9ffe7bcebb60c7ca9980cf88132841dc9d0ee6fa
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
14static char *user_format;
15static 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;
22static size_t builtin_formats_len;
23static size_t commit_formats_len;
24static size_t commit_formats_alloc;
25static struct cmt_fmt_map *find_commit_format(const char *sought);
26
27static void save_user_format(struct rev_info *rev, const char *cp, int is_tformat)
28{
29 free(user_format);
30 user_format = xstrdup(cp);
31 if (is_tformat)
32 rev->use_terminator = 1;
33 rev->commit_format = CMIT_FMT_USERFORMAT;
34}
35
36static int git_pretty_formats_config(const char *var, const char *value, void *cb)
37{
38 struct cmt_fmt_map *commit_format = NULL;
39 const char *name;
40 const char *fmt;
41 int i;
42
43 if (prefixcmp(var, "pretty."))
44 return 0;
45
46 name = var + strlen("pretty.");
47 for (i = 0; i < builtin_formats_len; i++) {
48 if (!strcmp(commit_formats[i].name, name))
49 return 0;
50 }
51
52 for (i = builtin_formats_len; i < commit_formats_len; i++) {
53 if (!strcmp(commit_formats[i].name, name)) {
54 commit_format = &commit_formats[i];
55 break;
56 }
57 }
58
59 if (!commit_format) {
60 ALLOC_GROW(commit_formats, commit_formats_len+1,
61 commit_formats_alloc);
62 commit_format = &commit_formats[commit_formats_len];
63 memset(commit_format, 0, sizeof(*commit_format));
64 commit_formats_len++;
65 }
66
67 commit_format->name = xstrdup(name);
68 commit_format->format = CMIT_FMT_USERFORMAT;
69 git_config_string(&fmt, var, value);
70 if (!prefixcmp(fmt, "format:") || !prefixcmp(fmt, "tformat:")) {
71 commit_format->is_tformat = fmt[0] == 't';
72 fmt = strchr(fmt, ':') + 1;
73 } else if (strchr(fmt, '%'))
74 commit_format->is_tformat = 1;
75 else
76 commit_format->is_alias = 1;
77 commit_format->user_format = fmt;
78
79 return 0;
80}
81
82static void setup_commit_formats(void)
83{
84 struct cmt_fmt_map builtin_formats[] = {
85 { "raw", CMIT_FMT_RAW, 0 },
86 { "medium", CMIT_FMT_MEDIUM, 0 },
87 { "short", CMIT_FMT_SHORT, 0 },
88 { "email", CMIT_FMT_EMAIL, 0 },
89 { "fuller", CMIT_FMT_FULLER, 0 },
90 { "full", CMIT_FMT_FULL, 0 },
91 { "oneline", CMIT_FMT_ONELINE, 1 }
92 };
93 commit_formats_len = ARRAY_SIZE(builtin_formats);
94 builtin_formats_len = commit_formats_len;
95 ALLOC_GROW(commit_formats, commit_formats_len, commit_formats_alloc);
96 memcpy(commit_formats, builtin_formats,
97 sizeof(*builtin_formats)*ARRAY_SIZE(builtin_formats));
98
99 git_config(git_pretty_formats_config, NULL);
100}
101
102static struct cmt_fmt_map *find_commit_format_recursive(const char *sought,
103 const char *original,
104 int num_redirections)
105{
106 struct cmt_fmt_map *found = NULL;
107 size_t found_match_len = 0;
108 int i;
109
110 if (num_redirections >= commit_formats_len)
111 die("invalid --pretty format: "
112 "'%s' references an alias which points to itself",
113 original);
114
115 for (i = 0; i < commit_formats_len; i++) {
116 size_t match_len;
117
118 if (prefixcmp(commit_formats[i].name, sought))
119 continue;
120
121 match_len = strlen(commit_formats[i].name);
122 if (found == NULL || found_match_len > match_len) {
123 found = &commit_formats[i];
124 found_match_len = match_len;
125 }
126 }
127
128 if (found && found->is_alias) {
129 found = find_commit_format_recursive(found->user_format,
130 original,
131 num_redirections+1);
132 }
133
134 return found;
135}
136
137static struct cmt_fmt_map *find_commit_format(const char *sought)
138{
139 if (!commit_formats)
140 setup_commit_formats();
141
142 return find_commit_format_recursive(sought, sought, 0);
143}
144
145void get_commit_format(const char *arg, struct rev_info *rev)
146{
147 struct cmt_fmt_map *commit_format;
148
149 rev->use_terminator = 0;
150 if (!arg || !*arg) {
151 rev->commit_format = CMIT_FMT_DEFAULT;
152 return;
153 }
154 if (!prefixcmp(arg, "format:") || !prefixcmp(arg, "tformat:")) {
155 save_user_format(rev, strchr(arg, ':') + 1, arg[0] == 't');
156 return;
157 }
158
159 if (strchr(arg, '%')) {
160 save_user_format(rev, arg, 1);
161 return;
162 }
163
164 commit_format = find_commit_format(arg);
165 if (!commit_format)
166 die("invalid --pretty format: %s", arg);
167
168 rev->commit_format = commit_format->format;
169 rev->use_terminator = commit_format->is_tformat;
170 if (commit_format->format == CMIT_FMT_USERFORMAT) {
171 save_user_format(rev, commit_format->user_format,
172 commit_format->is_tformat);
173 }
174}
175
176/*
177 * Generic support for pretty-printing the header
178 */
179static int get_one_line(const char *msg)
180{
181 int ret = 0;
182
183 for (;;) {
184 char c = *msg++;
185 if (!c)
186 break;
187 ret++;
188 if (c == '\n')
189 break;
190 }
191 return ret;
192}
193
194/* High bit set, or ISO-2022-INT */
195static int non_ascii(int ch)
196{
197 return !isascii(ch) || ch == '\033';
198}
199
200int has_non_ascii(const char *s)
201{
202 int ch;
203 if (!s)
204 return 0;
205 while ((ch = *s++) != '\0') {
206 if (non_ascii(ch))
207 return 1;
208 }
209 return 0;
210}
211
212static int is_rfc822_special(char ch)
213{
214 switch (ch) {
215 case '(':
216 case ')':
217 case '<':
218 case '>':
219 case '[':
220 case ']':
221 case ':':
222 case ';':
223 case '@':
224 case ',':
225 case '.':
226 case '"':
227 case '\\':
228 return 1;
229 default:
230 return 0;
231 }
232}
233
234static int needs_rfc822_quoting(const char *s, int len)
235{
236 int i;
237 for (i = 0; i < len; i++)
238 if (is_rfc822_special(s[i]))
239 return 1;
240 return 0;
241}
242
243static int last_line_length(struct strbuf *sb)
244{
245 int i;
246
247 /* How many bytes are already used on the last line? */
248 for (i = sb->len - 1; i >= 0; i--)
249 if (sb->buf[i] == '\n')
250 break;
251 return sb->len - (i + 1);
252}
253
254static void add_rfc822_quoted(struct strbuf *out, const char *s, int len)
255{
256 int i;
257
258 /* just a guess, we may have to also backslash-quote */
259 strbuf_grow(out, len + 2);
260
261 strbuf_addch(out, '"');
262 for (i = 0; i < len; i++) {
263 switch (s[i]) {
264 case '"':
265 case '\\':
266 strbuf_addch(out, '\\');
267 /* fall through */
268 default:
269 strbuf_addch(out, s[i]);
270 }
271 }
272 strbuf_addch(out, '"');
273}
274
275enum rfc2047_type {
276 RFC2047_SUBJECT,
277 RFC2047_ADDRESS,
278};
279
280static int is_rfc2047_special(char ch, enum rfc2047_type type)
281{
282 /*
283 * rfc2047, section 4.2:
284 *
285 * 8-bit values which correspond to printable ASCII characters other
286 * than "=", "?", and "_" (underscore), MAY be represented as those
287 * characters. (But see section 5 for restrictions.) In
288 * particular, SPACE and TAB MUST NOT be represented as themselves
289 * within encoded words.
290 */
291
292 /*
293 * rule out non-ASCII characters and non-printable characters (the
294 * non-ASCII check should be redundant as isprint() is not localized
295 * and only knows about ASCII, but be defensive about that)
296 */
297 if (non_ascii(ch) || !isprint(ch))
298 return 1;
299
300 /*
301 * rule out special printable characters (' ' should be the only
302 * whitespace character considered printable, but be defensive and use
303 * isspace())
304 */
305 if (isspace(ch) || ch == '=' || ch == '?' || ch == '_')
306 return 1;
307
308 /*
309 * rfc2047, section 5.3:
310 *
311 * As a replacement for a 'word' entity within a 'phrase', for example,
312 * one that precedes an address in a From, To, or Cc header. The ABNF
313 * definition for 'phrase' from RFC 822 thus becomes:
314 *
315 * phrase = 1*( encoded-word / word )
316 *
317 * In this case the set of characters that may be used in a "Q"-encoded
318 * 'encoded-word' is restricted to: <upper and lower case ASCII
319 * letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
320 * (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
321 * 'phrase' MUST be separated from any adjacent 'word', 'text' or
322 * 'special' by 'linear-white-space'.
323 */
324
325 if (type != RFC2047_ADDRESS)
326 return 0;
327
328 /* '=' and '_' are special cases and have been checked above */
329 return !(isalnum(ch) || ch == '!' || ch == '*' || ch == '+' || ch == '-' || ch == '/');
330}
331
332static int needs_rfc2047_encoding(const char *line, int len,
333 enum rfc2047_type type)
334{
335 int i;
336
337 for (i = 0; i < len; i++) {
338 int ch = line[i];
339 if (non_ascii(ch) || ch == '\n')
340 return 1;
341 if ((i + 1 < len) && (ch == '=' && line[i+1] == '?'))
342 return 1;
343 }
344
345 return 0;
346}
347
348static void add_rfc2047(struct strbuf *sb, const char *line, int len,
349 const char *encoding, enum rfc2047_type type)
350{
351 static const int max_encoded_length = 76; /* per rfc2047 */
352 int i;
353 int line_len = last_line_length(sb);
354
355 strbuf_grow(sb, len * 3 + strlen(encoding) + 100);
356 strbuf_addf(sb, "=?%s?q?", encoding);
357 line_len += strlen(encoding) + 5; /* 5 for =??q? */
358 for (i = 0; i < len; i++) {
359 unsigned ch = line[i] & 0xFF;
360 int is_special = is_rfc2047_special(ch, type);
361
362 /*
363 * According to RFC 2047, we could encode the special character
364 * ' ' (space) with '_' (underscore) for readability. But many
365 * programs do not understand this and just leave the
366 * underscore in place. Thus, we do nothing special here, which
367 * causes ' ' to be encoded as '=20', avoiding this problem.
368 */
369
370 if (line_len + 2 + (is_special ? 3 : 1) > max_encoded_length) {
371 strbuf_addf(sb, "?=\n =?%s?q?", encoding);
372 line_len = strlen(encoding) + 5 + 1; /* =??q? plus SP */
373 }
374
375 if (is_special) {
376 strbuf_addf(sb, "=%02X", ch);
377 line_len += 3;
378 } else {
379 strbuf_addch(sb, ch);
380 line_len++;
381 }
382 }
383 strbuf_addstr(sb, "?=");
384}
385
386void pp_user_info(const struct pretty_print_context *pp,
387 const char *what, struct strbuf *sb,
388 const char *line, const char *encoding)
389{
390 int max_length = 78; /* per rfc2822 */
391 char *date;
392 int namelen;
393 unsigned long time;
394 int tz;
395
396 if (pp->fmt == CMIT_FMT_ONELINE)
397 return;
398 date = strchr(line, '>');
399 if (!date)
400 return;
401 namelen = ++date - line;
402 time = strtoul(date, &date, 10);
403 tz = strtol(date, NULL, 10);
404
405 if (pp->fmt == CMIT_FMT_EMAIL) {
406 char *name_tail = strchr(line, '<');
407 int display_name_length;
408 if (!name_tail)
409 return;
410 while (line < name_tail && isspace(name_tail[-1]))
411 name_tail--;
412 display_name_length = name_tail - line;
413 strbuf_addstr(sb, "From: ");
414 if (needs_rfc2047_encoding(line, display_name_length, RFC2047_ADDRESS)) {
415 add_rfc2047(sb, line, display_name_length,
416 encoding, RFC2047_ADDRESS);
417 max_length = 76; /* per rfc2047 */
418 } else if (needs_rfc822_quoting(line, display_name_length)) {
419 struct strbuf quoted = STRBUF_INIT;
420 add_rfc822_quoted("ed, line, display_name_length);
421 strbuf_add_wrapped_bytes(sb, quoted.buf, quoted.len,
422 -6, 1, max_length);
423 strbuf_release("ed);
424 } else {
425 strbuf_add_wrapped_bytes(sb, line, display_name_length,
426 -6, 1, max_length);
427 }
428 if (namelen - display_name_length + last_line_length(sb) > max_length) {
429 strbuf_addch(sb, '\n');
430 if (!isspace(name_tail[0]))
431 strbuf_addch(sb, ' ');
432 }
433 strbuf_add(sb, name_tail, namelen - display_name_length);
434 strbuf_addch(sb, '\n');
435 } else {
436 strbuf_addf(sb, "%s: %.*s%.*s\n", what,
437 (pp->fmt == CMIT_FMT_FULLER) ? 4 : 0,
438 " ", namelen, line);
439 }
440 switch (pp->fmt) {
441 case CMIT_FMT_MEDIUM:
442 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, pp->date_mode));
443 break;
444 case CMIT_FMT_EMAIL:
445 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, DATE_RFC2822));
446 break;
447 case CMIT_FMT_FULLER:
448 strbuf_addf(sb, "%sDate: %s\n", what, show_date(time, tz, pp->date_mode));
449 break;
450 default:
451 /* notin' */
452 break;
453 }
454}
455
456static int is_empty_line(const char *line, int *len_p)
457{
458 int len = *len_p;
459 while (len && isspace(line[len-1]))
460 len--;
461 *len_p = len;
462 return !len;
463}
464
465static const char *skip_empty_lines(const char *msg)
466{
467 for (;;) {
468 int linelen = get_one_line(msg);
469 int ll = linelen;
470 if (!linelen)
471 break;
472 if (!is_empty_line(msg, &ll))
473 break;
474 msg += linelen;
475 }
476 return msg;
477}
478
479static void add_merge_info(const struct pretty_print_context *pp,
480 struct strbuf *sb, const struct commit *commit)
481{
482 struct commit_list *parent = commit->parents;
483
484 if ((pp->fmt == CMIT_FMT_ONELINE) || (pp->fmt == CMIT_FMT_EMAIL) ||
485 !parent || !parent->next)
486 return;
487
488 strbuf_addstr(sb, "Merge:");
489
490 while (parent) {
491 struct commit *p = parent->item;
492 const char *hex = NULL;
493 if (pp->abbrev)
494 hex = find_unique_abbrev(p->object.sha1, pp->abbrev);
495 if (!hex)
496 hex = sha1_to_hex(p->object.sha1);
497 parent = parent->next;
498
499 strbuf_addf(sb, " %s", hex);
500 }
501 strbuf_addch(sb, '\n');
502}
503
504static char *get_header(const struct commit *commit, const char *key)
505{
506 int key_len = strlen(key);
507 const char *line = commit->buffer;
508
509 while (line) {
510 const char *eol = strchr(line, '\n'), *next;
511
512 if (line == eol)
513 return NULL;
514 if (!eol) {
515 warning("malformed commit (header is missing newline): %s",
516 sha1_to_hex(commit->object.sha1));
517 eol = line + strlen(line);
518 next = NULL;
519 } else
520 next = eol + 1;
521 if (eol - line > key_len &&
522 !strncmp(line, key, key_len) &&
523 line[key_len] == ' ') {
524 return xmemdupz(line + key_len + 1, eol - line - key_len - 1);
525 }
526 line = next;
527 }
528 return NULL;
529}
530
531static char *replace_encoding_header(char *buf, const char *encoding)
532{
533 struct strbuf tmp = STRBUF_INIT;
534 size_t start, len;
535 char *cp = buf;
536
537 /* guess if there is an encoding header before a \n\n */
538 while (strncmp(cp, "encoding ", strlen("encoding "))) {
539 cp = strchr(cp, '\n');
540 if (!cp || *++cp == '\n')
541 return buf;
542 }
543 start = cp - buf;
544 cp = strchr(cp, '\n');
545 if (!cp)
546 return buf; /* should not happen but be defensive */
547 len = cp + 1 - (buf + start);
548
549 strbuf_attach(&tmp, buf, strlen(buf), strlen(buf) + 1);
550 if (is_encoding_utf8(encoding)) {
551 /* we have re-coded to UTF-8; drop the header */
552 strbuf_remove(&tmp, start, len);
553 } else {
554 /* just replaces XXXX in 'encoding XXXX\n' */
555 strbuf_splice(&tmp, start + strlen("encoding "),
556 len - strlen("encoding \n"),
557 encoding, strlen(encoding));
558 }
559 return strbuf_detach(&tmp, NULL);
560}
561
562char *logmsg_reencode(const struct commit *commit,
563 const char *output_encoding)
564{
565 static const char *utf8 = "UTF-8";
566 const char *use_encoding;
567 char *encoding;
568 char *out;
569
570 if (!output_encoding || !*output_encoding)
571 return NULL;
572 encoding = get_header(commit, "encoding");
573 use_encoding = encoding ? encoding : utf8;
574 if (same_encoding(use_encoding, output_encoding))
575 if (encoding) /* we'll strip encoding header later */
576 out = xstrdup(commit->buffer);
577 else
578 return NULL; /* nothing to do */
579 else
580 out = reencode_string(commit->buffer,
581 output_encoding, use_encoding);
582 if (out)
583 out = replace_encoding_header(out, output_encoding);
584
585 free(encoding);
586 return out;
587}
588
589static int mailmap_name(char *email, int email_len, char *name, int name_len)
590{
591 static struct string_list *mail_map;
592 if (!mail_map) {
593 mail_map = xcalloc(1, sizeof(*mail_map));
594 read_mailmap(mail_map, NULL);
595 }
596 return mail_map->nr && map_user(mail_map, email, email_len, name, name_len);
597}
598
599static size_t format_person_part(struct strbuf *sb, char part,
600 const char *msg, int len, enum date_mode dmode)
601{
602 /* currently all placeholders have same length */
603 const int placeholder_len = 2;
604 int tz;
605 unsigned long date = 0;
606 char person_name[1024];
607 char person_mail[1024];
608 struct ident_split s;
609 const char *name_start, *name_end, *mail_start, *mail_end;
610
611 if (split_ident_line(&s, msg, len) < 0)
612 goto skip;
613
614 name_start = s.name_begin;
615 name_end = s.name_end;
616 mail_start = s.mail_begin;
617 mail_end = s.mail_end;
618
619 if (part == 'N' || part == 'E') { /* mailmap lookup */
620 snprintf(person_name, sizeof(person_name), "%.*s",
621 (int)(name_end - name_start), name_start);
622 snprintf(person_mail, sizeof(person_mail), "%.*s",
623 (int)(mail_end - mail_start), mail_start);
624 mailmap_name(person_mail, sizeof(person_mail), person_name, sizeof(person_name));
625 name_start = person_name;
626 name_end = name_start + strlen(person_name);
627 mail_start = person_mail;
628 mail_end = mail_start + strlen(person_mail);
629 }
630 if (part == 'n' || part == 'N') { /* name */
631 strbuf_add(sb, name_start, name_end-name_start);
632 return placeholder_len;
633 }
634 if (part == 'e' || part == 'E') { /* email */
635 strbuf_add(sb, mail_start, mail_end-mail_start);
636 return placeholder_len;
637 }
638
639 if (!s.date_begin)
640 goto skip;
641
642 date = strtoul(s.date_begin, NULL, 10);
643
644 if (part == 't') { /* date, UNIX timestamp */
645 strbuf_add(sb, s.date_begin, s.date_end - s.date_begin);
646 return placeholder_len;
647 }
648
649 /* parse tz */
650 tz = strtoul(s.tz_begin + 1, NULL, 10);
651 if (*s.tz_begin == '-')
652 tz = -tz;
653
654 switch (part) {
655 case 'd': /* date */
656 strbuf_addstr(sb, show_date(date, tz, dmode));
657 return placeholder_len;
658 case 'D': /* date, RFC2822 style */
659 strbuf_addstr(sb, show_date(date, tz, DATE_RFC2822));
660 return placeholder_len;
661 case 'r': /* date, relative */
662 strbuf_addstr(sb, show_date(date, tz, DATE_RELATIVE));
663 return placeholder_len;
664 case 'i': /* date, ISO 8601 */
665 strbuf_addstr(sb, show_date(date, tz, DATE_ISO8601));
666 return placeholder_len;
667 }
668
669skip:
670 /*
671 * reading from either a bogus commit, or a reflog entry with
672 * %gn, %ge, etc.; 'sb' cannot be updated, but we still need
673 * to compute a valid return value.
674 */
675 if (part == 'n' || part == 'e' || part == 't' || part == 'd'
676 || part == 'D' || part == 'r' || part == 'i')
677 return placeholder_len;
678
679 return 0; /* unknown placeholder */
680}
681
682struct chunk {
683 size_t off;
684 size_t len;
685};
686
687struct format_commit_context {
688 const struct commit *commit;
689 const struct pretty_print_context *pretty_ctx;
690 unsigned commit_header_parsed:1;
691 unsigned commit_message_parsed:1;
692 unsigned commit_signature_parsed:1;
693 struct {
694 char *gpg_output;
695 char *gpg_status;
696 char good_bad;
697 char *signer;
698 } signature;
699 char *message;
700 size_t width, indent1, indent2;
701
702 /* These offsets are relative to the start of the commit message. */
703 struct chunk author;
704 struct chunk committer;
705 struct chunk encoding;
706 size_t message_off;
707 size_t subject_off;
708 size_t body_off;
709
710 /* The following ones are relative to the result struct strbuf. */
711 struct chunk abbrev_commit_hash;
712 struct chunk abbrev_tree_hash;
713 struct chunk abbrev_parent_hashes;
714 size_t wrap_start;
715};
716
717static int add_again(struct strbuf *sb, struct chunk *chunk)
718{
719 if (chunk->len) {
720 strbuf_adddup(sb, chunk->off, chunk->len);
721 return 1;
722 }
723
724 /*
725 * We haven't seen this chunk before. Our caller is surely
726 * going to add it the hard way now. Remember the most likely
727 * start of the to-be-added chunk: the current end of the
728 * struct strbuf.
729 */
730 chunk->off = sb->len;
731 return 0;
732}
733
734static void parse_commit_header(struct format_commit_context *context)
735{
736 const char *msg = context->message;
737 int i;
738
739 for (i = 0; msg[i]; i++) {
740 int eol;
741 for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
742 ; /* do nothing */
743
744 if (i == eol) {
745 break;
746 } else if (!prefixcmp(msg + i, "author ")) {
747 context->author.off = i + 7;
748 context->author.len = eol - i - 7;
749 } else if (!prefixcmp(msg + i, "committer ")) {
750 context->committer.off = i + 10;
751 context->committer.len = eol - i - 10;
752 } else if (!prefixcmp(msg + i, "encoding ")) {
753 context->encoding.off = i + 9;
754 context->encoding.len = eol - i - 9;
755 }
756 i = eol;
757 }
758 context->message_off = i;
759 context->commit_header_parsed = 1;
760}
761
762static int istitlechar(char c)
763{
764 return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') ||
765 (c >= '0' && c <= '9') || c == '.' || c == '_';
766}
767
768static void format_sanitized_subject(struct strbuf *sb, const char *msg)
769{
770 size_t trimlen;
771 size_t start_len = sb->len;
772 int space = 2;
773
774 for (; *msg && *msg != '\n'; msg++) {
775 if (istitlechar(*msg)) {
776 if (space == 1)
777 strbuf_addch(sb, '-');
778 space = 0;
779 strbuf_addch(sb, *msg);
780 if (*msg == '.')
781 while (*(msg+1) == '.')
782 msg++;
783 } else
784 space |= 1;
785 }
786
787 /* trim any trailing '.' or '-' characters */
788 trimlen = 0;
789 while (sb->len - trimlen > start_len &&
790 (sb->buf[sb->len - 1 - trimlen] == '.'
791 || sb->buf[sb->len - 1 - trimlen] == '-'))
792 trimlen++;
793 strbuf_remove(sb, sb->len - trimlen, trimlen);
794}
795
796const char *format_subject(struct strbuf *sb, const char *msg,
797 const char *line_separator)
798{
799 int first = 1;
800
801 for (;;) {
802 const char *line = msg;
803 int linelen = get_one_line(line);
804
805 msg += linelen;
806 if (!linelen || is_empty_line(line, &linelen))
807 break;
808
809 if (!sb)
810 continue;
811 strbuf_grow(sb, linelen + 2);
812 if (!first)
813 strbuf_addstr(sb, line_separator);
814 strbuf_add(sb, line, linelen);
815 first = 0;
816 }
817 return msg;
818}
819
820static void parse_commit_message(struct format_commit_context *c)
821{
822 const char *msg = c->message + c->message_off;
823 const char *start = c->message;
824
825 msg = skip_empty_lines(msg);
826 c->subject_off = msg - start;
827
828 msg = format_subject(NULL, msg, NULL);
829 msg = skip_empty_lines(msg);
830 c->body_off = msg - start;
831
832 c->commit_message_parsed = 1;
833}
834
835static void format_decoration(struct strbuf *sb, const struct commit *commit)
836{
837 struct name_decoration *d;
838 const char *prefix = " (";
839
840 load_ref_decorations(DECORATE_SHORT_REFS);
841 d = lookup_decoration(&name_decoration, &commit->object);
842 while (d) {
843 strbuf_addstr(sb, prefix);
844 prefix = ", ";
845 strbuf_addstr(sb, d->name);
846 d = d->next;
847 }
848 if (prefix[0] == ',')
849 strbuf_addch(sb, ')');
850}
851
852static void strbuf_wrap(struct strbuf *sb, size_t pos,
853 size_t width, size_t indent1, size_t indent2)
854{
855 struct strbuf tmp = STRBUF_INIT;
856
857 if (pos)
858 strbuf_add(&tmp, sb->buf, pos);
859 strbuf_add_wrapped_text(&tmp, sb->buf + pos,
860 (int) indent1, (int) indent2, (int) width);
861 strbuf_swap(&tmp, sb);
862 strbuf_release(&tmp);
863}
864
865static void rewrap_message_tail(struct strbuf *sb,
866 struct format_commit_context *c,
867 size_t new_width, size_t new_indent1,
868 size_t new_indent2)
869{
870 if (c->width == new_width && c->indent1 == new_indent1 &&
871 c->indent2 == new_indent2)
872 return;
873 if (c->wrap_start < sb->len)
874 strbuf_wrap(sb, c->wrap_start, c->width, c->indent1, c->indent2);
875 c->wrap_start = sb->len;
876 c->width = new_width;
877 c->indent1 = new_indent1;
878 c->indent2 = new_indent2;
879}
880
881static struct {
882 char result;
883 const char *check;
884} signature_check[] = {
885 { 'G', "\n[GNUPG:] GOODSIG " },
886 { 'B', "\n[GNUPG:] BADSIG " },
887};
888
889static void parse_signature_lines(struct format_commit_context *ctx)
890{
891 const char *buf = ctx->signature.gpg_status;
892 int i;
893
894 for (i = 0; i < ARRAY_SIZE(signature_check); i++) {
895 const char *found = strstr(buf, signature_check[i].check);
896 const char *next;
897 if (!found)
898 continue;
899 ctx->signature.good_bad = signature_check[i].result;
900 found += strlen(signature_check[i].check)+17;
901 next = strchrnul(found, '\n');
902 ctx->signature.signer = xmemdupz(found, next - found);
903 break;
904 }
905}
906
907static void parse_commit_signature(struct format_commit_context *ctx)
908{
909 struct strbuf payload = STRBUF_INIT;
910 struct strbuf signature = STRBUF_INIT;
911 struct strbuf gpg_output = STRBUF_INIT;
912 struct strbuf gpg_status = STRBUF_INIT;
913 int status;
914
915 ctx->commit_signature_parsed = 1;
916
917 if (parse_signed_commit(ctx->commit->object.sha1,
918 &payload, &signature) <= 0)
919 goto out;
920 status = verify_signed_buffer(payload.buf, payload.len,
921 signature.buf, signature.len,
922 &gpg_output, &gpg_status);
923 if (status && !gpg_output.len)
924 goto out;
925 ctx->signature.gpg_output = strbuf_detach(&gpg_output, NULL);
926 ctx->signature.gpg_status = strbuf_detach(&gpg_status, NULL);
927 parse_signature_lines(ctx);
928
929 out:
930 strbuf_release(&gpg_status);
931 strbuf_release(&gpg_output);
932 strbuf_release(&payload);
933 strbuf_release(&signature);
934}
935
936
937static int format_reflog_person(struct strbuf *sb,
938 char part,
939 struct reflog_walk_info *log,
940 enum date_mode dmode)
941{
942 const char *ident;
943
944 if (!log)
945 return 2;
946
947 ident = get_reflog_ident(log);
948 if (!ident)
949 return 2;
950
951 return format_person_part(sb, part, ident, strlen(ident), dmode);
952}
953
954static size_t format_commit_one(struct strbuf *sb, const char *placeholder,
955 void *context)
956{
957 struct format_commit_context *c = context;
958 const struct commit *commit = c->commit;
959 const char *msg = c->message;
960 struct commit_list *p;
961 int h1, h2;
962
963 /* these are independent of the commit */
964 switch (placeholder[0]) {
965 case 'C':
966 if (placeholder[1] == '(') {
967 const char *end = strchr(placeholder + 2, ')');
968 char color[COLOR_MAXLEN];
969 if (!end)
970 return 0;
971 color_parse_mem(placeholder + 2,
972 end - (placeholder + 2),
973 "--pretty format", color);
974 strbuf_addstr(sb, color);
975 return end - placeholder + 1;
976 }
977 if (!prefixcmp(placeholder + 1, "red")) {
978 strbuf_addstr(sb, GIT_COLOR_RED);
979 return 4;
980 } else if (!prefixcmp(placeholder + 1, "green")) {
981 strbuf_addstr(sb, GIT_COLOR_GREEN);
982 return 6;
983 } else if (!prefixcmp(placeholder + 1, "blue")) {
984 strbuf_addstr(sb, GIT_COLOR_BLUE);
985 return 5;
986 } else if (!prefixcmp(placeholder + 1, "reset")) {
987 strbuf_addstr(sb, GIT_COLOR_RESET);
988 return 6;
989 } else
990 return 0;
991 case 'n': /* newline */
992 strbuf_addch(sb, '\n');
993 return 1;
994 case 'x':
995 /* %x00 == NUL, %x0a == LF, etc. */
996 if (0 <= (h1 = hexval_table[0xff & placeholder[1]]) &&
997 h1 <= 16 &&
998 0 <= (h2 = hexval_table[0xff & placeholder[2]]) &&
999 h2 <= 16) {
1000 strbuf_addch(sb, (h1<<4)|h2);
1001 return 3;
1002 } else
1003 return 0;
1004 case 'w':
1005 if (placeholder[1] == '(') {
1006 unsigned long width = 0, indent1 = 0, indent2 = 0;
1007 char *next;
1008 const char *start = placeholder + 2;
1009 const char *end = strchr(start, ')');
1010 if (!end)
1011 return 0;
1012 if (end > start) {
1013 width = strtoul(start, &next, 10);
1014 if (*next == ',') {
1015 indent1 = strtoul(next + 1, &next, 10);
1016 if (*next == ',') {
1017 indent2 = strtoul(next + 1,
1018 &next, 10);
1019 }
1020 }
1021 if (*next != ')')
1022 return 0;
1023 }
1024 rewrap_message_tail(sb, c, width, indent1, indent2);
1025 return end - placeholder + 1;
1026 } else
1027 return 0;
1028 }
1029
1030 /* these depend on the commit */
1031 if (!commit->object.parsed)
1032 parse_object(commit->object.sha1);
1033
1034 switch (placeholder[0]) {
1035 case 'H': /* commit hash */
1036 strbuf_addstr(sb, sha1_to_hex(commit->object.sha1));
1037 return 1;
1038 case 'h': /* abbreviated commit hash */
1039 if (add_again(sb, &c->abbrev_commit_hash))
1040 return 1;
1041 strbuf_addstr(sb, find_unique_abbrev(commit->object.sha1,
1042 c->pretty_ctx->abbrev));
1043 c->abbrev_commit_hash.len = sb->len - c->abbrev_commit_hash.off;
1044 return 1;
1045 case 'T': /* tree hash */
1046 strbuf_addstr(sb, sha1_to_hex(commit->tree->object.sha1));
1047 return 1;
1048 case 't': /* abbreviated tree hash */
1049 if (add_again(sb, &c->abbrev_tree_hash))
1050 return 1;
1051 strbuf_addstr(sb, find_unique_abbrev(commit->tree->object.sha1,
1052 c->pretty_ctx->abbrev));
1053 c->abbrev_tree_hash.len = sb->len - c->abbrev_tree_hash.off;
1054 return 1;
1055 case 'P': /* parent hashes */
1056 for (p = commit->parents; p; p = p->next) {
1057 if (p != commit->parents)
1058 strbuf_addch(sb, ' ');
1059 strbuf_addstr(sb, sha1_to_hex(p->item->object.sha1));
1060 }
1061 return 1;
1062 case 'p': /* abbreviated parent hashes */
1063 if (add_again(sb, &c->abbrev_parent_hashes))
1064 return 1;
1065 for (p = commit->parents; p; p = p->next) {
1066 if (p != commit->parents)
1067 strbuf_addch(sb, ' ');
1068 strbuf_addstr(sb, find_unique_abbrev(
1069 p->item->object.sha1,
1070 c->pretty_ctx->abbrev));
1071 }
1072 c->abbrev_parent_hashes.len = sb->len -
1073 c->abbrev_parent_hashes.off;
1074 return 1;
1075 case 'm': /* left/right/bottom */
1076 strbuf_addstr(sb, get_revision_mark(NULL, commit));
1077 return 1;
1078 case 'd':
1079 format_decoration(sb, commit);
1080 return 1;
1081 case 'g': /* reflog info */
1082 switch(placeholder[1]) {
1083 case 'd': /* reflog selector */
1084 case 'D':
1085 if (c->pretty_ctx->reflog_info)
1086 get_reflog_selector(sb,
1087 c->pretty_ctx->reflog_info,
1088 c->pretty_ctx->date_mode,
1089 c->pretty_ctx->date_mode_explicit,
1090 (placeholder[1] == 'd'));
1091 return 2;
1092 case 's': /* reflog message */
1093 if (c->pretty_ctx->reflog_info)
1094 get_reflog_message(sb, c->pretty_ctx->reflog_info);
1095 return 2;
1096 case 'n':
1097 case 'N':
1098 case 'e':
1099 case 'E':
1100 return format_reflog_person(sb,
1101 placeholder[1],
1102 c->pretty_ctx->reflog_info,
1103 c->pretty_ctx->date_mode);
1104 }
1105 return 0; /* unknown %g placeholder */
1106 case 'N':
1107 if (c->pretty_ctx->notes_message) {
1108 strbuf_addstr(sb, c->pretty_ctx->notes_message);
1109 return 1;
1110 }
1111 return 0;
1112 }
1113
1114 if (placeholder[0] == 'G') {
1115 if (!c->commit_signature_parsed)
1116 parse_commit_signature(c);
1117 switch (placeholder[1]) {
1118 case 'G':
1119 if (c->signature.gpg_output)
1120 strbuf_addstr(sb, c->signature.gpg_output);
1121 break;
1122 case '?':
1123 switch (c->signature.good_bad) {
1124 case 'G':
1125 case 'B':
1126 strbuf_addch(sb, c->signature.good_bad);
1127 }
1128 break;
1129 case 'S':
1130 if (c->signature.signer)
1131 strbuf_addstr(sb, c->signature.signer);
1132 break;
1133 }
1134 return 2;
1135 }
1136
1137
1138 /* For the rest we have to parse the commit header. */
1139 if (!c->commit_header_parsed)
1140 parse_commit_header(c);
1141
1142 switch (placeholder[0]) {
1143 case 'a': /* author ... */
1144 return format_person_part(sb, placeholder[1],
1145 msg + c->author.off, c->author.len,
1146 c->pretty_ctx->date_mode);
1147 case 'c': /* committer ... */
1148 return format_person_part(sb, placeholder[1],
1149 msg + c->committer.off, c->committer.len,
1150 c->pretty_ctx->date_mode);
1151 case 'e': /* encoding */
1152 strbuf_add(sb, msg + c->encoding.off, c->encoding.len);
1153 return 1;
1154 case 'B': /* raw body */
1155 /* message_off is always left at the initial newline */
1156 strbuf_addstr(sb, msg + c->message_off + 1);
1157 return 1;
1158 }
1159
1160 /* Now we need to parse the commit message. */
1161 if (!c->commit_message_parsed)
1162 parse_commit_message(c);
1163
1164 switch (placeholder[0]) {
1165 case 's': /* subject */
1166 format_subject(sb, msg + c->subject_off, " ");
1167 return 1;
1168 case 'f': /* sanitized subject */
1169 format_sanitized_subject(sb, msg + c->subject_off);
1170 return 1;
1171 case 'b': /* body */
1172 strbuf_addstr(sb, msg + c->body_off);
1173 return 1;
1174 }
1175 return 0; /* unknown placeholder */
1176}
1177
1178static size_t format_commit_item(struct strbuf *sb, const char *placeholder,
1179 void *context)
1180{
1181 int consumed;
1182 size_t orig_len;
1183 enum {
1184 NO_MAGIC,
1185 ADD_LF_BEFORE_NON_EMPTY,
1186 DEL_LF_BEFORE_EMPTY,
1187 ADD_SP_BEFORE_NON_EMPTY
1188 } magic = NO_MAGIC;
1189
1190 switch (placeholder[0]) {
1191 case '-':
1192 magic = DEL_LF_BEFORE_EMPTY;
1193 break;
1194 case '+':
1195 magic = ADD_LF_BEFORE_NON_EMPTY;
1196 break;
1197 case ' ':
1198 magic = ADD_SP_BEFORE_NON_EMPTY;
1199 break;
1200 default:
1201 break;
1202 }
1203 if (magic != NO_MAGIC)
1204 placeholder++;
1205
1206 orig_len = sb->len;
1207 consumed = format_commit_one(sb, placeholder, context);
1208 if (magic == NO_MAGIC)
1209 return consumed;
1210
1211 if ((orig_len == sb->len) && magic == DEL_LF_BEFORE_EMPTY) {
1212 while (sb->len && sb->buf[sb->len - 1] == '\n')
1213 strbuf_setlen(sb, sb->len - 1);
1214 } else if (orig_len != sb->len) {
1215 if (magic == ADD_LF_BEFORE_NON_EMPTY)
1216 strbuf_insert(sb, orig_len, "\n", 1);
1217 else if (magic == ADD_SP_BEFORE_NON_EMPTY)
1218 strbuf_insert(sb, orig_len, " ", 1);
1219 }
1220 return consumed + 1;
1221}
1222
1223static size_t userformat_want_item(struct strbuf *sb, const char *placeholder,
1224 void *context)
1225{
1226 struct userformat_want *w = context;
1227
1228 if (*placeholder == '+' || *placeholder == '-' || *placeholder == ' ')
1229 placeholder++;
1230
1231 switch (*placeholder) {
1232 case 'N':
1233 w->notes = 1;
1234 break;
1235 }
1236 return 0;
1237}
1238
1239void userformat_find_requirements(const char *fmt, struct userformat_want *w)
1240{
1241 struct strbuf dummy = STRBUF_INIT;
1242
1243 if (!fmt) {
1244 if (!user_format)
1245 return;
1246 fmt = user_format;
1247 }
1248 strbuf_expand(&dummy, fmt, userformat_want_item, w);
1249 strbuf_release(&dummy);
1250}
1251
1252void format_commit_message(const struct commit *commit,
1253 const char *format, struct strbuf *sb,
1254 const struct pretty_print_context *pretty_ctx)
1255{
1256 struct format_commit_context context;
1257 const char *output_enc = pretty_ctx->output_encoding;
1258
1259 memset(&context, 0, sizeof(context));
1260 context.commit = commit;
1261 context.pretty_ctx = pretty_ctx;
1262 context.wrap_start = sb->len;
1263 context.message = logmsg_reencode(commit, output_enc);
1264 if (!context.message)
1265 context.message = commit->buffer;
1266
1267 strbuf_expand(sb, format, format_commit_item, &context);
1268 rewrap_message_tail(sb, &context, 0, 0, 0);
1269
1270 if (context.message != commit->buffer)
1271 free(context.message);
1272 free(context.signature.gpg_output);
1273 free(context.signature.signer);
1274}
1275
1276static void pp_header(const struct pretty_print_context *pp,
1277 const char *encoding,
1278 const struct commit *commit,
1279 const char **msg_p,
1280 struct strbuf *sb)
1281{
1282 int parents_shown = 0;
1283
1284 for (;;) {
1285 const char *line = *msg_p;
1286 int linelen = get_one_line(*msg_p);
1287
1288 if (!linelen)
1289 return;
1290 *msg_p += linelen;
1291
1292 if (linelen == 1)
1293 /* End of header */
1294 return;
1295
1296 if (pp->fmt == CMIT_FMT_RAW) {
1297 strbuf_add(sb, line, linelen);
1298 continue;
1299 }
1300
1301 if (!memcmp(line, "parent ", 7)) {
1302 if (linelen != 48)
1303 die("bad parent line in commit");
1304 continue;
1305 }
1306
1307 if (!parents_shown) {
1308 struct commit_list *parent;
1309 int num;
1310 for (parent = commit->parents, num = 0;
1311 parent;
1312 parent = parent->next, num++)
1313 ;
1314 /* with enough slop */
1315 strbuf_grow(sb, num * 50 + 20);
1316 add_merge_info(pp, sb, commit);
1317 parents_shown = 1;
1318 }
1319
1320 /*
1321 * MEDIUM == DEFAULT shows only author with dates.
1322 * FULL shows both authors but not dates.
1323 * FULLER shows both authors and dates.
1324 */
1325 if (!memcmp(line, "author ", 7)) {
1326 strbuf_grow(sb, linelen + 80);
1327 pp_user_info(pp, "Author", sb, line + 7, encoding);
1328 }
1329 if (!memcmp(line, "committer ", 10) &&
1330 (pp->fmt == CMIT_FMT_FULL || pp->fmt == CMIT_FMT_FULLER)) {
1331 strbuf_grow(sb, linelen + 80);
1332 pp_user_info(pp, "Commit", sb, line + 10, encoding);
1333 }
1334 }
1335}
1336
1337void pp_title_line(const struct pretty_print_context *pp,
1338 const char **msg_p,
1339 struct strbuf *sb,
1340 const char *encoding,
1341 int need_8bit_cte)
1342{
1343 static const int max_length = 78; /* per rfc2047 */
1344 struct strbuf title;
1345
1346 strbuf_init(&title, 80);
1347 *msg_p = format_subject(&title, *msg_p,
1348 pp->preserve_subject ? "\n" : " ");
1349
1350 strbuf_grow(sb, title.len + 1024);
1351 if (pp->subject) {
1352 strbuf_addstr(sb, pp->subject);
1353 if (needs_rfc2047_encoding(title.buf, title.len, RFC2047_SUBJECT))
1354 add_rfc2047(sb, title.buf, title.len,
1355 encoding, RFC2047_SUBJECT);
1356 else
1357 strbuf_add_wrapped_bytes(sb, title.buf, title.len,
1358 -last_line_length(sb), 1, max_length);
1359 } else {
1360 strbuf_addbuf(sb, &title);
1361 }
1362 strbuf_addch(sb, '\n');
1363
1364 if (need_8bit_cte > 0) {
1365 const char *header_fmt =
1366 "MIME-Version: 1.0\n"
1367 "Content-Type: text/plain; charset=%s\n"
1368 "Content-Transfer-Encoding: 8bit\n";
1369 strbuf_addf(sb, header_fmt, encoding);
1370 }
1371 if (pp->after_subject) {
1372 strbuf_addstr(sb, pp->after_subject);
1373 }
1374 if (pp->fmt == CMIT_FMT_EMAIL) {
1375 strbuf_addch(sb, '\n');
1376 }
1377 strbuf_release(&title);
1378}
1379
1380void pp_remainder(const struct pretty_print_context *pp,
1381 const char **msg_p,
1382 struct strbuf *sb,
1383 int indent)
1384{
1385 int first = 1;
1386 for (;;) {
1387 const char *line = *msg_p;
1388 int linelen = get_one_line(line);
1389 *msg_p += linelen;
1390
1391 if (!linelen)
1392 break;
1393
1394 if (is_empty_line(line, &linelen)) {
1395 if (first)
1396 continue;
1397 if (pp->fmt == CMIT_FMT_SHORT)
1398 break;
1399 }
1400 first = 0;
1401
1402 strbuf_grow(sb, linelen + indent + 20);
1403 if (indent) {
1404 memset(sb->buf + sb->len, ' ', indent);
1405 strbuf_setlen(sb, sb->len + indent);
1406 }
1407 strbuf_add(sb, line, linelen);
1408 strbuf_addch(sb, '\n');
1409 }
1410}
1411
1412void pretty_print_commit(const struct pretty_print_context *pp,
1413 const struct commit *commit,
1414 struct strbuf *sb)
1415{
1416 unsigned long beginning_of_body;
1417 int indent = 4;
1418 const char *msg = commit->buffer;
1419 char *reencoded;
1420 const char *encoding;
1421 int need_8bit_cte = pp->need_8bit_cte;
1422
1423 if (pp->fmt == CMIT_FMT_USERFORMAT) {
1424 format_commit_message(commit, user_format, sb, pp);
1425 return;
1426 }
1427
1428 encoding = get_log_output_encoding();
1429 reencoded = logmsg_reencode(commit, encoding);
1430 if (reencoded) {
1431 msg = reencoded;
1432 }
1433
1434 if (pp->fmt == CMIT_FMT_ONELINE || pp->fmt == CMIT_FMT_EMAIL)
1435 indent = 0;
1436
1437 /*
1438 * We need to check and emit Content-type: to mark it
1439 * as 8-bit if we haven't done so.
1440 */
1441 if (pp->fmt == CMIT_FMT_EMAIL && need_8bit_cte == 0) {
1442 int i, ch, in_body;
1443
1444 for (in_body = i = 0; (ch = msg[i]); i++) {
1445 if (!in_body) {
1446 /* author could be non 7-bit ASCII but
1447 * the log may be so; skip over the
1448 * header part first.
1449 */
1450 if (ch == '\n' && msg[i+1] == '\n')
1451 in_body = 1;
1452 }
1453 else if (non_ascii(ch)) {
1454 need_8bit_cte = 1;
1455 break;
1456 }
1457 }
1458 }
1459
1460 pp_header(pp, encoding, commit, &msg, sb);
1461 if (pp->fmt != CMIT_FMT_ONELINE && !pp->subject) {
1462 strbuf_addch(sb, '\n');
1463 }
1464
1465 /* Skip excess blank lines at the beginning of body, if any... */
1466 msg = skip_empty_lines(msg);
1467
1468 /* These formats treat the title line specially. */
1469 if (pp->fmt == CMIT_FMT_ONELINE || pp->fmt == CMIT_FMT_EMAIL)
1470 pp_title_line(pp, &msg, sb, encoding, need_8bit_cte);
1471
1472 beginning_of_body = sb->len;
1473 if (pp->fmt != CMIT_FMT_ONELINE)
1474 pp_remainder(pp, &msg, sb, indent);
1475 strbuf_rtrim(sb);
1476
1477 /* Make sure there is an EOLN for the non-oneline case */
1478 if (pp->fmt != CMIT_FMT_ONELINE)
1479 strbuf_addch(sb, '\n');
1480
1481 /*
1482 * The caller may append additional body text in e-mail
1483 * format. Make sure we did not strip the blank line
1484 * between the header and the body.
1485 */
1486 if (pp->fmt == CMIT_FMT_EMAIL && sb->len <= beginning_of_body)
1487 strbuf_addch(sb, '\n');
1488
1489 free(reencoded);
1490}
1491
1492void pp_commit_easy(enum cmit_fmt fmt, const struct commit *commit,
1493 struct strbuf *sb)
1494{
1495 struct pretty_print_context pp = {0};
1496 pp.fmt = fmt;
1497 pretty_print_commit(&pp, commit, sb);
1498}