6 #include "string-list.h"
10 static char *user_format;
12 void get_commit_format(const char *arg, struct rev_info *rev)
15 static struct cmt_fmt_map {
20 { "raw", 1, CMIT_FMT_RAW },
21 { "medium", 1, CMIT_FMT_MEDIUM },
22 { "short", 1, CMIT_FMT_SHORT },
23 { "email", 1, CMIT_FMT_EMAIL },
24 { "full", 5, CMIT_FMT_FULL },
25 { "fuller", 5, CMIT_FMT_FULLER },
26 { "oneline", 1, CMIT_FMT_ONELINE },
29 rev->use_terminator = 0;
31 rev->commit_format = CMIT_FMT_DEFAULT;
34 if (!prefixcmp(arg, "format:") || !prefixcmp(arg, "tformat:")) {
35 const char *cp = strchr(arg, ':') + 1;
37 user_format = xstrdup(cp);
39 rev->use_terminator = 1;
40 rev->commit_format = CMIT_FMT_USERFORMAT;
43 for (i = 0; i < ARRAY_SIZE(cmt_fmts); i++) {
44 if (!strncmp(arg, cmt_fmts[i].n, cmt_fmts[i].cmp_len) &&
45 !strncmp(arg, cmt_fmts[i].n, strlen(arg))) {
46 if (cmt_fmts[i].v == CMIT_FMT_ONELINE)
47 rev->use_terminator = 1;
48 rev->commit_format = cmt_fmts[i].v;
53 die("invalid --pretty format: %s", arg);
57 * Generic support for pretty-printing the header
59 static int get_one_line(const char *msg)
74 /* High bit set, or ISO-2022-INT */
78 return ((ch & 0x80) || (ch == 0x1b));
81 static int is_rfc2047_special(char ch)
83 return (non_ascii(ch) || (ch == '=') || (ch == '?') || (ch == '_'));
86 static void add_rfc2047(struct strbuf *sb, const char *line, int len,
91 for (i = 0; i < len; i++) {
95 if ((i + 1 < len) && (ch == '=' && line[i+1] == '?'))
98 strbuf_add(sb, line, len);
102 strbuf_grow(sb, len * 3 + strlen(encoding) + 100);
103 strbuf_addf(sb, "=?%s?q?", encoding);
104 for (i = last = 0; i < len; i++) {
105 unsigned ch = line[i] & 0xFF;
107 * We encode ' ' using '=20' even though rfc2047
108 * allows using '_' for readability. Unfortunately,
109 * many programs do not understand this and just
110 * leave the underscore in place.
112 if (is_rfc2047_special(ch) || ch == ' ') {
113 strbuf_add(sb, line + last, i - last);
114 strbuf_addf(sb, "=%02X", ch);
118 strbuf_add(sb, line + last, len - last);
119 strbuf_addstr(sb, "?=");
122 void pp_user_info(const char *what, enum cmit_fmt fmt, struct strbuf *sb,
123 const char *line, enum date_mode dmode,
124 const char *encoding)
130 const char *filler = " ";
132 if (fmt == CMIT_FMT_ONELINE)
134 date = strchr(line, '>');
137 namelen = ++date - line;
138 time = strtoul(date, &date, 10);
139 tz = strtol(date, NULL, 10);
141 if (fmt == CMIT_FMT_EMAIL) {
142 char *name_tail = strchr(line, '<');
143 int display_name_length;
146 while (line < name_tail && isspace(name_tail[-1]))
148 display_name_length = name_tail - line;
150 strbuf_addstr(sb, "From: ");
151 add_rfc2047(sb, line, display_name_length, encoding);
152 strbuf_add(sb, name_tail, namelen - display_name_length);
153 strbuf_addch(sb, '\n');
155 strbuf_addf(sb, "%s: %.*s%.*s\n", what,
156 (fmt == CMIT_FMT_FULLER) ? 4 : 0,
157 filler, namelen, line);
160 case CMIT_FMT_MEDIUM:
161 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, dmode));
164 strbuf_addf(sb, "Date: %s\n", show_date(time, tz, DATE_RFC2822));
166 case CMIT_FMT_FULLER:
167 strbuf_addf(sb, "%sDate: %s\n", what, show_date(time, tz, dmode));
175 static int is_empty_line(const char *line, int *len_p)
178 while (len && isspace(line[len-1]))
184 static void add_merge_info(enum cmit_fmt fmt, struct strbuf *sb,
185 const struct commit *commit, int abbrev)
187 struct commit_list *parent = commit->parents;
189 if ((fmt == CMIT_FMT_ONELINE) || (fmt == CMIT_FMT_EMAIL) ||
190 !parent || !parent->next)
193 strbuf_addstr(sb, "Merge:");
196 struct commit *p = parent->item;
197 const char *hex = NULL;
200 hex = find_unique_abbrev(p->object.sha1, abbrev);
202 hex = sha1_to_hex(p->object.sha1);
203 dots = (abbrev && strlen(hex) != 40) ? "..." : "";
204 parent = parent->next;
206 strbuf_addf(sb, " %s%s", hex, dots);
208 strbuf_addch(sb, '\n');
211 static char *get_header(const struct commit *commit, const char *key)
213 int key_len = strlen(key);
214 const char *line = commit->buffer;
217 const char *eol = strchr(line, '\n'), *next;
222 eol = line + strlen(line);
226 if (eol - line > key_len &&
227 !strncmp(line, key, key_len) &&
228 line[key_len] == ' ') {
229 return xmemdupz(line + key_len + 1, eol - line - key_len - 1);
235 static char *replace_encoding_header(char *buf, const char *encoding)
237 struct strbuf tmp = STRBUF_INIT;
241 /* guess if there is an encoding header before a \n\n */
242 while (strncmp(cp, "encoding ", strlen("encoding "))) {
243 cp = strchr(cp, '\n');
244 if (!cp || *++cp == '\n')
248 cp = strchr(cp, '\n');
250 return buf; /* should not happen but be defensive */
251 len = cp + 1 - (buf + start);
253 strbuf_attach(&tmp, buf, strlen(buf), strlen(buf) + 1);
254 if (is_encoding_utf8(encoding)) {
255 /* we have re-coded to UTF-8; drop the header */
256 strbuf_remove(&tmp, start, len);
258 /* just replaces XXXX in 'encoding XXXX\n' */
259 strbuf_splice(&tmp, start + strlen("encoding "),
260 len - strlen("encoding \n"),
261 encoding, strlen(encoding));
263 return strbuf_detach(&tmp, NULL);
266 static char *logmsg_reencode(const struct commit *commit,
267 const char *output_encoding)
269 static const char *utf8 = "utf-8";
270 const char *use_encoding;
274 if (!*output_encoding)
276 encoding = get_header(commit, "encoding");
277 use_encoding = encoding ? encoding : utf8;
278 if (!strcmp(use_encoding, output_encoding))
279 if (encoding) /* we'll strip encoding header later */
280 out = xstrdup(commit->buffer);
282 return NULL; /* nothing to do */
284 out = reencode_string(commit->buffer,
285 output_encoding, use_encoding);
287 out = replace_encoding_header(out, output_encoding);
293 static int mailmap_name(struct strbuf *sb, const char *email)
295 static struct string_list *mail_map;
299 mail_map = xcalloc(1, sizeof(*mail_map));
300 read_mailmap(mail_map, ".mailmap", NULL);
306 if (!map_email(mail_map, email, buffer, sizeof(buffer)))
308 strbuf_addstr(sb, buffer);
312 static size_t format_person_part(struct strbuf *sb, char part,
313 const char *msg, int len, enum date_mode dmode)
315 /* currently all placeholders have same length */
316 const int placeholder_len = 2;
317 int start, end, tz = 0;
318 unsigned long date = 0;
321 /* advance 'end' to point to email start delimiter */
322 for (end = 0; end < len && msg[end] != '<'; end++)
326 * When end points at the '<' that we found, it should have
327 * matching '>' later, which means 'end' must be strictly
333 if (part == 'n' || part == 'N') { /* name */
334 while (end > 0 && isspace(msg[end - 1]))
336 if (part != 'N' || !msg[end] || !msg[end + 1] ||
337 mailmap_name(sb, msg + end + 2) < 0)
338 strbuf_add(sb, msg, end);
339 return placeholder_len;
341 start = ++end; /* save email start position */
343 /* advance 'end' to point to email end delimiter */
344 for ( ; end < len && msg[end] != '>'; end++)
350 if (part == 'e') { /* email */
351 strbuf_add(sb, msg + start, end - start);
352 return placeholder_len;
355 /* advance 'start' to point to date start delimiter */
356 for (start = end + 1; start < len && isspace(msg[start]); start++)
360 date = strtoul(msg + start, &ep, 10);
361 if (msg + start == ep)
364 if (part == 't') { /* date, UNIX timestamp */
365 strbuf_add(sb, msg + start, ep - (msg + start));
366 return placeholder_len;
370 for (start = ep - msg + 1; start < len && isspace(msg[start]); start++)
372 if (start + 1 < len) {
373 tz = strtoul(msg + start + 1, NULL, 10);
374 if (msg[start] == '-')
380 strbuf_addstr(sb, show_date(date, tz, dmode));
381 return placeholder_len;
382 case 'D': /* date, RFC2822 style */
383 strbuf_addstr(sb, show_date(date, tz, DATE_RFC2822));
384 return placeholder_len;
385 case 'r': /* date, relative */
386 strbuf_addstr(sb, show_date(date, tz, DATE_RELATIVE));
387 return placeholder_len;
388 case 'i': /* date, ISO 8601 */
389 strbuf_addstr(sb, show_date(date, tz, DATE_ISO8601));
390 return placeholder_len;
395 * bogus commit, 'sb' cannot be updated, but we still need to
396 * compute a valid return value.
398 if (part == 'n' || part == 'e' || part == 't' || part == 'd'
399 || part == 'D' || part == 'r' || part == 'i')
400 return placeholder_len;
402 return 0; /* unknown placeholder */
410 struct format_commit_context {
411 const struct commit *commit;
412 enum date_mode dmode;
414 /* These offsets are relative to the start of the commit message. */
415 int commit_header_parsed;
416 struct chunk subject;
418 struct chunk committer;
419 struct chunk encoding;
422 /* The following ones are relative to the result struct strbuf. */
423 struct chunk abbrev_commit_hash;
424 struct chunk abbrev_tree_hash;
425 struct chunk abbrev_parent_hashes;
428 static int add_again(struct strbuf *sb, struct chunk *chunk)
431 strbuf_adddup(sb, chunk->off, chunk->len);
436 * We haven't seen this chunk before. Our caller is surely
437 * going to add it the hard way now. Remember the most likely
438 * start of the to-be-added chunk: the current end of the
441 chunk->off = sb->len;
445 static void parse_commit_header(struct format_commit_context *context)
447 const char *msg = context->commit->buffer;
449 enum { HEADER, SUBJECT, BODY } state;
451 for (i = 0, state = HEADER; msg[i] && state < BODY; i++) {
453 for (eol = i; msg[eol] && msg[eol] != '\n'; eol++)
456 if (state == SUBJECT) {
457 context->subject.off = i;
458 context->subject.len = eol - i;
463 /* strip empty lines */
464 while (msg[eol] == '\n' && msg[eol + 1] == '\n')
466 } else if (!prefixcmp(msg + i, "author ")) {
467 context->author.off = i + 7;
468 context->author.len = eol - i - 7;
469 } else if (!prefixcmp(msg + i, "committer ")) {
470 context->committer.off = i + 10;
471 context->committer.len = eol - i - 10;
472 } else if (!prefixcmp(msg + i, "encoding ")) {
473 context->encoding.off = i + 9;
474 context->encoding.len = eol - i - 9;
480 context->body_off = i;
481 context->commit_header_parsed = 1;
484 static void format_decoration(struct strbuf *sb, const struct commit *commit)
486 struct name_decoration *d;
487 const char *prefix = " (";
489 load_ref_decorations();
490 d = lookup_decoration(&name_decoration, &commit->object);
492 strbuf_addstr(sb, prefix);
494 strbuf_addstr(sb, d->name);
497 if (prefix[0] == ',')
498 strbuf_addch(sb, ')');
501 static size_t format_commit_item(struct strbuf *sb, const char *placeholder,
504 struct format_commit_context *c = context;
505 const struct commit *commit = c->commit;
506 const char *msg = commit->buffer;
507 struct commit_list *p;
510 /* these are independent of the commit */
511 switch (placeholder[0]) {
513 if (!prefixcmp(placeholder + 1, "red")) {
514 strbuf_addstr(sb, "\033[31m");
516 } else if (!prefixcmp(placeholder + 1, "green")) {
517 strbuf_addstr(sb, "\033[32m");
519 } else if (!prefixcmp(placeholder + 1, "blue")) {
520 strbuf_addstr(sb, "\033[34m");
522 } else if (!prefixcmp(placeholder + 1, "reset")) {
523 strbuf_addstr(sb, "\033[m");
527 case 'n': /* newline */
528 strbuf_addch(sb, '\n');
531 /* %x00 == NUL, %x0a == LF, etc. */
532 if (0 <= (h1 = hexval_table[0xff & placeholder[1]]) &&
534 0 <= (h2 = hexval_table[0xff & placeholder[2]]) &&
536 strbuf_addch(sb, (h1<<4)|h2);
542 /* these depend on the commit */
543 if (!commit->object.parsed)
544 parse_object(commit->object.sha1);
546 switch (placeholder[0]) {
547 case 'H': /* commit hash */
548 strbuf_addstr(sb, sha1_to_hex(commit->object.sha1));
550 case 'h': /* abbreviated commit hash */
551 if (add_again(sb, &c->abbrev_commit_hash))
553 strbuf_addstr(sb, find_unique_abbrev(commit->object.sha1,
555 c->abbrev_commit_hash.len = sb->len - c->abbrev_commit_hash.off;
557 case 'T': /* tree hash */
558 strbuf_addstr(sb, sha1_to_hex(commit->tree->object.sha1));
560 case 't': /* abbreviated tree hash */
561 if (add_again(sb, &c->abbrev_tree_hash))
563 strbuf_addstr(sb, find_unique_abbrev(commit->tree->object.sha1,
565 c->abbrev_tree_hash.len = sb->len - c->abbrev_tree_hash.off;
567 case 'P': /* parent hashes */
568 for (p = commit->parents; p; p = p->next) {
569 if (p != commit->parents)
570 strbuf_addch(sb, ' ');
571 strbuf_addstr(sb, sha1_to_hex(p->item->object.sha1));
574 case 'p': /* abbreviated parent hashes */
575 if (add_again(sb, &c->abbrev_parent_hashes))
577 for (p = commit->parents; p; p = p->next) {
578 if (p != commit->parents)
579 strbuf_addch(sb, ' ');
580 strbuf_addstr(sb, find_unique_abbrev(
581 p->item->object.sha1, DEFAULT_ABBREV));
583 c->abbrev_parent_hashes.len = sb->len -
584 c->abbrev_parent_hashes.off;
586 case 'm': /* left/right/bottom */
587 strbuf_addch(sb, (commit->object.flags & BOUNDARY)
589 : (commit->object.flags & SYMMETRIC_LEFT)
594 format_decoration(sb, commit);
598 /* For the rest we have to parse the commit header. */
599 if (!c->commit_header_parsed)
600 parse_commit_header(c);
602 switch (placeholder[0]) {
603 case 's': /* subject */
604 strbuf_add(sb, msg + c->subject.off, c->subject.len);
606 case 'a': /* author ... */
607 return format_person_part(sb, placeholder[1],
608 msg + c->author.off, c->author.len,
610 case 'c': /* committer ... */
611 return format_person_part(sb, placeholder[1],
612 msg + c->committer.off, c->committer.len,
614 case 'e': /* encoding */
615 strbuf_add(sb, msg + c->encoding.off, c->encoding.len);
618 strbuf_addstr(sb, msg + c->body_off);
621 return 0; /* unknown placeholder */
624 void format_commit_message(const struct commit *commit,
625 const void *format, struct strbuf *sb,
626 enum date_mode dmode)
628 struct format_commit_context context;
630 memset(&context, 0, sizeof(context));
631 context.commit = commit;
632 context.dmode = dmode;
633 strbuf_expand(sb, format, format_commit_item, &context);
636 static void pp_header(enum cmit_fmt fmt,
638 enum date_mode dmode,
639 const char *encoding,
640 const struct commit *commit,
644 int parents_shown = 0;
647 const char *line = *msg_p;
648 int linelen = get_one_line(*msg_p);
658 if (fmt == CMIT_FMT_RAW) {
659 strbuf_add(sb, line, linelen);
663 if (!memcmp(line, "parent ", 7)) {
665 die("bad parent line in commit");
669 if (!parents_shown) {
670 struct commit_list *parent;
672 for (parent = commit->parents, num = 0;
674 parent = parent->next, num++)
676 /* with enough slop */
677 strbuf_grow(sb, num * 50 + 20);
678 add_merge_info(fmt, sb, commit, abbrev);
683 * MEDIUM == DEFAULT shows only author with dates.
684 * FULL shows both authors but not dates.
685 * FULLER shows both authors and dates.
687 if (!memcmp(line, "author ", 7)) {
688 strbuf_grow(sb, linelen + 80);
689 pp_user_info("Author", fmt, sb, line + 7, dmode, encoding);
691 if (!memcmp(line, "committer ", 10) &&
692 (fmt == CMIT_FMT_FULL || fmt == CMIT_FMT_FULLER)) {
693 strbuf_grow(sb, linelen + 80);
694 pp_user_info("Commit", fmt, sb, line + 10, dmode, encoding);
699 void pp_title_line(enum cmit_fmt fmt,
703 const char *after_subject,
704 const char *encoding,
709 strbuf_init(&title, 80);
712 const char *line = *msg_p;
713 int linelen = get_one_line(line);
716 if (!linelen || is_empty_line(line, &linelen))
719 strbuf_grow(&title, linelen + 2);
721 if (fmt == CMIT_FMT_EMAIL) {
722 strbuf_addch(&title, '\n');
724 strbuf_addch(&title, ' ');
726 strbuf_add(&title, line, linelen);
729 strbuf_grow(sb, title.len + 1024);
731 strbuf_addstr(sb, subject);
732 add_rfc2047(sb, title.buf, title.len, encoding);
734 strbuf_addbuf(sb, &title);
736 strbuf_addch(sb, '\n');
738 if (need_8bit_cte > 0) {
739 const char *header_fmt =
740 "MIME-Version: 1.0\n"
741 "Content-Type: text/plain; charset=%s\n"
742 "Content-Transfer-Encoding: 8bit\n";
743 strbuf_addf(sb, header_fmt, encoding);
746 strbuf_addstr(sb, after_subject);
748 if (fmt == CMIT_FMT_EMAIL) {
749 strbuf_addch(sb, '\n');
751 strbuf_release(&title);
754 void pp_remainder(enum cmit_fmt fmt,
761 const char *line = *msg_p;
762 int linelen = get_one_line(line);
768 if (is_empty_line(line, &linelen)) {
771 if (fmt == CMIT_FMT_SHORT)
776 strbuf_grow(sb, linelen + indent + 20);
778 memset(sb->buf + sb->len, ' ', indent);
779 strbuf_setlen(sb, sb->len + indent);
781 strbuf_add(sb, line, linelen);
782 strbuf_addch(sb, '\n');
786 char *reencode_commit_message(const struct commit *commit, const char **encoding_p)
788 const char *encoding;
790 encoding = (git_log_output_encoding
791 ? git_log_output_encoding
792 : git_commit_encoding);
796 *encoding_p = encoding;
797 return logmsg_reencode(commit, encoding);
800 void pretty_print_commit(enum cmit_fmt fmt, const struct commit *commit,
801 struct strbuf *sb, int abbrev,
802 const char *subject, const char *after_subject,
803 enum date_mode dmode, int need_8bit_cte)
805 unsigned long beginning_of_body;
807 const char *msg = commit->buffer;
809 const char *encoding;
811 if (fmt == CMIT_FMT_USERFORMAT) {
812 format_commit_message(commit, user_format, sb, dmode);
816 reencoded = reencode_commit_message(commit, &encoding);
821 if (fmt == CMIT_FMT_ONELINE || fmt == CMIT_FMT_EMAIL)
825 * We need to check and emit Content-type: to mark it
826 * as 8-bit if we haven't done so.
828 if (fmt == CMIT_FMT_EMAIL && need_8bit_cte == 0) {
831 for (in_body = i = 0; (ch = msg[i]); i++) {
833 /* author could be non 7-bit ASCII but
834 * the log may be so; skip over the
837 if (ch == '\n' && msg[i+1] == '\n')
840 else if (non_ascii(ch)) {
847 pp_header(fmt, abbrev, dmode, encoding, commit, &msg, sb);
848 if (fmt != CMIT_FMT_ONELINE && !subject) {
849 strbuf_addch(sb, '\n');
852 /* Skip excess blank lines at the beginning of body, if any... */
854 int linelen = get_one_line(msg);
858 if (!is_empty_line(msg, &ll))
863 /* These formats treat the title line specially. */
864 if (fmt == CMIT_FMT_ONELINE || fmt == CMIT_FMT_EMAIL)
865 pp_title_line(fmt, &msg, sb, subject,
866 after_subject, encoding, need_8bit_cte);
868 beginning_of_body = sb->len;
869 if (fmt != CMIT_FMT_ONELINE)
870 pp_remainder(fmt, &msg, sb, indent);
873 /* Make sure there is an EOLN for the non-oneline case */
874 if (fmt != CMIT_FMT_ONELINE)
875 strbuf_addch(sb, '\n');
878 * The caller may append additional body text in e-mail
879 * format. Make sure we did not strip the blank line
880 * between the header and the body.
882 if (fmt == CMIT_FMT_EMAIL && sb->len <= beginning_of_body)
883 strbuf_addch(sb, '\n');