git-cvsimport-script: provide direct support for cvsps -z option
[git.git] / diff-tree.c
1 #include <ctype.h>
2 #include "cache.h"
3 #include "diff.h"
4 #include "commit.h"
5
6 static int show_root_diff = 0;
7 static int verbose_header = 0;
8 static int ignore_merges = 1;
9 static int recursive = 0;
10 static int show_tree_entry_in_recursive = 0;
11 static int read_stdin = 0;
12 static int diff_output_format = DIFF_FORMAT_HUMAN;
13 static int detect_rename = 0;
14 static int find_copies_harder = 0;
15 static int diff_setup_opt = 0;
16 static int diff_score_opt = 0;
17 static const char *pickaxe = NULL;
18 static int pickaxe_opts = 0;
19 static int diff_break_opt = -1;
20 static const char *orderfile = NULL;
21 static const char *diff_filter = NULL;
22 static const char *header = NULL;
23 static const char *header_prefix = "";
24 static enum cmit_fmt commit_format = CMIT_FMT_RAW;
25
26 // What paths are we interested in?
27 static int nr_paths = 0;
28 static const char **paths = NULL;
29 static int *pathlens = NULL;
30
31 static int diff_tree_sha1(const unsigned char *old, const unsigned char *new, const char *base);
32
33 static void update_tree_entry(void **bufp, unsigned long *sizep)
34 {
35         void *buf = *bufp;
36         unsigned long size = *sizep;
37         int len = strlen(buf) + 1 + 20;
38
39         if (size < len)
40                 die("corrupt tree file");
41         *bufp = buf + len;
42         *sizep = size - len;
43 }
44
45 static const unsigned char *extract(void *tree, unsigned long size, const char **pathp, unsigned int *modep)
46 {
47         int len = strlen(tree)+1;
48         const unsigned char *sha1 = tree + len;
49         const char *path = strchr(tree, ' ');
50         unsigned int mode;
51
52         if (!path || size < len + 20 || sscanf(tree, "%o", &mode) != 1)
53                 die("corrupt tree file");
54         *pathp = path+1;
55         *modep = DIFF_FILE_CANON_MODE(mode);
56         return sha1;
57 }
58
59 static char *malloc_base(const char *base, const char *path, int pathlen)
60 {
61         int baselen = strlen(base);
62         char *newbase = xmalloc(baselen + pathlen + 2);
63         memcpy(newbase, base, baselen);
64         memcpy(newbase + baselen, path, pathlen);
65         memcpy(newbase + baselen + pathlen, "/", 2);
66         return newbase;
67 }
68
69 static void show_file(const char *prefix, void *tree, unsigned long size, const char *base);
70 static void show_tree(const char *prefix, void *tree, unsigned long size, const char *base);
71
72 /* A file entry went away or appeared */
73 static void show_file(const char *prefix, void *tree, unsigned long size, const char *base)
74 {
75         unsigned mode;
76         const char *path;
77         const unsigned char *sha1 = extract(tree, size, &path, &mode);
78
79         if (recursive && S_ISDIR(mode)) {
80                 char type[20];
81                 unsigned long size;
82                 char *newbase = malloc_base(base, path, strlen(path));
83                 void *tree;
84
85                 tree = read_sha1_file(sha1, type, &size);
86                 if (!tree || strcmp(type, "tree"))
87                         die("corrupt tree sha %s", sha1_to_hex(sha1));
88
89                 show_tree(prefix, tree, size, newbase);
90
91                 free(tree);
92                 free(newbase);
93                 return;
94         }
95
96         diff_addremove(prefix[0], mode, sha1, base, path);
97 }
98
99 static int compare_tree_entry(void *tree1, unsigned long size1, void *tree2, unsigned long size2, const char *base)
100 {
101         unsigned mode1, mode2;
102         const char *path1, *path2;
103         const unsigned char *sha1, *sha2;
104         int cmp, pathlen1, pathlen2;
105
106         sha1 = extract(tree1, size1, &path1, &mode1);
107         sha2 = extract(tree2, size2, &path2, &mode2);
108
109         pathlen1 = strlen(path1);
110         pathlen2 = strlen(path2);
111         cmp = base_name_compare(path1, pathlen1, mode1, path2, pathlen2, mode2);
112         if (cmp < 0) {
113                 show_file("-", tree1, size1, base);
114                 return -1;
115         }
116         if (cmp > 0) {
117                 show_file("+", tree2, size2, base);
118                 return 1;
119         }
120         if (!find_copies_harder && !memcmp(sha1, sha2, 20) && mode1 == mode2)
121                 return 0;
122
123         /*
124          * If the filemode has changed to/from a directory from/to a regular
125          * file, we need to consider it a remove and an add.
126          */
127         if (S_ISDIR(mode1) != S_ISDIR(mode2)) {
128                 show_file("-", tree1, size1, base);
129                 show_file("+", tree2, size2, base);
130                 return 0;
131         }
132
133         if (recursive && S_ISDIR(mode1)) {
134                 int retval;
135                 char *newbase = malloc_base(base, path1, pathlen1);
136                 if (show_tree_entry_in_recursive)
137                         diff_change(mode1, mode2, sha1, sha2, base, path1);
138                 retval = diff_tree_sha1(sha1, sha2, newbase);
139                 free(newbase);
140                 return retval;
141         }
142
143         diff_change(mode1, mode2, sha1, sha2, base, path1);
144         return 0;
145 }
146
147 static int interesting(void *tree, unsigned long size, const char *base)
148 {
149         const char *path;
150         unsigned mode;
151         int i;
152         int baselen, pathlen;
153
154         if (!nr_paths)
155                 return 1;
156
157         (void)extract(tree, size, &path, &mode);
158
159         pathlen = strlen(path);
160         baselen = strlen(base);
161
162         for (i=0; i < nr_paths; i++) {
163                 const char *match = paths[i];
164                 int matchlen = pathlens[i];
165
166                 if (baselen >= matchlen) {
167                         /* If it doesn't match, move along... */
168                         if (strncmp(base, match, matchlen))
169                                 continue;
170
171                         /* The base is a subdirectory of a path which was specified. */
172                         return 1;
173                 }
174
175                 /* Does the base match? */
176                 if (strncmp(base, match, baselen))
177                         continue;
178
179                 match += baselen;
180                 matchlen -= baselen;
181
182                 if (pathlen > matchlen)
183                         continue;
184
185                 if (matchlen > pathlen) {
186                         if (match[pathlen] != '/')
187                                 continue;
188                         if (!S_ISDIR(mode))
189                                 continue;
190                 }
191
192                 if (strncmp(path, match, pathlen))
193                         continue;
194
195                 return 1;
196         }
197         return 0; /* No matches */
198 }
199
200 /* A whole sub-tree went away or appeared */
201 static void show_tree(const char *prefix, void *tree, unsigned long size, const char *base)
202 {
203         while (size) {
204                 if (interesting(tree, size, base))
205                         show_file(prefix, tree, size, base);
206                 update_tree_entry(&tree, &size);
207         }
208 }
209
210 static int diff_tree(void *tree1, unsigned long size1, void *tree2, unsigned long size2, const char *base)
211 {
212         while (size1 | size2) {
213                 if (nr_paths && size1 && !interesting(tree1, size1, base)) {
214                         update_tree_entry(&tree1, &size1);
215                         continue;
216                 }
217                 if (nr_paths && size2 && !interesting(tree2, size2, base)) {
218                         update_tree_entry(&tree2, &size2);
219                         continue;
220                 }
221                 if (!size1) {
222                         show_file("+", tree2, size2, base);
223                         update_tree_entry(&tree2, &size2);
224                         continue;
225                 }
226                 if (!size2) {
227                         show_file("-", tree1, size1, base);
228                         update_tree_entry(&tree1, &size1);
229                         continue;
230                 }
231                 switch (compare_tree_entry(tree1, size1, tree2, size2, base)) {
232                 case -1:
233                         update_tree_entry(&tree1, &size1);
234                         continue;
235                 case 0:
236                         update_tree_entry(&tree1, &size1);
237                         /* Fallthrough */
238                 case 1:
239                         update_tree_entry(&tree2, &size2);
240                         continue;
241                 }
242                 die("git-diff-tree: internal error");
243         }
244         return 0;
245 }
246
247 static int diff_tree_sha1(const unsigned char *old, const unsigned char *new, const char *base)
248 {
249         void *tree1, *tree2;
250         unsigned long size1, size2;
251         int retval;
252
253         tree1 = read_object_with_reference(old, "tree", &size1, NULL);
254         if (!tree1)
255                 die("unable to read source tree (%s)", sha1_to_hex(old));
256         tree2 = read_object_with_reference(new, "tree", &size2, NULL);
257         if (!tree2)
258                 die("unable to read destination tree (%s)", sha1_to_hex(new));
259         retval = diff_tree(tree1, size1, tree2, size2, base);
260         free(tree1);
261         free(tree2);
262         return retval;
263 }
264
265 static void call_diff_setup(void)
266 {
267         diff_setup(diff_setup_opt);
268 }
269
270 static int call_diff_flush(void)
271 {
272         diffcore_std(NULL,
273                      detect_rename, diff_score_opt,
274                      pickaxe, pickaxe_opts,
275                      diff_break_opt,
276                      orderfile,
277                      diff_filter);
278         if (diff_queue_is_empty()) {
279                 diff_flush(DIFF_FORMAT_NO_OUTPUT);
280                 return 0;
281         }
282         if (header) {
283                 printf("%s%c", header, diff_output_format == DIFF_FORMAT_MACHINE ? 0 : '\n');
284                 header = NULL;
285         }
286         diff_flush(diff_output_format);
287         return 1;
288 }
289
290 static int diff_tree_sha1_top(const unsigned char *old,
291                               const unsigned char *new, const char *base)
292 {
293         int ret;
294
295         call_diff_setup();
296         ret = diff_tree_sha1(old, new, base);
297         call_diff_flush();
298         return ret;
299 }
300
301 static int diff_root_tree(const unsigned char *new, const char *base)
302 {
303         int retval;
304         void *tree;
305         unsigned long size;
306
307         call_diff_setup();
308         tree = read_object_with_reference(new, "tree", &size, NULL);
309         if (!tree)
310                 die("unable to read root tree (%s)", sha1_to_hex(new));
311         retval = diff_tree("", 0, tree, size, base);
312         free(tree);
313         call_diff_flush();
314         return retval;
315 }
316
317 static const char *generate_header(const char *commit, const char *parent, const char *msg, unsigned long len)
318 {
319         static char this_header[16384];
320         int offset;
321
322         if (!verbose_header)
323                 return commit;
324
325         offset = sprintf(this_header, "%s%s (from %s)\n", header_prefix, commit, parent);
326         offset += pretty_print_commit(commit_format, msg, len, this_header + offset, sizeof(this_header) - offset);
327         return this_header;
328 }
329
330 static int diff_tree_commit(const unsigned char *commit, const char *name)
331 {
332         unsigned long size, offset;
333         char *buf = read_object_with_reference(commit, "commit", &size, NULL);
334
335         if (!buf)
336                 return -1;
337
338         if (!name) {
339                 static char commit_name[60];
340                 strcpy(commit_name, sha1_to_hex(commit));
341                 name = commit_name;
342         }
343
344         /* Root commit? */
345         if (show_root_diff && memcmp(buf + 46, "parent ", 7)) {
346                 header = generate_header(name, "root", buf, size);
347                 diff_root_tree(commit, "");
348         }
349
350         /* More than one parent? */
351         if (ignore_merges) {
352                 if (!memcmp(buf + 46 + 48, "parent ", 7))
353                         return 0;
354         }
355
356         offset = 46;
357         while (offset + 48 < size && !memcmp(buf + offset, "parent ", 7)) {
358                 unsigned char parent[20];
359                 if (get_sha1_hex(buf + offset + 7, parent))
360                         return -1;
361                 header = generate_header(name, sha1_to_hex(parent), buf, size);
362                 diff_tree_sha1_top(parent, commit, "");
363                 if (!header && verbose_header) {
364                         header_prefix = "\ndiff-tree ";
365                         /*
366                          * Don't print multiple merge entries if we
367                          * don't print the diffs.
368                          */
369                 }
370                 offset += 48;
371         }
372         return 0;
373 }
374
375 static int diff_tree_stdin(char *line)
376 {
377         int len = strlen(line);
378         unsigned char commit[20], parent[20];
379         static char this_header[1000];
380
381         if (!len || line[len-1] != '\n')
382                 return -1;
383         line[len-1] = 0;
384         if (get_sha1_hex(line, commit))
385                 return -1;
386         if (isspace(line[40]) && !get_sha1_hex(line+41, parent)) {
387                 line[40] = 0;
388                 line[81] = 0;
389                 sprintf(this_header, "%s (from %s)\n", line, line+41);
390                 header = this_header;
391                 return diff_tree_sha1_top(parent, commit, "");
392         }
393         line[40] = 0;
394         return diff_tree_commit(commit, line);
395 }
396
397 static char *diff_tree_usage =
398 "git-diff-tree [-p] [-r] [-z] [--stdin] [-m] [-s] [-v] [--pretty] [-t] [-R] [-B] [-M] [-C] [--find-copies-header] [-O<orderfile>] [-S<string>] [--pickaxe-all] <tree-ish> <tree-ish>";
399
400 int main(int argc, const char **argv)
401 {
402         int nr_sha1;
403         char line[1000];
404         unsigned char sha1[2][20];
405
406         nr_sha1 = 0;
407         for (;;) {
408                 const char *arg;
409
410                 argv++;
411                 argc--;
412                 arg = *argv;
413                 if (!arg)
414                         break;
415
416                 if (*arg != '-') {
417                         if (nr_sha1 < 2 && !get_sha1(arg, sha1[nr_sha1])) {
418                                 nr_sha1++;
419                                 continue;
420                         }
421                         break;
422                 }
423
424                 if (!strcmp(arg, "--")) {
425                         argv++;
426                         argc--;
427                         break;
428                 }
429                 if (!strcmp(arg, "-r")) {
430                         recursive = 1;
431                         continue;
432                 }
433                 if (!strcmp(arg, "-t")) {
434                         recursive = show_tree_entry_in_recursive = 1;
435                         continue;
436                 }
437                 if (!strcmp(arg, "-R")) {
438                         diff_setup_opt |= DIFF_SETUP_REVERSE;
439                         continue;
440                 }
441                 if (!strcmp(arg, "-p")) {
442                         diff_output_format = DIFF_FORMAT_PATCH;
443                         recursive = 1;
444                         continue;
445                 }
446                 if (!strncmp(arg, "-S", 2)) {
447                         pickaxe = arg + 2;
448                         continue;
449                 }
450                 if (!strncmp(arg, "-O", 2)) {
451                         orderfile = arg + 2;
452                         continue;
453                 }
454                 if (!strncmp(arg, "--diff-filter=", 14)) {
455                         diff_filter = arg + 14;
456                         continue;
457                 }
458                 if (!strcmp(arg, "--pickaxe-all")) {
459                         pickaxe_opts = DIFF_PICKAXE_ALL;
460                         continue;
461                 }
462                 if (!strncmp(arg, "-M", 2)) {
463                         detect_rename = DIFF_DETECT_RENAME;
464                         if ((diff_score_opt = diff_scoreopt_parse(arg)) == -1)
465                                 usage(diff_tree_usage);
466                         continue;
467                 }
468                 if (!strncmp(arg, "-C", 2)) {
469                         detect_rename = DIFF_DETECT_COPY;
470                         if ((diff_score_opt = diff_scoreopt_parse(arg)) == -1)
471                                 usage(diff_tree_usage);
472                         continue;
473                 }
474                 if (!strncmp(arg, "-B", 2)) {
475                         if ((diff_break_opt = diff_scoreopt_parse(arg)) == -1)
476                                 usage(diff_tree_usage);
477                         continue;
478                 }
479                 if (!strcmp(arg, "--find-copies-harder")) {
480                         find_copies_harder = 1;
481                         continue;
482                 }
483                 if (!strcmp(arg, "-z")) {
484                         diff_output_format = DIFF_FORMAT_MACHINE;
485                         continue;
486                 }
487                 if (!strcmp(arg, "-m")) {
488                         ignore_merges = 0;
489                         continue;
490                 }
491                 if (!strcmp(arg, "-s")) {
492                         diff_output_format = DIFF_FORMAT_NO_OUTPUT;
493                         continue;
494                 }
495                 if (!strcmp(arg, "-v")) {
496                         verbose_header = 1;
497                         header_prefix = "diff-tree ";
498                         continue;
499                 }
500                 if (!strncmp(arg, "--pretty", 8)) {
501                         verbose_header = 1;
502                         header_prefix = "diff-tree ";
503                         commit_format = get_commit_format(arg+8);
504                         continue;
505                 }
506                 if (!strcmp(arg, "--stdin")) {
507                         read_stdin = 1;
508                         continue;
509                 }
510                 if (!strcmp(arg, "--root")) {
511                         show_root_diff = 1;
512                         continue;
513                 }
514                 usage(diff_tree_usage);
515         }
516         if (find_copies_harder && detect_rename != DIFF_DETECT_COPY)
517                 usage(diff_tree_usage);
518
519         if (argc > 0) {
520                 int i;
521
522                 paths = argv;
523                 nr_paths = argc;
524                 pathlens = xmalloc(nr_paths * sizeof(int));
525                 for (i=0; i<nr_paths; i++)
526                         pathlens[i] = strlen(paths[i]);
527         }
528
529         switch (nr_sha1) {
530         case 0:
531                 if (!read_stdin)
532                         usage(diff_tree_usage);
533                 break;
534         case 1:
535                 diff_tree_commit(sha1[0], NULL);
536                 break;
537         case 2:
538                 diff_tree_sha1_top(sha1[0], sha1[1], "");
539                 break;
540         }
541
542         if (!read_stdin)
543                 return 0;
544
545         if (detect_rename)
546                 diff_setup_opt |= (DIFF_SETUP_USE_SIZE_CACHE |
547                                    DIFF_SETUP_USE_CACHE);
548         while (fgets(line, sizeof(line), stdin))
549                 diff_tree_stdin(line);
550
551         return 0;
552 }