mem-pool: add convenience functions for strdup and strndup
[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 "strvec.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 #include "apply.h"
14
15 struct patch_util {
16         /* For the search for an exact match */
17         struct hashmap_entry e;
18         const char *diff, *patch;
19
20         int i, shown;
21         int diffsize;
22         size_t diff_offset;
23         /* the index of the matching item in the other branch, or -1 */
24         int matching;
25         struct object_id oid;
26 };
27
28 static size_t find_end_of_line(char *buffer, unsigned long size)
29 {
30         char *eol = memchr(buffer, '\n', size);
31
32         if (!eol)
33                 return size;
34
35         *eol = '\0';
36         return eol + 1 - buffer;
37 }
38
39 /*
40  * Reads the patches into a string list, with the `util` field being populated
41  * as struct object_id (will need to be free()d).
42  */
43 static int read_patches(const char *range, struct string_list *list,
44                         const struct strvec *other_arg)
45 {
46         struct child_process cp = CHILD_PROCESS_INIT;
47         struct strbuf buf = STRBUF_INIT, contents = STRBUF_INIT;
48         struct patch_util *util = NULL;
49         int in_header = 1;
50         char *line, *current_filename = NULL;
51         int offset, len;
52         size_t size;
53
54         strvec_pushl(&cp.args, "log", "--no-color", "-p", "--no-merges",
55                      "--reverse", "--date-order", "--decorate=no",
56                      "--no-prefix",
57                      /*
58                       * Choose indicators that are not used anywhere
59                       * else in diffs, but still look reasonable
60                       * (e.g. will not be confusing when debugging)
61                       */
62                      "--output-indicator-new=>",
63                      "--output-indicator-old=<",
64                      "--output-indicator-context=#",
65                      "--no-abbrev-commit",
66                      "--pretty=medium",
67                      "--notes",
68                      NULL);
69         if (other_arg)
70                 strvec_pushv(&cp.args, other_arg->v);
71         strvec_push(&cp.args, range);
72         cp.out = -1;
73         cp.no_stdin = 1;
74         cp.git_cmd = 1;
75
76         if (start_command(&cp))
77                 return error_errno(_("could not start `log`"));
78         if (strbuf_read(&contents, cp.out, 0) < 0) {
79                 error_errno(_("could not read `log` output"));
80                 finish_command(&cp);
81                 return -1;
82         }
83
84         line = contents.buf;
85         size = contents.len;
86         for (offset = 0; size > 0; offset += len, size -= len, line += len) {
87                 const char *p;
88
89                 len = find_end_of_line(line, size);
90                 line[len - 1] = '\0';
91                 if (skip_prefix(line, "commit ", &p)) {
92                         if (util) {
93                                 string_list_append(list, buf.buf)->util = util;
94                                 strbuf_reset(&buf);
95                         }
96                         util = xcalloc(sizeof(*util), 1);
97                         if (get_oid(p, &util->oid)) {
98                                 error(_("could not parse commit '%s'"), p);
99                                 free(util);
100                                 string_list_clear(list, 1);
101                                 strbuf_release(&buf);
102                                 strbuf_release(&contents);
103                                 finish_command(&cp);
104                                 return -1;
105                         }
106                         util->matching = -1;
107                         in_header = 1;
108                         continue;
109                 }
110
111                 if (!util) {
112                         error(_("could not parse first line of `log` output: "
113                                 "did not start with 'commit ': '%s'"),
114                               line);
115                         string_list_clear(list, 1);
116                         strbuf_release(&buf);
117                         strbuf_release(&contents);
118                         finish_command(&cp);
119                         return -1;
120                 }
121
122                 if (starts_with(line, "diff --git")) {
123                         struct patch patch = { 0 };
124                         struct strbuf root = STRBUF_INIT;
125                         int linenr = 0;
126                         int orig_len;
127
128                         in_header = 0;
129                         strbuf_addch(&buf, '\n');
130                         if (!util->diff_offset)
131                                 util->diff_offset = buf.len;
132                         line[len - 1] = '\n';
133                         orig_len = len;
134                         len = parse_git_diff_header(&root, &linenr, 0, line,
135                                                     len, size, &patch);
136                         if (len < 0)
137                                 die(_("could not parse git header '%.*s'"),
138                                     orig_len, line);
139                         strbuf_addstr(&buf, " ## ");
140                         if (patch.is_new > 0)
141                                 strbuf_addf(&buf, "%s (new)", patch.new_name);
142                         else if (patch.is_delete > 0)
143                                 strbuf_addf(&buf, "%s (deleted)", patch.old_name);
144                         else if (patch.is_rename)
145                                 strbuf_addf(&buf, "%s => %s", patch.old_name, patch.new_name);
146                         else
147                                 strbuf_addstr(&buf, patch.new_name);
148
149                         free(current_filename);
150                         if (patch.is_delete > 0)
151                                 current_filename = xstrdup(patch.old_name);
152                         else
153                                 current_filename = xstrdup(patch.new_name);
154
155                         if (patch.new_mode && patch.old_mode &&
156                             patch.old_mode != patch.new_mode)
157                                 strbuf_addf(&buf, " (mode change %06o => %06o)",
158                                             patch.old_mode, patch.new_mode);
159
160                         strbuf_addstr(&buf, " ##");
161                 } else if (in_header) {
162                         if (starts_with(line, "Author: ")) {
163                                 strbuf_addstr(&buf, " ## Metadata ##\n");
164                                 strbuf_addstr(&buf, line);
165                                 strbuf_addstr(&buf, "\n\n");
166                                 strbuf_addstr(&buf, " ## Commit message ##\n");
167                         } else if (starts_with(line, "Notes") &&
168                                    line[strlen(line) - 1] == ':') {
169                                 strbuf_addstr(&buf, "\n\n");
170                                 /* strip the trailing colon */
171                                 strbuf_addf(&buf, " ## %.*s ##\n",
172                                             (int)(strlen(line) - 1), line);
173                         } else if (starts_with(line, "    ")) {
174                                 p = line + len - 2;
175                                 while (isspace(*p) && p >= line)
176                                         p--;
177                                 strbuf_add(&buf, line, p - line + 1);
178                                 strbuf_addch(&buf, '\n');
179                         }
180                         continue;
181                 } else if (skip_prefix(line, "@@ ", &p)) {
182                         p = strstr(p, "@@");
183                         strbuf_addstr(&buf, "@@");
184                         if (current_filename && p[2])
185                                 strbuf_addf(&buf, " %s:", current_filename);
186                         if (p)
187                                 strbuf_addstr(&buf, p + 2);
188                 } else if (!line[0])
189                         /*
190                          * A completely blank (not ' \n', which is context)
191                          * line is not valid in a diff.  We skip it
192                          * silently, because this neatly handles the blank
193                          * separator line between commits in git-log
194                          * output.
195                          */
196                         continue;
197                 else if (line[0] == '>') {
198                         strbuf_addch(&buf, '+');
199                         strbuf_addstr(&buf, line + 1);
200                 } else if (line[0] == '<') {
201                         strbuf_addch(&buf, '-');
202                         strbuf_addstr(&buf, line + 1);
203                 } else if (line[0] == '#') {
204                         strbuf_addch(&buf, ' ');
205                         strbuf_addstr(&buf, line + 1);
206                 } else {
207                         strbuf_addch(&buf, ' ');
208                         strbuf_addstr(&buf, line);
209                 }
210
211                 strbuf_addch(&buf, '\n');
212                 util->diffsize++;
213         }
214         strbuf_release(&contents);
215
216         if (util)
217                 string_list_append(list, buf.buf)->util = util;
218         strbuf_release(&buf);
219         free(current_filename);
220
221         if (finish_command(&cp))
222                 return -1;
223
224         return 0;
225 }
226
227 static int patch_util_cmp(const void *dummy, const struct patch_util *a,
228                           const struct patch_util *b, const char *keydata)
229 {
230         return strcmp(a->diff, keydata ? keydata : b->diff);
231 }
232
233 static void find_exact_matches(struct string_list *a, struct string_list *b)
234 {
235         struct hashmap map;
236         int i;
237
238         hashmap_init(&map, (hashmap_cmp_fn)patch_util_cmp, NULL, 0);
239
240         /* First, add the patches of a to a hash map */
241         for (i = 0; i < a->nr; i++) {
242                 struct patch_util *util = a->items[i].util;
243
244                 util->i = i;
245                 util->patch = a->items[i].string;
246                 util->diff = util->patch + util->diff_offset;
247                 hashmap_entry_init(&util->e, strhash(util->diff));
248                 hashmap_add(&map, &util->e);
249         }
250
251         /* Now try to find exact matches in b */
252         for (i = 0; i < b->nr; i++) {
253                 struct patch_util *util = b->items[i].util, *other;
254
255                 util->i = i;
256                 util->patch = b->items[i].string;
257                 util->diff = util->patch + util->diff_offset;
258                 hashmap_entry_init(&util->e, strhash(util->diff));
259                 other = hashmap_remove_entry(&map, util, e, NULL);
260                 if (other) {
261                         if (other->matching >= 0)
262                                 BUG("already assigned!");
263
264                         other->matching = i;
265                         util->matching = other->i;
266                 }
267         }
268
269         hashmap_free(&map);
270 }
271
272 static void diffsize_consume(void *data, char *line, unsigned long len)
273 {
274         (*(int *)data)++;
275 }
276
277 static void diffsize_hunk(void *data, long ob, long on, long nb, long nn,
278                           const char *funcline, long funclen)
279 {
280         diffsize_consume(data, NULL, 0);
281 }
282
283 static int diffsize(const char *a, const char *b)
284 {
285         xpparam_t pp = { 0 };
286         xdemitconf_t cfg = { 0 };
287         mmfile_t mf1, mf2;
288         int count = 0;
289
290         mf1.ptr = (char *)a;
291         mf1.size = strlen(a);
292         mf2.ptr = (char *)b;
293         mf2.size = strlen(b);
294
295         cfg.ctxlen = 3;
296         if (!xdi_diff_outf(&mf1, &mf2,
297                            diffsize_hunk, diffsize_consume, &count,
298                            &pp, &cfg))
299                 return count;
300
301         error(_("failed to generate diff"));
302         return COST_MAX;
303 }
304
305 static void get_correspondences(struct string_list *a, struct string_list *b,
306                                 int creation_factor)
307 {
308         int n = a->nr + b->nr;
309         int *cost, c, *a2b, *b2a;
310         int i, j;
311
312         ALLOC_ARRAY(cost, st_mult(n, n));
313         ALLOC_ARRAY(a2b, n);
314         ALLOC_ARRAY(b2a, n);
315
316         for (i = 0; i < a->nr; i++) {
317                 struct patch_util *a_util = a->items[i].util;
318
319                 for (j = 0; j < b->nr; j++) {
320                         struct patch_util *b_util = b->items[j].util;
321
322                         if (a_util->matching == j)
323                                 c = 0;
324                         else if (a_util->matching < 0 && b_util->matching < 0)
325                                 c = diffsize(a_util->diff, b_util->diff);
326                         else
327                                 c = COST_MAX;
328                         cost[i + n * j] = c;
329                 }
330
331                 c = a_util->matching < 0 ?
332                         a_util->diffsize * creation_factor / 100 : COST_MAX;
333                 for (j = b->nr; j < n; j++)
334                         cost[i + n * j] = c;
335         }
336
337         for (j = 0; j < b->nr; j++) {
338                 struct patch_util *util = b->items[j].util;
339
340                 c = util->matching < 0 ?
341                         util->diffsize * creation_factor / 100 : COST_MAX;
342                 for (i = a->nr; i < n; i++)
343                         cost[i + n * j] = c;
344         }
345
346         for (i = a->nr; i < n; i++)
347                 for (j = b->nr; j < n; j++)
348                         cost[i + n * j] = 0;
349
350         compute_assignment(n, n, cost, a2b, b2a);
351
352         for (i = 0; i < a->nr; i++)
353                 if (a2b[i] >= 0 && a2b[i] < b->nr) {
354                         struct patch_util *a_util = a->items[i].util;
355                         struct patch_util *b_util = b->items[a2b[i]].util;
356
357                         a_util->matching = a2b[i];
358                         b_util->matching = i;
359                 }
360
361         free(cost);
362         free(a2b);
363         free(b2a);
364 }
365
366 static void output_pair_header(struct diff_options *diffopt,
367                                int patch_no_width,
368                                struct strbuf *buf,
369                                struct strbuf *dashes,
370                                struct patch_util *a_util,
371                                struct patch_util *b_util)
372 {
373         struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
374         struct commit *commit;
375         char status;
376         const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
377         const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
378         const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
379         const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
380         const char *color;
381
382         if (!dashes->len)
383                 strbuf_addchars(dashes, '-',
384                                 strlen(find_unique_abbrev(oid,
385                                                           DEFAULT_ABBREV)));
386
387         if (!b_util) {
388                 color = color_old;
389                 status = '<';
390         } else if (!a_util) {
391                 color = color_new;
392                 status = '>';
393         } else if (strcmp(a_util->patch, b_util->patch)) {
394                 color = color_commit;
395                 status = '!';
396         } else {
397                 color = color_commit;
398                 status = '=';
399         }
400
401         strbuf_reset(buf);
402         strbuf_addstr(buf, status == '!' ? color_old : color);
403         if (!a_util)
404                 strbuf_addf(buf, "%*s:  %s ", patch_no_width, "-", dashes->buf);
405         else
406                 strbuf_addf(buf, "%*d:  %s ", patch_no_width, a_util->i + 1,
407                             find_unique_abbrev(&a_util->oid, DEFAULT_ABBREV));
408
409         if (status == '!')
410                 strbuf_addf(buf, "%s%s", color_reset, color);
411         strbuf_addch(buf, status);
412         if (status == '!')
413                 strbuf_addf(buf, "%s%s", color_reset, color_new);
414
415         if (!b_util)
416                 strbuf_addf(buf, " %*s:  %s", patch_no_width, "-", dashes->buf);
417         else
418                 strbuf_addf(buf, " %*d:  %s", patch_no_width, b_util->i + 1,
419                             find_unique_abbrev(&b_util->oid, DEFAULT_ABBREV));
420
421         commit = lookup_commit_reference(the_repository, oid);
422         if (commit) {
423                 if (status == '!')
424                         strbuf_addf(buf, "%s%s", color_reset, color);
425
426                 strbuf_addch(buf, ' ');
427                 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
428         }
429         strbuf_addf(buf, "%s\n", color_reset);
430
431         fwrite(buf->buf, buf->len, 1, diffopt->file);
432 }
433
434 static struct userdiff_driver section_headers = {
435         .funcname = { "^ ## (.*) ##$\n"
436                       "^.?@@ (.*)$", REG_EXTENDED }
437 };
438
439 static struct diff_filespec *get_filespec(const char *name, const char *p)
440 {
441         struct diff_filespec *spec = alloc_filespec(name);
442
443         fill_filespec(spec, &null_oid, 0, 0100644);
444         spec->data = (char *)p;
445         spec->size = strlen(p);
446         spec->should_munmap = 0;
447         spec->is_stdin = 1;
448         spec->driver = &section_headers;
449
450         return spec;
451 }
452
453 static void patch_diff(const char *a, const char *b,
454                        struct diff_options *diffopt)
455 {
456         diff_queue(&diff_queued_diff,
457                    get_filespec("a", a), get_filespec("b", b));
458
459         diffcore_std(diffopt);
460         diff_flush(diffopt);
461 }
462
463 static void output(struct string_list *a, struct string_list *b,
464                    struct diff_options *diffopt)
465 {
466         struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
467         int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
468         int i = 0, j = 0;
469
470         /*
471          * We assume the user is really more interested in the second argument
472          * ("newer" version). To that end, we print the output in the order of
473          * the RHS (the `b` parameter). To put the LHS (the `a` parameter)
474          * commits that are no longer in the RHS into a good place, we place
475          * them once we have shown all of their predecessors in the LHS.
476          */
477
478         while (i < a->nr || j < b->nr) {
479                 struct patch_util *a_util, *b_util;
480                 a_util = i < a->nr ? a->items[i].util : NULL;
481                 b_util = j < b->nr ? b->items[j].util : NULL;
482
483                 /* Skip all the already-shown commits from the LHS. */
484                 while (i < a->nr && a_util->shown)
485                         a_util = ++i < a->nr ? a->items[i].util : NULL;
486
487                 /* Show unmatched LHS commit whose predecessors were shown. */
488                 if (i < a->nr && a_util->matching < 0) {
489                         output_pair_header(diffopt, patch_no_width,
490                                            &buf, &dashes, a_util, NULL);
491                         i++;
492                         continue;
493                 }
494
495                 /* Show unmatched RHS commits. */
496                 while (j < b->nr && b_util->matching < 0) {
497                         output_pair_header(diffopt, patch_no_width,
498                                            &buf, &dashes, NULL, b_util);
499                         b_util = ++j < b->nr ? b->items[j].util : NULL;
500                 }
501
502                 /* Show matching LHS/RHS pair. */
503                 if (j < b->nr) {
504                         a_util = a->items[b_util->matching].util;
505                         output_pair_header(diffopt, patch_no_width,
506                                            &buf, &dashes, a_util, b_util);
507                         if (!(diffopt->output_format & DIFF_FORMAT_NO_OUTPUT))
508                                 patch_diff(a->items[b_util->matching].string,
509                                            b->items[j].string, diffopt);
510                         a_util->shown = 1;
511                         j++;
512                 }
513         }
514         strbuf_release(&buf);
515         strbuf_release(&dashes);
516 }
517
518 static struct strbuf *output_prefix_cb(struct diff_options *opt, void *data)
519 {
520         return data;
521 }
522
523 int show_range_diff(const char *range1, const char *range2,
524                     int creation_factor, int dual_color,
525                     const struct diff_options *diffopt,
526                     const struct strvec *other_arg)
527 {
528         int res = 0;
529
530         struct string_list branch1 = STRING_LIST_INIT_DUP;
531         struct string_list branch2 = STRING_LIST_INIT_DUP;
532
533         if (read_patches(range1, &branch1, other_arg))
534                 res = error(_("could not parse log for '%s'"), range1);
535         if (!res && read_patches(range2, &branch2, other_arg))
536                 res = error(_("could not parse log for '%s'"), range2);
537
538         if (!res) {
539                 struct diff_options opts;
540                 struct strbuf indent = STRBUF_INIT;
541
542                 if (diffopt)
543                         memcpy(&opts, diffopt, sizeof(opts));
544                 else
545                         diff_setup(&opts);
546
547                 if (!opts.output_format)
548                         opts.output_format = DIFF_FORMAT_PATCH;
549                 opts.flags.suppress_diff_headers = 1;
550                 opts.flags.dual_color_diffed_diffs = dual_color;
551                 opts.flags.suppress_hunk_header_line_count = 1;
552                 opts.output_prefix = output_prefix_cb;
553                 strbuf_addstr(&indent, "    ");
554                 opts.output_prefix_data = &indent;
555                 diff_setup_done(&opts);
556
557                 find_exact_matches(&branch1, &branch2);
558                 get_correspondences(&branch1, &branch2, creation_factor);
559                 output(&branch1, &branch2, &opts);
560
561                 strbuf_release(&indent);
562         }
563
564         string_list_clear(&branch1, 1);
565         string_list_clear(&branch2, 1);
566
567         return res;
568 }