[PATCH] Optimize diff-tree -[CM] --stdin
[git] / diff-tree.c
1 #include <ctype.h>
2 #include "cache.h"
3 #include "diff.h"
4
5 static int show_root_diff = 0;
6 static int verbose_header = 0;
7 static int ignore_merges = 1;
8 static int recursive = 0;
9 static int show_tree_entry_in_recursive = 0;
10 static int read_stdin = 0;
11 static int diff_output_format = DIFF_FORMAT_HUMAN;
12 static int detect_rename = 0;
13 static int diff_setup_opt = 0;
14 static int diff_score_opt = 0;
15 static const char *pickaxe = NULL;
16 static int pickaxe_opts = 0;
17 static const char *header = NULL;
18 static const char *header_prefix = "";
19
20 // What paths are we interested in?
21 static int nr_paths = 0;
22 static const char **paths = NULL;
23 static int *pathlens = NULL;
24
25 static int diff_tree_sha1(const unsigned char *old, const unsigned char *new, const char *base);
26
27 static void update_tree_entry(void **bufp, unsigned long *sizep)
28 {
29         void *buf = *bufp;
30         unsigned long size = *sizep;
31         int len = strlen(buf) + 1 + 20;
32
33         if (size < len)
34                 die("corrupt tree file");
35         *bufp = buf + len;
36         *sizep = size - len;
37 }
38
39 static const unsigned char *extract(void *tree, unsigned long size, const char **pathp, unsigned int *modep)
40 {
41         int len = strlen(tree)+1;
42         const unsigned char *sha1 = tree + len;
43         const char *path = strchr(tree, ' ');
44
45         if (!path || size < len + 20 || sscanf(tree, "%o", modep) != 1)
46                 die("corrupt tree file");
47         *pathp = path+1;
48         return sha1;
49 }
50
51 static char *malloc_base(const char *base, const char *path, int pathlen)
52 {
53         int baselen = strlen(base);
54         char *newbase = xmalloc(baselen + pathlen + 2);
55         memcpy(newbase, base, baselen);
56         memcpy(newbase + baselen, path, pathlen);
57         memcpy(newbase + baselen + pathlen, "/", 2);
58         return newbase;
59 }
60
61 static void show_file(const char *prefix, void *tree, unsigned long size, const char *base);
62 static void show_tree(const char *prefix, void *tree, unsigned long size, const char *base);
63
64 /* A file entry went away or appeared */
65 static void show_file(const char *prefix, void *tree, unsigned long size, const char *base)
66 {
67         unsigned mode;
68         const char *path;
69         const unsigned char *sha1 = extract(tree, size, &path, &mode);
70
71         if (recursive && S_ISDIR(mode)) {
72                 char type[20];
73                 unsigned long size;
74                 char *newbase = malloc_base(base, path, strlen(path));
75                 void *tree;
76
77                 tree = read_sha1_file(sha1, type, &size);
78                 if (!tree || strcmp(type, "tree"))
79                         die("corrupt tree sha %s", sha1_to_hex(sha1));
80
81                 show_tree(prefix, tree, size, newbase);
82
83                 free(tree);
84                 free(newbase);
85                 return;
86         }
87
88         diff_addremove(prefix[0], mode, sha1, base, path);
89 }
90
91 static int compare_tree_entry(void *tree1, unsigned long size1, void *tree2, unsigned long size2, const char *base)
92 {
93         unsigned mode1, mode2;
94         const char *path1, *path2;
95         const unsigned char *sha1, *sha2;
96         int cmp, pathlen1, pathlen2;
97
98         sha1 = extract(tree1, size1, &path1, &mode1);
99         sha2 = extract(tree2, size2, &path2, &mode2);
100
101         pathlen1 = strlen(path1);
102         pathlen2 = strlen(path2);
103         cmp = base_name_compare(path1, pathlen1, mode1, path2, pathlen2, mode2);
104         if (cmp < 0) {
105                 show_file("-", tree1, size1, base);
106                 return -1;
107         }
108         if (cmp > 0) {
109                 show_file("+", tree2, size2, base);
110                 return 1;
111         }
112         if (!memcmp(sha1, sha2, 20) && mode1 == mode2)
113                 return 0;
114
115         /*
116          * If the filemode has changed to/from a directory from/to a regular
117          * file, we need to consider it a remove and an add.
118          */
119         if (S_ISDIR(mode1) != S_ISDIR(mode2)) {
120                 show_file("-", tree1, size1, base);
121                 show_file("+", tree2, size2, base);
122                 return 0;
123         }
124
125         if (recursive && S_ISDIR(mode1)) {
126                 int retval;
127                 char *newbase = malloc_base(base, path1, pathlen1);
128                 if (show_tree_entry_in_recursive)
129                         diff_change(mode1, mode2, sha1, sha2, base, path1);
130                 retval = diff_tree_sha1(sha1, sha2, newbase);
131                 free(newbase);
132                 return retval;
133         }
134
135         diff_change(mode1, mode2, sha1, sha2, base, path1);
136         return 0;
137 }
138
139 static int interesting(void *tree, unsigned long size, const char *base)
140 {
141         const char *path;
142         unsigned mode;
143         int i;
144         int baselen, pathlen;
145
146         if (!nr_paths)
147                 return 1;
148
149         (void)extract(tree, size, &path, &mode);
150
151         pathlen = strlen(path);
152         baselen = strlen(base);
153
154         for (i=0; i < nr_paths; i++) {
155                 const char *match = paths[i];
156                 int matchlen = pathlens[i];
157
158                 if (baselen >= matchlen) {
159                         /* If it doesn't match, move along... */
160                         if (strncmp(base, match, matchlen))
161                                 continue;
162
163                         /* The base is a subdirectory of a path which was specified. */
164                         return 1;
165                 }
166
167                 /* Does the base match? */
168                 if (strncmp(base, match, baselen))
169                         continue;
170
171                 match += baselen;
172                 matchlen -= baselen;
173
174                 if (pathlen > matchlen)
175                         continue;
176
177                 if (matchlen > pathlen) {
178                         if (match[pathlen] != '/')
179                                 continue;
180                         if (!S_ISDIR(mode))
181                                 continue;
182                 }
183
184                 if (strncmp(path, match, pathlen))
185                         continue;
186
187                 return 1;
188         }
189         return 0; /* No matches */
190 }
191
192 /* A whole sub-tree went away or appeared */
193 static void show_tree(const char *prefix, void *tree, unsigned long size, const char *base)
194 {
195         while (size) {
196                 if (interesting(tree, size, base))
197                         show_file(prefix, tree, size, base);
198                 update_tree_entry(&tree, &size);
199         }
200 }
201
202 static int diff_tree(void *tree1, unsigned long size1, void *tree2, unsigned long size2, const char *base)
203 {
204         while (size1 | size2) {
205                 if (nr_paths && size1 && !interesting(tree1, size1, base)) {
206                         update_tree_entry(&tree1, &size1);
207                         continue;
208                 }
209                 if (nr_paths && size2 && !interesting(tree2, size2, base)) {
210                         update_tree_entry(&tree2, &size2);
211                         continue;
212                 }
213                 if (!size1) {
214                         show_file("+", tree2, size2, base);
215                         update_tree_entry(&tree2, &size2);
216                         continue;
217                 }
218                 if (!size2) {
219                         show_file("-", tree1, size1, base);
220                         update_tree_entry(&tree1, &size1);
221                         continue;
222                 }
223                 switch (compare_tree_entry(tree1, size1, tree2, size2, base)) {
224                 case -1:
225                         update_tree_entry(&tree1, &size1);
226                         continue;
227                 case 0:
228                         update_tree_entry(&tree1, &size1);
229                         /* Fallthrough */
230                 case 1:
231                         update_tree_entry(&tree2, &size2);
232                         continue;
233                 }
234                 die("git-diff-tree: internal error");
235         }
236         return 0;
237 }
238
239 static int diff_tree_sha1(const unsigned char *old, const unsigned char *new, const char *base)
240 {
241         void *tree1, *tree2;
242         unsigned long size1, size2;
243         int retval;
244
245         tree1 = read_object_with_reference(old, "tree", &size1, NULL);
246         if (!tree1)
247                 die("unable to read source tree (%s)", sha1_to_hex(old));
248         tree2 = read_object_with_reference(new, "tree", &size2, NULL);
249         if (!tree2)
250                 die("unable to read destination tree (%s)", sha1_to_hex(new));
251         retval = diff_tree(tree1, size1, tree2, size2, base);
252         free(tree1);
253         free(tree2);
254         return retval;
255 }
256
257 static void call_diff_setup(void)
258 {
259         diff_setup(diff_setup_opt);
260 }
261
262 static int call_diff_flush(void)
263 {
264         if (detect_rename)
265                 diffcore_rename(detect_rename, diff_score_opt);
266         if (pickaxe)
267                 diffcore_pickaxe(pickaxe, pickaxe_opts);
268         if (diff_queue_is_empty()) {
269                 diff_flush(DIFF_FORMAT_NO_OUTPUT, 0);
270                 return 0;
271         }
272         if (header) {
273                 const char *fmt = "%s";
274                 if (diff_output_format == DIFF_FORMAT_MACHINE)
275                         fmt = "%s%c";
276                 
277                 printf(fmt, header, 0);
278                 header = NULL;
279         }
280         diff_flush(diff_output_format, 1);
281         return 1;
282 }
283
284 static int diff_tree_sha1_top(const unsigned char *old,
285                               const unsigned char *new, const char *base)
286 {
287         int ret;
288
289         call_diff_setup();
290         ret = diff_tree_sha1(old, new, base);
291         call_diff_flush();
292         return ret;
293 }
294
295 static int diff_root_tree(const unsigned char *new, const char *base)
296 {
297         int retval;
298         void *tree;
299         unsigned long size;
300
301         call_diff_setup();
302         tree = read_object_with_reference(new, "tree", &size, NULL);
303         if (!tree)
304                 die("unable to read root tree (%s)", sha1_to_hex(new));
305         retval = diff_tree("", 0, tree, size, base);
306         free(tree);
307         call_diff_flush();
308         return retval;
309 }
310
311 static int get_one_line(const char *msg, unsigned long len)
312 {
313         int ret = 0;
314
315         while (len--) {
316                 ret++;
317                 if (*msg++ == '\n')
318                         break;
319         }
320         return ret;
321 }
322
323 static int add_author_info(char *buf, const char *line, int len)
324 {
325         char *date;
326         unsigned int namelen;
327         unsigned long time;
328         int tz;
329
330         line += strlen("author ");
331         date = strchr(line, '>');
332         if (!date)
333                 return 0;
334         namelen = ++date - line;
335         time = strtoul(date, &date, 10);
336         tz = strtol(date, NULL, 10);
337
338         return sprintf(buf, "Author: %.*s\nDate:   %s\n",
339                 namelen, line,
340                 show_date(time, tz));
341 }
342
343 static char *generate_header(const char *commit, const char *parent, const char *msg, unsigned long len)
344 {
345         static char this_header[16384];
346         int offset;
347
348         offset = sprintf(this_header, "%s%s (from %s)\n", header_prefix, commit, parent);
349         if (verbose_header) {
350                 int hdr = 1;
351
352                 for (;;) {
353                         const char *line = msg;
354                         int linelen = get_one_line(msg, len);
355
356                         if (!linelen)
357                                 break;
358
359                         /*
360                          * We want some slop for indentation and a possible
361                          * final "...". Thus the "+ 20".
362                          */
363                         if (offset + linelen + 20 > sizeof(this_header)) {
364                                 memcpy(this_header + offset, "    ...\n", 8);
365                                 offset += 8;
366                                 break;
367                         }
368
369                         msg += linelen;
370                         len -= linelen;
371                         if (linelen == 1)
372                                 hdr = 0;
373                         if (hdr) {
374                                 if (!memcmp(line, "author ", 7))
375                                         offset += add_author_info(this_header + offset, line, linelen);
376                                 continue;
377                         }
378                         memset(this_header + offset, ' ', 4);
379                         memcpy(this_header + offset + 4, line, linelen);
380                         offset += linelen + 4;
381                 }
382                 /* Make sure there is an EOLN */
383                 if (this_header[offset-1] != '\n')
384                         this_header[offset++] = '\n';
385                 /* Add _another_ EOLN if we are doing diff output */
386                 this_header[offset++] = '\n';
387                 this_header[offset] = 0;
388         }
389
390         return this_header;
391 }
392
393 static int diff_tree_commit(const unsigned char *commit, const char *name)
394 {
395         unsigned long size, offset;
396         char *buf = read_object_with_reference(commit, "commit", &size, NULL);
397
398         if (!buf)
399                 return -1;
400
401         if (!name) {
402                 static char commit_name[60];
403                 strcpy(commit_name, sha1_to_hex(commit));
404                 name = commit_name;
405         }
406
407         /* Root commit? */
408         if (show_root_diff && memcmp(buf + 46, "parent ", 7)) {
409                 header = generate_header(name, "root", buf, size);
410                 diff_root_tree(commit, "");
411         }
412
413         /* More than one parent? */
414         if (ignore_merges) {
415                 if (!memcmp(buf + 46 + 48, "parent ", 7))
416                         return 0;
417         }
418
419         offset = 46;
420         while (offset + 48 < size && !memcmp(buf + offset, "parent ", 7)) {
421                 unsigned char parent[20];
422                 if (get_sha1_hex(buf + offset + 7, parent))
423                         return -1;
424                 header = generate_header(name, sha1_to_hex(parent), buf, size);
425                 diff_tree_sha1_top(parent, commit, "");
426                 if (!header && verbose_header) {
427                         header_prefix = "\ndiff-tree ";
428                         /*
429                          * Don't print multiple merge entries if we
430                          * don't print the diffs.
431                          */
432                 }
433                 offset += 48;
434         }
435         return 0;
436 }
437
438 static int diff_tree_stdin(char *line)
439 {
440         int len = strlen(line);
441         unsigned char commit[20], parent[20];
442         static char this_header[1000];
443
444         if (!len || line[len-1] != '\n')
445                 return -1;
446         line[len-1] = 0;
447         if (get_sha1_hex(line, commit))
448                 return -1;
449         if (isspace(line[40]) && !get_sha1_hex(line+41, parent)) {
450                 line[40] = 0;
451                 line[81] = 0;
452                 sprintf(this_header, "%s (from %s)\n", line, line+41);
453                 header = this_header;
454                 return diff_tree_sha1_top(parent, commit, "");
455         }
456         line[40] = 0;
457         return diff_tree_commit(commit, line);
458 }
459
460 static char *diff_tree_usage =
461 "git-diff-tree [-p] [-r] [-z] [--stdin] [-M] [-C] [-R] [-S<string>] [-m] [-s] [-v] [-t] <tree-ish> <tree-ish>";
462
463 int main(int argc, const char **argv)
464 {
465         int nr_sha1;
466         char line[1000];
467         unsigned char sha1[2][20];
468
469         nr_sha1 = 0;
470         for (;;) {
471                 const char *arg;
472
473                 argv++;
474                 argc--;
475                 arg = *argv;
476                 if (!arg)
477                         break;
478
479                 if (*arg != '-') {
480                         if (nr_sha1 < 2 && !get_sha1(arg, sha1[nr_sha1])) {
481                                 nr_sha1++;
482                                 continue;
483                         }
484                         break;
485                 }
486
487                 if (!strcmp(arg, "--")) {
488                         argv++;
489                         argc--;
490                         break;
491                 }
492                 if (!strcmp(arg, "-r")) {
493                         recursive = 1;
494                         continue;
495                 }
496                 if (!strcmp(arg, "-t")) {
497                         recursive = show_tree_entry_in_recursive = 1;
498                         continue;
499                 }
500                 if (!strcmp(arg, "-R")) {
501                         diff_setup_opt |= DIFF_SETUP_REVERSE;
502                         continue;
503                 }
504                 if (!strcmp(arg, "-p")) {
505                         diff_output_format = DIFF_FORMAT_PATCH;
506                         recursive = 1;
507                         continue;
508                 }
509                 if (!strncmp(arg, "-S", 2)) {
510                         pickaxe = arg + 2;
511                         continue;
512                 }
513                 if (!strcmp(arg, "--pickaxe-all")) {
514                         pickaxe_opts = DIFF_PICKAXE_ALL;
515                         continue;
516                 }
517                 if (!strncmp(arg, "-M", 2)) {
518                         detect_rename = DIFF_DETECT_RENAME;
519                         diff_score_opt = diff_scoreopt_parse(arg);
520                         continue;
521                 }
522                 if (!strncmp(arg, "-C", 2)) {
523                         detect_rename = DIFF_DETECT_COPY;
524                         diff_score_opt = diff_scoreopt_parse(arg);
525                         continue;
526                 }
527                 if (!strcmp(arg, "-z")) {
528                         diff_output_format = DIFF_FORMAT_MACHINE;
529                         continue;
530                 }
531                 if (!strcmp(arg, "-m")) {
532                         ignore_merges = 0;
533                         continue;
534                 }
535                 if (!strcmp(arg, "-s")) {
536                         diff_output_format = DIFF_FORMAT_NO_OUTPUT;
537                         continue;
538                 }
539                 if (!strcmp(arg, "-v")) {
540                         verbose_header = 1;
541                         header_prefix = "diff-tree ";
542                         continue;
543                 }
544                 if (!strcmp(arg, "--stdin")) {
545                         read_stdin = 1;
546                         continue;
547                 }
548                 if (!strcmp(arg, "--root")) {
549                         show_root_diff = 1;
550                         continue;
551                 }
552                 usage(diff_tree_usage);
553         }
554
555         if (argc > 0) {
556                 int i;
557
558                 paths = argv;
559                 nr_paths = argc;
560                 pathlens = xmalloc(nr_paths * sizeof(int));
561                 for (i=0; i<nr_paths; i++)
562                         pathlens[i] = strlen(paths[i]);
563         }
564
565         switch (nr_sha1) {
566         case 0:
567                 if (!read_stdin)
568                         usage(diff_tree_usage);
569                 break;
570         case 1:
571                 diff_tree_commit(sha1[0], NULL);
572                 break;
573         case 2:
574                 diff_tree_sha1_top(sha1[0], sha1[1], "");
575                 break;
576         }
577
578         if (!read_stdin)
579                 return 0;
580
581         if (detect_rename)
582                 diff_setup_opt |= (DIFF_SETUP_USE_SIZE_CACHE |
583                                    DIFF_SETUP_USE_CACHE);
584         while (fgets(line, sizeof(line), stdin))
585                 diff_tree_stdin(line);
586
587         return 0;
588 }