diff-tree: fix up comparison of "interesting" sub-trees
[git.git] / diff-tree.c
1 #include <ctype.h>
2 #include "cache.h"
3 #include "diff.h"
4
5 static int silent = 0;
6 static int verbose_header = 0;
7 static int ignore_merges = 1;
8 static int recursive = 0;
9 static int read_stdin = 0;
10 static int line_termination = '\n';
11 static int generate_patch = 0;
12 static const char *header = NULL;
13 static const char *header_prefix = "";
14
15 // What paths are we interested in?
16 static int nr_paths = 0;
17 static char **paths = NULL;
18 static int *pathlens = NULL;
19
20 static int diff_tree_sha1(const unsigned char *old, const unsigned char *new, const char *base);
21
22 static void update_tree_entry(void **bufp, unsigned long *sizep)
23 {
24         void *buf = *bufp;
25         unsigned long size = *sizep;
26         int len = strlen(buf) + 1 + 20;
27
28         if (size < len)
29                 die("corrupt tree file");
30         *bufp = buf + len;
31         *sizep = size - len;
32 }
33
34 static const unsigned char *extract(void *tree, unsigned long size, const char **pathp, unsigned int *modep)
35 {
36         int len = strlen(tree)+1;
37         const unsigned char *sha1 = tree + len;
38         const char *path = strchr(tree, ' ');
39
40         if (!path || size < len + 20 || sscanf(tree, "%o", modep) != 1)
41                 die("corrupt tree file");
42         *pathp = path+1;
43         return sha1;
44 }
45
46 static char *malloc_base(const char *base, const char *path, int pathlen)
47 {
48         int baselen = strlen(base);
49         char *newbase = xmalloc(baselen + pathlen + 2);
50         memcpy(newbase, base, baselen);
51         memcpy(newbase + baselen, path, pathlen);
52         memcpy(newbase + baselen + pathlen, "/", 2);
53         return newbase;
54 }
55
56 static void show_file(const char *prefix, void *tree, unsigned long size, const char *base);
57
58 /* A whole sub-tree went away or appeared */
59 static void show_tree(const char *prefix, void *tree, unsigned long size, const char *base)
60 {
61         while (size) {
62                 show_file(prefix, tree, size, base);
63                 update_tree_entry(&tree, &size);
64         }
65 }
66
67 /* A file entry went away or appeared */
68 static void show_file(const char *prefix, void *tree, unsigned long size, const char *base)
69 {
70         unsigned mode;
71         const char *path;
72         const unsigned char *sha1 = extract(tree, size, &path, &mode);
73
74         if (header) {
75                 printf("%s", header);
76                 header = NULL;
77         }
78
79         if (silent)
80                 return;
81
82         if (recursive && S_ISDIR(mode)) {
83                 char type[20];
84                 unsigned long size;
85                 char *newbase = malloc_base(base, path, strlen(path));
86                 void *tree;
87
88                 tree = read_sha1_file(sha1, type, &size);
89                 if (!tree || strcmp(type, "tree"))
90                         die("corrupt tree sha %s", sha1_to_hex(sha1));
91
92                 show_tree(prefix, tree, size, newbase);
93                 
94                 free(tree);
95                 free(newbase);
96                 return;
97         }
98
99         if (generate_patch) {
100                 if (!S_ISDIR(mode))
101                         diff_addremove(prefix[0], mode, sha1, base, path);
102         }
103         else
104                 printf("%s%06o\t%s\t%s\t%s%s%c", prefix, mode,
105                        S_ISDIR(mode) ? "tree" : "blob",
106                        sha1_to_hex(sha1), base, path,
107                        line_termination);
108 }
109
110 static int compare_tree_entry(void *tree1, unsigned long size1, void *tree2, unsigned long size2, const char *base)
111 {
112         unsigned mode1, mode2;
113         const char *path1, *path2;
114         const unsigned char *sha1, *sha2;
115         int cmp, pathlen1, pathlen2;
116         char old_sha1_hex[50];
117
118         sha1 = extract(tree1, size1, &path1, &mode1);
119         sha2 = extract(tree2, size2, &path2, &mode2);
120
121         pathlen1 = strlen(path1);
122         pathlen2 = strlen(path2);
123         cmp = cache_name_compare(path1, pathlen1, path2, pathlen2);
124         if (cmp < 0) {
125                 show_file("-", tree1, size1, base);
126                 return -1;
127         }
128         if (cmp > 0) {
129                 show_file("+", tree2, size2, base);
130                 return 1;
131         }
132         if (!memcmp(sha1, sha2, 20) && mode1 == mode2)
133                 return 0;
134
135         /*
136          * If the filemode has changed to/from a directory from/to a regular
137          * file, we need to consider it a remove and an add.
138          */
139         if (S_ISDIR(mode1) != S_ISDIR(mode2)) {
140                 show_file("-", tree1, size1, base);
141                 show_file("+", tree2, size2, base);
142                 return 0;
143         }
144
145         if (recursive && S_ISDIR(mode1)) {
146                 int retval;
147                 char *newbase = malloc_base(base, path1, pathlen1);
148                 retval = diff_tree_sha1(sha1, sha2, newbase);
149                 free(newbase);
150                 return retval;
151         }
152
153         if (header) {
154                 printf("%s", header);
155                 header = NULL;
156         }
157         if (silent)
158                 return 0;
159
160         if (generate_patch) {
161                 if (!S_ISDIR(mode1))
162                         diff_change(mode1, mode2, sha1, sha2, base, path1);
163         }
164         else {
165                 strcpy(old_sha1_hex, sha1_to_hex(sha1));
166                 printf("*%06o->%06o\t%s\t%s->%s\t%s%s%c", mode1, mode2,
167                        S_ISDIR(mode1) ? "tree" : "blob",
168                        old_sha1_hex, sha1_to_hex(sha2), base, path1,
169                        line_termination);
170         }
171         return 0;
172 }
173
174 static int interesting(void *tree, unsigned long size, const char *base)
175 {
176         const char *path;
177         unsigned mode;
178         int i;
179         int baselen, pathlen;
180
181         if (!nr_paths)
182                 return 1;
183
184         (void)extract(tree, size, &path, &mode);
185
186         pathlen = strlen(path);
187         baselen = strlen(base);
188
189         for (i=0; i < nr_paths; i++) {
190                 const char *match = paths[i];
191                 int matchlen = pathlens[i];
192
193                 if (baselen >= matchlen) {
194                         /* If it doesn't match, move along... */
195                         if (strncmp(base, match, matchlen))
196                                 continue;
197
198                         /* The base is a subdirectory of a path which was specified. */
199                         return 1;
200                 }
201
202                 /* Does the base match? */
203                 if (strncmp(base, match, baselen))
204                         continue;
205
206                 match += baselen;
207                 matchlen -= baselen;
208
209                 if (pathlen > matchlen)
210                         continue;
211
212                 if (matchlen > pathlen) {
213                         if (match[pathlen] != '/')
214                                 continue;
215                 }
216
217                 if (strncmp(path, match, pathlen))
218                         continue;
219
220                 return 1;
221         }
222         return 0; /* No matches */
223 }
224
225 static int diff_tree(void *tree1, unsigned long size1, void *tree2, unsigned long size2, const char *base)
226 {
227         while (size1 | size2) {
228                 if (nr_paths && size1 && !interesting(tree1, size1, base)) {
229                         update_tree_entry(&tree1, &size1);
230                         continue;
231                 }
232                 if (nr_paths && size2 && !interesting(tree2, size2, base)) {
233                         update_tree_entry(&tree2, &size2);
234                         continue;
235                 }
236                 if (!size1) {
237                         show_file("+", tree2, size2, base);
238                         update_tree_entry(&tree2, &size2);
239                         continue;
240                 }
241                 if (!size2) {
242                         show_file("-", tree1, size1, base);
243                         update_tree_entry(&tree1, &size1);
244                         continue;
245                 }
246                 switch (compare_tree_entry(tree1, size1, tree2, size2, base)) {
247                 case -1:
248                         update_tree_entry(&tree1, &size1);
249                         continue;
250                 case 0:
251                         update_tree_entry(&tree1, &size1);
252                         /* Fallthrough */
253                 case 1:
254                         update_tree_entry(&tree2, &size2);
255                         continue;
256                 }
257                 die("diff-tree: internal error");
258         }
259         return 0;
260 }
261
262 static int diff_tree_sha1(const unsigned char *old, const unsigned char *new, const char *base)
263 {
264         void *tree1, *tree2;
265         unsigned long size1, size2;
266         int retval;
267
268         tree1 = read_object_with_reference(old, "tree", &size1, 0);
269         if (!tree1)
270                 die("unable to read source tree (%s)", sha1_to_hex(old));
271         tree2 = read_object_with_reference(new, "tree", &size2, 0);
272         if (!tree2)
273                 die("unable to read destination tree (%s)", sha1_to_hex(new));
274         retval = diff_tree(tree1, size1, tree2, size2, base);
275         free(tree1);
276         free(tree2);
277         return retval;
278 }
279
280 static int get_one_line(const char *msg, unsigned long len)
281 {
282         int ret = 0;
283
284         while (len--) {
285                 ret++;
286                 if (*msg++ == '\n')
287                         break;
288         }
289         return ret;
290 }
291
292 static int add_author_info(char *buf, const char *line, int len)
293 {
294         char *date;
295         unsigned int namelen;
296         unsigned long time;
297         int tz;
298
299         line += strlen("author ");
300         date = strchr(line, '>');
301         if (!date)
302                 return 0;
303         namelen = ++date - line;
304         time = strtoul(date, &date, 10);
305         tz = strtol(date, NULL, 10);
306
307         return sprintf(buf, "Author: %.*s\nDate:   %s\n",
308                 namelen, line,
309                 show_date(time, tz));
310 }
311
312 static char *generate_header(const char *commit, const char *parent, const char *msg, unsigned long len)
313 {
314         static char this_header[1000];
315         int offset;
316
317         offset = sprintf(this_header, "%s%s (from %s)\n", header_prefix, commit, parent);
318         if (verbose_header) {
319                 int hdr = 1;
320
321                 for (;;) {
322                         const char *line = msg;
323                         int linelen = get_one_line(msg, len);
324
325                         if (!linelen)
326                                 break;
327                         if (offset + linelen + 10 > sizeof(this_header))
328                                 break;
329
330                         msg += linelen;
331                         len -= linelen;
332                         if (linelen == 1)
333                                 hdr = 0;
334                         if (hdr) {
335                                 if (!memcmp(line, "author ", 7))
336                                         offset += add_author_info(this_header + offset, line, linelen);
337                                 continue;
338                         }
339                         memset(this_header + offset, ' ', 4);
340                         memcpy(this_header + offset + 4, line, linelen);
341                         offset += linelen + 4;
342                 }
343                 this_header[offset++] = '\n';
344                 this_header[offset] = 0;
345         }
346
347         return this_header;
348 }
349
350 static int diff_tree_commit(const unsigned char *commit, const char *name)
351 {
352         unsigned long size, offset;
353         char *buf = read_object_with_reference(commit, "commit", &size, NULL);
354
355         if (!buf)
356                 return -1;
357
358         /* More than one parent? */
359         if (ignore_merges) {
360                 if (!memcmp(buf + 46 + 48, "parent ", 7))
361                         return 0;
362         }
363
364         if (!name) {
365                 static char commit_name[60];
366                 strcpy(commit_name, sha1_to_hex(commit));
367                 name = commit_name;
368         }
369
370         offset = 46;
371         while (offset + 48 < size && !memcmp(buf + offset, "parent ", 7)) {
372                 unsigned char parent[20];
373                 if (get_sha1_hex(buf + offset + 7, parent))
374                         return -1;
375                 header = generate_header(name, sha1_to_hex(parent), buf, size);
376                 diff_tree_sha1(parent, commit, "");
377                 if (!header && verbose_header)
378                         header_prefix = "\ndiff-tree ";
379                 offset += 48;
380         }
381         return 0;
382 }
383
384 static int diff_tree_stdin(char *line)
385 {
386         int len = strlen(line);
387         unsigned char commit[20], parent[20];
388         static char this_header[1000];
389
390         if (!len || line[len-1] != '\n')
391                 return -1;
392         line[len-1] = 0;
393         if (get_sha1_hex(line, commit))
394                 return -1;
395         if (isspace(line[40]) && !get_sha1_hex(line+41, parent)) {
396                 line[40] = 0;
397                 line[81] = 0;
398                 sprintf(this_header, "%s (from %s)\n", line, line+41);
399                 header = this_header;
400                 return diff_tree_sha1(parent, commit, "");
401         }
402         line[40] = 0;
403         return diff_tree_commit(commit, line);
404 }
405
406 static char *diff_tree_usage =
407 "diff-tree [-p] [-r] [-z] [--stdin] [-m] [-s] [-v] <tree sha1> <tree sha1>";
408
409 int main(int argc, char **argv)
410 {
411         int nr_sha1;
412         char line[1000];
413         unsigned char sha1[2][20];
414
415         nr_sha1 = 0;
416         for (;;) {
417                 char *arg;
418
419                 argv++;
420                 argc--;
421                 arg = *argv;
422                 if (!arg)
423                         break;
424
425                 if (*arg != '-') {
426                         if (nr_sha1 < 2 && !get_sha1(arg, sha1[nr_sha1])) {
427                                 nr_sha1++;
428                                 continue;
429                         }
430                         break;
431                 }
432
433                 if (!strcmp(arg, "--")) {
434                         argv++;
435                         argc--;
436                         break;
437                 }
438                 if (!strcmp(arg, "-r")) {
439                         recursive = 1;
440                         continue;
441                 }
442                 if (!strcmp(arg, "-p")) {
443                         recursive = generate_patch = 1;
444                         continue;
445                 }
446                 if (!strcmp(arg, "-z")) {
447                         line_termination = '\0';
448                         continue;
449                 }
450                 if (!strcmp(arg, "-m")) {
451                         ignore_merges = 0;
452                         continue;
453                 }
454                 if (!strcmp(arg, "-s")) {
455                         silent = 1;
456                         continue;
457                 }
458                 if (!strcmp(arg, "-v")) {
459                         verbose_header = 1;
460                         header_prefix = "diff-tree ";
461                         continue;
462                 }
463                 if (!strcmp(arg, "--stdin")) {
464                         read_stdin = 1;
465                         continue;
466                 }
467                 usage(diff_tree_usage);
468         }
469
470         if (argc > 0) {
471                 int i;
472
473                 paths = argv;
474                 nr_paths = argc;
475                 pathlens = xmalloc(nr_paths * sizeof(int));
476                 for (i=0; i<nr_paths; i++)
477                         pathlens[i] = strlen(paths[i]);
478         }
479
480         switch (nr_sha1) {
481         case 0:
482                 if (!read_stdin)
483                         usage(diff_tree_usage);
484                 break;
485         case 1:
486                 diff_tree_commit(sha1[0], NULL);
487                 break;
488         case 2:
489                 diff_tree_sha1(sha1[0], sha1[1], "");
490                 break;
491         }
492
493         if (!read_stdin)
494                 return 0;
495
496         while (fgets(line, sizeof(line), stdin))
497                 diff_tree_stdin(line);
498
499         return 0;
500 }