xdiff: generate "anti-diffs" aka what is common to two files
[git/git-svn.git] / commit.c
blobe51ffa1c6cf5948c0e6c6d0b905fc868f4464ccf
1 #include "cache.h"
2 #include "tag.h"
3 #include "commit.h"
5 int save_commit_buffer = 1;
7 struct sort_node
9 /*
10 * the number of children of the associated commit
11 * that also occur in the list being sorted.
13 unsigned int indegree;
16 * reference to original list item that we will re-use
17 * on output.
19 struct commit_list * list_item;
23 const char *commit_type = "commit";
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 },
39 enum cmit_fmt get_commit_format(const char *arg)
41 int i;
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;
52 die("invalid --pretty format: %s", arg);
55 static struct commit *check_commit(struct object *obj,
56 const unsigned char *sha1,
57 int quiet)
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;
65 return (struct commit *) obj;
68 struct commit *lookup_commit_reference_gently(const unsigned char *sha1,
69 int quiet)
71 struct object *obj = deref_tag(parse_object(sha1), NULL, 0);
73 if (!obj)
74 return NULL;
75 return check_commit(obj, sha1, quiet);
78 struct commit *lookup_commit_reference(const unsigned char *sha1)
80 return lookup_commit_reference_gently(sha1, 0);
83 struct commit *lookup_commit(const unsigned char *sha1)
85 struct object *obj = lookup_object(sha1);
86 if (!obj) {
87 struct commit *ret = alloc_commit_node();
88 created_object(sha1, &ret->object);
89 ret->object.type = TYPE_COMMIT;
90 return ret;
92 if (!obj->type)
93 obj->type = TYPE_COMMIT;
94 return check_commit(obj, sha1, 0);
97 static unsigned long parse_commit_date(const char *buf)
99 unsigned long date;
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;
115 static struct commit_graft **commit_graft;
116 static int commit_graft_alloc, commit_graft_nr;
118 static int commit_graft_pos(const unsigned char *sha1)
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;
134 return -lo - 1;
137 int register_commit_graft(struct commit_graft *graft, int ignore_dups)
139 int pos = commit_graft_pos(graft->sha1);
141 if (0 <= pos) {
142 if (ignore_dups)
143 free(graft);
144 else {
145 free(commit_graft[pos]);
146 commit_graft[pos] = graft;
148 return 1;
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);
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;
166 struct commit_graft *read_graft_line(char *buf, int len)
168 /* The format is just "Commit Parent1 Parent2 ...\n" */
169 int i;
170 struct commit_graft *graft = NULL;
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;
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;
193 return graft;
196 int read_graft_file(const char *graft_file)
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);
211 fclose(fp);
212 return 0;
215 static void prepare_commit_graft(void)
217 static int commit_graft_prepared;
218 char *graft_file;
220 if (commit_graft_prepared)
221 return;
222 graft_file = get_graft_file();
223 read_graft_file(graft_file);
224 commit_graft_prepared = 1;
227 static struct commit_graft *lookup_commit_graft(const unsigned char *sha1)
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];
237 int parse_commit_buffer(struct commit *item, void *buffer, unsigned long size)
239 char *tail = buffer;
240 char *bufptr = buffer;
241 unsigned char parent[20];
242 struct commit_list **pptr;
243 struct commit_graft *graft;
244 unsigned n_refs = 0;
246 if (item->object.parsed)
247 return 0;
248 item->object.parsed = 1;
249 tail += size;
250 if (tail <= bufptr + 5 || memcmp(bufptr, "tree ", 5))
251 return error("bogus commit object %s", sha1_to_hex(item->object.sha1));
252 if (tail <= bufptr + 45 || get_sha1_hex(bufptr + 5, parent) < 0)
253 return error("bad tree pointer in commit %s",
254 sha1_to_hex(item->object.sha1));
255 item->tree = lookup_tree(parent);
256 if (item->tree)
257 n_refs++;
258 bufptr += 46; /* "tree " + "hex sha1" + "\n" */
259 pptr = &item->parents;
261 graft = lookup_commit_graft(item->object.sha1);
262 while (bufptr + 48 < tail && !memcmp(bufptr, "parent ", 7)) {
263 struct commit *new_parent;
265 if (tail <= bufptr + 48 ||
266 get_sha1_hex(bufptr + 7, parent) ||
267 bufptr[47] != '\n')
268 return error("bad parents in commit %s", sha1_to_hex(item->object.sha1));
269 bufptr += 48;
270 if (graft)
271 continue;
272 new_parent = lookup_commit(parent);
273 if (new_parent) {
274 pptr = &commit_list_insert(new_parent, pptr)->next;
275 n_refs++;
278 if (graft) {
279 int i;
280 struct commit *new_parent;
281 for (i = 0; i < graft->nr_parent; i++) {
282 new_parent = lookup_commit(graft->parent[i]);
283 if (!new_parent)
284 continue;
285 pptr = &commit_list_insert(new_parent, pptr)->next;
286 n_refs++;
289 item->date = parse_commit_date(bufptr);
291 if (track_object_refs) {
292 unsigned i = 0;
293 struct commit_list *p;
294 struct object_refs *refs = alloc_object_refs(n_refs);
295 if (item->tree)
296 refs->ref[i++] = &item->tree->object;
297 for (p = item->parents; p; p = p->next)
298 refs->ref[i++] = &p->item->object;
299 set_object_refs(&item->object, refs);
302 return 0;
305 int parse_commit(struct commit *item)
307 char type[20];
308 void *buffer;
309 unsigned long size;
310 int ret;
312 if (item->object.parsed)
313 return 0;
314 buffer = read_sha1_file(item->object.sha1, type, &size);
315 if (!buffer)
316 return error("Could not read %s",
317 sha1_to_hex(item->object.sha1));
318 if (strcmp(type, commit_type)) {
319 free(buffer);
320 return error("Object %s not a commit",
321 sha1_to_hex(item->object.sha1));
323 ret = parse_commit_buffer(item, buffer, size);
324 if (save_commit_buffer && !ret) {
325 item->buffer = buffer;
326 return 0;
328 free(buffer);
329 return ret;
332 struct commit_list *commit_list_insert(struct commit *item, struct commit_list **list_p)
334 struct commit_list *new_list = xmalloc(sizeof(struct commit_list));
335 new_list->item = item;
336 new_list->next = *list_p;
337 *list_p = new_list;
338 return new_list;
341 void free_commit_list(struct commit_list *list)
343 while (list) {
344 struct commit_list *temp = list;
345 list = temp->next;
346 free(temp);
350 struct commit_list * insert_by_date(struct commit *item, struct commit_list **list)
352 struct commit_list **pp = list;
353 struct commit_list *p;
354 while ((p = *pp) != NULL) {
355 if (p->item->date < item->date) {
356 break;
358 pp = &p->next;
360 return commit_list_insert(item, pp);
364 void sort_by_date(struct commit_list **list)
366 struct commit_list *ret = NULL;
367 while (*list) {
368 insert_by_date((*list)->item, &ret);
369 *list = (*list)->next;
371 *list = ret;
374 struct commit *pop_most_recent_commit(struct commit_list **list,
375 unsigned int mark)
377 struct commit *ret = (*list)->item;
378 struct commit_list *parents = ret->parents;
379 struct commit_list *old = *list;
381 *list = (*list)->next;
382 free(old);
384 while (parents) {
385 struct commit *commit = parents->item;
386 parse_commit(commit);
387 if (!(commit->object.flags & mark)) {
388 commit->object.flags |= mark;
389 insert_by_date(commit, list);
391 parents = parents->next;
393 return ret;
396 void clear_commit_marks(struct commit *commit, unsigned int mark)
398 struct commit_list *parents;
400 parents = commit->parents;
401 commit->object.flags &= ~mark;
402 while (parents) {
403 struct commit *parent = parents->item;
404 if (parent && parent->object.parsed &&
405 (parent->object.flags & mark))
406 clear_commit_marks(parent, mark);
407 parents = parents->next;
412 * Generic support for pretty-printing the header
414 static int get_one_line(const char *msg, unsigned long len)
416 int ret = 0;
418 while (len--) {
419 char c = *msg++;
420 if (!c)
421 break;
422 ret++;
423 if (c == '\n')
424 break;
426 return ret;
429 static int is_rfc2047_special(char ch)
431 return ((ch & 0x80) || (ch == '=') || (ch == '?') || (ch == '_'));
434 static int add_rfc2047(char *buf, const char *line, int len)
436 char *bp = buf;
437 int i, needquote;
438 static const char q_utf8[] = "=?utf-8?q?";
440 for (i = needquote = 0; !needquote && i < len; i++) {
441 unsigned ch = line[i];
442 if (ch & 0x80)
443 needquote++;
444 if ((i + 1 < len) &&
445 (ch == '=' && line[i+1] == '?'))
446 needquote++;
448 if (!needquote)
449 return sprintf(buf, "%.*s", len, line);
451 memcpy(bp, q_utf8, sizeof(q_utf8)-1);
452 bp += sizeof(q_utf8)-1;
453 for (i = 0; i < len; i++) {
454 unsigned ch = line[i] & 0xFF;
455 if (is_rfc2047_special(ch)) {
456 sprintf(bp, "=%02X", ch);
457 bp += 3;
459 else if (ch == ' ')
460 *bp++ = '_';
461 else
462 *bp++ = ch;
464 memcpy(bp, "?=", 2);
465 bp += 2;
466 return bp - buf;
469 static int add_user_info(const char *what, enum cmit_fmt fmt, char *buf, const char *line)
471 char *date;
472 int namelen;
473 unsigned long time;
474 int tz, ret;
475 const char *filler = " ";
477 if (fmt == CMIT_FMT_ONELINE)
478 return 0;
479 date = strchr(line, '>');
480 if (!date)
481 return 0;
482 namelen = ++date - line;
483 time = strtoul(date, &date, 10);
484 tz = strtol(date, NULL, 10);
486 if (fmt == CMIT_FMT_EMAIL) {
487 char *name_tail = strchr(line, '<');
488 int display_name_length;
489 if (!name_tail)
490 return 0;
491 while (line < name_tail && isspace(name_tail[-1]))
492 name_tail--;
493 display_name_length = name_tail - line;
494 filler = "";
495 strcpy(buf, "From: ");
496 ret = strlen(buf);
497 ret += add_rfc2047(buf + ret, line, display_name_length);
498 memcpy(buf + ret, name_tail, namelen - display_name_length);
499 ret += namelen - display_name_length;
500 buf[ret++] = '\n';
502 else {
503 ret = sprintf(buf, "%s: %.*s%.*s\n", what,
504 (fmt == CMIT_FMT_FULLER) ? 4 : 0,
505 filler, namelen, line);
507 switch (fmt) {
508 case CMIT_FMT_MEDIUM:
509 ret += sprintf(buf + ret, "Date: %s\n", show_date(time, tz));
510 break;
511 case CMIT_FMT_EMAIL:
512 ret += sprintf(buf + ret, "Date: %s\n",
513 show_rfc2822_date(time, tz));
514 break;
515 case CMIT_FMT_FULLER:
516 ret += sprintf(buf + ret, "%sDate: %s\n", what, show_date(time, tz));
517 break;
518 default:
519 /* notin' */
520 break;
522 return ret;
525 static int is_empty_line(const char *line, int *len_p)
527 int len = *len_p;
528 while (len && isspace(line[len-1]))
529 len--;
530 *len_p = len;
531 return !len;
534 static int add_merge_info(enum cmit_fmt fmt, char *buf, const struct commit *commit, int abbrev)
536 struct commit_list *parent = commit->parents;
537 int offset;
539 if ((fmt == CMIT_FMT_ONELINE) || (fmt == CMIT_FMT_EMAIL) ||
540 !parent || !parent->next)
541 return 0;
543 offset = sprintf(buf, "Merge:");
545 while (parent) {
546 struct commit *p = parent->item;
547 const char *hex = abbrev
548 ? find_unique_abbrev(p->object.sha1, abbrev)
549 : sha1_to_hex(p->object.sha1);
550 const char *dots = (abbrev && strlen(hex) != 40) ? "..." : "";
551 parent = parent->next;
553 offset += sprintf(buf + offset, " %s%s", hex, dots);
555 buf[offset++] = '\n';
556 return offset;
559 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)
561 int hdr = 1, body = 0;
562 unsigned long offset = 0;
563 int indent = 4;
564 int parents_shown = 0;
565 const char *msg = commit->buffer;
566 int plain_non_ascii = 0;
568 if (fmt == CMIT_FMT_ONELINE || fmt == CMIT_FMT_EMAIL)
569 indent = 0;
571 /* After-subject is used to pass in Content-Type: multipart
572 * MIME header; in that case we do not have to do the
573 * plaintext content type even if the commit message has
574 * non 7-bit ASCII character. Otherwise, check if we need
575 * to say this is not a 7-bit ASCII.
577 if (fmt == CMIT_FMT_EMAIL && !after_subject) {
578 int i, ch, in_body;
580 for (in_body = i = 0; (ch = msg[i]) && i < len; i++) {
581 if (!in_body) {
582 /* author could be non 7-bit ASCII but
583 * the log may so; skip over the
584 * header part first.
586 if (ch == '\n' &&
587 i + 1 < len && msg[i+1] == '\n')
588 in_body = 1;
590 else if (ch & 0x80) {
591 plain_non_ascii = 1;
592 break;
597 for (;;) {
598 const char *line = msg;
599 int linelen = get_one_line(msg, len);
601 if (!linelen)
602 break;
605 * We want some slop for indentation and a possible
606 * final "...". Thus the "+ 20".
608 if (offset + linelen + 20 > space) {
609 memcpy(buf + offset, " ...\n", 8);
610 offset += 8;
611 break;
614 msg += linelen;
615 len -= linelen;
616 if (hdr) {
617 if (linelen == 1) {
618 hdr = 0;
619 if ((fmt != CMIT_FMT_ONELINE) && !subject)
620 buf[offset++] = '\n';
621 continue;
623 if (fmt == CMIT_FMT_RAW) {
624 memcpy(buf + offset, line, linelen);
625 offset += linelen;
626 continue;
628 if (!memcmp(line, "parent ", 7)) {
629 if (linelen != 48)
630 die("bad parent line in commit");
631 continue;
634 if (!parents_shown) {
635 offset += add_merge_info(fmt, buf + offset,
636 commit, abbrev);
637 parents_shown = 1;
638 continue;
641 * MEDIUM == DEFAULT shows only author with dates.
642 * FULL shows both authors but not dates.
643 * FULLER shows both authors and dates.
645 if (!memcmp(line, "author ", 7))
646 offset += add_user_info("Author", fmt,
647 buf + offset,
648 line + 7);
649 if (!memcmp(line, "committer ", 10) &&
650 (fmt == CMIT_FMT_FULL || fmt == CMIT_FMT_FULLER))
651 offset += add_user_info("Commit", fmt,
652 buf + offset,
653 line + 10);
654 continue;
657 if (is_empty_line(line, &linelen)) {
658 if (!body)
659 continue;
660 if (subject)
661 continue;
662 if (fmt == CMIT_FMT_SHORT)
663 break;
664 } else {
665 body = 1;
668 if (subject) {
669 int slen = strlen(subject);
670 memcpy(buf + offset, subject, slen);
671 offset += slen;
672 offset += add_rfc2047(buf + offset, line, linelen);
674 else {
675 memset(buf + offset, ' ', indent);
676 memcpy(buf + offset + indent, line, linelen);
677 offset += linelen + indent;
679 buf[offset++] = '\n';
680 if (fmt == CMIT_FMT_ONELINE)
681 break;
682 if (subject && plain_non_ascii) {
683 static const char header[] =
684 "Content-Type: text/plain; charset=UTF-8\n"
685 "Content-Transfer-Encoding: 8bit\n";
686 memcpy(buf + offset, header, sizeof(header)-1);
687 offset += sizeof(header)-1;
689 if (after_subject) {
690 int slen = strlen(after_subject);
691 if (slen > space - offset - 1)
692 slen = space - offset - 1;
693 memcpy(buf + offset, after_subject, slen);
694 offset += slen;
695 after_subject = NULL;
697 subject = NULL;
699 while (offset && isspace(buf[offset-1]))
700 offset--;
701 /* Make sure there is an EOLN for the non-oneline case */
702 if (fmt != CMIT_FMT_ONELINE)
703 buf[offset++] = '\n';
704 buf[offset] = '\0';
705 return offset;
708 struct commit *pop_commit(struct commit_list **stack)
710 struct commit_list *top = *stack;
711 struct commit *item = top ? top->item : NULL;
713 if (top) {
714 *stack = top->next;
715 free(top);
717 return item;
720 int count_parents(struct commit * commit)
722 int count = 0;
723 struct commit_list * parents = commit->parents;
724 for (count=0;parents; parents=parents->next,count++)
726 return count;
729 void topo_sort_default_setter(struct commit *c, void *data)
731 c->util = data;
734 void *topo_sort_default_getter(struct commit *c)
736 return c->util;
740 * Performs an in-place topological sort on the list supplied.
742 void sort_in_topological_order(struct commit_list ** list, int lifo)
744 sort_in_topological_order_fn(list, lifo, topo_sort_default_setter,
745 topo_sort_default_getter);
748 void sort_in_topological_order_fn(struct commit_list ** list, int lifo,
749 topo_sort_set_fn_t setter,
750 topo_sort_get_fn_t getter)
752 struct commit_list * next = *list;
753 struct commit_list * work = NULL, **insert;
754 struct commit_list ** pptr = list;
755 struct sort_node * nodes;
756 struct sort_node * next_nodes;
757 int count = 0;
759 /* determine the size of the list */
760 while (next) {
761 next = next->next;
762 count++;
765 if (!count)
766 return;
767 /* allocate an array to help sort the list */
768 nodes = xcalloc(count, sizeof(*nodes));
769 /* link the list to the array */
770 next_nodes = nodes;
771 next=*list;
772 while (next) {
773 next_nodes->list_item = next;
774 setter(next->item, next_nodes);
775 next_nodes++;
776 next = next->next;
778 /* update the indegree */
779 next=*list;
780 while (next) {
781 struct commit_list * parents = next->item->parents;
782 while (parents) {
783 struct commit * parent=parents->item;
784 struct sort_node * pn = (struct sort_node *) getter(parent);
786 if (pn)
787 pn->indegree++;
788 parents=parents->next;
790 next=next->next;
793 * find the tips
795 * tips are nodes not reachable from any other node in the list
797 * the tips serve as a starting set for the work queue.
799 next=*list;
800 insert = &work;
801 while (next) {
802 struct sort_node * node = (struct sort_node *) getter(next->item);
804 if (node->indegree == 0) {
805 insert = &commit_list_insert(next->item, insert)->next;
807 next=next->next;
810 /* process the list in topological order */
811 if (!lifo)
812 sort_by_date(&work);
813 while (work) {
814 struct commit * work_item = pop_commit(&work);
815 struct sort_node * work_node = (struct sort_node *) getter(work_item);
816 struct commit_list * parents = work_item->parents;
818 while (parents) {
819 struct commit * parent=parents->item;
820 struct sort_node * pn = (struct sort_node *) getter(parent);
822 if (pn) {
824 * parents are only enqueued for emission
825 * when all their children have been emitted thereby
826 * guaranteeing topological order.
828 pn->indegree--;
829 if (!pn->indegree) {
830 if (!lifo)
831 insert_by_date(parent, &work);
832 else
833 commit_list_insert(parent, &work);
836 parents=parents->next;
839 * work_item is a commit all of whose children
840 * have already been emitted. we can emit it now.
842 *pptr = work_node->list_item;
843 pptr = &(*pptr)->next;
844 *pptr = NULL;
845 setter(work_item, NULL);
847 free(nodes);