Merge branch 'ap/maint-diff-rename-avoid-overlap' into maint
[git] / tree-diff.c
1 /*
2  * Helper functions for tree diff generation
3  */
4 #include "cache.h"
5 #include "diff.h"
6 #include "diffcore.h"
7 #include "tree.h"
8
9 static void show_entry(struct diff_options *opt, const char *prefix,
10                        struct tree_desc *desc, struct strbuf *base);
11
12 static int compare_tree_entry(struct tree_desc *t1, struct tree_desc *t2,
13                               struct strbuf *base, struct diff_options *opt)
14 {
15         unsigned mode1, mode2;
16         const char *path1, *path2;
17         const unsigned char *sha1, *sha2;
18         int cmp, pathlen1, pathlen2;
19         int old_baselen = base->len;
20
21         sha1 = tree_entry_extract(t1, &path1, &mode1);
22         sha2 = tree_entry_extract(t2, &path2, &mode2);
23
24         pathlen1 = tree_entry_len(&t1->entry);
25         pathlen2 = tree_entry_len(&t2->entry);
26         cmp = base_name_compare(path1, pathlen1, mode1, path2, pathlen2, mode2);
27         if (cmp < 0) {
28                 show_entry(opt, "-", t1, base);
29                 return -1;
30         }
31         if (cmp > 0) {
32                 show_entry(opt, "+", t2, base);
33                 return 1;
34         }
35         if (!DIFF_OPT_TST(opt, FIND_COPIES_HARDER) && !hashcmp(sha1, sha2) && mode1 == mode2)
36                 return 0;
37
38         /*
39          * If the filemode has changed to/from a directory from/to a regular
40          * file, we need to consider it a remove and an add.
41          */
42         if (S_ISDIR(mode1) != S_ISDIR(mode2)) {
43                 show_entry(opt, "-", t1, base);
44                 show_entry(opt, "+", t2, base);
45                 return 0;
46         }
47
48         strbuf_add(base, path1, pathlen1);
49         if (DIFF_OPT_TST(opt, RECURSIVE) && S_ISDIR(mode1)) {
50                 if (DIFF_OPT_TST(opt, TREE_IN_RECURSIVE)) {
51                         opt->change(opt, mode1, mode2,
52                                     sha1, sha2, 1, 1, base->buf, 0, 0);
53                 }
54                 strbuf_addch(base, '/');
55                 diff_tree_sha1(sha1, sha2, base->buf, opt);
56         } else {
57                 opt->change(opt, mode1, mode2, sha1, sha2, 1, 1, base->buf, 0, 0);
58         }
59         strbuf_setlen(base, old_baselen);
60         return 0;
61 }
62
63 /* A whole sub-tree went away or appeared */
64 static void show_tree(struct diff_options *opt, const char *prefix,
65                       struct tree_desc *desc, struct strbuf *base)
66 {
67         enum interesting match = entry_not_interesting;
68         for (; desc->size; update_tree_entry(desc)) {
69                 if (match != all_entries_interesting) {
70                         match = tree_entry_interesting(&desc->entry, base, 0,
71                                                        &opt->pathspec);
72                         if (match == all_entries_not_interesting)
73                                 break;
74                         if (match == entry_not_interesting)
75                                 continue;
76                 }
77                 show_entry(opt, prefix, desc, base);
78         }
79 }
80
81 /* A file entry went away or appeared */
82 static void show_entry(struct diff_options *opt, const char *prefix,
83                        struct tree_desc *desc, struct strbuf *base)
84 {
85         unsigned mode;
86         const char *path;
87         const unsigned char *sha1 = tree_entry_extract(desc, &path, &mode);
88         int pathlen = tree_entry_len(&desc->entry);
89         int old_baselen = base->len;
90
91         strbuf_add(base, path, pathlen);
92         if (DIFF_OPT_TST(opt, RECURSIVE) && S_ISDIR(mode)) {
93                 enum object_type type;
94                 struct tree_desc inner;
95                 void *tree;
96                 unsigned long size;
97
98                 tree = read_sha1_file(sha1, &type, &size);
99                 if (!tree || type != OBJ_TREE)
100                         die("corrupt tree sha %s", sha1_to_hex(sha1));
101
102                 if (DIFF_OPT_TST(opt, TREE_IN_RECURSIVE))
103                         opt->add_remove(opt, *prefix, mode, sha1, 1, base->buf, 0);
104
105                 strbuf_addch(base, '/');
106
107                 init_tree_desc(&inner, tree, size);
108                 show_tree(opt, prefix, &inner, base);
109                 free(tree);
110         } else
111                 opt->add_remove(opt, prefix[0], mode, sha1, 1, base->buf, 0);
112
113         strbuf_setlen(base, old_baselen);
114 }
115
116 static void skip_uninteresting(struct tree_desc *t, struct strbuf *base,
117                                struct diff_options *opt,
118                                enum interesting *match)
119 {
120         while (t->size) {
121                 *match = tree_entry_interesting(&t->entry, base, 0, &opt->pathspec);
122                 if (*match) {
123                         if (*match == all_entries_not_interesting)
124                                 t->size = 0;
125                         break;
126                 }
127                 update_tree_entry(t);
128         }
129 }
130
131 int diff_tree(struct tree_desc *t1, struct tree_desc *t2,
132               const char *base_str, struct diff_options *opt)
133 {
134         struct strbuf base;
135         int baselen = strlen(base_str);
136         enum interesting t1_match = entry_not_interesting;
137         enum interesting t2_match = entry_not_interesting;
138
139         /* Enable recursion indefinitely */
140         opt->pathspec.recursive = DIFF_OPT_TST(opt, RECURSIVE);
141         opt->pathspec.max_depth = -1;
142
143         strbuf_init(&base, PATH_MAX);
144         strbuf_add(&base, base_str, baselen);
145
146         for (;;) {
147                 if (diff_can_quit_early(opt))
148                         break;
149                 if (opt->pathspec.nr) {
150                         skip_uninteresting(t1, &base, opt, &t1_match);
151                         skip_uninteresting(t2, &base, opt, &t2_match);
152                 }
153                 if (!t1->size) {
154                         if (!t2->size)
155                                 break;
156                         show_entry(opt, "+", t2, &base);
157                         update_tree_entry(t2);
158                         continue;
159                 }
160                 if (!t2->size) {
161                         show_entry(opt, "-", t1, &base);
162                         update_tree_entry(t1);
163                         continue;
164                 }
165                 switch (compare_tree_entry(t1, t2, &base, opt)) {
166                 case -1:
167                         update_tree_entry(t1);
168                         continue;
169                 case 0:
170                         update_tree_entry(t1);
171                         /* Fallthrough */
172                 case 1:
173                         update_tree_entry(t2);
174                         continue;
175                 }
176                 die("git diff-tree: internal error");
177         }
178
179         strbuf_release(&base);
180         return 0;
181 }
182
183 /*
184  * Does it look like the resulting diff might be due to a rename?
185  *  - single entry
186  *  - not a valid previous file
187  */
188 static inline int diff_might_be_rename(void)
189 {
190         return diff_queued_diff.nr == 1 &&
191                 !DIFF_FILE_VALID(diff_queued_diff.queue[0]->one);
192 }
193
194 static void try_to_follow_renames(struct tree_desc *t1, struct tree_desc *t2, const char *base, struct diff_options *opt)
195 {
196         struct diff_options diff_opts;
197         struct diff_queue_struct *q = &diff_queued_diff;
198         struct diff_filepair *choice;
199         const char *paths[1];
200         int i;
201
202         /* Remove the file creation entry from the diff queue, and remember it */
203         choice = q->queue[0];
204         q->nr = 0;
205
206         diff_setup(&diff_opts);
207         DIFF_OPT_SET(&diff_opts, RECURSIVE);
208         DIFF_OPT_SET(&diff_opts, FIND_COPIES_HARDER);
209         diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT;
210         diff_opts.single_follow = opt->pathspec.raw[0];
211         diff_opts.break_opt = opt->break_opt;
212         diff_opts.rename_score = opt->rename_score;
213         paths[0] = NULL;
214         diff_tree_setup_paths(paths, &diff_opts);
215         diff_setup_done(&diff_opts);
216         diff_tree(t1, t2, base, &diff_opts);
217         diffcore_std(&diff_opts);
218         diff_tree_release_paths(&diff_opts);
219
220         /* Go through the new set of filepairing, and see if we find a more interesting one */
221         opt->found_follow = 0;
222         for (i = 0; i < q->nr; i++) {
223                 struct diff_filepair *p = q->queue[i];
224
225                 /*
226                  * Found a source? Not only do we use that for the new
227                  * diff_queued_diff, we will also use that as the path in
228                  * the future!
229                  */
230                 if ((p->status == 'R' || p->status == 'C') &&
231                     !strcmp(p->two->path, opt->pathspec.raw[0])) {
232                         /* Switch the file-pairs around */
233                         q->queue[i] = choice;
234                         choice = p;
235
236                         /* Update the path we use from now on.. */
237                         diff_tree_release_paths(opt);
238                         opt->pathspec.raw[0] = xstrdup(p->one->path);
239                         diff_tree_setup_paths(opt->pathspec.raw, opt);
240
241                         /*
242                          * The caller expects us to return a set of vanilla
243                          * filepairs to let a later call to diffcore_std()
244                          * it makes to sort the renames out (among other
245                          * things), but we already have found renames
246                          * ourselves; signal diffcore_std() not to muck with
247                          * rename information.
248                          */
249                         opt->found_follow = 1;
250                         break;
251                 }
252         }
253
254         /*
255          * Then, discard all the non-relevant file pairs...
256          */
257         for (i = 0; i < q->nr; i++) {
258                 struct diff_filepair *p = q->queue[i];
259                 diff_free_filepair(p);
260         }
261
262         /*
263          * .. and re-instate the one we want (which might be either the
264          * original one, or the rename/copy we found)
265          */
266         q->queue[0] = choice;
267         q->nr = 1;
268 }
269
270 int diff_tree_sha1(const unsigned char *old, const unsigned char *new, const char *base, struct diff_options *opt)
271 {
272         void *tree1, *tree2;
273         struct tree_desc t1, t2;
274         unsigned long size1, size2;
275         int retval;
276
277         tree1 = read_object_with_reference(old, tree_type, &size1, NULL);
278         if (!tree1)
279                 die("unable to read source tree (%s)", sha1_to_hex(old));
280         tree2 = read_object_with_reference(new, tree_type, &size2, NULL);
281         if (!tree2)
282                 die("unable to read destination tree (%s)", sha1_to_hex(new));
283         init_tree_desc(&t1, tree1, size1);
284         init_tree_desc(&t2, tree2, size2);
285         retval = diff_tree(&t1, &t2, base, opt);
286         if (!*base && DIFF_OPT_TST(opt, FOLLOW_RENAMES) && diff_might_be_rename()) {
287                 init_tree_desc(&t1, tree1, size1);
288                 init_tree_desc(&t2, tree2, size2);
289                 try_to_follow_renames(&t1, &t2, base, opt);
290         }
291         free(tree1);
292         free(tree2);
293         return retval;
294 }
295
296 int diff_root_tree_sha1(const unsigned char *new, const char *base, struct diff_options *opt)
297 {
298         int retval;
299         void *tree;
300         unsigned long size;
301         struct tree_desc empty, real;
302
303         tree = read_object_with_reference(new, tree_type, &size, NULL);
304         if (!tree)
305                 die("unable to read root tree (%s)", sha1_to_hex(new));
306         init_tree_desc(&real, tree, size);
307
308         init_tree_desc(&empty, "", 0);
309         retval = diff_tree(&empty, &real, base, opt);
310         free(tree);
311         return retval;
312 }
313
314 void diff_tree_release_paths(struct diff_options *opt)
315 {
316         free_pathspec(&opt->pathspec);
317 }
318
319 void diff_tree_setup_paths(const char **p, struct diff_options *opt)
320 {
321         init_pathspec(&opt->pathspec, p);
322 }