Merge branch 'en/rerere-multi-stage-1-fix' into maint
[git] / range-diff.c
1 #include "cache.h"
2 #include "range-diff.h"
3 #include "string-list.h"
4 #include "run-command.h"
5 #include "argv-array.h"
6 #include "hashmap.h"
7 #include "xdiff-interface.h"
8 #include "linear-assignment.h"
9 #include "diffcore.h"
10 #include "commit.h"
11 #include "pretty.h"
12 #include "userdiff.h"
13
14 struct patch_util {
15         /* For the search for an exact match */
16         struct hashmap_entry e;
17         const char *diff, *patch;
18
19         int i, shown;
20         int diffsize;
21         size_t diff_offset;
22         /* the index of the matching item in the other branch, or -1 */
23         int matching;
24         struct object_id oid;
25 };
26
27 /*
28  * Reads the patches into a string list, with the `util` field being populated
29  * as struct object_id (will need to be free()d).
30  */
31 static int read_patches(const char *range, struct string_list *list)
32 {
33         struct child_process cp = CHILD_PROCESS_INIT;
34         FILE *in;
35         struct strbuf buf = STRBUF_INIT, line = STRBUF_INIT;
36         struct patch_util *util = NULL;
37         int in_header = 1;
38
39         argv_array_pushl(&cp.args, "log", "--no-color", "-p", "--no-merges",
40                         "--reverse", "--date-order", "--decorate=no",
41                         "--no-abbrev-commit", range,
42                         NULL);
43         cp.out = -1;
44         cp.no_stdin = 1;
45         cp.git_cmd = 1;
46
47         if (start_command(&cp))
48                 return error_errno(_("could not start `log`"));
49         in = fdopen(cp.out, "r");
50         if (!in) {
51                 error_errno(_("could not read `log` output"));
52                 finish_command(&cp);
53                 return -1;
54         }
55
56         while (strbuf_getline(&line, in) != EOF) {
57                 const char *p;
58
59                 if (skip_prefix(line.buf, "commit ", &p)) {
60                         if (util) {
61                                 string_list_append(list, buf.buf)->util = util;
62                                 strbuf_reset(&buf);
63                         }
64                         util = xcalloc(sizeof(*util), 1);
65                         if (get_oid(p, &util->oid)) {
66                                 error(_("could not parse commit '%s'"), p);
67                                 free(util);
68                                 string_list_clear(list, 1);
69                                 strbuf_release(&buf);
70                                 strbuf_release(&line);
71                                 fclose(in);
72                                 finish_command(&cp);
73                                 return -1;
74                         }
75                         util->matching = -1;
76                         in_header = 1;
77                         continue;
78                 }
79
80                 if (starts_with(line.buf, "diff --git")) {
81                         in_header = 0;
82                         strbuf_addch(&buf, '\n');
83                         if (!util->diff_offset)
84                                 util->diff_offset = buf.len;
85                         strbuf_addbuf(&buf, &line);
86                 } else if (in_header) {
87                         if (starts_with(line.buf, "Author: ")) {
88                                 strbuf_addbuf(&buf, &line);
89                                 strbuf_addstr(&buf, "\n\n");
90                         } else if (starts_with(line.buf, "    ")) {
91                                 strbuf_rtrim(&line);
92                                 strbuf_addbuf(&buf, &line);
93                                 strbuf_addch(&buf, '\n');
94                         }
95                         continue;
96                 } else if (starts_with(line.buf, "@@ "))
97                         strbuf_addstr(&buf, "@@");
98                 else if (!line.buf[0] || starts_with(line.buf, "index "))
99                         /*
100                          * A completely blank (not ' \n', which is context)
101                          * line is not valid in a diff.  We skip it
102                          * silently, because this neatly handles the blank
103                          * separator line between commits in git-log
104                          * output.
105                          *
106                          * We also want to ignore the diff's `index` lines
107                          * because they contain exact blob hashes in which
108                          * we are not interested.
109                          */
110                         continue;
111                 else
112                         strbuf_addbuf(&buf, &line);
113
114                 strbuf_addch(&buf, '\n');
115                 util->diffsize++;
116         }
117         fclose(in);
118         strbuf_release(&line);
119
120         if (util)
121                 string_list_append(list, buf.buf)->util = util;
122         strbuf_release(&buf);
123
124         if (finish_command(&cp))
125                 return -1;
126
127         return 0;
128 }
129
130 static int patch_util_cmp(const void *dummy, const struct patch_util *a,
131                      const struct patch_util *b, const char *keydata)
132 {
133         return strcmp(a->diff, keydata ? keydata : b->diff);
134 }
135
136 static void find_exact_matches(struct string_list *a, struct string_list *b)
137 {
138         struct hashmap map;
139         int i;
140
141         hashmap_init(&map, (hashmap_cmp_fn)patch_util_cmp, NULL, 0);
142
143         /* First, add the patches of a to a hash map */
144         for (i = 0; i < a->nr; i++) {
145                 struct patch_util *util = a->items[i].util;
146
147                 util->i = i;
148                 util->patch = a->items[i].string;
149                 util->diff = util->patch + util->diff_offset;
150                 hashmap_entry_init(util, strhash(util->diff));
151                 hashmap_add(&map, util);
152         }
153
154         /* Now try to find exact matches in b */
155         for (i = 0; i < b->nr; i++) {
156                 struct patch_util *util = b->items[i].util, *other;
157
158                 util->i = i;
159                 util->patch = b->items[i].string;
160                 util->diff = util->patch + util->diff_offset;
161                 hashmap_entry_init(util, strhash(util->diff));
162                 other = hashmap_remove(&map, util, NULL);
163                 if (other) {
164                         if (other->matching >= 0)
165                                 BUG("already assigned!");
166
167                         other->matching = i;
168                         util->matching = other->i;
169                 }
170         }
171
172         hashmap_free(&map, 0);
173 }
174
175 static void diffsize_consume(void *data, char *line, unsigned long len)
176 {
177         (*(int *)data)++;
178 }
179
180 static int diffsize(const char *a, const char *b)
181 {
182         xpparam_t pp = { 0 };
183         xdemitconf_t cfg = { 0 };
184         mmfile_t mf1, mf2;
185         int count = 0;
186
187         mf1.ptr = (char *)a;
188         mf1.size = strlen(a);
189         mf2.ptr = (char *)b;
190         mf2.size = strlen(b);
191
192         cfg.ctxlen = 3;
193         if (!xdi_diff_outf(&mf1, &mf2, diffsize_consume, &count, &pp, &cfg))
194                 return count;
195
196         error(_("failed to generate diff"));
197         return COST_MAX;
198 }
199
200 static void get_correspondences(struct string_list *a, struct string_list *b,
201                                 int creation_factor)
202 {
203         int n = a->nr + b->nr;
204         int *cost, c, *a2b, *b2a;
205         int i, j;
206
207         ALLOC_ARRAY(cost, st_mult(n, n));
208         ALLOC_ARRAY(a2b, n);
209         ALLOC_ARRAY(b2a, n);
210
211         for (i = 0; i < a->nr; i++) {
212                 struct patch_util *a_util = a->items[i].util;
213
214                 for (j = 0; j < b->nr; j++) {
215                         struct patch_util *b_util = b->items[j].util;
216
217                         if (a_util->matching == j)
218                                 c = 0;
219                         else if (a_util->matching < 0 && b_util->matching < 0)
220                                 c = diffsize(a_util->diff, b_util->diff);
221                         else
222                                 c = COST_MAX;
223                         cost[i + n * j] = c;
224                 }
225
226                 c = a_util->matching < 0 ?
227                         a_util->diffsize * creation_factor / 100 : COST_MAX;
228                 for (j = b->nr; j < n; j++)
229                         cost[i + n * j] = c;
230         }
231
232         for (j = 0; j < b->nr; j++) {
233                 struct patch_util *util = b->items[j].util;
234
235                 c = util->matching < 0 ?
236                         util->diffsize * creation_factor / 100 : COST_MAX;
237                 for (i = a->nr; i < n; i++)
238                         cost[i + n * j] = c;
239         }
240
241         for (i = a->nr; i < n; i++)
242                 for (j = b->nr; j < n; j++)
243                         cost[i + n * j] = 0;
244
245         compute_assignment(n, n, cost, a2b, b2a);
246
247         for (i = 0; i < a->nr; i++)
248                 if (a2b[i] >= 0 && a2b[i] < b->nr) {
249                         struct patch_util *a_util = a->items[i].util;
250                         struct patch_util *b_util = b->items[a2b[i]].util;
251
252                         a_util->matching = a2b[i];
253                         b_util->matching = i;
254                 }
255
256         free(cost);
257         free(a2b);
258         free(b2a);
259 }
260
261 static void output_pair_header(struct diff_options *diffopt,
262                                int patch_no_width,
263                                struct strbuf *buf,
264                                struct strbuf *dashes,
265                                struct patch_util *a_util,
266                                struct patch_util *b_util)
267 {
268         struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
269         struct commit *commit;
270         char status;
271         const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
272         const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
273         const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
274         const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
275         const char *color;
276
277         if (!dashes->len)
278                 strbuf_addchars(dashes, '-',
279                                 strlen(find_unique_abbrev(oid,
280                                                           DEFAULT_ABBREV)));
281
282         if (!b_util) {
283                 color = color_old;
284                 status = '<';
285         } else if (!a_util) {
286                 color = color_new;
287                 status = '>';
288         } else if (strcmp(a_util->patch, b_util->patch)) {
289                 color = color_commit;
290                 status = '!';
291         } else {
292                 color = color_commit;
293                 status = '=';
294         }
295
296         strbuf_reset(buf);
297         strbuf_addstr(buf, status == '!' ? color_old : color);
298         if (!a_util)
299                 strbuf_addf(buf, "%*s:  %s ", patch_no_width, "-", dashes->buf);
300         else
301                 strbuf_addf(buf, "%*d:  %s ", patch_no_width, a_util->i + 1,
302                             find_unique_abbrev(&a_util->oid, DEFAULT_ABBREV));
303
304         if (status == '!')
305                 strbuf_addf(buf, "%s%s", color_reset, color);
306         strbuf_addch(buf, status);
307         if (status == '!')
308                 strbuf_addf(buf, "%s%s", color_reset, color_new);
309
310         if (!b_util)
311                 strbuf_addf(buf, " %*s:  %s", patch_no_width, "-", dashes->buf);
312         else
313                 strbuf_addf(buf, " %*d:  %s", patch_no_width, b_util->i + 1,
314                             find_unique_abbrev(&b_util->oid, DEFAULT_ABBREV));
315
316         commit = lookup_commit_reference(the_repository, oid);
317         if (commit) {
318                 if (status == '!')
319                         strbuf_addf(buf, "%s%s", color_reset, color);
320
321                 strbuf_addch(buf, ' ');
322                 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
323         }
324         strbuf_addf(buf, "%s\n", color_reset);
325
326         fwrite(buf->buf, buf->len, 1, stdout);
327 }
328
329 static struct userdiff_driver no_func_name = {
330         .funcname = { "$^", 0 }
331 };
332
333 static struct diff_filespec *get_filespec(const char *name, const char *p)
334 {
335         struct diff_filespec *spec = alloc_filespec(name);
336
337         fill_filespec(spec, &null_oid, 0, 0644);
338         spec->data = (char *)p;
339         spec->size = strlen(p);
340         spec->should_munmap = 0;
341         spec->is_stdin = 1;
342         spec->driver = &no_func_name;
343
344         return spec;
345 }
346
347 static void patch_diff(const char *a, const char *b,
348                               struct diff_options *diffopt)
349 {
350         diff_queue(&diff_queued_diff,
351                    get_filespec("a", a), get_filespec("b", b));
352
353         diffcore_std(diffopt);
354         diff_flush(diffopt);
355 }
356
357 static void output(struct string_list *a, struct string_list *b,
358                    struct diff_options *diffopt)
359 {
360         struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
361         int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
362         int i = 0, j = 0;
363
364         /*
365          * We assume the user is really more interested in the second argument
366          * ("newer" version). To that end, we print the output in the order of
367          * the RHS (the `b` parameter). To put the LHS (the `a` parameter)
368          * commits that are no longer in the RHS into a good place, we place
369          * them once we have shown all of their predecessors in the LHS.
370          */
371
372         while (i < a->nr || j < b->nr) {
373                 struct patch_util *a_util, *b_util;
374                 a_util = i < a->nr ? a->items[i].util : NULL;
375                 b_util = j < b->nr ? b->items[j].util : NULL;
376
377                 /* Skip all the already-shown commits from the LHS. */
378                 while (i < a->nr && a_util->shown)
379                         a_util = ++i < a->nr ? a->items[i].util : NULL;
380
381                 /* Show unmatched LHS commit whose predecessors were shown. */
382                 if (i < a->nr && a_util->matching < 0) {
383                         output_pair_header(diffopt, patch_no_width,
384                                            &buf, &dashes, a_util, NULL);
385                         i++;
386                         continue;
387                 }
388
389                 /* Show unmatched RHS commits. */
390                 while (j < b->nr && b_util->matching < 0) {
391                         output_pair_header(diffopt, patch_no_width,
392                                            &buf, &dashes, NULL, b_util);
393                         b_util = ++j < b->nr ? b->items[j].util : NULL;
394                 }
395
396                 /* Show matching LHS/RHS pair. */
397                 if (j < b->nr) {
398                         a_util = a->items[b_util->matching].util;
399                         output_pair_header(diffopt, patch_no_width,
400                                            &buf, &dashes, a_util, b_util);
401                         if (!(diffopt->output_format & DIFF_FORMAT_NO_OUTPUT))
402                                 patch_diff(a->items[b_util->matching].string,
403                                            b->items[j].string, diffopt);
404                         a_util->shown = 1;
405                         j++;
406                 }
407         }
408         strbuf_release(&buf);
409         strbuf_release(&dashes);
410 }
411
412 int show_range_diff(const char *range1, const char *range2,
413                     int creation_factor, struct diff_options *diffopt)
414 {
415         int res = 0;
416
417         struct string_list branch1 = STRING_LIST_INIT_DUP;
418         struct string_list branch2 = STRING_LIST_INIT_DUP;
419
420         if (read_patches(range1, &branch1))
421                 res = error(_("could not parse log for '%s'"), range1);
422         if (!res && read_patches(range2, &branch2))
423                 res = error(_("could not parse log for '%s'"), range2);
424
425         if (!res) {
426                 find_exact_matches(&branch1, &branch2);
427                 get_correspondences(&branch1, &branch2, creation_factor);
428                 output(&branch1, &branch2, diffopt);
429         }
430
431         string_list_clear(&branch1, 1);
432         string_list_clear(&branch2, 1);
433
434         return res;
435 }