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
14struct 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 */
31static 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 /*
42 * Choose indicators that are not used anywhere
43 * else in diffs, but still look reasonable
44 * (e.g. will not be confusing when debugging)
45 */
46 "--output-indicator-new=>",
47 "--output-indicator-old=<",
48 "--output-indicator-context=#",
49 "--no-abbrev-commit", range,
50 NULL);
51 cp.out = -1;
52 cp.no_stdin = 1;
53 cp.git_cmd = 1;
54
55 if (start_command(&cp))
56 return error_errno(_("could not start `log`"));
57 in = fdopen(cp.out, "r");
58 if (!in) {
59 error_errno(_("could not read `log` output"));
60 finish_command(&cp);
61 return -1;
62 }
63
64 while (strbuf_getline(&line, in) != EOF) {
65 const char *p;
66
67 if (skip_prefix(line.buf, "commit ", &p)) {
68 if (util) {
69 string_list_append(list, buf.buf)->util = util;
70 strbuf_reset(&buf);
71 }
72 util = xcalloc(sizeof(*util), 1);
73 if (get_oid(p, &util->oid)) {
74 error(_("could not parse commit '%s'"), p);
75 free(util);
76 string_list_clear(list, 1);
77 strbuf_release(&buf);
78 strbuf_release(&line);
79 fclose(in);
80 finish_command(&cp);
81 return -1;
82 }
83 util->matching = -1;
84 in_header = 1;
85 continue;
86 }
87
88 if (starts_with(line.buf, "diff --git")) {
89 in_header = 0;
90 strbuf_addch(&buf, '\n');
91 if (!util->diff_offset)
92 util->diff_offset = buf.len;
93 strbuf_addch(&buf, ' ');
94 strbuf_addbuf(&buf, &line);
95 } else if (in_header) {
96 if (starts_with(line.buf, "Author: ")) {
97 strbuf_addbuf(&buf, &line);
98 strbuf_addstr(&buf, "\n\n");
99 } else if (starts_with(line.buf, " ")) {
100 strbuf_rtrim(&line);
101 strbuf_addbuf(&buf, &line);
102 strbuf_addch(&buf, '\n');
103 }
104 continue;
105 } else if (starts_with(line.buf, "@@ "))
106 strbuf_addstr(&buf, "@@");
107 else if (!line.buf[0] || starts_with(line.buf, "index "))
108 /*
109 * A completely blank (not ' \n', which is context)
110 * line is not valid in a diff. We skip it
111 * silently, because this neatly handles the blank
112 * separator line between commits in git-log
113 * output.
114 *
115 * We also want to ignore the diff's `index` lines
116 * because they contain exact blob hashes in which
117 * we are not interested.
118 */
119 continue;
120 else if (line.buf[0] == '>') {
121 strbuf_addch(&buf, '+');
122 strbuf_add(&buf, line.buf + 1, line.len - 1);
123 } else if (line.buf[0] == '<') {
124 strbuf_addch(&buf, '-');
125 strbuf_add(&buf, line.buf + 1, line.len - 1);
126 } else if (line.buf[0] == '#') {
127 strbuf_addch(&buf, ' ');
128 strbuf_add(&buf, line.buf + 1, line.len - 1);
129 } else {
130 strbuf_addch(&buf, ' ');
131 strbuf_addbuf(&buf, &line);
132 }
133
134 strbuf_addch(&buf, '\n');
135 util->diffsize++;
136 }
137 fclose(in);
138 strbuf_release(&line);
139
140 if (util)
141 string_list_append(list, buf.buf)->util = util;
142 strbuf_release(&buf);
143
144 if (finish_command(&cp))
145 return -1;
146
147 return 0;
148}
149
150static int patch_util_cmp(const void *dummy, const struct patch_util *a,
151 const struct patch_util *b, const char *keydata)
152{
153 return strcmp(a->diff, keydata ? keydata : b->diff);
154}
155
156static void find_exact_matches(struct string_list *a, struct string_list *b)
157{
158 struct hashmap map;
159 int i;
160
161 hashmap_init(&map, (hashmap_cmp_fn)patch_util_cmp, NULL, 0);
162
163 /* First, add the patches of a to a hash map */
164 for (i = 0; i < a->nr; i++) {
165 struct patch_util *util = a->items[i].util;
166
167 util->i = i;
168 util->patch = a->items[i].string;
169 util->diff = util->patch + util->diff_offset;
170 hashmap_entry_init(util, strhash(util->diff));
171 hashmap_add(&map, util);
172 }
173
174 /* Now try to find exact matches in b */
175 for (i = 0; i < b->nr; i++) {
176 struct patch_util *util = b->items[i].util, *other;
177
178 util->i = i;
179 util->patch = b->items[i].string;
180 util->diff = util->patch + util->diff_offset;
181 hashmap_entry_init(util, strhash(util->diff));
182 other = hashmap_remove(&map, util, NULL);
183 if (other) {
184 if (other->matching >= 0)
185 BUG("already assigned!");
186
187 other->matching = i;
188 util->matching = other->i;
189 }
190 }
191
192 hashmap_free(&map, 0);
193}
194
195static void diffsize_consume(void *data, char *line, unsigned long len)
196{
197 (*(int *)data)++;
198}
199
200static void diffsize_hunk(void *data, long ob, long on, long nb, long nn,
201 const char *funcline, long funclen)
202{
203 diffsize_consume(data, NULL, 0);
204}
205
206static int diffsize(const char *a, const char *b)
207{
208 xpparam_t pp = { 0 };
209 xdemitconf_t cfg = { 0 };
210 mmfile_t mf1, mf2;
211 int count = 0;
212
213 mf1.ptr = (char *)a;
214 mf1.size = strlen(a);
215 mf2.ptr = (char *)b;
216 mf2.size = strlen(b);
217
218 cfg.ctxlen = 3;
219 if (!xdi_diff_outf(&mf1, &mf2,
220 diffsize_hunk, diffsize_consume, &count,
221 &pp, &cfg))
222 return count;
223
224 error(_("failed to generate diff"));
225 return COST_MAX;
226}
227
228static void get_correspondences(struct string_list *a, struct string_list *b,
229 int creation_factor)
230{
231 int n = a->nr + b->nr;
232 int *cost, c, *a2b, *b2a;
233 int i, j;
234
235 ALLOC_ARRAY(cost, st_mult(n, n));
236 ALLOC_ARRAY(a2b, n);
237 ALLOC_ARRAY(b2a, n);
238
239 for (i = 0; i < a->nr; i++) {
240 struct patch_util *a_util = a->items[i].util;
241
242 for (j = 0; j < b->nr; j++) {
243 struct patch_util *b_util = b->items[j].util;
244
245 if (a_util->matching == j)
246 c = 0;
247 else if (a_util->matching < 0 && b_util->matching < 0)
248 c = diffsize(a_util->diff, b_util->diff);
249 else
250 c = COST_MAX;
251 cost[i + n * j] = c;
252 }
253
254 c = a_util->matching < 0 ?
255 a_util->diffsize * creation_factor / 100 : COST_MAX;
256 for (j = b->nr; j < n; j++)
257 cost[i + n * j] = c;
258 }
259
260 for (j = 0; j < b->nr; j++) {
261 struct patch_util *util = b->items[j].util;
262
263 c = util->matching < 0 ?
264 util->diffsize * creation_factor / 100 : COST_MAX;
265 for (i = a->nr; i < n; i++)
266 cost[i + n * j] = c;
267 }
268
269 for (i = a->nr; i < n; i++)
270 for (j = b->nr; j < n; j++)
271 cost[i + n * j] = 0;
272
273 compute_assignment(n, n, cost, a2b, b2a);
274
275 for (i = 0; i < a->nr; i++)
276 if (a2b[i] >= 0 && a2b[i] < b->nr) {
277 struct patch_util *a_util = a->items[i].util;
278 struct patch_util *b_util = b->items[a2b[i]].util;
279
280 a_util->matching = a2b[i];
281 b_util->matching = i;
282 }
283
284 free(cost);
285 free(a2b);
286 free(b2a);
287}
288
289static void output_pair_header(struct diff_options *diffopt,
290 int patch_no_width,
291 struct strbuf *buf,
292 struct strbuf *dashes,
293 struct patch_util *a_util,
294 struct patch_util *b_util)
295{
296 struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
297 struct commit *commit;
298 char status;
299 const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
300 const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
301 const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
302 const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
303 const char *color;
304
305 if (!dashes->len)
306 strbuf_addchars(dashes, '-',
307 strlen(find_unique_abbrev(oid,
308 DEFAULT_ABBREV)));
309
310 if (!b_util) {
311 color = color_old;
312 status = '<';
313 } else if (!a_util) {
314 color = color_new;
315 status = '>';
316 } else if (strcmp(a_util->patch, b_util->patch)) {
317 color = color_commit;
318 status = '!';
319 } else {
320 color = color_commit;
321 status = '=';
322 }
323
324 strbuf_reset(buf);
325 strbuf_addstr(buf, status == '!' ? color_old : color);
326 if (!a_util)
327 strbuf_addf(buf, "%*s: %s ", patch_no_width, "-", dashes->buf);
328 else
329 strbuf_addf(buf, "%*d: %s ", patch_no_width, a_util->i + 1,
330 find_unique_abbrev(&a_util->oid, DEFAULT_ABBREV));
331
332 if (status == '!')
333 strbuf_addf(buf, "%s%s", color_reset, color);
334 strbuf_addch(buf, status);
335 if (status == '!')
336 strbuf_addf(buf, "%s%s", color_reset, color_new);
337
338 if (!b_util)
339 strbuf_addf(buf, " %*s: %s", patch_no_width, "-", dashes->buf);
340 else
341 strbuf_addf(buf, " %*d: %s", patch_no_width, b_util->i + 1,
342 find_unique_abbrev(&b_util->oid, DEFAULT_ABBREV));
343
344 commit = lookup_commit_reference(the_repository, oid);
345 if (commit) {
346 if (status == '!')
347 strbuf_addf(buf, "%s%s", color_reset, color);
348
349 strbuf_addch(buf, ' ');
350 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
351 }
352 strbuf_addf(buf, "%s\n", color_reset);
353
354 fwrite(buf->buf, buf->len, 1, diffopt->file);
355}
356
357static struct userdiff_driver no_func_name = {
358 .funcname = { "$^", 0 }
359};
360
361static struct diff_filespec *get_filespec(const char *name, const char *p)
362{
363 struct diff_filespec *spec = alloc_filespec(name);
364
365 fill_filespec(spec, &null_oid, 0, 0100644);
366 spec->data = (char *)p;
367 spec->size = strlen(p);
368 spec->should_munmap = 0;
369 spec->is_stdin = 1;
370 spec->driver = &no_func_name;
371
372 return spec;
373}
374
375static void patch_diff(const char *a, const char *b,
376 struct diff_options *diffopt)
377{
378 diff_queue(&diff_queued_diff,
379 get_filespec("a", a), get_filespec("b", b));
380
381 diffcore_std(diffopt);
382 diff_flush(diffopt);
383}
384
385static void output(struct string_list *a, struct string_list *b,
386 struct diff_options *diffopt)
387{
388 struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
389 int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
390 int i = 0, j = 0;
391
392 /*
393 * We assume the user is really more interested in the second argument
394 * ("newer" version). To that end, we print the output in the order of
395 * the RHS (the `b` parameter). To put the LHS (the `a` parameter)
396 * commits that are no longer in the RHS into a good place, we place
397 * them once we have shown all of their predecessors in the LHS.
398 */
399
400 while (i < a->nr || j < b->nr) {
401 struct patch_util *a_util, *b_util;
402 a_util = i < a->nr ? a->items[i].util : NULL;
403 b_util = j < b->nr ? b->items[j].util : NULL;
404
405 /* Skip all the already-shown commits from the LHS. */
406 while (i < a->nr && a_util->shown)
407 a_util = ++i < a->nr ? a->items[i].util : NULL;
408
409 /* Show unmatched LHS commit whose predecessors were shown. */
410 if (i < a->nr && a_util->matching < 0) {
411 output_pair_header(diffopt, patch_no_width,
412 &buf, &dashes, a_util, NULL);
413 i++;
414 continue;
415 }
416
417 /* Show unmatched RHS commits. */
418 while (j < b->nr && b_util->matching < 0) {
419 output_pair_header(diffopt, patch_no_width,
420 &buf, &dashes, NULL, b_util);
421 b_util = ++j < b->nr ? b->items[j].util : NULL;
422 }
423
424 /* Show matching LHS/RHS pair. */
425 if (j < b->nr) {
426 a_util = a->items[b_util->matching].util;
427 output_pair_header(diffopt, patch_no_width,
428 &buf, &dashes, a_util, b_util);
429 if (!(diffopt->output_format & DIFF_FORMAT_NO_OUTPUT))
430 patch_diff(a->items[b_util->matching].string,
431 b->items[j].string, diffopt);
432 a_util->shown = 1;
433 j++;
434 }
435 }
436 strbuf_release(&buf);
437 strbuf_release(&dashes);
438}
439
440static struct strbuf *output_prefix_cb(struct diff_options *opt, void *data)
441{
442 return data;
443}
444
445int show_range_diff(const char *range1, const char *range2,
446 int creation_factor, int dual_color,
447 struct diff_options *diffopt)
448{
449 int res = 0;
450
451 struct string_list branch1 = STRING_LIST_INIT_DUP;
452 struct string_list branch2 = STRING_LIST_INIT_DUP;
453
454 if (read_patches(range1, &branch1))
455 res = error(_("could not parse log for '%s'"), range1);
456 if (!res && read_patches(range2, &branch2))
457 res = error(_("could not parse log for '%s'"), range2);
458
459 if (!res) {
460 struct diff_options opts;
461 struct strbuf indent = STRBUF_INIT;
462
463 memcpy(&opts, diffopt, sizeof(opts));
464 if (!opts.output_format)
465 opts.output_format = DIFF_FORMAT_PATCH;
466 opts.flags.suppress_diff_headers = 1;
467 opts.flags.dual_color_diffed_diffs = dual_color;
468 opts.output_prefix = output_prefix_cb;
469 strbuf_addstr(&indent, " ");
470 opts.output_prefix_data = &indent;
471 diff_setup_done(&opts);
472
473 find_exact_matches(&branch1, &branch2);
474 get_correspondences(&branch1, &branch2, creation_factor);
475 output(&branch1, &branch2, &opts);
476
477 strbuf_release(&indent);
478 }
479
480 string_list_clear(&branch1, 1);
481 string_list_clear(&branch2, 1);
482
483 return res;
484}