Shrink "struct object" a bit
[git/git.git] / commit.c
1 #include "cache.h"
2 #include "tag.h"
3 #include "commit.h"
4
5 int save_commit_buffer = 1;
6
7 struct sort_node
8 {
9 /*
10 * the number of children of the associated commit
11 * that also occur in the list being sorted.
12 */
13 unsigned int indegree;
14
15 /*
16 * reference to original list item that we will re-use
17 * on output.
18 */
19 struct commit_list * list_item;
20
21 };
22
23 const char *commit_type = "commit";
24
25 struct cmt_fmt_map {
26 const char *n;
27 size_t cmp_len;
28 enum cmit_fmt v;
29 } cmt_fmts[] = {
30 { "raw", 1, CMIT_FMT_RAW },
31 { "medium", 1, CMIT_FMT_MEDIUM },
32 { "short", 1, CMIT_FMT_SHORT },
33 { "email", 1, CMIT_FMT_EMAIL },
34 { "full", 5, CMIT_FMT_FULL },
35 { "fuller", 5, CMIT_FMT_FULLER },
36 { "oneline", 1, CMIT_FMT_ONELINE },
37 };
38
39 enum cmit_fmt get_commit_format(const char *arg)
40 {
41 int i;
42
43 if (!arg || !*arg)
44 return CMIT_FMT_DEFAULT;
45 if (*arg == '=')
46 arg++;
47 for (i = 0; i < ARRAY_SIZE(cmt_fmts); i++) {
48 if (!strncmp(arg, cmt_fmts[i].n, cmt_fmts[i].cmp_len))
49 return cmt_fmts[i].v;
50 }
51
52 die("invalid --pretty format: %s", arg);
53 }
54
55 static struct commit *check_commit(struct object *obj,
56 const unsigned char *sha1,
57 int quiet)
58 {
59 if (obj->type != TYPE_COMMIT) {
60 if (!quiet)
61 error("Object %s is a %s, not a commit",
62 sha1_to_hex(sha1), typename(obj->type));
63 return NULL;
64 }
65 return (struct commit *) obj;
66 }
67
68 struct commit *lookup_commit_reference_gently(const unsigned char *sha1,
69 int quiet)
70 {
71 struct object *obj = deref_tag(parse_object(sha1), NULL, 0);
72
73 if (!obj)
74 return NULL;
75 return check_commit(obj, sha1, quiet);
76 }
77
78 struct commit *lookup_commit_reference(const unsigned char *sha1)
79 {
80 return lookup_commit_reference_gently(sha1, 0);
81 }
82
83 struct commit *lookup_commit(const unsigned char *sha1)
84 {
85 struct object *obj = lookup_object(sha1);
86 if (!obj) {
87 struct commit *ret = xcalloc(1, sizeof(struct commit));
88 created_object(sha1, &ret->object);
89 ret->object.type = TYPE_COMMIT;
90 return ret;
91 }
92 if (!obj->type)
93 obj->type = TYPE_COMMIT;
94 return check_commit(obj, sha1, 0);
95 }
96
97 static unsigned long parse_commit_date(const char *buf)
98 {
99 unsigned long date;
100
101 if (memcmp(buf, "author", 6))
102 return 0;
103 while (*buf++ != '\n')
104 /* nada */;
105 if (memcmp(buf, "committer", 9))
106 return 0;
107 while (*buf++ != '>')
108 /* nada */;
109 date = strtoul(buf, NULL, 10);
110 if (date == ULONG_MAX)
111 date = 0;
112 return date;
113 }
114
115 static struct commit_graft **commit_graft;
116 static int commit_graft_alloc, commit_graft_nr;
117
118 static int commit_graft_pos(const unsigned char *sha1)
119 {
120 int lo, hi;
121 lo = 0;
122 hi = commit_graft_nr;
123 while (lo < hi) {
124 int mi = (lo + hi) / 2;
125 struct commit_graft *graft = commit_graft[mi];
126 int cmp = memcmp(sha1, graft->sha1, 20);
127 if (!cmp)
128 return mi;
129 if (cmp < 0)
130 hi = mi;
131 else
132 lo = mi + 1;
133 }
134 return -lo - 1;
135 }
136
137 int register_commit_graft(struct commit_graft *graft, int ignore_dups)
138 {
139 int pos = commit_graft_pos(graft->sha1);
140
141 if (0 <= pos) {
142 if (ignore_dups)
143 free(graft);
144 else {
145 free(commit_graft[pos]);
146 commit_graft[pos] = graft;
147 }
148 return 1;
149 }
150 pos = -pos - 1;
151 if (commit_graft_alloc <= ++commit_graft_nr) {
152 commit_graft_alloc = alloc_nr(commit_graft_alloc);
153 commit_graft = xrealloc(commit_graft,
154 sizeof(*commit_graft) *
155 commit_graft_alloc);
156 }
157 if (pos < commit_graft_nr)
158 memmove(commit_graft + pos + 1,
159 commit_graft + pos,
160 (commit_graft_nr - pos - 1) *
161 sizeof(*commit_graft));
162 commit_graft[pos] = graft;
163 return 0;
164 }
165
166 struct commit_graft *read_graft_line(char *buf, int len)
167 {
168 /* The format is just "Commit Parent1 Parent2 ...\n" */
169 int i;
170 struct commit_graft *graft = NULL;
171
172 if (buf[len-1] == '\n')
173 buf[--len] = 0;
174 if (buf[0] == '#' || buf[0] == '\0')
175 return NULL;
176 if ((len + 1) % 41) {
177 bad_graft_data:
178 error("bad graft data: %s", buf);
179 free(graft);
180 return NULL;
181 }
182 i = (len + 1) / 41 - 1;
183 graft = xmalloc(sizeof(*graft) + 20 * i);
184 graft->nr_parent = i;
185 if (get_sha1_hex(buf, graft->sha1))
186 goto bad_graft_data;
187 for (i = 40; i < len; i += 41) {
188 if (buf[i] != ' ')
189 goto bad_graft_data;
190 if (get_sha1_hex(buf + i + 1, graft->parent[i/41]))
191 goto bad_graft_data;
192 }
193 return graft;
194 }
195
196 int read_graft_file(const char *graft_file)
197 {
198 FILE *fp = fopen(graft_file, "r");
199 char buf[1024];
200 if (!fp)
201 return -1;
202 while (fgets(buf, sizeof(buf), fp)) {
203 /* The format is just "Commit Parent1 Parent2 ...\n" */
204 int len = strlen(buf);
205 struct commit_graft *graft = read_graft_line(buf, len);
206 if (!graft)
207 continue;
208 if (register_commit_graft(graft, 1))
209 error("duplicate graft data: %s", buf);
210 }
211 fclose(fp);
212 return 0;
213 }
214
215 static void prepare_commit_graft(void)
216 {
217 static int commit_graft_prepared;
218 char *graft_file;
219
220 if (commit_graft_prepared)
221 return;
222 graft_file = get_graft_file();
223 read_graft_file(graft_file);
224 commit_graft_prepared = 1;
225 }
226
227 static struct commit_graft *lookup_commit_graft(const unsigned char *sha1)
228 {
229 int pos;
230 prepare_commit_graft();
231 pos = commit_graft_pos(sha1);
232 if (pos < 0)
233 return NULL;
234 return commit_graft[pos];
235 }
236
237 int parse_commit_buffer(struct commit *item, void *buffer, unsigned long size)
238 {
239 char *bufptr = buffer;
240 unsigned char parent[20];
241 struct commit_list **pptr;
242 struct commit_graft *graft;
243 unsigned n_refs = 0;
244
245 if (item->object.parsed)
246 return 0;
247 item->object.parsed = 1;
248 if (memcmp(bufptr, "tree ", 5))
249 return error("bogus commit object %s", sha1_to_hex(item->object.sha1));
250 if (get_sha1_hex(bufptr + 5, parent) < 0)
251 return error("bad tree pointer in commit %s",
252 sha1_to_hex(item->object.sha1));
253 item->tree = lookup_tree(parent);
254 if (item->tree)
255 n_refs++;
256 bufptr += 46; /* "tree " + "hex sha1" + "\n" */
257 pptr = &item->parents;
258
259 graft = lookup_commit_graft(item->object.sha1);
260 while (!memcmp(bufptr, "parent ", 7)) {
261 struct commit *new_parent;
262
263 if (get_sha1_hex(bufptr + 7, parent) || bufptr[47] != '\n')
264 return error("bad parents in commit %s", sha1_to_hex(item->object.sha1));
265 bufptr += 48;
266 if (graft)
267 continue;
268 new_parent = lookup_commit(parent);
269 if (new_parent) {
270 pptr = &commit_list_insert(new_parent, pptr)->next;
271 n_refs++;
272 }
273 }
274 if (graft) {
275 int i;
276 struct commit *new_parent;
277 for (i = 0; i < graft->nr_parent; i++) {
278 new_parent = lookup_commit(graft->parent[i]);
279 if (!new_parent)
280 continue;
281 pptr = &commit_list_insert(new_parent, pptr)->next;
282 n_refs++;
283 }
284 }
285 item->date = parse_commit_date(bufptr);
286
287 if (track_object_refs) {
288 unsigned i = 0;
289 struct commit_list *p;
290 struct object_refs *refs = alloc_object_refs(n_refs);
291 if (item->tree)
292 refs->ref[i++] = &item->tree->object;
293 for (p = item->parents; p; p = p->next)
294 refs->ref[i++] = &p->item->object;
295 set_object_refs(&item->object, refs);
296 }
297
298 return 0;
299 }
300
301 int parse_commit(struct commit *item)
302 {
303 char type[20];
304 void *buffer;
305 unsigned long size;
306 int ret;
307
308 if (item->object.parsed)
309 return 0;
310 buffer = read_sha1_file(item->object.sha1, type, &size);
311 if (!buffer)
312 return error("Could not read %s",
313 sha1_to_hex(item->object.sha1));
314 if (strcmp(type, commit_type)) {
315 free(buffer);
316 return error("Object %s not a commit",
317 sha1_to_hex(item->object.sha1));
318 }
319 ret = parse_commit_buffer(item, buffer, size);
320 if (save_commit_buffer && !ret) {
321 item->buffer = buffer;
322 return 0;
323 }
324 free(buffer);
325 return ret;
326 }
327
328 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
329 {
330 struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
331 new_list->item = item;
332 new_list->next = *list_p;
333 *list_p = new_list;
334 return new_list;
335 }
336
337 void free_commit_list(struct commit_list *list)
338 {
339 while (list) {
340 struct commit_list *temp = list;
341 list = temp->next;
342 free(temp);
343 }
344 }
345
346 struct commit_list * insert_by_date(struct commit *item, struct commit_list **list)
347 {
348 struct commit_list **pp = list;
349 struct commit_list *p;
350 while ((p = *pp) != NULL) {
351 if (p->item->date < item->date) {
352 break;
353 }
354 pp = &p->next;
355 }
356 return commit_list_insert(item, pp);
357 }
358
359
360 void sort_by_date(struct commit_list **list)
361 {
362 struct commit_list *ret = NULL;
363 while (*list) {
364 insert_by_date((*list)->item, &ret);
365 *list = (*list)->next;
366 }
367 *list = ret;
368 }
369
370 struct commit *pop_most_recent_commit(struct commit_list **list,
371 unsigned int mark)
372 {
373 struct commit *ret = (*list)->item;
374 struct commit_list *parents = ret->parents;
375 struct commit_list *old = *list;
376
377 *list = (*list)->next;
378 free(old);
379
380 while (parents) {
381 struct commit *commit = parents->item;
382 parse_commit(commit);
383 if (!(commit->object.flags & mark)) {
384 commit->object.flags |= mark;
385 insert_by_date(commit, list);
386 }
387 parents = parents->next;
388 }
389 return ret;
390 }
391
392 void clear_commit_marks(struct commit *commit, unsigned int mark)
393 {
394 struct commit_list *parents;
395
396 parents = commit->parents;
397 commit->object.flags &= ~mark;
398 while (parents) {
399 struct commit *parent = parents->item;
400 if (parent && parent->object.parsed &&
401 (parent->object.flags & mark))
402 clear_commit_marks(parent, mark);
403 parents = parents->next;
404 }
405 }
406
407 /*
408 * Generic support for pretty-printing the header
409 */
410 static int get_one_line(const char *msg, unsigned long len)
411 {
412 int ret = 0;
413
414 while (len--) {
415 char c = *msg++;
416 if (!c)
417 break;
418 ret++;
419 if (c == '\n')
420 break;
421 }
422 return ret;
423 }
424
425 static int is_rfc2047_special(char ch)
426 {
427 return ((ch & 0x80) || (ch == '=') || (ch == '?') || (ch == '_'));
428 }
429
430 static int add_rfc2047(char *buf, const char *line, int len)
431 {
432 char *bp = buf;
433 int i, needquote;
434 static const char q_utf8[] = "=?utf-8?q?";
435
436 for (i = needquote = 0; !needquote && i < len; i++) {
437 unsigned ch = line[i];
438 if (ch & 0x80)
439 needquote++;
440 if ((i + 1 < len) &&
441 (ch == '=' && line[i+1] == '?'))
442 needquote++;
443 }
444 if (!needquote)
445 return sprintf(buf, "%.*s", len, line);
446
447 memcpy(bp, q_utf8, sizeof(q_utf8)-1);
448 bp += sizeof(q_utf8)-1;
449 for (i = 0; i < len; i++) {
450 unsigned ch = line[i];
451 if (is_rfc2047_special(ch)) {
452 sprintf(bp, "=%02X", ch);
453 bp += 3;
454 }
455 else if (ch == ' ')
456 *bp++ = '_';
457 else
458 *bp++ = ch;
459 }
460 memcpy(bp, "?=", 2);
461 bp += 2;
462 return bp - buf;
463 }
464
465 static int add_user_info(const char *what, enum cmit_fmt fmt, char *buf, const char *line)
466 {
467 char *date;
468 int namelen;
469 unsigned long time;
470 int tz, ret;
471 const char *filler = " ";
472
473 if (fmt == CMIT_FMT_ONELINE)
474 return 0;
475 date = strchr(line, '>');
476 if (!date)
477 return 0;
478 namelen = ++date - line;
479 time = strtoul(date, &date, 10);
480 tz = strtol(date, NULL, 10);
481
482 if (fmt == CMIT_FMT_EMAIL) {
483 char *name_tail = strchr(line, '<');
484 int display_name_length;
485 if (!name_tail)
486 return 0;
487 while (line < name_tail && isspace(name_tail[-1]))
488 name_tail--;
489 display_name_length = name_tail - line;
490 filler = "";
491 strcpy(buf, "From: ");
492 ret = strlen(buf);
493 ret += add_rfc2047(buf + ret, line, display_name_length);
494 memcpy(buf + ret, name_tail, namelen - display_name_length);
495 ret += namelen - display_name_length;
496 buf[ret++] = '\n';
497 }
498 else {
499 ret = sprintf(buf, "%s: %.*s%.*s\n", what,
500 (fmt == CMIT_FMT_FULLER) ? 4 : 0,
501 filler, namelen, line);
502 }
503 switch (fmt) {
504 case CMIT_FMT_MEDIUM:
505 ret += sprintf(buf + ret, "Date: %s\n", show_date(time, tz));
506 break;
507 case CMIT_FMT_EMAIL:
508 ret += sprintf(buf + ret, "Date: %s\n",
509 show_rfc2822_date(time, tz));
510 break;
511 case CMIT_FMT_FULLER:
512 ret += sprintf(buf + ret, "%sDate: %s\n", what, show_date(time, tz));
513 break;
514 default:
515 /* notin' */
516 break;
517 }
518 return ret;
519 }
520
521 static int is_empty_line(const char *line, int *len_p)
522 {
523 int len = *len_p;
524 while (len && isspace(line[len-1]))
525 len--;
526 *len_p = len;
527 return !len;
528 }
529
530 static int add_merge_info(enum cmit_fmt fmt, char *buf, const struct commit *commit, int abbrev)
531 {
532 struct commit_list *parent = commit->parents;
533 int offset;
534
535 if ((fmt == CMIT_FMT_ONELINE) || (fmt == CMIT_FMT_EMAIL) ||
536 !parent || !parent->next)
537 return 0;
538
539 offset = sprintf(buf, "Merge:");
540
541 while (parent) {
542 struct commit *p = parent->item;
543 const char *hex = abbrev
544 ? find_unique_abbrev(p->object.sha1, abbrev)
545 : sha1_to_hex(p->object.sha1);
546 char *dots = (abbrev && strlen(hex) != 40) ? "..." : "";
547 parent = parent->next;
548
549 offset += sprintf(buf + offset, " %s%s", hex, dots);
550 }
551 buf[offset++] = '\n';
552 return offset;
553 }
554
555 unsigned long pretty_print_commit(enum cmit_fmt fmt, const struct commit *commit, unsigned long len, char *buf, unsigned long space, int abbrev, const char *subject, const char *after_subject)
556 {
557 int hdr = 1, body = 0;
558 unsigned long offset = 0;
559 int indent = 4;
560 int parents_shown = 0;
561 const char *msg = commit->buffer;
562 int plain_non_ascii = 0;
563
564 if (fmt == CMIT_FMT_ONELINE || fmt == CMIT_FMT_EMAIL)
565 indent = 0;
566
567 /* After-subject is used to pass in Content-Type: multipart
568 * MIME header; in that case we do not have to do the
569 * plaintext content type even if the commit message has
570 * non 7-bit ASCII character. Otherwise, check if we need
571 * to say this is not a 7-bit ASCII.
572 */
573 if (fmt == CMIT_FMT_EMAIL && !after_subject) {
574 int i;
575 for (i = 0; !plain_non_ascii && msg[i] && i < len; i++)
576 if (msg[i] & 0x80)
577 plain_non_ascii = 1;
578 }
579
580 for (;;) {
581 const char *line = msg;
582 int linelen = get_one_line(msg, len);
583
584 if (!linelen)
585 break;
586
587 /*
588 * We want some slop for indentation and a possible
589 * final "...". Thus the "+ 20".
590 */
591 if (offset + linelen + 20 > space) {
592 memcpy(buf + offset, " ...\n", 8);
593 offset += 8;
594 break;
595 }
596
597 msg += linelen;
598 len -= linelen;
599 if (hdr) {
600 if (linelen == 1) {
601 hdr = 0;
602 if ((fmt != CMIT_FMT_ONELINE) && !subject)
603 buf[offset++] = '\n';
604 continue;
605 }
606 if (fmt == CMIT_FMT_RAW) {
607 memcpy(buf + offset, line, linelen);
608 offset += linelen;
609 continue;
610 }
611 if (!memcmp(line, "parent ", 7)) {
612 if (linelen != 48)
613 die("bad parent line in commit");
614 continue;
615 }
616
617 if (!parents_shown) {
618 offset += add_merge_info(fmt, buf + offset,
619 commit, abbrev);
620 parents_shown = 1;
621 continue;
622 }
623 /*
624 * MEDIUM == DEFAULT shows only author with dates.
625 * FULL shows both authors but not dates.
626 * FULLER shows both authors and dates.
627 */
628 if (!memcmp(line, "author ", 7))
629 offset += add_user_info("Author", fmt,
630 buf + offset,
631 line + 7);
632 if (!memcmp(line, "committer ", 10) &&
633 (fmt == CMIT_FMT_FULL || fmt == CMIT_FMT_FULLER))
634 offset += add_user_info("Commit", fmt,
635 buf + offset,
636 line + 10);
637 continue;
638 }
639
640 if (is_empty_line(line, &linelen)) {
641 if (!body)
642 continue;
643 if (subject)
644 continue;
645 if (fmt == CMIT_FMT_SHORT)
646 break;
647 } else {
648 body = 1;
649 }
650
651 if (subject) {
652 int slen = strlen(subject);
653 memcpy(buf + offset, subject, slen);
654 offset += slen;
655 offset += add_rfc2047(buf + offset, line, linelen);
656 }
657 else {
658 memset(buf + offset, ' ', indent);
659 memcpy(buf + offset + indent, line, linelen);
660 offset += linelen + indent;
661 }
662 buf[offset++] = '\n';
663 if (fmt == CMIT_FMT_ONELINE)
664 break;
665 if (subject && plain_non_ascii) {
666 static const char header[] =
667 "Content-Type: text/plain; charset=UTF-8\n"
668 "Content-Transfer-Encoding: 8bit\n";
669 memcpy(buf + offset, header, sizeof(header)-1);
670 offset += sizeof(header)-1;
671 }
672 if (after_subject) {
673 int slen = strlen(after_subject);
674 if (slen > space - offset - 1)
675 slen = space - offset - 1;
676 memcpy(buf + offset, after_subject, slen);
677 offset += slen;
678 after_subject = NULL;
679 }
680 subject = NULL;
681 }
682 while (offset && isspace(buf[offset-1]))
683 offset--;
684 /* Make sure there is an EOLN for the non-oneline case */
685 if (fmt != CMIT_FMT_ONELINE)
686 buf[offset++] = '\n';
687 buf[offset] = '\0';
688 return offset;
689 }
690
691 struct commit *pop_commit(struct commit_list **stack)
692 {
693 struct commit_list *top = *stack;
694 struct commit *item = top ? top->item : NULL;
695
696 if (top) {
697 *stack = top->next;
698 free(top);
699 }
700 return item;
701 }
702
703 int count_parents(struct commit * commit)
704 {
705 int count = 0;
706 struct commit_list * parents = commit->parents;
707 for (count=0;parents; parents=parents->next,count++)
708 ;
709 return count;
710 }
711
712 void topo_sort_default_setter(struct commit *c, void *data)
713 {
714 c->object.util = data;
715 }
716
717 void *topo_sort_default_getter(struct commit *c)
718 {
719 return c->object.util;
720 }
721
722 /*
723 * Performs an in-place topological sort on the list supplied.
724 */
725 void sort_in_topological_order(struct commit_list ** list, int lifo)
726 {
727 sort_in_topological_order_fn(list, lifo, topo_sort_default_setter,
728 topo_sort_default_getter);
729 }
730
731 void sort_in_topological_order_fn(struct commit_list ** list, int lifo,
732 topo_sort_set_fn_t setter,
733 topo_sort_get_fn_t getter)
734 {
735 struct commit_list * next = *list;
736 struct commit_list * work = NULL, **insert;
737 struct commit_list ** pptr = list;
738 struct sort_node * nodes;
739 struct sort_node * next_nodes;
740 int count = 0;
741
742 /* determine the size of the list */
743 while (next) {
744 next = next->next;
745 count++;
746 }
747
748 if (!count)
749 return;
750 /* allocate an array to help sort the list */
751 nodes = xcalloc(count, sizeof(*nodes));
752 /* link the list to the array */
753 next_nodes = nodes;
754 next=*list;
755 while (next) {
756 next_nodes->list_item = next;
757 setter(next->item, next_nodes);
758 next_nodes++;
759 next = next->next;
760 }
761 /* update the indegree */
762 next=*list;
763 while (next) {
764 struct commit_list * parents = next->item->parents;
765 while (parents) {
766 struct commit * parent=parents->item;
767 struct sort_node * pn = (struct sort_node *) getter(parent);
768
769 if (pn)
770 pn->indegree++;
771 parents=parents->next;
772 }
773 next=next->next;
774 }
775 /*
776 * find the tips
777 *
778 * tips are nodes not reachable from any other node in the list
779 *
780 * the tips serve as a starting set for the work queue.
781 */
782 next=*list;
783 insert = &work;
784 while (next) {
785 struct sort_node * node = (struct sort_node *) getter(next->item);
786
787 if (node->indegree == 0) {
788 insert = &commit_list_insert(next->item, insert)->next;
789 }
790 next=next->next;
791 }
792
793 /* process the list in topological order */
794 if (!lifo)
795 sort_by_date(&work);
796 while (work) {
797 struct commit * work_item = pop_commit(&work);
798 struct sort_node * work_node = (struct sort_node *) getter(work_item);
799 struct commit_list * parents = work_item->parents;
800
801 while (parents) {
802 struct commit * parent=parents->item;
803 struct sort_node * pn = (struct sort_node *) getter(parent);
804
805 if (pn) {
806 /*
807 * parents are only enqueued for emission
808 * when all their children have been emitted thereby
809 * guaranteeing topological order.
810 */
811 pn->indegree--;
812 if (!pn->indegree) {
813 if (!lifo)
814 insert_by_date(parent, &work);
815 else
816 commit_list_insert(parent, &work);
817 }
818 }
819 parents=parents->next;
820 }
821 /*
822 * work_item is a commit all of whose children
823 * have already been emitted. we can emit it now.
824 */
825 *pptr = work_node->list_item;
826 pptr = &(*pptr)->next;
827 *pptr = NULL;
828 setter(work_item, NULL);
829 }
830 free(nodes);
831 }