Merge branch 'cb/pcre2-chartables-leakfix'
[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 #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 {
45         struct child_process cp = CHILD_PROCESS_INIT;
46         struct strbuf buf = STRBUF_INIT, contents = STRBUF_INIT;
47         struct patch_util *util = NULL;
48         int in_header = 1;
49         char *line, *current_filename = NULL;
50         int offset, len;
51         size_t size;
52
53         argv_array_pushl(&cp.args, "log", "--no-color", "-p", "--no-merges",
54                         "--reverse", "--date-order", "--decorate=no",
55                         "--no-prefix",
56                         /*
57                          * Choose indicators that are not used anywhere
58                          * else in diffs, but still look reasonable
59                          * (e.g. will not be confusing when debugging)
60                          */
61                         "--output-indicator-new=>",
62                         "--output-indicator-old=<",
63                         "--output-indicator-context=#",
64                         "--no-abbrev-commit", range,
65                         NULL);
66         cp.out = -1;
67         cp.no_stdin = 1;
68         cp.git_cmd = 1;
69
70         if (start_command(&cp))
71                 return error_errno(_("could not start `log`"));
72         if (strbuf_read(&contents, cp.out, 0) < 0) {
73                 error_errno(_("could not read `log` output"));
74                 finish_command(&cp);
75                 return -1;
76         }
77
78         line = contents.buf;
79         size = contents.len;
80         for (offset = 0; size > 0; offset += len, size -= len, line += len) {
81                 const char *p;
82
83                 len = find_end_of_line(line, size);
84                 line[len - 1] = '\0';
85                 if (skip_prefix(line, "commit ", &p)) {
86                         if (util) {
87                                 string_list_append(list, buf.buf)->util = util;
88                                 strbuf_reset(&buf);
89                         }
90                         util = xcalloc(sizeof(*util), 1);
91                         if (get_oid(p, &util->oid)) {
92                                 error(_("could not parse commit '%s'"), p);
93                                 free(util);
94                                 string_list_clear(list, 1);
95                                 strbuf_release(&buf);
96                                 strbuf_release(&contents);
97                                 finish_command(&cp);
98                                 return -1;
99                         }
100                         util->matching = -1;
101                         in_header = 1;
102                         continue;
103                 }
104
105                 if (starts_with(line, "diff --git")) {
106                         struct patch patch = { 0 };
107                         struct strbuf root = STRBUF_INIT;
108                         int linenr = 0;
109
110                         in_header = 0;
111                         strbuf_addch(&buf, '\n');
112                         if (!util->diff_offset)
113                                 util->diff_offset = buf.len;
114                         line[len - 1] = '\n';
115                         len = parse_git_diff_header(&root, &linenr, 0, line,
116                                                     len, size, &patch);
117                         if (len < 0)
118                                 die(_("could not parse git header '%.*s'"), (int)len, line);
119                         strbuf_addstr(&buf, " ## ");
120                         if (patch.is_new > 0)
121                                 strbuf_addf(&buf, "%s (new)", patch.new_name);
122                         else if (patch.is_delete > 0)
123                                 strbuf_addf(&buf, "%s (deleted)", patch.old_name);
124                         else if (patch.is_rename)
125                                 strbuf_addf(&buf, "%s => %s", patch.old_name, patch.new_name);
126                         else
127                                 strbuf_addstr(&buf, patch.new_name);
128
129                         free(current_filename);
130                         if (patch.is_delete > 0)
131                                 current_filename = xstrdup(patch.old_name);
132                         else
133                                 current_filename = xstrdup(patch.new_name);
134
135                         if (patch.new_mode && patch.old_mode &&
136                             patch.old_mode != patch.new_mode)
137                                 strbuf_addf(&buf, " (mode change %06o => %06o)",
138                                             patch.old_mode, patch.new_mode);
139
140                         strbuf_addstr(&buf, " ##");
141                 } else if (in_header) {
142                         if (starts_with(line, "Author: ")) {
143                                 strbuf_addstr(&buf, " ## Metadata ##\n");
144                                 strbuf_addstr(&buf, line);
145                                 strbuf_addstr(&buf, "\n\n");
146                                 strbuf_addstr(&buf, " ## Commit message ##\n");
147                         } else if (starts_with(line, "    ")) {
148                                 p = line + len - 2;
149                                 while (isspace(*p) && p >= line)
150                                         p--;
151                                 strbuf_add(&buf, line, p - line + 1);
152                                 strbuf_addch(&buf, '\n');
153                         }
154                         continue;
155                 } else if (skip_prefix(line, "@@ ", &p)) {
156                         p = strstr(p, "@@");
157                         strbuf_addstr(&buf, "@@");
158                         if (current_filename && p[2])
159                                 strbuf_addf(&buf, " %s:", current_filename);
160                         if (p)
161                                 strbuf_addstr(&buf, p + 2);
162                 } else if (!line[0])
163                         /*
164                          * A completely blank (not ' \n', which is context)
165                          * line is not valid in a diff.  We skip it
166                          * silently, because this neatly handles the blank
167                          * separator line between commits in git-log
168                          * output.
169                          */
170                         continue;
171                 else if (line[0] == '>') {
172                         strbuf_addch(&buf, '+');
173                         strbuf_addstr(&buf, line + 1);
174                 } else if (line[0] == '<') {
175                         strbuf_addch(&buf, '-');
176                         strbuf_addstr(&buf, line + 1);
177                 } else if (line[0] == '#') {
178                         strbuf_addch(&buf, ' ');
179                         strbuf_addstr(&buf, line + 1);
180                 } else {
181                         strbuf_addch(&buf, ' ');
182                         strbuf_addstr(&buf, line);
183                 }
184
185                 strbuf_addch(&buf, '\n');
186                 util->diffsize++;
187         }
188         strbuf_release(&contents);
189
190         if (util)
191                 string_list_append(list, buf.buf)->util = util;
192         strbuf_release(&buf);
193         free(current_filename);
194
195         if (finish_command(&cp))
196                 return -1;
197
198         return 0;
199 }
200
201 static int patch_util_cmp(const void *dummy, const struct patch_util *a,
202                           const struct patch_util *b, const char *keydata)
203 {
204         return strcmp(a->diff, keydata ? keydata : b->diff);
205 }
206
207 static void find_exact_matches(struct string_list *a, struct string_list *b)
208 {
209         struct hashmap map;
210         int i;
211
212         hashmap_init(&map, (hashmap_cmp_fn)patch_util_cmp, NULL, 0);
213
214         /* First, add the patches of a to a hash map */
215         for (i = 0; i < a->nr; i++) {
216                 struct patch_util *util = a->items[i].util;
217
218                 util->i = i;
219                 util->patch = a->items[i].string;
220                 util->diff = util->patch + util->diff_offset;
221                 hashmap_entry_init(&util->e, strhash(util->diff));
222                 hashmap_add(&map, &util->e);
223         }
224
225         /* Now try to find exact matches in b */
226         for (i = 0; i < b->nr; i++) {
227                 struct patch_util *util = b->items[i].util, *other;
228
229                 util->i = i;
230                 util->patch = b->items[i].string;
231                 util->diff = util->patch + util->diff_offset;
232                 hashmap_entry_init(&util->e, strhash(util->diff));
233                 other = hashmap_remove_entry(&map, util, e, NULL);
234                 if (other) {
235                         if (other->matching >= 0)
236                                 BUG("already assigned!");
237
238                         other->matching = i;
239                         util->matching = other->i;
240                 }
241         }
242
243         hashmap_free(&map);
244 }
245
246 static void diffsize_consume(void *data, char *line, unsigned long len)
247 {
248         (*(int *)data)++;
249 }
250
251 static void diffsize_hunk(void *data, long ob, long on, long nb, long nn,
252                           const char *funcline, long funclen)
253 {
254         diffsize_consume(data, NULL, 0);
255 }
256
257 static int diffsize(const char *a, const char *b)
258 {
259         xpparam_t pp = { 0 };
260         xdemitconf_t cfg = { 0 };
261         mmfile_t mf1, mf2;
262         int count = 0;
263
264         mf1.ptr = (char *)a;
265         mf1.size = strlen(a);
266         mf2.ptr = (char *)b;
267         mf2.size = strlen(b);
268
269         cfg.ctxlen = 3;
270         if (!xdi_diff_outf(&mf1, &mf2,
271                            diffsize_hunk, diffsize_consume, &count,
272                            &pp, &cfg))
273                 return count;
274
275         error(_("failed to generate diff"));
276         return COST_MAX;
277 }
278
279 static void get_correspondences(struct string_list *a, struct string_list *b,
280                                 int creation_factor)
281 {
282         int n = a->nr + b->nr;
283         int *cost, c, *a2b, *b2a;
284         int i, j;
285
286         ALLOC_ARRAY(cost, st_mult(n, n));
287         ALLOC_ARRAY(a2b, n);
288         ALLOC_ARRAY(b2a, n);
289
290         for (i = 0; i < a->nr; i++) {
291                 struct patch_util *a_util = a->items[i].util;
292
293                 for (j = 0; j < b->nr; j++) {
294                         struct patch_util *b_util = b->items[j].util;
295
296                         if (a_util->matching == j)
297                                 c = 0;
298                         else if (a_util->matching < 0 && b_util->matching < 0)
299                                 c = diffsize(a_util->diff, b_util->diff);
300                         else
301                                 c = COST_MAX;
302                         cost[i + n * j] = c;
303                 }
304
305                 c = a_util->matching < 0 ?
306                         a_util->diffsize * creation_factor / 100 : COST_MAX;
307                 for (j = b->nr; j < n; j++)
308                         cost[i + n * j] = c;
309         }
310
311         for (j = 0; j < b->nr; j++) {
312                 struct patch_util *util = b->items[j].util;
313
314                 c = util->matching < 0 ?
315                         util->diffsize * creation_factor / 100 : COST_MAX;
316                 for (i = a->nr; i < n; i++)
317                         cost[i + n * j] = c;
318         }
319
320         for (i = a->nr; i < n; i++)
321                 for (j = b->nr; j < n; j++)
322                         cost[i + n * j] = 0;
323
324         compute_assignment(n, n, cost, a2b, b2a);
325
326         for (i = 0; i < a->nr; i++)
327                 if (a2b[i] >= 0 && a2b[i] < b->nr) {
328                         struct patch_util *a_util = a->items[i].util;
329                         struct patch_util *b_util = b->items[a2b[i]].util;
330
331                         a_util->matching = a2b[i];
332                         b_util->matching = i;
333                 }
334
335         free(cost);
336         free(a2b);
337         free(b2a);
338 }
339
340 static void output_pair_header(struct diff_options *diffopt,
341                                int patch_no_width,
342                                struct strbuf *buf,
343                                struct strbuf *dashes,
344                                struct patch_util *a_util,
345                                struct patch_util *b_util)
346 {
347         struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
348         struct commit *commit;
349         char status;
350         const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
351         const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
352         const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
353         const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
354         const char *color;
355
356         if (!dashes->len)
357                 strbuf_addchars(dashes, '-',
358                                 strlen(find_unique_abbrev(oid,
359                                                           DEFAULT_ABBREV)));
360
361         if (!b_util) {
362                 color = color_old;
363                 status = '<';
364         } else if (!a_util) {
365                 color = color_new;
366                 status = '>';
367         } else if (strcmp(a_util->patch, b_util->patch)) {
368                 color = color_commit;
369                 status = '!';
370         } else {
371                 color = color_commit;
372                 status = '=';
373         }
374
375         strbuf_reset(buf);
376         strbuf_addstr(buf, status == '!' ? color_old : color);
377         if (!a_util)
378                 strbuf_addf(buf, "%*s:  %s ", patch_no_width, "-", dashes->buf);
379         else
380                 strbuf_addf(buf, "%*d:  %s ", patch_no_width, a_util->i + 1,
381                             find_unique_abbrev(&a_util->oid, DEFAULT_ABBREV));
382
383         if (status == '!')
384                 strbuf_addf(buf, "%s%s", color_reset, color);
385         strbuf_addch(buf, status);
386         if (status == '!')
387                 strbuf_addf(buf, "%s%s", color_reset, color_new);
388
389         if (!b_util)
390                 strbuf_addf(buf, " %*s:  %s", patch_no_width, "-", dashes->buf);
391         else
392                 strbuf_addf(buf, " %*d:  %s", patch_no_width, b_util->i + 1,
393                             find_unique_abbrev(&b_util->oid, DEFAULT_ABBREV));
394
395         commit = lookup_commit_reference(the_repository, oid);
396         if (commit) {
397                 if (status == '!')
398                         strbuf_addf(buf, "%s%s", color_reset, color);
399
400                 strbuf_addch(buf, ' ');
401                 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
402         }
403         strbuf_addf(buf, "%s\n", color_reset);
404
405         fwrite(buf->buf, buf->len, 1, diffopt->file);
406 }
407
408 static struct userdiff_driver section_headers = {
409         .funcname = { "^ ## (.*) ##$\n"
410                       "^.?@@ (.*)$", REG_EXTENDED }
411 };
412
413 static struct diff_filespec *get_filespec(const char *name, const char *p)
414 {
415         struct diff_filespec *spec = alloc_filespec(name);
416
417         fill_filespec(spec, &null_oid, 0, 0100644);
418         spec->data = (char *)p;
419         spec->size = strlen(p);
420         spec->should_munmap = 0;
421         spec->is_stdin = 1;
422         spec->driver = &section_headers;
423
424         return spec;
425 }
426
427 static void patch_diff(const char *a, const char *b,
428                        struct diff_options *diffopt)
429 {
430         diff_queue(&diff_queued_diff,
431                    get_filespec("a", a), get_filespec("b", b));
432
433         diffcore_std(diffopt);
434         diff_flush(diffopt);
435 }
436
437 static void output(struct string_list *a, struct string_list *b,
438                    struct diff_options *diffopt)
439 {
440         struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
441         int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
442         int i = 0, j = 0;
443
444         /*
445          * We assume the user is really more interested in the second argument
446          * ("newer" version). To that end, we print the output in the order of
447          * the RHS (the `b` parameter). To put the LHS (the `a` parameter)
448          * commits that are no longer in the RHS into a good place, we place
449          * them once we have shown all of their predecessors in the LHS.
450          */
451
452         while (i < a->nr || j < b->nr) {
453                 struct patch_util *a_util, *b_util;
454                 a_util = i < a->nr ? a->items[i].util : NULL;
455                 b_util = j < b->nr ? b->items[j].util : NULL;
456
457                 /* Skip all the already-shown commits from the LHS. */
458                 while (i < a->nr && a_util->shown)
459                         a_util = ++i < a->nr ? a->items[i].util : NULL;
460
461                 /* Show unmatched LHS commit whose predecessors were shown. */
462                 if (i < a->nr && a_util->matching < 0) {
463                         output_pair_header(diffopt, patch_no_width,
464                                            &buf, &dashes, a_util, NULL);
465                         i++;
466                         continue;
467                 }
468
469                 /* Show unmatched RHS commits. */
470                 while (j < b->nr && b_util->matching < 0) {
471                         output_pair_header(diffopt, patch_no_width,
472                                            &buf, &dashes, NULL, b_util);
473                         b_util = ++j < b->nr ? b->items[j].util : NULL;
474                 }
475
476                 /* Show matching LHS/RHS pair. */
477                 if (j < b->nr) {
478                         a_util = a->items[b_util->matching].util;
479                         output_pair_header(diffopt, patch_no_width,
480                                            &buf, &dashes, a_util, b_util);
481                         if (!(diffopt->output_format & DIFF_FORMAT_NO_OUTPUT))
482                                 patch_diff(a->items[b_util->matching].string,
483                                            b->items[j].string, diffopt);
484                         a_util->shown = 1;
485                         j++;
486                 }
487         }
488         strbuf_release(&buf);
489         strbuf_release(&dashes);
490 }
491
492 static struct strbuf *output_prefix_cb(struct diff_options *opt, void *data)
493 {
494         return data;
495 }
496
497 int show_range_diff(const char *range1, const char *range2,
498                     int creation_factor, int dual_color,
499                     struct diff_options *diffopt)
500 {
501         int res = 0;
502
503         struct string_list branch1 = STRING_LIST_INIT_DUP;
504         struct string_list branch2 = STRING_LIST_INIT_DUP;
505
506         if (read_patches(range1, &branch1))
507                 res = error(_("could not parse log for '%s'"), range1);
508         if (!res && read_patches(range2, &branch2))
509                 res = error(_("could not parse log for '%s'"), range2);
510
511         if (!res) {
512                 struct diff_options opts;
513                 struct strbuf indent = STRBUF_INIT;
514
515                 if (diffopt)
516                         memcpy(&opts, diffopt, sizeof(opts));
517                 else
518                         diff_setup(&opts);
519
520                 if (!opts.output_format)
521                         opts.output_format = DIFF_FORMAT_PATCH;
522                 opts.flags.suppress_diff_headers = 1;
523                 opts.flags.dual_color_diffed_diffs = dual_color;
524                 opts.flags.suppress_hunk_header_line_count = 1;
525                 opts.output_prefix = output_prefix_cb;
526                 strbuf_addstr(&indent, "    ");
527                 opts.output_prefix_data = &indent;
528                 diff_setup_done(&opts);
529
530                 find_exact_matches(&branch1, &branch2);
531                 get_correspondences(&branch1, &branch2, creation_factor);
532                 output(&branch1, &branch2, &opts);
533
534                 strbuf_release(&indent);
535         }
536
537         string_list_clear(&branch1, 1);
538         string_list_clear(&branch2, 1);
539
540         return res;
541 }