6 #include "string-list.h"
11 #include "reflog-walk.h"
12 #include "gpg-interface.h"
14 static char *user_format
;
15 static struct cmt_fmt_map
{
20 const char *user_format
;
22 static size_t builtin_formats_len
;
23 static size_t commit_formats_len
;
24 static size_t commit_formats_alloc
;
25 static struct cmt_fmt_map
*find_commit_format(const char *sought
);
27 static void save_user_format(struct rev_info
*rev
, const char *cp
, int is_tformat
)
30 user_format
= xstrdup(cp
);
32 rev
->use_terminator
= 1;
33 rev
->commit_format
= CMIT_FMT_USERFORMAT
;
36 static int git_pretty_formats_config(const char *var
, const char *value
, void *cb
)
38 struct cmt_fmt_map
*commit_format
= NULL
;
43 if (!skip_prefix(var
, "pretty.", &name
))
46 for (i
= 0; i
< builtin_formats_len
; i
++) {
47 if (!strcmp(commit_formats
[i
].name
, name
))
51 for (i
= builtin_formats_len
; i
< commit_formats_len
; i
++) {
52 if (!strcmp(commit_formats
[i
].name
, name
)) {
53 commit_format
= &commit_formats
[i
];
59 ALLOC_GROW(commit_formats
, commit_formats_len
+1,
60 commit_formats_alloc
);
61 commit_format
= &commit_formats
[commit_formats_len
];
62 memset(commit_format
, 0, sizeof(*commit_format
));
66 commit_format
->name
= xstrdup(name
);
67 commit_format
->format
= CMIT_FMT_USERFORMAT
;
68 git_config_string(&fmt
, var
, value
);
69 if (starts_with(fmt
, "format:") || starts_with(fmt
, "tformat:")) {
70 commit_format
->is_tformat
= fmt
[0] == 't';
71 fmt
= strchr(fmt
, ':') + 1;
72 } else if (strchr(fmt
, '%'))
73 commit_format
->is_tformat
= 1;
75 commit_format
->is_alias
= 1;
76 commit_format
->user_format
= fmt
;
81 static void setup_commit_formats(void)
83 struct cmt_fmt_map builtin_formats
[] = {
84 { "raw", CMIT_FMT_RAW
, 0 },
85 { "medium", CMIT_FMT_MEDIUM
, 0 },
86 { "short", CMIT_FMT_SHORT
, 0 },
87 { "email", CMIT_FMT_EMAIL
, 0 },
88 { "fuller", CMIT_FMT_FULLER
, 0 },
89 { "full", CMIT_FMT_FULL
, 0 },
90 { "oneline", CMIT_FMT_ONELINE
, 1 }
92 commit_formats_len
= ARRAY_SIZE(builtin_formats
);
93 builtin_formats_len
= commit_formats_len
;
94 ALLOC_GROW(commit_formats
, commit_formats_len
, commit_formats_alloc
);
95 memcpy(commit_formats
, builtin_formats
,
96 sizeof(*builtin_formats
)*ARRAY_SIZE(builtin_formats
));
98 git_config(git_pretty_formats_config
, NULL
);
101 static struct cmt_fmt_map
*find_commit_format_recursive(const char *sought
,
102 const char *original
,
103 int num_redirections
)
105 struct cmt_fmt_map
*found
= NULL
;
106 size_t found_match_len
= 0;
109 if (num_redirections
>= commit_formats_len
)
110 die("invalid --pretty format: "
111 "'%s' references an alias which points to itself",
114 for (i
= 0; i
< commit_formats_len
; i
++) {
117 if (!starts_with(commit_formats
[i
].name
, sought
))
120 match_len
= strlen(commit_formats
[i
].name
);
121 if (found
== NULL
|| found_match_len
> match_len
) {
122 found
= &commit_formats
[i
];
123 found_match_len
= match_len
;
127 if (found
&& found
->is_alias
) {
128 found
= find_commit_format_recursive(found
->user_format
,
136 static struct cmt_fmt_map
*find_commit_format(const char *sought
)
139 setup_commit_formats();
141 return find_commit_format_recursive(sought
, sought
, 0);
144 void get_commit_format(const char *arg
, struct rev_info
*rev
)
146 struct cmt_fmt_map
*commit_format
;
148 rev
->use_terminator
= 0;
150 rev
->commit_format
= CMIT_FMT_DEFAULT
;
153 if (starts_with(arg
, "format:") || starts_with(arg
, "tformat:")) {
154 save_user_format(rev
, strchr(arg
, ':') + 1, arg
[0] == 't');
158 if (strchr(arg
, '%')) {
159 save_user_format(rev
, arg
, 1);
163 commit_format
= find_commit_format(arg
);
165 die("invalid --pretty format: %s", arg
);
167 rev
->commit_format
= commit_format
->format
;
168 rev
->use_terminator
= commit_format
->is_tformat
;
169 if (commit_format
->format
== CMIT_FMT_USERFORMAT
) {
170 save_user_format(rev
, commit_format
->user_format
,
171 commit_format
->is_tformat
);
176 * Generic support for pretty-printing the header
178 static int get_one_line(const char *msg
)
193 /* High bit set, or ISO-2022-INT */
194 static int non_ascii(int ch
)
196 return !isascii(ch
) || ch
== '\033';
199 int has_non_ascii(const char *s
)
204 while ((ch
= *s
++) != '\0') {
211 static int is_rfc822_special(char ch
)
233 static int needs_rfc822_quoting(const char *s
, int len
)
236 for (i
= 0; i
< len
; i
++)
237 if (is_rfc822_special(s
[i
]))
242 static int last_line_length(struct strbuf
*sb
)
246 /* How many bytes are already used on the last line? */
247 for (i
= sb
->len
- 1; i
>= 0; i
--)
248 if (sb
->buf
[i
] == '\n')
250 return sb
->len
- (i
+ 1);
253 static void add_rfc822_quoted(struct strbuf
*out
, const char *s
, int len
)
257 /* just a guess, we may have to also backslash-quote */
258 strbuf_grow(out
, len
+ 2);
260 strbuf_addch(out
, '"');
261 for (i
= 0; i
< len
; i
++) {
265 strbuf_addch(out
, '\\');
268 strbuf_addch(out
, s
[i
]);
271 strbuf_addch(out
, '"');
279 static int is_rfc2047_special(char ch
, enum rfc2047_type type
)
282 * rfc2047, section 4.2:
284 * 8-bit values which correspond to printable ASCII characters other
285 * than "=", "?", and "_" (underscore), MAY be represented as those
286 * characters. (But see section 5 for restrictions.) In
287 * particular, SPACE and TAB MUST NOT be represented as themselves
288 * within encoded words.
292 * rule out non-ASCII characters and non-printable characters (the
293 * non-ASCII check should be redundant as isprint() is not localized
294 * and only knows about ASCII, but be defensive about that)
296 if (non_ascii(ch
) || !isprint(ch
))
300 * rule out special printable characters (' ' should be the only
301 * whitespace character considered printable, but be defensive and use
304 if (isspace(ch
) || ch
== '=' || ch
== '?' || ch
== '_')
308 * rfc2047, section 5.3:
310 * As a replacement for a 'word' entity within a 'phrase', for example,
311 * one that precedes an address in a From, To, or Cc header. The ABNF
312 * definition for 'phrase' from RFC 822 thus becomes:
314 * phrase = 1*( encoded-word / word )
316 * In this case the set of characters that may be used in a "Q"-encoded
317 * 'encoded-word' is restricted to: <upper and lower case ASCII
318 * letters, decimal digits, "!", "*", "+", "-", "/", "=", and "_"
319 * (underscore, ASCII 95.)>. An 'encoded-word' that appears within a
320 * 'phrase' MUST be separated from any adjacent 'word', 'text' or
321 * 'special' by 'linear-white-space'.
324 if (type
!= RFC2047_ADDRESS
)
327 /* '=' and '_' are special cases and have been checked above */
328 return !(isalnum(ch
) || ch
== '!' || ch
== '*' || ch
== '+' || ch
== '-' || ch
== '/');
331 static int needs_rfc2047_encoding(const char *line
, int len
,
332 enum rfc2047_type type
)
336 for (i
= 0; i
< len
; i
++) {
338 if (non_ascii(ch
) || ch
== '\n')
340 if ((i
+ 1 < len
) && (ch
== '=' && line
[i
+1] == '?'))
347 static void add_rfc2047(struct strbuf
*sb
, const char *line
, size_t len
,
348 const char *encoding
, enum rfc2047_type type
)
350 static const int max_encoded_length
= 76; /* per rfc2047 */
352 int line_len
= last_line_length(sb
);
354 strbuf_grow(sb
, len
* 3 + strlen(encoding
) + 100);
355 strbuf_addf(sb
, "=?%s?q?", encoding
);
356 line_len
+= strlen(encoding
) + 5; /* 5 for =??q? */
360 * RFC 2047, section 5 (3):
362 * Each 'encoded-word' MUST represent an integral number of
363 * characters. A multi-octet character may not be split across
364 * adjacent 'encoded- word's.
366 const unsigned char *p
= (const unsigned char *)line
;
367 int chrlen
= mbs_chrlen(&line
, &len
, encoding
);
368 int is_special
= (chrlen
> 1) || is_rfc2047_special(*p
, type
);
370 /* "=%02X" * chrlen, or the byte itself */
371 const char *encoded_fmt
= is_special
? "=%02X" : "%c";
372 int encoded_len
= is_special
? 3 * chrlen
: 1;
375 * According to RFC 2047, we could encode the special character
376 * ' ' (space) with '_' (underscore) for readability. But many
377 * programs do not understand this and just leave the
378 * underscore in place. Thus, we do nothing special here, which
379 * causes ' ' to be encoded as '=20', avoiding this problem.
382 if (line_len
+ encoded_len
+ 2 > max_encoded_length
) {
383 /* It won't fit with trailing "?=" --- break the line */
384 strbuf_addf(sb
, "?=\n =?%s?q?", encoding
);
385 line_len
= strlen(encoding
) + 5 + 1; /* =??q? plus SP */
388 for (i
= 0; i
< chrlen
; i
++)
389 strbuf_addf(sb
, encoded_fmt
, p
[i
]);
390 line_len
+= encoded_len
;
392 strbuf_addstr(sb
, "?=");
395 const char *show_ident_date(const struct ident_split
*ident
,
398 unsigned long date
= 0;
401 if (ident
->date_begin
&& ident
->date_end
)
402 date
= strtoul(ident
->date_begin
, NULL
, 10);
403 if (date_overflows(date
))
406 if (ident
->tz_begin
&& ident
->tz_end
)
407 tz
= strtol(ident
->tz_begin
, NULL
, 10);
408 if (tz
>= INT_MAX
|| tz
<= INT_MIN
)
411 return show_date(date
, tz
, mode
);
414 void pp_user_info(struct pretty_print_context
*pp
,
415 const char *what
, struct strbuf
*sb
,
416 const char *line
, const char *encoding
)
418 struct ident_split ident
;
420 const char *mailbuf
, *namebuf
;
421 size_t namelen
, maillen
;
422 int max_length
= 78; /* per rfc2822 */
424 if (pp
->fmt
== CMIT_FMT_ONELINE
)
427 line_end
= strchrnul(line
, '\n');
428 if (split_ident_line(&ident
, line
, line_end
- line
))
431 mailbuf
= ident
.mail_begin
;
432 maillen
= ident
.mail_end
- ident
.mail_begin
;
433 namebuf
= ident
.name_begin
;
434 namelen
= ident
.name_end
- ident
.name_begin
;
437 map_user(pp
->mailmap
, &mailbuf
, &maillen
, &namebuf
, &namelen
);
439 if (pp
->fmt
== CMIT_FMT_EMAIL
) {
440 if (pp
->from_ident
&& ident_cmp(pp
->from_ident
, &ident
)) {
441 struct strbuf buf
= STRBUF_INIT
;
443 strbuf_addstr(&buf
, "From: ");
444 strbuf_add(&buf
, namebuf
, namelen
);
445 strbuf_addstr(&buf
, " <");
446 strbuf_add(&buf
, mailbuf
, maillen
);
447 strbuf_addstr(&buf
, ">\n");
448 string_list_append(&pp
->in_body_headers
,
449 strbuf_detach(&buf
, NULL
));
451 mailbuf
= pp
->from_ident
->mail_begin
;
452 maillen
= pp
->from_ident
->mail_end
- mailbuf
;
453 namebuf
= pp
->from_ident
->name_begin
;
454 namelen
= pp
->from_ident
->name_end
- namebuf
;
457 strbuf_addstr(sb
, "From: ");
458 if (needs_rfc2047_encoding(namebuf
, namelen
, RFC2047_ADDRESS
)) {
459 add_rfc2047(sb
, namebuf
, namelen
,
460 encoding
, RFC2047_ADDRESS
);
461 max_length
= 76; /* per rfc2047 */
462 } else if (needs_rfc822_quoting(namebuf
, namelen
)) {
463 struct strbuf quoted
= STRBUF_INIT
;
464 add_rfc822_quoted("ed
, namebuf
, namelen
);
465 strbuf_add_wrapped_bytes(sb
, quoted
.buf
, quoted
.len
,
467 strbuf_release("ed
);
469 strbuf_add_wrapped_bytes(sb
, namebuf
, namelen
,
474 last_line_length(sb
) + strlen(" <") + maillen
+ strlen(">"))
475 strbuf_addch(sb
, '\n');
476 strbuf_addf(sb
, " <%.*s>\n", (int)maillen
, mailbuf
);
478 strbuf_addf(sb
, "%s: %.*s%.*s <%.*s>\n", what
,
479 (pp
->fmt
== CMIT_FMT_FULLER
) ? 4 : 0, " ",
480 (int)namelen
, namebuf
, (int)maillen
, mailbuf
);
484 case CMIT_FMT_MEDIUM
:
485 strbuf_addf(sb
, "Date: %s\n",
486 show_ident_date(&ident
, pp
->date_mode
));
489 strbuf_addf(sb
, "Date: %s\n",
490 show_ident_date(&ident
, DATE_RFC2822
));
492 case CMIT_FMT_FULLER
:
493 strbuf_addf(sb
, "%sDate: %s\n", what
,
494 show_ident_date(&ident
, pp
->date_mode
));
502 static int is_empty_line(const char *line
, int *len_p
)
505 while (len
&& isspace(line
[len
- 1]))
511 static const char *skip_empty_lines(const char *msg
)
514 int linelen
= get_one_line(msg
);
518 if (!is_empty_line(msg
, &ll
))
525 static void add_merge_info(const struct pretty_print_context
*pp
,
526 struct strbuf
*sb
, const struct commit
*commit
)
528 struct commit_list
*parent
= commit
->parents
;
530 if ((pp
->fmt
== CMIT_FMT_ONELINE
) || (pp
->fmt
== CMIT_FMT_EMAIL
) ||
531 !parent
|| !parent
->next
)
534 strbuf_addstr(sb
, "Merge:");
537 struct commit
*p
= parent
->item
;
538 const char *hex
= NULL
;
540 hex
= find_unique_abbrev(p
->object
.sha1
, pp
->abbrev
);
542 hex
= sha1_to_hex(p
->object
.sha1
);
543 parent
= parent
->next
;
545 strbuf_addf(sb
, " %s", hex
);
547 strbuf_addch(sb
, '\n');
550 static char *get_header(const struct commit
*commit
, const char *msg
,
553 int key_len
= strlen(key
);
554 const char *line
= msg
;
557 const char *eol
= strchrnul(line
, '\n'), *next
;
562 warning("malformed commit (header is missing newline): %s",
563 sha1_to_hex(commit
->object
.sha1
));
567 if (eol
- line
> key_len
&&
568 !strncmp(line
, key
, key_len
) &&
569 line
[key_len
] == ' ') {
570 return xmemdupz(line
+ key_len
+ 1, eol
- line
- key_len
- 1);
577 static char *replace_encoding_header(char *buf
, const char *encoding
)
579 struct strbuf tmp
= STRBUF_INIT
;
583 /* guess if there is an encoding header before a \n\n */
584 while (strncmp(cp
, "encoding ", strlen("encoding "))) {
585 cp
= strchr(cp
, '\n');
586 if (!cp
|| *++cp
== '\n')
590 cp
= strchr(cp
, '\n');
592 return buf
; /* should not happen but be defensive */
593 len
= cp
+ 1 - (buf
+ start
);
595 strbuf_attach(&tmp
, buf
, strlen(buf
), strlen(buf
) + 1);
596 if (is_encoding_utf8(encoding
)) {
597 /* we have re-coded to UTF-8; drop the header */
598 strbuf_remove(&tmp
, start
, len
);
600 /* just replaces XXXX in 'encoding XXXX\n' */
601 strbuf_splice(&tmp
, start
+ strlen("encoding "),
602 len
- strlen("encoding \n"),
603 encoding
, strlen(encoding
));
605 return strbuf_detach(&tmp
, NULL
);
608 const char *logmsg_reencode(const struct commit
*commit
,
609 char **commit_encoding
,
610 const char *output_encoding
)
612 static const char *utf8
= "UTF-8";
613 const char *use_encoding
;
615 const char *msg
= get_commit_buffer(commit
, NULL
);
618 if (!output_encoding
|| !*output_encoding
) {
621 get_header(commit
, msg
, "encoding");
624 encoding
= get_header(commit
, msg
, "encoding");
626 *commit_encoding
= encoding
;
627 use_encoding
= encoding
? encoding
: utf8
;
628 if (same_encoding(use_encoding
, output_encoding
)) {
630 * No encoding work to be done. If we have no encoding header
631 * at all, then there's nothing to do, and we can return the
632 * message verbatim (whether newly allocated or not).
638 * Otherwise, we still want to munge the encoding header in the
639 * result, which will be done by modifying the buffer. If we
640 * are using a fresh copy, we can reuse it. But if we are using
641 * the cached copy from get_commit_buffer, we need to duplicate it
642 * to avoid munging the cached copy.
644 if (msg
== get_cached_commit_buffer(commit
, NULL
))
651 * There's actual encoding work to do. Do the reencoding, which
652 * still leaves the header to be replaced in the next step. At
653 * this point, we are done with msg. If we allocated a fresh
654 * copy, we can free it.
656 out
= reencode_string(msg
, output_encoding
, use_encoding
);
658 unuse_commit_buffer(commit
, msg
);
662 * This replacement actually consumes the buffer we hand it, so we do
663 * not have to worry about freeing the old "out" here.
666 out
= replace_encoding_header(out
, output_encoding
);
668 if (!commit_encoding
)
671 * If the re-encoding failed, out might be NULL here; in that
672 * case we just return the commit message verbatim.
674 return out
? out
: msg
;
677 static int mailmap_name(const char **email
, size_t *email_len
,
678 const char **name
, size_t *name_len
)
680 static struct string_list
*mail_map
;
682 mail_map
= xcalloc(1, sizeof(*mail_map
));
683 read_mailmap(mail_map
, NULL
);
685 return mail_map
->nr
&& map_user(mail_map
, email
, email_len
, name
, name_len
);
688 static size_t format_person_part(struct strbuf
*sb
, char part
,
689 const char *msg
, int len
, enum date_mode dmode
)
691 /* currently all placeholders have same length */
692 const int placeholder_len
= 2;
693 struct ident_split s
;
694 const char *name
, *mail
;
695 size_t maillen
, namelen
;
697 if (split_ident_line(&s
, msg
, len
) < 0)
701 namelen
= s
.name_end
- s
.name_begin
;
703 maillen
= s
.mail_end
- s
.mail_begin
;
705 if (part
== 'N' || part
== 'E') /* mailmap lookup */
706 mailmap_name(&mail
, &maillen
, &name
, &namelen
);
707 if (part
== 'n' || part
== 'N') { /* name */
708 strbuf_add(sb
, name
, namelen
);
709 return placeholder_len
;
711 if (part
== 'e' || part
== 'E') { /* email */
712 strbuf_add(sb
, mail
, maillen
);
713 return placeholder_len
;
719 if (part
== 't') { /* date, UNIX timestamp */
720 strbuf_add(sb
, s
.date_begin
, s
.date_end
- s
.date_begin
);
721 return placeholder_len
;
726 strbuf_addstr(sb
, show_ident_date(&s
, dmode
));
727 return placeholder_len
;
728 case 'D': /* date, RFC2822 style */
729 strbuf_addstr(sb
, show_ident_date(&s
, DATE_RFC2822
));
730 return placeholder_len
;
731 case 'r': /* date, relative */
732 strbuf_addstr(sb
, show_ident_date(&s
, DATE_RELATIVE
));
733 return placeholder_len
;
734 case 'i': /* date, ISO 8601 */
735 strbuf_addstr(sb
, show_ident_date(&s
, DATE_ISO8601
));
736 return placeholder_len
;
741 * reading from either a bogus commit, or a reflog entry with
742 * %gn, %ge, etc.; 'sb' cannot be updated, but we still need
743 * to compute a valid return value.
745 if (part
== 'n' || part
== 'e' || part
== 't' || part
== 'd'
746 || part
== 'D' || part
== 'r' || part
== 'i')
747 return placeholder_len
;
749 return 0; /* unknown placeholder */
761 flush_left_and_steal
,
772 struct format_commit_context
{
773 const struct commit
*commit
;
774 const struct pretty_print_context
*pretty_ctx
;
775 unsigned commit_header_parsed
:1;
776 unsigned commit_message_parsed
:1;
777 struct signature_check signature_check
;
778 enum flush_type flush_type
;
779 enum trunc_type truncate
;
781 char *commit_encoding
;
782 size_t width
, indent1
, indent2
;
786 /* These offsets are relative to the start of the commit message. */
788 struct chunk committer
;
793 /* The following ones are relative to the result struct strbuf. */
794 struct chunk abbrev_commit_hash
;
795 struct chunk abbrev_tree_hash
;
796 struct chunk abbrev_parent_hashes
;
800 static int add_again(struct strbuf
*sb
, struct chunk
*chunk
)
803 strbuf_adddup(sb
, chunk
->off
, chunk
->len
);
808 * We haven't seen this chunk before. Our caller is surely
809 * going to add it the hard way now. Remember the most likely
810 * start of the to-be-added chunk: the current end of the
813 chunk
->off
= sb
->len
;
817 static void parse_commit_header(struct format_commit_context
*context
)
819 const char *msg
= context
->message
;
822 for (i
= 0; msg
[i
]; i
++) {
824 for (eol
= i
; msg
[eol
] && msg
[eol
] != '\n'; eol
++)
829 } else if (starts_with(msg
+ i
, "author ")) {
830 context
->author
.off
= i
+ 7;
831 context
->author
.len
= eol
- i
- 7;
832 } else if (starts_with(msg
+ i
, "committer ")) {
833 context
->committer
.off
= i
+ 10;
834 context
->committer
.len
= eol
- i
- 10;
838 context
->message_off
= i
;
839 context
->commit_header_parsed
= 1;
842 static int istitlechar(char c
)
844 return (c
>= 'a' && c
<= 'z') || (c
>= 'A' && c
<= 'Z') ||
845 (c
>= '0' && c
<= '9') || c
== '.' || c
== '_';
848 static void format_sanitized_subject(struct strbuf
*sb
, const char *msg
)
851 size_t start_len
= sb
->len
;
854 for (; *msg
&& *msg
!= '\n'; msg
++) {
855 if (istitlechar(*msg
)) {
857 strbuf_addch(sb
, '-');
859 strbuf_addch(sb
, *msg
);
861 while (*(msg
+1) == '.')
867 /* trim any trailing '.' or '-' characters */
869 while (sb
->len
- trimlen
> start_len
&&
870 (sb
->buf
[sb
->len
- 1 - trimlen
] == '.'
871 || sb
->buf
[sb
->len
- 1 - trimlen
] == '-'))
873 strbuf_remove(sb
, sb
->len
- trimlen
, trimlen
);
876 const char *format_subject(struct strbuf
*sb
, const char *msg
,
877 const char *line_separator
)
882 const char *line
= msg
;
883 int linelen
= get_one_line(line
);
886 if (!linelen
|| is_empty_line(line
, &linelen
))
891 strbuf_grow(sb
, linelen
+ 2);
893 strbuf_addstr(sb
, line_separator
);
894 strbuf_add(sb
, line
, linelen
);
900 static void parse_commit_message(struct format_commit_context
*c
)
902 const char *msg
= c
->message
+ c
->message_off
;
903 const char *start
= c
->message
;
905 msg
= skip_empty_lines(msg
);
906 c
->subject_off
= msg
- start
;
908 msg
= format_subject(NULL
, msg
, NULL
);
909 msg
= skip_empty_lines(msg
);
910 c
->body_off
= msg
- start
;
912 c
->commit_message_parsed
= 1;
915 static void strbuf_wrap(struct strbuf
*sb
, size_t pos
,
916 size_t width
, size_t indent1
, size_t indent2
)
918 struct strbuf tmp
= STRBUF_INIT
;
921 strbuf_add(&tmp
, sb
->buf
, pos
);
922 strbuf_add_wrapped_text(&tmp
, sb
->buf
+ pos
,
923 (int) indent1
, (int) indent2
, (int) width
);
924 strbuf_swap(&tmp
, sb
);
925 strbuf_release(&tmp
);
928 static void rewrap_message_tail(struct strbuf
*sb
,
929 struct format_commit_context
*c
,
930 size_t new_width
, size_t new_indent1
,
933 if (c
->width
== new_width
&& c
->indent1
== new_indent1
&&
934 c
->indent2
== new_indent2
)
936 if (c
->wrap_start
< sb
->len
)
937 strbuf_wrap(sb
, c
->wrap_start
, c
->width
, c
->indent1
, c
->indent2
);
938 c
->wrap_start
= sb
->len
;
939 c
->width
= new_width
;
940 c
->indent1
= new_indent1
;
941 c
->indent2
= new_indent2
;
944 static int format_reflog_person(struct strbuf
*sb
,
946 struct reflog_walk_info
*log
,
947 enum date_mode dmode
)
954 ident
= get_reflog_ident(log
);
958 return format_person_part(sb
, part
, ident
, strlen(ident
), dmode
);
961 static size_t parse_color(struct strbuf
*sb
, /* in UTF-8 */
962 const char *placeholder
,
963 struct format_commit_context
*c
)
965 if (placeholder
[1] == '(') {
966 const char *begin
= placeholder
+ 2;
967 const char *end
= strchr(begin
, ')');
968 char color
[COLOR_MAXLEN
];
972 if (starts_with(begin
, "auto,")) {
973 if (!want_color(c
->pretty_ctx
->color
))
974 return end
- placeholder
+ 1;
977 color_parse_mem(begin
,
979 "--pretty format", color
);
980 strbuf_addstr(sb
, color
);
981 return end
- placeholder
+ 1;
983 if (starts_with(placeholder
+ 1, "red")) {
984 strbuf_addstr(sb
, GIT_COLOR_RED
);
986 } else if (starts_with(placeholder
+ 1, "green")) {
987 strbuf_addstr(sb
, GIT_COLOR_GREEN
);
989 } else if (starts_with(placeholder
+ 1, "blue")) {
990 strbuf_addstr(sb
, GIT_COLOR_BLUE
);
992 } else if (starts_with(placeholder
+ 1, "reset")) {
993 strbuf_addstr(sb
, GIT_COLOR_RESET
);
999 static size_t parse_padding_placeholder(struct strbuf
*sb
,
1000 const char *placeholder
,
1001 struct format_commit_context
*c
)
1003 const char *ch
= placeholder
;
1004 enum flush_type flush_type
;
1009 flush_type
= flush_right
;
1013 flush_type
= flush_both
;
1015 } else if (*ch
== '>') {
1016 flush_type
= flush_left_and_steal
;
1019 flush_type
= flush_left
;
1025 /* the next value means "wide enough to that column" */
1032 const char *start
= ch
+ 1;
1033 const char *end
= start
+ strcspn(start
, ",)");
1036 if (!end
|| end
== start
)
1038 width
= strtoul(start
, &next
, 10);
1039 if (next
== start
|| width
== 0)
1041 c
->padding
= to_column
? -width
: width
;
1042 c
->flush_type
= flush_type
;
1046 end
= strchr(start
, ')');
1047 if (!end
|| end
== start
)
1049 if (starts_with(start
, "trunc)"))
1050 c
->truncate
= trunc_right
;
1051 else if (starts_with(start
, "ltrunc)"))
1052 c
->truncate
= trunc_left
;
1053 else if (starts_with(start
, "mtrunc)"))
1054 c
->truncate
= trunc_middle
;
1058 c
->truncate
= trunc_none
;
1060 return end
- placeholder
+ 1;
1065 static size_t format_commit_one(struct strbuf
*sb
, /* in UTF-8 */
1066 const char *placeholder
,
1069 struct format_commit_context
*c
= context
;
1070 const struct commit
*commit
= c
->commit
;
1071 const char *msg
= c
->message
;
1072 struct commit_list
*p
;
1075 /* these are independent of the commit */
1076 switch (placeholder
[0]) {
1078 if (starts_with(placeholder
+ 1, "(auto)")) {
1080 return 7; /* consumed 7 bytes, "C(auto)" */
1082 int ret
= parse_color(sb
, placeholder
, c
);
1086 * Otherwise, we decided to treat %C<unknown>
1087 * as a literal string, and the previous
1088 * %C(auto) is still valid.
1092 case 'n': /* newline */
1093 strbuf_addch(sb
, '\n');
1096 /* %x00 == NUL, %x0a == LF, etc. */
1097 if (0 <= (h1
= hexval_table
[0xff & placeholder
[1]]) &&
1099 0 <= (h2
= hexval_table
[0xff & placeholder
[2]]) &&
1101 strbuf_addch(sb
, (h1
<<4)|h2
);
1106 if (placeholder
[1] == '(') {
1107 unsigned long width
= 0, indent1
= 0, indent2
= 0;
1109 const char *start
= placeholder
+ 2;
1110 const char *end
= strchr(start
, ')');
1114 width
= strtoul(start
, &next
, 10);
1116 indent1
= strtoul(next
+ 1, &next
, 10);
1118 indent2
= strtoul(next
+ 1,
1125 rewrap_message_tail(sb
, c
, width
, indent1
, indent2
);
1126 return end
- placeholder
+ 1;
1132 return parse_padding_placeholder(sb
, placeholder
, c
);
1135 /* these depend on the commit */
1136 if (!commit
->object
.parsed
)
1137 parse_object(commit
->object
.sha1
);
1139 switch (placeholder
[0]) {
1140 case 'H': /* commit hash */
1141 strbuf_addstr(sb
, diff_get_color(c
->auto_color
, DIFF_COMMIT
));
1142 strbuf_addstr(sb
, sha1_to_hex(commit
->object
.sha1
));
1143 strbuf_addstr(sb
, diff_get_color(c
->auto_color
, DIFF_RESET
));
1145 case 'h': /* abbreviated commit hash */
1146 strbuf_addstr(sb
, diff_get_color(c
->auto_color
, DIFF_COMMIT
));
1147 if (add_again(sb
, &c
->abbrev_commit_hash
)) {
1148 strbuf_addstr(sb
, diff_get_color(c
->auto_color
, DIFF_RESET
));
1151 strbuf_addstr(sb
, find_unique_abbrev(commit
->object
.sha1
,
1152 c
->pretty_ctx
->abbrev
));
1153 strbuf_addstr(sb
, diff_get_color(c
->auto_color
, DIFF_RESET
));
1154 c
->abbrev_commit_hash
.len
= sb
->len
- c
->abbrev_commit_hash
.off
;
1156 case 'T': /* tree hash */
1157 strbuf_addstr(sb
, sha1_to_hex(commit
->tree
->object
.sha1
));
1159 case 't': /* abbreviated tree hash */
1160 if (add_again(sb
, &c
->abbrev_tree_hash
))
1162 strbuf_addstr(sb
, find_unique_abbrev(commit
->tree
->object
.sha1
,
1163 c
->pretty_ctx
->abbrev
));
1164 c
->abbrev_tree_hash
.len
= sb
->len
- c
->abbrev_tree_hash
.off
;
1166 case 'P': /* parent hashes */
1167 for (p
= commit
->parents
; p
; p
= p
->next
) {
1168 if (p
!= commit
->parents
)
1169 strbuf_addch(sb
, ' ');
1170 strbuf_addstr(sb
, sha1_to_hex(p
->item
->object
.sha1
));
1173 case 'p': /* abbreviated parent hashes */
1174 if (add_again(sb
, &c
->abbrev_parent_hashes
))
1176 for (p
= commit
->parents
; p
; p
= p
->next
) {
1177 if (p
!= commit
->parents
)
1178 strbuf_addch(sb
, ' ');
1179 strbuf_addstr(sb
, find_unique_abbrev(
1180 p
->item
->object
.sha1
,
1181 c
->pretty_ctx
->abbrev
));
1183 c
->abbrev_parent_hashes
.len
= sb
->len
-
1184 c
->abbrev_parent_hashes
.off
;
1186 case 'm': /* left/right/bottom */
1187 strbuf_addstr(sb
, get_revision_mark(NULL
, commit
));
1190 load_ref_decorations(DECORATE_SHORT_REFS
);
1191 format_decorations(sb
, commit
, c
->auto_color
);
1193 case 'g': /* reflog info */
1194 switch(placeholder
[1]) {
1195 case 'd': /* reflog selector */
1197 if (c
->pretty_ctx
->reflog_info
)
1198 get_reflog_selector(sb
,
1199 c
->pretty_ctx
->reflog_info
,
1200 c
->pretty_ctx
->date_mode
,
1201 c
->pretty_ctx
->date_mode_explicit
,
1202 (placeholder
[1] == 'd'));
1204 case 's': /* reflog message */
1205 if (c
->pretty_ctx
->reflog_info
)
1206 get_reflog_message(sb
, c
->pretty_ctx
->reflog_info
);
1212 return format_reflog_person(sb
,
1214 c
->pretty_ctx
->reflog_info
,
1215 c
->pretty_ctx
->date_mode
);
1217 return 0; /* unknown %g placeholder */
1219 if (c
->pretty_ctx
->notes_message
) {
1220 strbuf_addstr(sb
, c
->pretty_ctx
->notes_message
);
1226 if (placeholder
[0] == 'G') {
1227 if (!c
->signature_check
.result
)
1228 check_commit_signature(c
->commit
, &(c
->signature_check
));
1229 switch (placeholder
[1]) {
1231 if (c
->signature_check
.gpg_output
)
1232 strbuf_addstr(sb
, c
->signature_check
.gpg_output
);
1235 switch (c
->signature_check
.result
) {
1240 strbuf_addch(sb
, c
->signature_check
.result
);
1244 if (c
->signature_check
.signer
)
1245 strbuf_addstr(sb
, c
->signature_check
.signer
);
1248 if (c
->signature_check
.key
)
1249 strbuf_addstr(sb
, c
->signature_check
.key
);
1258 /* For the rest we have to parse the commit header. */
1259 if (!c
->commit_header_parsed
)
1260 parse_commit_header(c
);
1262 switch (placeholder
[0]) {
1263 case 'a': /* author ... */
1264 return format_person_part(sb
, placeholder
[1],
1265 msg
+ c
->author
.off
, c
->author
.len
,
1266 c
->pretty_ctx
->date_mode
);
1267 case 'c': /* committer ... */
1268 return format_person_part(sb
, placeholder
[1],
1269 msg
+ c
->committer
.off
, c
->committer
.len
,
1270 c
->pretty_ctx
->date_mode
);
1271 case 'e': /* encoding */
1272 if (c
->commit_encoding
)
1273 strbuf_addstr(sb
, c
->commit_encoding
);
1275 case 'B': /* raw body */
1276 /* message_off is always left at the initial newline */
1277 strbuf_addstr(sb
, msg
+ c
->message_off
+ 1);
1281 /* Now we need to parse the commit message. */
1282 if (!c
->commit_message_parsed
)
1283 parse_commit_message(c
);
1285 switch (placeholder
[0]) {
1286 case 's': /* subject */
1287 format_subject(sb
, msg
+ c
->subject_off
, " ");
1289 case 'f': /* sanitized subject */
1290 format_sanitized_subject(sb
, msg
+ c
->subject_off
);
1292 case 'b': /* body */
1293 strbuf_addstr(sb
, msg
+ c
->body_off
);
1296 return 0; /* unknown placeholder */
1299 static size_t format_and_pad_commit(struct strbuf
*sb
, /* in UTF-8 */
1300 const char *placeholder
,
1301 struct format_commit_context
*c
)
1303 struct strbuf local_sb
= STRBUF_INIT
;
1304 int total_consumed
= 0, len
, padding
= c
->padding
;
1306 const char *start
= strrchr(sb
->buf
, '\n');
1310 occupied
= utf8_strnwidth(start
, -1, 1);
1311 padding
= (-padding
) - occupied
;
1314 int modifier
= *placeholder
== 'C';
1315 int consumed
= format_commit_one(&local_sb
, placeholder
, c
);
1316 total_consumed
+= consumed
;
1321 placeholder
+= consumed
;
1322 if (*placeholder
!= '%')
1327 len
= utf8_strnwidth(local_sb
.buf
, -1, 1);
1329 if (c
->flush_type
== flush_left_and_steal
) {
1330 const char *ch
= sb
->buf
+ sb
->len
- 1;
1331 while (len
> padding
&& ch
> sb
->buf
) {
1338 /* check for trailing ansi sequences */
1342 while (ch
- p
< 10 && *p
!= '\033')
1345 ch
+ 1 - p
!= display_mode_esc_sequence_len(p
))
1348 * got a good ansi sequence, put it back to
1349 * local_sb as we're cutting sb
1351 strbuf_insert(&local_sb
, 0, p
, ch
+ 1 - p
);
1354 strbuf_setlen(sb
, ch
+ 1 - sb
->buf
);
1355 c
->flush_type
= flush_left
;
1358 if (len
> padding
) {
1359 switch (c
->truncate
) {
1361 strbuf_utf8_replace(&local_sb
,
1362 0, len
- (padding
- 2),
1366 strbuf_utf8_replace(&local_sb
,
1368 len
- (padding
- 2),
1372 strbuf_utf8_replace(&local_sb
,
1373 padding
- 2, len
- (padding
- 2),
1379 strbuf_addbuf(sb
, &local_sb
);
1381 int sb_len
= sb
->len
, offset
= 0;
1382 if (c
->flush_type
== flush_left
)
1383 offset
= padding
- len
;
1384 else if (c
->flush_type
== flush_both
)
1385 offset
= (padding
- len
) / 2;
1387 * we calculate padding in columns, now
1388 * convert it back to chars
1390 padding
= padding
- len
+ local_sb
.len
;
1391 strbuf_grow(sb
, padding
);
1392 strbuf_setlen(sb
, sb_len
+ padding
);
1393 memset(sb
->buf
+ sb_len
, ' ', sb
->len
- sb_len
);
1394 memcpy(sb
->buf
+ sb_len
+ offset
, local_sb
.buf
,
1397 strbuf_release(&local_sb
);
1398 c
->flush_type
= no_flush
;
1399 return total_consumed
;
1402 static size_t format_commit_item(struct strbuf
*sb
, /* in UTF-8 */
1403 const char *placeholder
,
1410 ADD_LF_BEFORE_NON_EMPTY
,
1411 DEL_LF_BEFORE_EMPTY
,
1412 ADD_SP_BEFORE_NON_EMPTY
1415 switch (placeholder
[0]) {
1417 magic
= DEL_LF_BEFORE_EMPTY
;
1420 magic
= ADD_LF_BEFORE_NON_EMPTY
;
1423 magic
= ADD_SP_BEFORE_NON_EMPTY
;
1428 if (magic
!= NO_MAGIC
)
1432 if (((struct format_commit_context
*)context
)->flush_type
!= no_flush
)
1433 consumed
= format_and_pad_commit(sb
, placeholder
, context
);
1435 consumed
= format_commit_one(sb
, placeholder
, context
);
1436 if (magic
== NO_MAGIC
)
1439 if ((orig_len
== sb
->len
) && magic
== DEL_LF_BEFORE_EMPTY
) {
1440 while (sb
->len
&& sb
->buf
[sb
->len
- 1] == '\n')
1441 strbuf_setlen(sb
, sb
->len
- 1);
1442 } else if (orig_len
!= sb
->len
) {
1443 if (magic
== ADD_LF_BEFORE_NON_EMPTY
)
1444 strbuf_insert(sb
, orig_len
, "\n", 1);
1445 else if (magic
== ADD_SP_BEFORE_NON_EMPTY
)
1446 strbuf_insert(sb
, orig_len
, " ", 1);
1448 return consumed
+ 1;
1451 static size_t userformat_want_item(struct strbuf
*sb
, const char *placeholder
,
1454 struct userformat_want
*w
= context
;
1456 if (*placeholder
== '+' || *placeholder
== '-' || *placeholder
== ' ')
1459 switch (*placeholder
) {
1467 void userformat_find_requirements(const char *fmt
, struct userformat_want
*w
)
1469 struct strbuf dummy
= STRBUF_INIT
;
1476 strbuf_expand(&dummy
, fmt
, userformat_want_item
, w
);
1477 strbuf_release(&dummy
);
1480 void format_commit_message(const struct commit
*commit
,
1481 const char *format
, struct strbuf
*sb
,
1482 const struct pretty_print_context
*pretty_ctx
)
1484 struct format_commit_context context
;
1485 const char *output_enc
= pretty_ctx
->output_encoding
;
1486 const char *utf8
= "UTF-8";
1488 memset(&context
, 0, sizeof(context
));
1489 context
.commit
= commit
;
1490 context
.pretty_ctx
= pretty_ctx
;
1491 context
.wrap_start
= sb
->len
;
1493 * convert a commit message to UTF-8 first
1494 * as far as 'format_commit_item' assumes it in UTF-8
1496 context
.message
= logmsg_reencode(commit
,
1497 &context
.commit_encoding
,
1500 strbuf_expand(sb
, format
, format_commit_item
, &context
);
1501 rewrap_message_tail(sb
, &context
, 0, 0, 0);
1503 /* then convert a commit message to an actual output encoding */
1505 if (same_encoding(utf8
, output_enc
))
1508 if (context
.commit_encoding
&&
1509 !same_encoding(context
.commit_encoding
, utf8
))
1510 output_enc
= context
.commit_encoding
;
1515 char *out
= reencode_string_len(sb
->buf
, sb
->len
,
1516 output_enc
, utf8
, &outsz
);
1518 strbuf_attach(sb
, out
, outsz
, outsz
+ 1);
1521 free(context
.commit_encoding
);
1522 unuse_commit_buffer(commit
, context
.message
);
1525 static void pp_header(struct pretty_print_context
*pp
,
1526 const char *encoding
,
1527 const struct commit
*commit
,
1531 int parents_shown
= 0;
1534 const char *line
= *msg_p
;
1535 int linelen
= get_one_line(*msg_p
);
1545 if (pp
->fmt
== CMIT_FMT_RAW
) {
1546 strbuf_add(sb
, line
, linelen
);
1550 if (starts_with(line
, "parent ")) {
1552 die("bad parent line in commit");
1556 if (!parents_shown
) {
1557 unsigned num
= commit_list_count(commit
->parents
);
1558 /* with enough slop */
1559 strbuf_grow(sb
, num
* 50 + 20);
1560 add_merge_info(pp
, sb
, commit
);
1565 * MEDIUM == DEFAULT shows only author with dates.
1566 * FULL shows both authors but not dates.
1567 * FULLER shows both authors and dates.
1569 if (starts_with(line
, "author ")) {
1570 strbuf_grow(sb
, linelen
+ 80);
1571 pp_user_info(pp
, "Author", sb
, line
+ 7, encoding
);
1573 if (starts_with(line
, "committer ") &&
1574 (pp
->fmt
== CMIT_FMT_FULL
|| pp
->fmt
== CMIT_FMT_FULLER
)) {
1575 strbuf_grow(sb
, linelen
+ 80);
1576 pp_user_info(pp
, "Commit", sb
, line
+ 10, encoding
);
1581 void pp_title_line(struct pretty_print_context
*pp
,
1584 const char *encoding
,
1587 static const int max_length
= 78; /* per rfc2047 */
1588 struct strbuf title
;
1590 strbuf_init(&title
, 80);
1591 *msg_p
= format_subject(&title
, *msg_p
,
1592 pp
->preserve_subject
? "\n" : " ");
1594 strbuf_grow(sb
, title
.len
+ 1024);
1596 strbuf_addstr(sb
, pp
->subject
);
1597 if (needs_rfc2047_encoding(title
.buf
, title
.len
, RFC2047_SUBJECT
))
1598 add_rfc2047(sb
, title
.buf
, title
.len
,
1599 encoding
, RFC2047_SUBJECT
);
1601 strbuf_add_wrapped_bytes(sb
, title
.buf
, title
.len
,
1602 -last_line_length(sb
), 1, max_length
);
1604 strbuf_addbuf(sb
, &title
);
1606 strbuf_addch(sb
, '\n');
1608 if (need_8bit_cte
== 0) {
1610 for (i
= 0; i
< pp
->in_body_headers
.nr
; i
++) {
1611 if (has_non_ascii(pp
->in_body_headers
.items
[i
].string
)) {
1618 if (need_8bit_cte
> 0) {
1619 const char *header_fmt
=
1620 "MIME-Version: 1.0\n"
1621 "Content-Type: text/plain; charset=%s\n"
1622 "Content-Transfer-Encoding: 8bit\n";
1623 strbuf_addf(sb
, header_fmt
, encoding
);
1625 if (pp
->after_subject
) {
1626 strbuf_addstr(sb
, pp
->after_subject
);
1628 if (pp
->fmt
== CMIT_FMT_EMAIL
) {
1629 strbuf_addch(sb
, '\n');
1632 if (pp
->in_body_headers
.nr
) {
1634 for (i
= 0; i
< pp
->in_body_headers
.nr
; i
++) {
1635 strbuf_addstr(sb
, pp
->in_body_headers
.items
[i
].string
);
1636 free(pp
->in_body_headers
.items
[i
].string
);
1638 string_list_clear(&pp
->in_body_headers
, 0);
1639 strbuf_addch(sb
, '\n');
1642 strbuf_release(&title
);
1645 void pp_remainder(struct pretty_print_context
*pp
,
1652 const char *line
= *msg_p
;
1653 int linelen
= get_one_line(line
);
1659 if (is_empty_line(line
, &linelen
)) {
1662 if (pp
->fmt
== CMIT_FMT_SHORT
)
1667 strbuf_grow(sb
, linelen
+ indent
+ 20);
1669 memset(sb
->buf
+ sb
->len
, ' ', indent
);
1670 strbuf_setlen(sb
, sb
->len
+ indent
);
1672 strbuf_add(sb
, line
, linelen
);
1673 strbuf_addch(sb
, '\n');
1677 void pretty_print_commit(struct pretty_print_context
*pp
,
1678 const struct commit
*commit
,
1681 unsigned long beginning_of_body
;
1684 const char *reencoded
;
1685 const char *encoding
;
1686 int need_8bit_cte
= pp
->need_8bit_cte
;
1688 if (pp
->fmt
== CMIT_FMT_USERFORMAT
) {
1689 format_commit_message(commit
, user_format
, sb
, pp
);
1693 encoding
= get_log_output_encoding();
1694 msg
= reencoded
= logmsg_reencode(commit
, NULL
, encoding
);
1696 if (pp
->fmt
== CMIT_FMT_ONELINE
|| pp
->fmt
== CMIT_FMT_EMAIL
)
1700 * We need to check and emit Content-type: to mark it
1701 * as 8-bit if we haven't done so.
1703 if (pp
->fmt
== CMIT_FMT_EMAIL
&& need_8bit_cte
== 0) {
1706 for (in_body
= i
= 0; (ch
= msg
[i
]); i
++) {
1708 /* author could be non 7-bit ASCII but
1709 * the log may be so; skip over the
1710 * header part first.
1712 if (ch
== '\n' && msg
[i
+1] == '\n')
1715 else if (non_ascii(ch
)) {
1722 pp_header(pp
, encoding
, commit
, &msg
, sb
);
1723 if (pp
->fmt
!= CMIT_FMT_ONELINE
&& !pp
->subject
) {
1724 strbuf_addch(sb
, '\n');
1727 /* Skip excess blank lines at the beginning of body, if any... */
1728 msg
= skip_empty_lines(msg
);
1730 /* These formats treat the title line specially. */
1731 if (pp
->fmt
== CMIT_FMT_ONELINE
|| pp
->fmt
== CMIT_FMT_EMAIL
)
1732 pp_title_line(pp
, &msg
, sb
, encoding
, need_8bit_cte
);
1734 beginning_of_body
= sb
->len
;
1735 if (pp
->fmt
!= CMIT_FMT_ONELINE
)
1736 pp_remainder(pp
, &msg
, sb
, indent
);
1739 /* Make sure there is an EOLN for the non-oneline case */
1740 if (pp
->fmt
!= CMIT_FMT_ONELINE
)
1741 strbuf_addch(sb
, '\n');
1744 * The caller may append additional body text in e-mail
1745 * format. Make sure we did not strip the blank line
1746 * between the header and the body.
1748 if (pp
->fmt
== CMIT_FMT_EMAIL
&& sb
->len
<= beginning_of_body
)
1749 strbuf_addch(sb
, '\n');
1751 unuse_commit_buffer(commit
, reencoded
);
1754 void pp_commit_easy(enum cmit_fmt fmt
, const struct commit
*commit
,
1757 struct pretty_print_context pp
= {0};
1759 pretty_print_commit(&pp
, commit
, sb
);