utf8_width(&start, &remain);
2176 if (n < 0 || !start)
2177 return -1;
2178 width += n;
2179 }
2180 return width;
2181 }
2182
2183 static void strbuf_add_tabexpand(struct strbuf *sb, struct grep_opt *opt,
2184 enum git_colorbool color, int tabwidth, const char *line,
2185 int linelen)
2186 {
2187 const char *tab;
2188
2189 while ((tab = memchr(line, '\t', linelen)) != NULL) {
2190 int width = pp_utf8_width(line, tab);
2191
2192 /*
2193 * If it wasn't well-formed utf8, or it
2194 * had characters with badly defined
2195 * width (control characters etc), just
2196 * give up on trying to align things.
2197 */
2198 if (width < 0)
2199 break;
2200
2201 /* Output the data .. */
2202 append_line_with_color(sb, opt, line, tab - line, color,
2203 GREP_CONTEXT_BODY,
2204 GREP_HEADER_FIELD_MAX);
2205
2206 /* .. and the de-tabified tab */
2207 strbuf_addchars(sb, ' ', tabwidth - (width % tabwidth));
2208
2209 /* Skip over the printed part .. */
2210 linelen -= tab + 1 - line;
2211 line = tab + 1;
2212 }
2213
2214 /*
2215 * Print out everything after the last tab without
2216 * worrying about width - there's nothing more to
2217 * align.
2218 */
2219 append_line_with_color(sb, opt, line, linelen, color, GREP_CONTEXT_BODY,
2220 GREP_HEADER_FIELD_MAX);
2221 }
2222
2223 /*
2224 * pp_handle_indent() prints out the indentation, and
2225 * the whole line (without the final newline), after
2226 * de-tabifying.
2227 */
2228 static void pp_handle_indent(struct pretty_print_context *pp,
2229 struct strbuf *sb, int indent,
2230 const char *line, int linelen)
2231 {
2232 struct grep_opt *opt = pp->rev ? &pp->rev->grep_filter : NULL;
2233
2234 strbuf_addchars(sb, ' ', indent);
2235 if (pp->expand_tabs_in_log)
2236 strbuf_add_tabexpand(sb, opt, pp->color, pp->expand_tabs_in_log,
2237 line, linelen);
2238 else
2239 append_line_with_color(sb, opt, line, linelen, pp->color,
2240 GREP_CONTEXT_BODY,
2241 GREP_HEADER_FIELD_MAX);
2242 }
2243
2244 static int is_mboxrd_from(const char *line, int len)
2245 {
2246 /*
2247 * a line matching /^From $/ here would only have len == 4
2248 * at this point because is_empty_line would've trimmed all
2249 * trailing space
2250 */
2251 return len > 4 && starts_with(line + strspn(line, ">"), "From ");
2252 }
2253
2254 void pp_remainder(struct pretty_print_context *pp,
2255 const char **msg_p,
2256 struct strbuf *sb,
2257 int indent)
2258 {
2259 struct grep_opt *opt = pp->rev ? &pp->rev->grep_filter : NULL;
2260 int first = 1;
2261
2262 for (;;) {
2263 const char *line = *msg_p;
2264 int linelen = get_one_line(line);
2265 *msg_p += linelen;
2266
2267 if (!linelen)
2268 break;
2269
2270 if (is_blank_line(line, &linelen)) {
2271 if (first)
2272 continue;
2273 if (pp->fmt == CMIT_FMT_SHORT)
2274 break;
2275 }
2276 first = 0;
2277
2278 strbuf_grow(sb, linelen + indent + 20);
2279 if (indent)
2280 pp_handle_indent(pp, sb, indent, line, linelen);
2281 else if (pp->expand_tabs_in_log)
2282 strbuf_add_tabexpand(sb, opt, pp->color,
2283 pp->expand_tabs_in_log, line,
2284 linelen);
2285 else {
2286 if (pp->fmt == CMIT_FMT_MBOXRD &&
2287 is_mboxrd_from(line, linelen))
2288 strbuf_addch(sb, '>');
2289
2290 append_line_with_color(sb, opt, line, linelen,
2291 pp->color, GREP_CONTEXT_BODY,
2292 GREP_HEADER_FIELD_MAX);
2293 }
2294 strbuf_addch(sb, '\n');
2295 }
2296 }
2297
2298 void pretty_print_commit(struct pretty_print_context *pp,
2299 const struct commit *commit,
2300 struct strbuf *sb)
2301 {
2302 unsigned long beginning_of_body;
2303 int indent = 4;
2304 const char *msg;
2305 const char *reencoded;
2306 const char *encoding;
2307 int need_8bit_cte = pp->need_8bit_cte;
2308
2309 if (pp->fmt == CMIT_FMT_USERFORMAT) {
2310 repo_format_commit_message(the_repository, commit,
2311 user_format, sb, pp);
2312 return;
2313 }
2314
2315 encoding = get_log_output_encoding();
2316 msg = reencoded = repo_logmsg_reencode(the_repository, commit, NULL,
2317 encoding);
2318
2319 if (pp->fmt == CMIT_FMT_ONELINE || cmit_fmt_is_mail(pp->fmt))
2320 indent = 0;
2321
2322 /*
2323 * We need to check and emit Content-type: to mark it
2324 * as 8-bit if we haven't done so.
2325 */
2326 if (cmit_fmt_is_mail(pp->fmt) && need_8bit_cte == 0) {
2327 int i, ch, in_body;
2328
2329 for (in_body = i = 0; (ch = msg[i]); i++) {
2330 if (!in_body) {
2331 /* author could be non 7-bit ASCII but
2332 * the log may be so; skip over the
2333 * header part first.
2334 */
2335 if (ch == '\n' && msg[i+1] == '\n')
2336 in_body = 1;
2337 }
2338 else if (non_ascii(ch)) {
2339 need_8bit_cte = 1;
2340 break;
2341 }
2342 }
2343 }
2344
2345 pp_header(pp, encoding, commit, &msg, sb);
2346 if (pp->fmt != CMIT_FMT_ONELINE && !cmit_fmt_is_mail(pp->fmt)) {
2347 strbuf_addch(sb, '\n');
2348 }
2349
2350 /* Skip excess blank lines at the beginning of body, if any... */
2351 msg = skip_blank_lines(msg);
2352
2353 /* These formats treat the title line specially. */
2354 if (pp->fmt == CMIT_FMT_ONELINE) {
2355 msg = format_subject(sb, msg, " ");
2356 strbuf_addch(sb, '\n');
2357 } else if (cmit_fmt_is_mail(pp->fmt))
2358 pp_email_subject(pp, &msg, sb, encoding, need_8bit_cte);
2359
2360 beginning_of_body = sb->len;
2361 if (pp->fmt != CMIT_FMT_ONELINE)
2362 pp_remainder(pp, &msg, sb, indent);
2363 strbuf_rtrim(sb);
2364
2365 /* Make sure there is an EOLN for the non-oneline case */
2366 if (pp->fmt != CMIT_FMT_ONELINE)
2367 strbuf_addch(sb, '\n');
2368
2369 /*
2370 * The caller may append additional body text in e-mail
2371 * format. Make sure we did not strip the blank line
2372 * between the header and the body.
2373 */
2374 if (cmit_fmt_is_mail(pp->fmt) && sb->len <= beginning_of_body)
2375 strbuf_addch(sb, '\n');
2376
2377 repo_unuse_commit_buffer(the_repository, commit, reencoded);
2378 }
2379
2380 void pp_commit_easy(enum cmit_fmt fmt, const struct commit *commit,
2381 struct strbuf *sb)
2382 {
2383 struct pretty_print_context pp = {0};
2384 pp.fmt = fmt;
2385 pretty_print_commit(&pp, commit, sb);
2386 }