range-diff.c

   1 #include "cache.h"
   2 #include "range-diff.h"
   3 #include "string-list.h"
   4 #include "run-command.h"
   5 #include "strvec.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 #include "revision.h"
  15
  16 struct patch_util {
  17         /* For the search for an exact match */
  18         struct hashmap_entry e;
  19         const char *diff, *patch;
  20
  21         int i, shown;
  22         int diffsize;
  23         size_t diff_offset;
  24         /* the index of the matching item in the other branch, or -1 */
  25         int matching;
  26         struct object_id oid;
  27 };
  28
  29 /*
  30  * Reads the patches into a string list, with the `util` field being populated
  31  * as struct object_id (will need to be free()d).
  32  */
  33 static int read_patches(const char *range, struct string_list *list,
  34                         const struct strvec *other_arg)
  35 {
  36         struct child_process cp = CHILD_PROCESS_INIT;
  37         struct strbuf buf = STRBUF_INIT, contents = STRBUF_INIT;
  38         struct patch_util *util = NULL;
  39         int in_header = 1;
  40         char *line, *current_filename = NULL;
  41         ssize_t len;
  42         size_t size;
  43         int ret = -1;
  44
  45         strvec_pushl(&cp.args, "log", "--no-color", "-p", "--no-merges",
  46                      "--reverse", "--date-order", "--decorate=no",
  47                      "--no-prefix", "--submodule=short",
  48                      /*
  49                       * Choose indicators that are not used anywhere
  50                       * else in diffs, but still look reasonable
  51                       * (e.g. will not be confusing when debugging)
  52                       */
  53                      "--output-indicator-new=>",
  54                      "--output-indicator-old=<",
  55                      "--output-indicator-context=#",
  56                      "--no-abbrev-commit",
  57                      "--pretty=medium",
  58                      "--notes",
  59                      NULL);
  60         strvec_push(&cp.args, range);
  61         if (other_arg)
  62                 strvec_pushv(&cp.args, other_arg->v);
  63         cp.out = -1;
  64         cp.no_stdin = 1;
  65         cp.git_cmd = 1;
  66
  67         if (start_command(&cp))
  68                 return error_errno(_("could not start `log`"));
  69         if (strbuf_read(&contents, cp.out, 0) < 0) {
  70                 error_errno(_("could not read `log` output"));
  71                 finish_command(&cp);
  72                 goto cleanup;
  73         }
  74         if (finish_command(&cp))
  75                 goto cleanup;
  76
  77         line = contents.buf;
  78         size = contents.len;
  79         for (; size > 0; size -= len, line += len) {
  80                 const char *p;
  81                 char *eol;
  82
  83                 eol = memchr(line, '\n', size);
  84                 if (eol) {
  85                         *eol = '\0';
  86                         len = eol + 1 - line;
  87                 } else {
  88                         len = size;
  89                 }
  90
  91                 if (skip_prefix(line, "commit ", &p)) {
  92                         if (util) {
  93                                 string_list_append(list, buf.buf)->util = util;
  94                                 strbuf_reset(&buf);
  95                         }
  96                         CALLOC_ARRAY(util, 1);
  97                         if (get_oid(p, &util->oid)) {
  98                                 error(_("could not parse commit '%s'"), p);
  99                                 FREE_AND_NULL(util);
 100                                 string_list_clear(list, 1);
 101                                 goto cleanup;
 102                         }
 103                         util->matching = -1;
 104                         in_header = 1;
 105                         continue;
 106                 }
 107
 108                 if (!util) {
 109                         error(_("could not parse first line of `log` output: "
 110                                 "did not start with 'commit ': '%s'"),
 111                               line);
 112                         string_list_clear(list, 1);
 113                         goto cleanup;
 114                 }
 115
 116                 if (starts_with(line, "diff --git")) {
 117                         struct patch patch = { 0 };
 118                         struct strbuf root = STRBUF_INIT;
 119                         int linenr = 0;
 120                         int orig_len;
 121
 122                         in_header = 0;
 123                         strbuf_addch(&buf, '\n');
 124                         if (!util->diff_offset)
 125                                 util->diff_offset = buf.len;
 126                         if (eol)
 127                                 *eol = '\n';
 128                         orig_len = len;
 129                         len = parse_git_diff_header(&root, &linenr, 0, line,
 130                                                     len, size, &patch);
 131                         if (len < 0) {
 132                                 error(_("could not parse git header '%.*s'"),
 133                                       orig_len, line);
 134                                 FREE_AND_NULL(util);
 135                                 string_list_clear(list, 1);
 136                                 goto cleanup;
 137                         }
 138                         strbuf_addstr(&buf, " ## ");
 139                         if (patch.is_new > 0)
 140                                 strbuf_addf(&buf, "%s (new)", patch.new_name);
 141                         else if (patch.is_delete > 0)
 142                                 strbuf_addf(&buf, "%s (deleted)", patch.old_name);
 143                         else if (patch.is_rename)
 144                                 strbuf_addf(&buf, "%s => %s", patch.old_name, patch.new_name);
 145                         else
 146                                 strbuf_addstr(&buf, patch.new_name);
 147
 148                         free(current_filename);
 149                         if (patch.is_delete > 0)
 150                                 current_filename = xstrdup(patch.old_name);
 151                         else
 152                                 current_filename = xstrdup(patch.new_name);
 153
 154                         if (patch.new_mode && patch.old_mode &&
 155                             patch.old_mode != patch.new_mode)
 156                                 strbuf_addf(&buf, " (mode change %06o => %06o)",
 157                                             patch.old_mode, patch.new_mode);
 158
 159                         strbuf_addstr(&buf, " ##");
 160                         release_patch(&patch);
 161                 } else if (in_header) {
 162                         if (starts_with(line, "Author: ")) {
 163                                 strbuf_addstr(&buf, " ## Metadata ##\n");
 164                                 strbuf_addstr(&buf, line);
 165                                 strbuf_addstr(&buf, "\n\n");
 166                                 strbuf_addstr(&buf, " ## Commit message ##\n");
 167                         } else if (starts_with(line, "Notes") &&
 168                                    line[strlen(line) - 1] == ':') {
 169                                 strbuf_addstr(&buf, "\n\n");
 170                                 /* strip the trailing colon */
 171                                 strbuf_addf(&buf, " ## %.*s ##\n",
 172                                             (int)(strlen(line) - 1), line);
 173                         } else if (starts_with(line, "    ")) {
 174                                 p = line + len - 2;
 175                                 while (isspace(*p) && p >= line)
 176                                         p--;
 177                                 strbuf_add(&buf, line, p - line + 1);
 178                                 strbuf_addch(&buf, '\n');
 179                         }
 180                         continue;
 181                 } else if (skip_prefix(line, "@@ ", &p)) {
 182                         p = strstr(p, "@@");
 183                         strbuf_addstr(&buf, "@@");
 184                         if (current_filename && p[2])
 185                                 strbuf_addf(&buf, " %s:", current_filename);
 186                         if (p)
 187                                 strbuf_addstr(&buf, p + 2);
 188                 } else if (!line[0])
 189                         /*
 190                          * A completely blank (not ' \n', which is context)
 191                          * line is not valid in a diff.  We skip it
 192                          * silently, because this neatly handles the blank
 193                          * separator line between commits in git-log
 194                          * output.
 195                          */
 196                         continue;
 197                 else if (line[0] == '>') {
 198                         strbuf_addch(&buf, '+');
 199                         strbuf_addstr(&buf, line + 1);
 200                 } else if (line[0] == '<') {
 201                         strbuf_addch(&buf, '-');
 202                         strbuf_addstr(&buf, line + 1);
 203                 } else if (line[0] == '#') {
 204                         strbuf_addch(&buf, ' ');
 205                         strbuf_addstr(&buf, line + 1);
 206                 } else {
 207                         strbuf_addch(&buf, ' ');
 208                         strbuf_addstr(&buf, line);
 209                 }
 210
 211                 strbuf_addch(&buf, '\n');
 212                 util->diffsize++;
 213         }
 214
 215         ret = 0;
 216 cleanup:
 217         strbuf_release(&contents);
 218
 219         if (util)
 220                 string_list_append(list, buf.buf)->util = util;
 221         strbuf_release(&buf);
 222         free(current_filename);
 223
 224         return ret;
 225 }
 226
 227 static int patch_util_cmp(const void *cmp_data UNUSED,
 228                           const struct patch_util *a,
 229                           const struct patch_util *b,
 230                           const char *keydata)
 231 {
 232         return strcmp(a->diff, keydata ? keydata : b->diff);
 233 }
 234
 235 static void find_exact_matches(struct string_list *a, struct string_list *b)
 236 {
 237         struct hashmap map = HASHMAP_INIT((hashmap_cmp_fn)patch_util_cmp, NULL);
 238         int i;
 239
 240         /* First, add the patches of a to a hash map */
 241         for (i = 0; i < a->nr; i++) {
 242                 struct patch_util *util = a->items[i].util;
 243
 244                 util->i = i;
 245                 util->patch = a->items[i].string;
 246                 util->diff = util->patch + util->diff_offset;
 247                 hashmap_entry_init(&util->e, strhash(util->diff));
 248                 hashmap_add(&map, &util->e);
 249         }
 250
 251         /* Now try to find exact matches in b */
 252         for (i = 0; i < b->nr; i++) {
 253                 struct patch_util *util = b->items[i].util, *other;
 254
 255                 util->i = i;
 256                 util->patch = b->items[i].string;
 257                 util->diff = util->patch + util->diff_offset;
 258                 hashmap_entry_init(&util->e, strhash(util->diff));
 259                 other = hashmap_remove_entry(&map, util, e, NULL);
 260                 if (other) {
 261                         if (other->matching >= 0)
 262                                 BUG("already assigned!");
 263
 264                         other->matching = i;
 265                         util->matching = other->i;
 266                 }
 267         }
 268
 269         hashmap_clear(&map);
 270 }
 271
 272 static int diffsize_consume(void *data,
 273                              char *line UNUSED,
 274                              unsigned long len UNUSED)
 275 {
 276         (*(int *)data)++;
 277         return 0;
 278 }
 279
 280 static void diffsize_hunk(void *data,
 281                           long ob UNUSED, long on UNUSED,
 282                           long nb UNUSED, long nn UNUSED,
 283                           const char *func UNUSED, long funclen UNUSED)
 284 {
 285         diffsize_consume(data, NULL, 0);
 286 }
 287
 288 static int diffsize(const char *a, const char *b)
 289 {
 290         xpparam_t pp = { 0 };
 291         xdemitconf_t cfg = { 0 };
 292         mmfile_t mf1, mf2;
 293         int count = 0;
 294
 295         mf1.ptr = (char *)a;
 296         mf1.size = strlen(a);
 297         mf2.ptr = (char *)b;
 298         mf2.size = strlen(b);
 299
 300         cfg.ctxlen = 3;
 301         if (!xdi_diff_outf(&mf1, &mf2,
 302                            diffsize_hunk, diffsize_consume, &count,
 303                            &pp, &cfg))
 304                 return count;
 305
 306         error(_("failed to generate diff"));
 307         return COST_MAX;
 308 }
 309
 310 static void get_correspondences(struct string_list *a, struct string_list *b,
 311                                 int creation_factor)
 312 {
 313         int n = a->nr + b->nr;
 314         int *cost, c, *a2b, *b2a;
 315         int i, j;
 316
 317         ALLOC_ARRAY(cost, st_mult(n, n));
 318         ALLOC_ARRAY(a2b, n);
 319         ALLOC_ARRAY(b2a, n);
 320
 321         for (i = 0; i < a->nr; i++) {
 322                 struct patch_util *a_util = a->items[i].util;
 323
 324                 for (j = 0; j < b->nr; j++) {
 325                         struct patch_util *b_util = b->items[j].util;
 326
 327                         if (a_util->matching == j)
 328                                 c = 0;
 329                         else if (a_util->matching < 0 && b_util->matching < 0)
 330                                 c = diffsize(a_util->diff, b_util->diff);
 331                         else
 332                                 c = COST_MAX;
 333                         cost[i + n * j] = c;
 334                 }
 335
 336                 c = a_util->matching < 0 ?
 337                         a_util->diffsize * creation_factor / 100 : COST_MAX;
 338                 for (j = b->nr; j < n; j++)
 339                         cost[i + n * j] = c;
 340         }
 341
 342         for (j = 0; j < b->nr; j++) {
 343                 struct patch_util *util = b->items[j].util;
 344
 345                 c = util->matching < 0 ?
 346                         util->diffsize * creation_factor / 100 : COST_MAX;
 347                 for (i = a->nr; i < n; i++)
 348                         cost[i + n * j] = c;
 349         }
 350
 351         for (i = a->nr; i < n; i++)
 352                 for (j = b->nr; j < n; j++)
 353                         cost[i + n * j] = 0;
 354
 355         compute_assignment(n, n, cost, a2b, b2a);
 356
 357         for (i = 0; i < a->nr; i++)
 358                 if (a2b[i] >= 0 && a2b[i] < b->nr) {
 359                         struct patch_util *a_util = a->items[i].util;
 360                         struct patch_util *b_util = b->items[a2b[i]].util;
 361
 362                         a_util->matching = a2b[i];
 363                         b_util->matching = i;
 364                 }
 365
 366         free(cost);
 367         free(a2b);
 368         free(b2a);
 369 }
 370
 371 static void output_pair_header(struct diff_options *diffopt,
 372                                int patch_no_width,
 373                                struct strbuf *buf,
 374                                struct strbuf *dashes,
 375                                struct patch_util *a_util,
 376                                struct patch_util *b_util)
 377 {
 378         struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
 379         struct commit *commit;
 380         char status;
 381         const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
 382         const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
 383         const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
 384         const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
 385         const char *color;
 386         int abbrev = diffopt->abbrev;
 387
 388         if (abbrev < 0)
 389                 abbrev = DEFAULT_ABBREV;
 390
 391         if (!dashes->len)
 392                 strbuf_addchars(dashes, '-',
 393                                 strlen(find_unique_abbrev(oid, abbrev)));
 394
 395         if (!b_util) {
 396                 color = color_old;
 397                 status = '<';
 398         } else if (!a_util) {
 399                 color = color_new;
 400                 status = '>';
 401         } else if (strcmp(a_util->patch, b_util->patch)) {
 402                 color = color_commit;
 403                 status = '!';
 404         } else {
 405                 color = color_commit;
 406                 status = '=';
 407         }
 408
 409         strbuf_reset(buf);
 410         strbuf_addstr(buf, status == '!' ? color_old : color);
 411         if (!a_util)
 412                 strbuf_addf(buf, "%*s:  %s ", patch_no_width, "-", dashes->buf);
 413         else
 414                 strbuf_addf(buf, "%*d:  %s ", patch_no_width, a_util->i + 1,
 415                             find_unique_abbrev(&a_util->oid, abbrev));
 416
 417         if (status == '!')
 418                 strbuf_addf(buf, "%s%s", color_reset, color);
 419         strbuf_addch(buf, status);
 420         if (status == '!')
 421                 strbuf_addf(buf, "%s%s", color_reset, color_new);
 422
 423         if (!b_util)
 424                 strbuf_addf(buf, " %*s:  %s", patch_no_width, "-", dashes->buf);
 425         else
 426                 strbuf_addf(buf, " %*d:  %s", patch_no_width, b_util->i + 1,
 427                             find_unique_abbrev(&b_util->oid, abbrev));
 428
 429         commit = lookup_commit_reference(the_repository, oid);
 430         if (commit) {
 431                 if (status == '!')
 432                         strbuf_addf(buf, "%s%s", color_reset, color);
 433
 434                 strbuf_addch(buf, ' ');
 435                 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
 436         }
 437         strbuf_addf(buf, "%s\n", color_reset);
 438
 439         fwrite(buf->buf, buf->len, 1, diffopt->file);
 440 }
 441
 442 static struct userdiff_driver section_headers = {
 443         .funcname = { "^ ## (.*) ##$\n"
 444                       "^.?@@ (.*)$", REG_EXTENDED }
 445 };
 446
 447 static struct diff_filespec *get_filespec(const char *name, const char *p)
 448 {
 449         struct diff_filespec *spec = alloc_filespec(name);
 450
 451         fill_filespec(spec, null_oid(), 0, 0100644);
 452         spec->data = (char *)p;
 453         spec->size = strlen(p);
 454         spec->should_munmap = 0;
 455         spec->is_stdin = 1;
 456         spec->driver = &section_headers;
 457
 458         return spec;
 459 }
 460
 461 static void patch_diff(const char *a, const char *b,
 462                        struct diff_options *diffopt)
 463 {
 464         diff_queue(&diff_queued_diff,
 465                    get_filespec("a", a), get_filespec("b", b));
 466
 467         diffcore_std(diffopt);
 468         diff_flush(diffopt);
 469 }
 470
 471 static struct strbuf *output_prefix_cb(struct diff_options *opt UNUSED, void *data)
 472 {
 473         return data;
 474 }
 475
 476 static void output(struct string_list *a, struct string_list *b,
 477                    struct range_diff_options *range_diff_opts)
 478 {
 479         struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
 480         int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
 481         int i = 0, j = 0;
 482         struct diff_options opts;
 483         struct strbuf indent = STRBUF_INIT;
 484
 485         if (range_diff_opts->diffopt)
 486                 memcpy(&opts, range_diff_opts->diffopt, sizeof(opts));
 487         else
 488                 diff_setup(&opts);
 489
 490         opts.no_free = 1;
 491         if (!opts.output_format)
 492                 opts.output_format = DIFF_FORMAT_PATCH;
 493         opts.flags.suppress_diff_headers = 1;
 494         opts.flags.dual_color_diffed_diffs =
 495                 range_diff_opts->dual_color;
 496         opts.flags.suppress_hunk_header_line_count = 1;
 497         opts.output_prefix = output_prefix_cb;
 498         strbuf_addstr(&indent, "    ");
 499         opts.output_prefix_data = &indent;
 500         diff_setup_done(&opts);
 501
 502         /*
 503          * We assume the user is really more interested in the second argument
 504          * ("newer" version). To that end, we print the output in the order of
 505          * the RHS (the `b` parameter). To put the LHS (the `a` parameter)
 506          * commits that are no longer in the RHS into a good place, we place
 507          * them once we have shown all of their predecessors in the LHS.
 508          */
 509
 510         while (i < a->nr || j < b->nr) {
 511                 struct patch_util *a_util, *b_util;
 512                 a_util = i < a->nr ? a->items[i].util : NULL;
 513                 b_util = j < b->nr ? b->items[j].util : NULL;
 514
 515                 /* Skip all the already-shown commits from the LHS. */
 516                 while (i < a->nr && a_util->shown)
 517                         a_util = ++i < a->nr ? a->items[i].util : NULL;
 518
 519                 /* Show unmatched LHS commit whose predecessors were shown. */
 520                 if (i < a->nr && a_util->matching < 0) {
 521                         if (!range_diff_opts->right_only)
 522                                 output_pair_header(&opts, patch_no_width,
 523                                            &buf, &dashes, a_util, NULL);
 524                         i++;
 525                         continue;
 526                 }
 527
 528                 /* Show unmatched RHS commits. */
 529                 while (j < b->nr && b_util->matching < 0) {
 530                         if (!range_diff_opts->left_only)
 531                                 output_pair_header(&opts, patch_no_width,
 532                                            &buf, &dashes, NULL, b_util);
 533                         b_util = ++j < b->nr ? b->items[j].util : NULL;
 534                 }
 535
 536                 /* Show matching LHS/RHS pair. */
 537                 if (j < b->nr) {
 538                         a_util = a->items[b_util->matching].util;
 539                         output_pair_header(&opts, patch_no_width,
 540                                            &buf, &dashes, a_util, b_util);
 541                         if (!(opts.output_format & DIFF_FORMAT_NO_OUTPUT))
 542                                 patch_diff(a->items[b_util->matching].string,
 543                                            b->items[j].string, &opts);
 544                         a_util->shown = 1;
 545                         j++;
 546                 }
 547         }
 548         strbuf_release(&buf);
 549         strbuf_release(&dashes);
 550         strbuf_release(&indent);
 551         opts.no_free = 0;
 552         diff_free(&opts);
 553 }
 554
 555 int show_range_diff(const char *range1, const char *range2,
 556                     struct range_diff_options *range_diff_opts)
 557 {
 558         int res = 0;
 559
 560         struct string_list branch1 = STRING_LIST_INIT_DUP;
 561         struct string_list branch2 = STRING_LIST_INIT_DUP;
 562
 563         if (range_diff_opts->left_only && range_diff_opts->right_only)
 564                 res = error(_("options '%s' and '%s' cannot be used together"), "--left-only", "--right-only");
 565
 566         if (!res && read_patches(range1, &branch1, range_diff_opts->other_arg))
 567                 res = error(_("could not parse log for '%s'"), range1);
 568         if (!res && read_patches(range2, &branch2, range_diff_opts->other_arg))
 569                 res = error(_("could not parse log for '%s'"), range2);
 570
 571         if (!res) {
 572                 find_exact_matches(&branch1, &branch2);
 573                 get_correspondences(&branch1, &branch2,
 574                                     range_diff_opts->creation_factor);
 575                 output(&branch1, &branch2, range_diff_opts);
 576         }
 577
 578         string_list_clear(&branch1, 1);
 579         string_list_clear(&branch2, 1);
 580
 581         return res;
 582 }
 583
 584 int is_range_diff_range(const char *arg)
 585 {
 586         char *copy = xstrdup(arg); /* setup_revisions() modifies it */
 587         const char *argv[] = { "", copy, "--", NULL };
 588         int i, positive = 0, negative = 0;
 589         struct rev_info revs;
 590
 591         init_revisions(&revs, NULL);
 592         if (setup_revisions(3, argv, &revs, NULL) == 1) {
 593                 for (i = 0; i < revs.pending.nr; i++)
 594                         if (revs.pending.objects[i].item->flags & UNINTERESTING)
 595                                 negative++;
 596                         else
 597                                 positive++;
 598                 for (i = 0; i < revs.pending.nr; i++) {
 599                         struct object *obj = revs.pending.objects[i].item;
 600
 601                         if (obj->type == OBJ_COMMIT)
 602                                 clear_commit_marks((struct commit *)obj,
 603                                                    ALL_REV_FLAGS);
 604                 }
 605         }
 606
 607         free(copy);
 608         release_revisions(&revs);
 609         return negative > 0 && positive > 0;
 610 }