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