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