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