7 static char *user_format
;
9 void get_commit_format(const char *arg
, struct rev_info
*rev
)
12 static struct cmt_fmt_map
{
17 { "raw", 1, CMIT_FMT_RAW
},
18 { "medium", 1, CMIT_FMT_MEDIUM
},
19 { "short", 1, CMIT_FMT_SHORT
},
20 { "email", 1, CMIT_FMT_EMAIL
},
21 { "full", 5, CMIT_FMT_FULL
},
22 { "fuller", 5, CMIT_FMT_FULLER
},
23 { "oneline", 1, CMIT_FMT_ONELINE
},
26 rev
->use_terminator
= 0;
28 rev
->commit_format
= CMIT_FMT_DEFAULT
;
31 if (!prefixcmp(arg
, "format:") || !prefixcmp(arg
, "tformat:")) {
32 const char *cp
= strchr(arg
, ':') + 1;
34 user_format
= xstrdup(cp
);
36 rev
->use_terminator
= 1;
37 rev
->commit_format
= CMIT_FMT_USERFORMAT
;
40 for (i
= 0; i
< ARRAY_SIZE(cmt_fmts
); i
++) {
41 if (!strncmp(arg
, cmt_fmts
[i
].n
, cmt_fmts
[i
].cmp_len
) &&
42 !strncmp(arg
, cmt_fmts
[i
].n
, strlen(arg
))) {
43 if (cmt_fmts
[i
].v
== CMIT_FMT_ONELINE
)
44 rev
->use_terminator
= 1;
45 rev
->commit_format
= cmt_fmts
[i
].v
;
50 die("invalid --pretty format: %s", arg
);
54 * Generic support for pretty-printing the header
56 static int get_one_line(const char *msg
)
71 /* High bit set, or ISO-2022-INT */
75 return ((ch
& 0x80) || (ch
== 0x1b));
78 static int is_rfc2047_special(char ch
)
80 return (non_ascii(ch
) || (ch
== '=') || (ch
== '?') || (ch
== '_'));
83 static void add_rfc2047(struct strbuf
*sb
, const char *line
, int len
,
88 for (i
= 0; i
< len
; i
++) {
92 if ((i
+ 1 < len
) && (ch
== '=' && line
[i
+1] == '?'))
95 strbuf_add(sb
, line
, len
);
99 strbuf_grow(sb
, len
* 3 + strlen(encoding
) + 100);
100 strbuf_addf(sb
, "=?%s?q?", encoding
);
101 for (i
= last
= 0; i
< len
; i
++) {
102 unsigned ch
= line
[i
] & 0xFF;
104 * We encode ' ' using '=20' even though rfc2047
105 * allows using '_' for readability. Unfortunately,
106 * many programs do not understand this and just
107 * leave the underscore in place.
109 if (is_rfc2047_special(ch
) || ch
== ' ') {
110 strbuf_add(sb
, line
+ last
, i
- last
);
111 strbuf_addf(sb
, "=%02X", ch
);
115 strbuf_add(sb
, line
+ last
, len
- last
);
116 strbuf_addstr(sb
, "?=");
119 void pp_user_info(const char *what
, enum cmit_fmt fmt
, struct strbuf
*sb
,
120 const char *line
, enum date_mode dmode
,
121 const char *encoding
)
127 const char *filler
= " ";
129 if (fmt
== CMIT_FMT_ONELINE
)
131 date
= strchr(line
, '>');
134 namelen
= ++date
- line
;
135 time
= strtoul(date
, &date
, 10);
136 tz
= strtol(date
, NULL
, 10);
138 if (fmt
== CMIT_FMT_EMAIL
) {
139 char *name_tail
= strchr(line
, '<');
140 int display_name_length
;
143 while (line
< name_tail
&& isspace(name_tail
[-1]))
145 display_name_length
= name_tail
- line
;
147 strbuf_addstr(sb
, "From: ");
148 add_rfc2047(sb
, line
, display_name_length
, encoding
);
149 strbuf_add(sb
, name_tail
, namelen
- display_name_length
);
150 strbuf_addch(sb
, '\n');
152 strbuf_addf(sb
, "%s: %.*s%.*s\n", what
,
153 (fmt
== CMIT_FMT_FULLER
) ? 4 : 0,
154 filler
, namelen
, line
);
157 case CMIT_FMT_MEDIUM
:
158 strbuf_addf(sb
, "Date: %s\n", show_date(time
, tz
, dmode
));
161 strbuf_addf(sb
, "Date: %s\n", show_date(time
, tz
, DATE_RFC2822
));
163 case CMIT_FMT_FULLER
:
164 strbuf_addf(sb
, "%sDate: %s\n", what
, show_date(time
, tz
, dmode
));
172 static int is_empty_line(const char *line
, int *len_p
)
175 while (len
&& isspace(line
[len
-1]))
181 static void add_merge_info(enum cmit_fmt fmt
, struct strbuf
*sb
,
182 const struct commit
*commit
, int abbrev
)
184 struct commit_list
*parent
= commit
->parents
;
186 if ((fmt
== CMIT_FMT_ONELINE
) || (fmt
== CMIT_FMT_EMAIL
) ||
187 !parent
|| !parent
->next
)
190 strbuf_addstr(sb
, "Merge:");
193 struct commit
*p
= parent
->item
;
194 const char *hex
= NULL
;
197 hex
= find_unique_abbrev(p
->object
.sha1
, abbrev
);
199 hex
= sha1_to_hex(p
->object
.sha1
);
200 dots
= (abbrev
&& strlen(hex
) != 40) ? "..." : "";
201 parent
= parent
->next
;
203 strbuf_addf(sb
, " %s%s", hex
, dots
);
205 strbuf_addch(sb
, '\n');
208 static char *get_header(const struct commit
*commit
, const char *key
)
210 int key_len
= strlen(key
);
211 const char *line
= commit
->buffer
;
214 const char *eol
= strchr(line
, '\n'), *next
;
219 eol
= line
+ strlen(line
);
223 if (eol
- line
> key_len
&&
224 !strncmp(line
, key
, key_len
) &&
225 line
[key_len
] == ' ') {
226 return xmemdupz(line
+ key_len
+ 1, eol
- line
- key_len
- 1);
232 static char *replace_encoding_header(char *buf
, const char *encoding
)
238 /* guess if there is an encoding header before a \n\n */
239 while (strncmp(cp
, "encoding ", strlen("encoding "))) {
240 cp
= strchr(cp
, '\n');
241 if (!cp
|| *++cp
== '\n')
245 cp
= strchr(cp
, '\n');
247 return buf
; /* should not happen but be defensive */
248 len
= cp
+ 1 - (buf
+ start
);
250 strbuf_init(&tmp
, 0);
251 strbuf_attach(&tmp
, buf
, strlen(buf
), strlen(buf
) + 1);
252 if (is_encoding_utf8(encoding
)) {
253 /* we have re-coded to UTF-8; drop the header */
254 strbuf_remove(&tmp
, start
, len
);
256 /* just replaces XXXX in 'encoding XXXX\n' */
257 strbuf_splice(&tmp
, start
+ strlen("encoding "),
258 len
- strlen("encoding \n"),
259 encoding
, strlen(encoding
));
261 return strbuf_detach(&tmp
, NULL
);
264 static char *logmsg_reencode(const struct commit
*commit
,
265 const char *output_encoding
)
267 static const char *utf8
= "utf-8";
268 const char *use_encoding
;
272 if (!*output_encoding
)
274 encoding
= get_header(commit
, "encoding");
275 use_encoding
= encoding
? encoding
: utf8
;
276 if (!strcmp(use_encoding
, output_encoding
))
277 if (encoding
) /* we'll strip encoding header later */
278 out
= xstrdup(commit
->buffer
);
280 return NULL
; /* nothing to do */
282 out
= reencode_string(commit
->buffer
,
283 output_encoding
, use_encoding
);
285 out
= replace_encoding_header(out
, output_encoding
);
291 static size_t format_person_part(struct strbuf
*sb
, char part
,
292 const char *msg
, int len
)
294 /* currently all placeholders have same length */
295 const int placeholder_len
= 2;
296 int start
, end
, tz
= 0;
297 unsigned long date
= 0;
300 /* advance 'end' to point to email start delimiter */
301 for (end
= 0; end
< len
&& msg
[end
] != '<'; end
++)
305 * When end points at the '<' that we found, it should have
306 * matching '>' later, which means 'end' must be strictly
312 if (part
== 'n') { /* name */
313 while (end
> 0 && isspace(msg
[end
- 1]))
315 strbuf_add(sb
, msg
, end
);
316 return placeholder_len
;
318 start
= ++end
; /* save email start position */
320 /* advance 'end' to point to email end delimiter */
321 for ( ; end
< len
&& msg
[end
] != '>'; end
++)
327 if (part
== 'e') { /* email */
328 strbuf_add(sb
, msg
+ start
, end
- start
);
329 return placeholder_len
;
332 /* advance 'start' to point to date start delimiter */
333 for (start
= end
+ 1; start
< len
&& isspace(msg
[start
]); start
++)
337 date
= strtoul(msg
+ start
, &ep
, 10);
338 if (msg
+ start
== ep
)
341 if (part
== 't') { /* date, UNIX timestamp */
342 strbuf_add(sb
, msg
+ start
, ep
- (msg
+ start
));
343 return placeholder_len
;
347 for (start
= ep
- msg
+ 1; start
< len
&& isspace(msg
[start
]); start
++)
349 if (start
+ 1 < len
) {
350 tz
= strtoul(msg
+ start
+ 1, NULL
, 10);
351 if (msg
[start
] == '-')
357 strbuf_addstr(sb
, show_date(date
, tz
, DATE_NORMAL
));
358 return placeholder_len
;
359 case 'D': /* date, RFC2822 style */
360 strbuf_addstr(sb
, show_date(date
, tz
, DATE_RFC2822
));
361 return placeholder_len
;
362 case 'r': /* date, relative */
363 strbuf_addstr(sb
, show_date(date
, tz
, DATE_RELATIVE
));
364 return placeholder_len
;
365 case 'i': /* date, ISO 8601 */
366 strbuf_addstr(sb
, show_date(date
, tz
, DATE_ISO8601
));
367 return placeholder_len
;
372 * bogus commit, 'sb' cannot be updated, but we still need to
373 * compute a valid return value.
375 if (part
== 'n' || part
== 'e' || part
== 't' || part
== 'd'
376 || part
== 'D' || part
== 'r' || part
== 'i')
377 return placeholder_len
;
379 return 0; /* unknown placeholder */
387 struct format_commit_context
{
388 const struct commit
*commit
;
390 /* These offsets are relative to the start of the commit message. */
391 int commit_header_parsed
;
392 struct chunk subject
;
394 struct chunk committer
;
395 struct chunk encoding
;
398 /* The following ones are relative to the result struct strbuf. */
399 struct chunk abbrev_commit_hash
;
400 struct chunk abbrev_tree_hash
;
401 struct chunk abbrev_parent_hashes
;
404 static int add_again(struct strbuf
*sb
, struct chunk
*chunk
)
407 strbuf_adddup(sb
, chunk
->off
, chunk
->len
);
412 * We haven't seen this chunk before. Our caller is surely
413 * going to add it the hard way now. Remember the most likely
414 * start of the to-be-added chunk: the current end of the
417 chunk
->off
= sb
->len
;
421 static void parse_commit_header(struct format_commit_context
*context
)
423 const char *msg
= context
->commit
->buffer
;
425 enum { HEADER
, SUBJECT
, BODY
} state
;
427 for (i
= 0, state
= HEADER
; msg
[i
] && state
< BODY
; i
++) {
429 for (eol
= i
; msg
[eol
] && msg
[eol
] != '\n'; eol
++)
432 if (state
== SUBJECT
) {
433 context
->subject
.off
= i
;
434 context
->subject
.len
= eol
- i
;
439 /* strip empty lines */
440 while (msg
[eol
] == '\n' && msg
[eol
+ 1] == '\n')
442 } else if (!prefixcmp(msg
+ i
, "author ")) {
443 context
->author
.off
= i
+ 7;
444 context
->author
.len
= eol
- i
- 7;
445 } else if (!prefixcmp(msg
+ i
, "committer ")) {
446 context
->committer
.off
= i
+ 10;
447 context
->committer
.len
= eol
- i
- 10;
448 } else if (!prefixcmp(msg
+ i
, "encoding ")) {
449 context
->encoding
.off
= i
+ 9;
450 context
->encoding
.len
= eol
- i
- 9;
456 context
->body_off
= i
;
457 context
->commit_header_parsed
= 1;
460 static size_t format_commit_item(struct strbuf
*sb
, const char *placeholder
,
463 struct format_commit_context
*c
= context
;
464 const struct commit
*commit
= c
->commit
;
465 const char *msg
= commit
->buffer
;
466 struct commit_list
*p
;
469 /* these are independent of the commit */
470 switch (placeholder
[0]) {
472 if (!prefixcmp(placeholder
+ 1, "red")) {
473 strbuf_addstr(sb
, "\033[31m");
475 } else if (!prefixcmp(placeholder
+ 1, "green")) {
476 strbuf_addstr(sb
, "\033[32m");
478 } else if (!prefixcmp(placeholder
+ 1, "blue")) {
479 strbuf_addstr(sb
, "\033[34m");
481 } else if (!prefixcmp(placeholder
+ 1, "reset")) {
482 strbuf_addstr(sb
, "\033[m");
486 case 'n': /* newline */
487 strbuf_addch(sb
, '\n');
490 /* %x00 == NUL, %x0a == LF, etc. */
491 if (0 <= (h1
= hexval_table
[0xff & placeholder
[1]]) &&
493 0 <= (h2
= hexval_table
[0xff & placeholder
[2]]) &&
495 strbuf_addch(sb
, (h1
<<4)|h2
);
501 /* these depend on the commit */
502 if (!commit
->object
.parsed
)
503 parse_object(commit
->object
.sha1
);
505 switch (placeholder
[0]) {
506 case 'H': /* commit hash */
507 strbuf_addstr(sb
, sha1_to_hex(commit
->object
.sha1
));
509 case 'h': /* abbreviated commit hash */
510 if (add_again(sb
, &c
->abbrev_commit_hash
))
512 strbuf_addstr(sb
, find_unique_abbrev(commit
->object
.sha1
,
514 c
->abbrev_commit_hash
.len
= sb
->len
- c
->abbrev_commit_hash
.off
;
516 case 'T': /* tree hash */
517 strbuf_addstr(sb
, sha1_to_hex(commit
->tree
->object
.sha1
));
519 case 't': /* abbreviated tree hash */
520 if (add_again(sb
, &c
->abbrev_tree_hash
))
522 strbuf_addstr(sb
, find_unique_abbrev(commit
->tree
->object
.sha1
,
524 c
->abbrev_tree_hash
.len
= sb
->len
- c
->abbrev_tree_hash
.off
;
526 case 'P': /* parent hashes */
527 for (p
= commit
->parents
; p
; p
= p
->next
) {
528 if (p
!= commit
->parents
)
529 strbuf_addch(sb
, ' ');
530 strbuf_addstr(sb
, sha1_to_hex(p
->item
->object
.sha1
));
533 case 'p': /* abbreviated parent hashes */
534 if (add_again(sb
, &c
->abbrev_parent_hashes
))
536 for (p
= commit
->parents
; p
; p
= p
->next
) {
537 if (p
!= commit
->parents
)
538 strbuf_addch(sb
, ' ');
539 strbuf_addstr(sb
, find_unique_abbrev(
540 p
->item
->object
.sha1
, DEFAULT_ABBREV
));
542 c
->abbrev_parent_hashes
.len
= sb
->len
-
543 c
->abbrev_parent_hashes
.off
;
545 case 'm': /* left/right/bottom */
546 strbuf_addch(sb
, (commit
->object
.flags
& BOUNDARY
)
548 : (commit
->object
.flags
& SYMMETRIC_LEFT
)
554 /* For the rest we have to parse the commit header. */
555 if (!c
->commit_header_parsed
)
556 parse_commit_header(c
);
558 switch (placeholder
[0]) {
559 case 's': /* subject */
560 strbuf_add(sb
, msg
+ c
->subject
.off
, c
->subject
.len
);
562 case 'a': /* author ... */
563 return format_person_part(sb
, placeholder
[1],
564 msg
+ c
->author
.off
, c
->author
.len
);
565 case 'c': /* committer ... */
566 return format_person_part(sb
, placeholder
[1],
567 msg
+ c
->committer
.off
, c
->committer
.len
);
568 case 'e': /* encoding */
569 strbuf_add(sb
, msg
+ c
->encoding
.off
, c
->encoding
.len
);
572 strbuf_addstr(sb
, msg
+ c
->body_off
);
575 return 0; /* unknown placeholder */
578 void format_commit_message(const struct commit
*commit
,
579 const void *format
, struct strbuf
*sb
)
581 struct format_commit_context context
;
583 memset(&context
, 0, sizeof(context
));
584 context
.commit
= commit
;
585 strbuf_expand(sb
, format
, format_commit_item
, &context
);
588 static void pp_header(enum cmit_fmt fmt
,
590 enum date_mode dmode
,
591 const char *encoding
,
592 const struct commit
*commit
,
596 int parents_shown
= 0;
599 const char *line
= *msg_p
;
600 int linelen
= get_one_line(*msg_p
);
610 if (fmt
== CMIT_FMT_RAW
) {
611 strbuf_add(sb
, line
, linelen
);
615 if (!memcmp(line
, "parent ", 7)) {
617 die("bad parent line in commit");
621 if (!parents_shown
) {
622 struct commit_list
*parent
;
624 for (parent
= commit
->parents
, num
= 0;
626 parent
= parent
->next
, num
++)
628 /* with enough slop */
629 strbuf_grow(sb
, num
* 50 + 20);
630 add_merge_info(fmt
, sb
, commit
, abbrev
);
635 * MEDIUM == DEFAULT shows only author with dates.
636 * FULL shows both authors but not dates.
637 * FULLER shows both authors and dates.
639 if (!memcmp(line
, "author ", 7)) {
640 strbuf_grow(sb
, linelen
+ 80);
641 pp_user_info("Author", fmt
, sb
, line
+ 7, dmode
, encoding
);
643 if (!memcmp(line
, "committer ", 10) &&
644 (fmt
== CMIT_FMT_FULL
|| fmt
== CMIT_FMT_FULLER
)) {
645 strbuf_grow(sb
, linelen
+ 80);
646 pp_user_info("Commit", fmt
, sb
, line
+ 10, dmode
, encoding
);
651 void pp_title_line(enum cmit_fmt fmt
,
655 const char *after_subject
,
656 const char *encoding
,
661 strbuf_init(&title
, 80);
664 const char *line
= *msg_p
;
665 int linelen
= get_one_line(line
);
668 if (!linelen
|| is_empty_line(line
, &linelen
))
671 strbuf_grow(&title
, linelen
+ 2);
673 if (fmt
== CMIT_FMT_EMAIL
) {
674 strbuf_addch(&title
, '\n');
676 strbuf_addch(&title
, ' ');
678 strbuf_add(&title
, line
, linelen
);
681 strbuf_grow(sb
, title
.len
+ 1024);
683 strbuf_addstr(sb
, subject
);
684 add_rfc2047(sb
, title
.buf
, title
.len
, encoding
);
686 strbuf_addbuf(sb
, &title
);
688 strbuf_addch(sb
, '\n');
690 if (need_8bit_cte
> 0) {
691 const char *header_fmt
=
692 "MIME-Version: 1.0\n"
693 "Content-Type: text/plain; charset=%s\n"
694 "Content-Transfer-Encoding: 8bit\n";
695 strbuf_addf(sb
, header_fmt
, encoding
);
698 strbuf_addstr(sb
, after_subject
);
700 if (fmt
== CMIT_FMT_EMAIL
) {
701 strbuf_addch(sb
, '\n');
703 strbuf_release(&title
);
706 void pp_remainder(enum cmit_fmt fmt
,
713 const char *line
= *msg_p
;
714 int linelen
= get_one_line(line
);
720 if (is_empty_line(line
, &linelen
)) {
723 if (fmt
== CMIT_FMT_SHORT
)
728 strbuf_grow(sb
, linelen
+ indent
+ 20);
730 memset(sb
->buf
+ sb
->len
, ' ', indent
);
731 strbuf_setlen(sb
, sb
->len
+ indent
);
733 strbuf_add(sb
, line
, linelen
);
734 strbuf_addch(sb
, '\n');
738 void pretty_print_commit(enum cmit_fmt fmt
, const struct commit
*commit
,
739 struct strbuf
*sb
, int abbrev
,
740 const char *subject
, const char *after_subject
,
741 enum date_mode dmode
, int need_8bit_cte
)
743 unsigned long beginning_of_body
;
745 const char *msg
= commit
->buffer
;
747 const char *encoding
;
749 if (fmt
== CMIT_FMT_USERFORMAT
) {
750 format_commit_message(commit
, user_format
, sb
);
754 encoding
= (git_log_output_encoding
755 ? git_log_output_encoding
756 : git_commit_encoding
);
759 reencoded
= logmsg_reencode(commit
, encoding
);
764 if (fmt
== CMIT_FMT_ONELINE
|| fmt
== CMIT_FMT_EMAIL
)
768 * We need to check and emit Content-type: to mark it
769 * as 8-bit if we haven't done so.
771 if (fmt
== CMIT_FMT_EMAIL
&& need_8bit_cte
== 0) {
774 for (in_body
= i
= 0; (ch
= msg
[i
]); i
++) {
776 /* author could be non 7-bit ASCII but
777 * the log may be so; skip over the
780 if (ch
== '\n' && msg
[i
+1] == '\n')
783 else if (non_ascii(ch
)) {
790 pp_header(fmt
, abbrev
, dmode
, encoding
, commit
, &msg
, sb
);
791 if (fmt
!= CMIT_FMT_ONELINE
&& !subject
) {
792 strbuf_addch(sb
, '\n');
795 /* Skip excess blank lines at the beginning of body, if any... */
797 int linelen
= get_one_line(msg
);
801 if (!is_empty_line(msg
, &ll
))
806 /* These formats treat the title line specially. */
807 if (fmt
== CMIT_FMT_ONELINE
|| fmt
== CMIT_FMT_EMAIL
)
808 pp_title_line(fmt
, &msg
, sb
, subject
,
809 after_subject
, encoding
, need_8bit_cte
);
811 beginning_of_body
= sb
->len
;
812 if (fmt
!= CMIT_FMT_ONELINE
)
813 pp_remainder(fmt
, &msg
, sb
, indent
);
816 /* Make sure there is an EOLN for the non-oneline case */
817 if (fmt
!= CMIT_FMT_ONELINE
)
818 strbuf_addch(sb
, '\n');
821 * The caller may append additional body text in e-mail
822 * format. Make sure we did not strip the blank line
823 * between the header and the body.
825 if (fmt
== CMIT_FMT_EMAIL
&& sb
->len
<= beginning_of_body
)
826 strbuf_addch(sb
, '\n');