range-diff.c

   1 #include "cache.h"
   2 #include "environment.h"
   3 #include "gettext.h"
   4 #include "range-diff.h"
   5 #include "string-list.h"
   6 #include "run-command.h"
   7 #include "strvec.h"
   8 #include "hashmap.h"
   9 #include "xdiff-interface.h"
  10 #include "linear-assignment.h"
  11 #include "diffcore.h"
  12 #include "commit.h"
  13 #include "pretty.h"
  14 #include "userdiff.h"
  15 #include "apply.h"
  16 #include "revision.h"
  17
  18 struct patch_util {
  19         /* For the search for an exact match */
  20         struct hashmap_entry e;
  21         const char *diff, *patch;
  22
  23         int i, shown;
  24         int diffsize;
  25         size_t diff_offset;
  26         /* the index of the matching item in the other branch, or -1 */
  27         int matching;
  28         struct object_id oid;
  29 };
  30
  31 /*
  32  * Reads the patches into a string list, with the `util` field being populated
  33  * as struct object_id (will need to be free()d).
  34  */
  35 static int read_patches(const char *range, struct string_list *list,
  36                         const struct strvec *other_arg)
  37 {
  38         struct child_process cp = CHILD_PROCESS_INIT;
  39         struct strbuf buf = STRBUF_INIT, contents = STRBUF_INIT;
  40         struct patch_util *util = NULL;
  41         int in_header = 1;
  42         char *line, *current_filename = NULL;
  43         ssize_t len;
  44         size_t size;
  45         int ret = -1;
  46
  47         strvec_pushl(&cp.args, "log", "--no-color", "-p", "--no-merges",
  48                      "--reverse", "--date-order", "--decorate=no",
  49                      "--no-prefix", "--submodule=short",
  50                      /*
  51                       * Choose indicators that are not used anywhere
  52                       * else in diffs, but still look reasonable
  53                       * (e.g. will not be confusing when debugging)
  54                       */
  55                      "--output-indicator-new=>",
  56                      "--output-indicator-old=<",
  57                      "--output-indicator-context=#",
  58                      "--no-abbrev-commit",
  59                      "--pretty=medium",
  60                      "--notes",
  61                      NULL);
  62         strvec_push(&cp.args, range);
  63         if (other_arg)
  64                 strvec_pushv(&cp.args, other_arg->v);
  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                 goto cleanup;
  75         }
  76         if (finish_command(&cp))
  77                 goto cleanup;
  78
  79         line = contents.buf;
  80         size = contents.len;
  81         for (; size > 0; size -= len, line += len) {
  82                 const char *p;
  83                 char *eol;
  84
  85                 eol = memchr(line, '\n', size);
  86                 if (eol) {
  87                         *eol = '\0';
  88                         len = eol + 1 - line;
  89                 } else {
  90                         len = size;
  91                 }
  92
  93                 if (skip_prefix(line, "commit ", &p)) {
  94                         if (util) {
  95                                 string_list_append(list, buf.buf)->util = util;
  96                                 strbuf_reset(&buf);
  97                         }
  98                         CALLOC_ARRAY(util, 1);
  99                         if (repo_get_oid(the_repository, p, &util->oid)) {
 100                                 error(_("could not parse commit '%s'"), p);
 101                                 FREE_AND_NULL(util);
 102                                 string_list_clear(list, 1);
 103                                 goto cleanup;
 104                         }
 105                         util->matching = -1;
 106                         in_header = 1;
 107                         continue;
 108                 }
 109
 110                 if (!util) {
 111                         error(_("could not parse first line of `log` output: "
 112                                 "did not start with 'commit ': '%s'"),
 113                               line);
 114                         string_list_clear(list, 1);
 115                         goto cleanup;
 116                 }
 117
 118                 if (starts_with(line, "diff --git")) {
 119                         struct patch patch = { 0 };
 120                         struct strbuf root = STRBUF_INIT;
 121                         int linenr = 0;
 122                         int orig_len;
 123
 124                         in_header = 0;
 125                         strbuf_addch(&buf, '\n');
 126                         if (!util->diff_offset)
 127                                 util->diff_offset = buf.len;
 128                         if (eol)
 129                                 *eol = '\n';
 130                         orig_len = len;
 131                         len = parse_git_diff_header(&root, &linenr, 0, line,
 132                                                     len, size, &patch);
 133                         if (len < 0) {
 134                                 error(_("could not parse git header '%.*s'"),
 135                                       orig_len, line);
 136                                 FREE_AND_NULL(util);
 137                                 string_list_clear(list, 1);
 138                                 goto cleanup;
 139                         }
 140                         strbuf_addstr(&buf, " ## ");
 141                         if (patch.is_new > 0)
 142                                 strbuf_addf(&buf, "%s (new)", patch.new_name);
 143                         else if (patch.is_delete > 0)
 144                                 strbuf_addf(&buf, "%s (deleted)", patch.old_name);
 145                         else if (patch.is_rename)
 146                                 strbuf_addf(&buf, "%s => %s", patch.old_name, patch.new_name);
 147                         else
 148                                 strbuf_addstr(&buf, patch.new_name);
 149
 150                         free(current_filename);
 151                         if (patch.is_delete > 0)
 152                                 current_filename = xstrdup(patch.old_name);
 153                         else
 154                                 current_filename = xstrdup(patch.new_name);
 155
 156                         if (patch.new_mode && patch.old_mode &&
 157                             patch.old_mode != patch.new_mode)
 158                                 strbuf_addf(&buf, " (mode change %06o => %06o)",
 159                                             patch.old_mode, patch.new_mode);
 160
 161                         strbuf_addstr(&buf, " ##");
 162                         release_patch(&patch);
 163                 } else if (in_header) {
 164                         if (starts_with(line, "Author: ")) {
 165                                 strbuf_addstr(&buf, " ## Metadata ##\n");
 166                                 strbuf_addstr(&buf, line);
 167                                 strbuf_addstr(&buf, "\n\n");
 168                                 strbuf_addstr(&buf, " ## Commit message ##\n");
 169                         } else if (starts_with(line, "Notes") &&
 170                                    line[strlen(line) - 1] == ':') {
 171                                 strbuf_addstr(&buf, "\n\n");
 172                                 /* strip the trailing colon */
 173                                 strbuf_addf(&buf, " ## %.*s ##\n",
 174                                             (int)(strlen(line) - 1), line);
 175                         } else if (starts_with(line, "    ")) {
 176                                 p = line + len - 2;
 177                                 while (isspace(*p) && p >= line)
 178                                         p--;
 179                                 strbuf_add(&buf, line, p - line + 1);
 180                                 strbuf_addch(&buf, '\n');
 181                         }
 182                         continue;
 183                 } else if (skip_prefix(line, "@@ ", &p)) {
 184                         p = strstr(p, "@@");
 185                         strbuf_addstr(&buf, "@@");
 186                         if (current_filename && p[2])
 187                                 strbuf_addf(&buf, " %s:", current_filename);
 188                         if (p)
 189                                 strbuf_addstr(&buf, p + 2);
 190                 } else if (!line[0])
 191                         /*
 192                          * A completely blank (not ' \n', which is context)
 193                          * line is not valid in a diff.  We skip it
 194                          * silently, because this neatly handles the blank
 195                          * separator line between commits in git-log
 196                          * output.
 197                          */
 198                         continue;
 199                 else if (line[0] == '>') {
 200                         strbuf_addch(&buf, '+');
 201                         strbuf_addstr(&buf, line + 1);
 202                 } else if (line[0] == '<') {
 203                         strbuf_addch(&buf, '-');
 204                         strbuf_addstr(&buf, line + 1);
 205                 } else if (line[0] == '#') {
 206                         strbuf_addch(&buf, ' ');
 207                         strbuf_addstr(&buf, line + 1);
 208                 } else {
 209                         strbuf_addch(&buf, ' ');
 210                         strbuf_addstr(&buf, line);
 211                 }
 212
 213                 strbuf_addch(&buf, '\n');
 214                 util->diffsize++;
 215         }
 216
 217         ret = 0;
 218 cleanup:
 219         strbuf_release(&contents);
 220
 221         if (util)
 222                 string_list_append(list, buf.buf)->util = util;
 223         strbuf_release(&buf);
 224         free(current_filename);
 225
 226         return ret;
 227 }
 228
 229 static int patch_util_cmp(const void *cmp_data UNUSED,
 230                           const struct patch_util *a,
 231                           const struct patch_util *b,
 232                           const char *keydata)
 233 {
 234         return strcmp(a->diff, keydata ? keydata : b->diff);
 235 }
 236
 237 static void find_exact_matches(struct string_list *a, struct string_list *b)
 238 {
 239         struct hashmap map = HASHMAP_INIT((hashmap_cmp_fn)patch_util_cmp, NULL);
 240         int i;
 241
 242         /* First, add the patches of a to a hash map */
 243         for (i = 0; i < a->nr; i++) {
 244                 struct patch_util *util = a->items[i].util;
 245
 246                 util->i = i;
 247                 util->patch = a->items[i].string;
 248                 util->diff = util->patch + util->diff_offset;
 249                 hashmap_entry_init(&util->e, strhash(util->diff));
 250                 hashmap_add(&map, &util->e);
 251         }
 252
 253         /* Now try to find exact matches in b */
 254         for (i = 0; i < b->nr; i++) {
 255                 struct patch_util *util = b->items[i].util, *other;
 256
 257                 util->i = i;
 258                 util->patch = b->items[i].string;
 259                 util->diff = util->patch + util->diff_offset;
 260                 hashmap_entry_init(&util->e, strhash(util->diff));
 261                 other = hashmap_remove_entry(&map, util, e, NULL);
 262                 if (other) {
 263                         if (other->matching >= 0)
 264                                 BUG("already assigned!");
 265
 266                         other->matching = i;
 267                         util->matching = other->i;
 268                 }
 269         }
 270
 271         hashmap_clear(&map);
 272 }
 273
 274 static int diffsize_consume(void *data,
 275                              char *line UNUSED,
 276                              unsigned long len UNUSED)
 277 {
 278         (*(int *)data)++;
 279         return 0;
 280 }
 281
 282 static void diffsize_hunk(void *data,
 283                           long ob UNUSED, long on UNUSED,
 284                           long nb UNUSED, long nn UNUSED,
 285                           const char *func UNUSED, long funclen UNUSED)
 286 {
 287         diffsize_consume(data, NULL, 0);
 288 }
 289
 290 static int diffsize(const char *a, const char *b)
 291 {
 292         xpparam_t pp = { 0 };
 293         xdemitconf_t cfg = { 0 };
 294         mmfile_t mf1, mf2;
 295         int count = 0;
 296
 297         mf1.ptr = (char *)a;
 298         mf1.size = strlen(a);
 299         mf2.ptr = (char *)b;
 300         mf2.size = strlen(b);
 301
 302         cfg.ctxlen = 3;
 303         if (!xdi_diff_outf(&mf1, &mf2,
 304                            diffsize_hunk, diffsize_consume, &count,
 305                            &pp, &cfg))
 306                 return count;
 307
 308         error(_("failed to generate diff"));
 309         return COST_MAX;
 310 }
 311
 312 static void get_correspondences(struct string_list *a, struct string_list *b,
 313                                 int creation_factor)
 314 {
 315         int n = a->nr + b->nr;
 316         int *cost, c, *a2b, *b2a;
 317         int i, j;
 318
 319         ALLOC_ARRAY(cost, st_mult(n, n));
 320         ALLOC_ARRAY(a2b, n);
 321         ALLOC_ARRAY(b2a, n);
 322
 323         for (i = 0; i < a->nr; i++) {
 324                 struct patch_util *a_util = a->items[i].util;
 325
 326                 for (j = 0; j < b->nr; j++) {
 327                         struct patch_util *b_util = b->items[j].util;
 328
 329                         if (a_util->matching == j)
 330                                 c = 0;
 331                         else if (a_util->matching < 0 && b_util->matching < 0)
 332                                 c = diffsize(a_util->diff, b_util->diff);
 333                         else
 334                                 c = COST_MAX;
 335                         cost[i + n * j] = c;
 336                 }
 337
 338                 c = a_util->matching < 0 ?
 339                         a_util->diffsize * creation_factor / 100 : COST_MAX;
 340                 for (j = b->nr; j < n; j++)
 341                         cost[i + n * j] = c;
 342         }
 343
 344         for (j = 0; j < b->nr; j++) {
 345                 struct patch_util *util = b->items[j].util;
 346
 347                 c = util->matching < 0 ?
 348                         util->diffsize * creation_factor / 100 : COST_MAX;
 349                 for (i = a->nr; i < n; i++)
 350                         cost[i + n * j] = c;
 351         }
 352
 353         for (i = a->nr; i < n; i++)
 354                 for (j = b->nr; j < n; j++)
 355                         cost[i + n * j] = 0;
 356
 357         compute_assignment(n, n, cost, a2b, b2a);
 358
 359         for (i = 0; i < a->nr; i++)
 360                 if (a2b[i] >= 0 && a2b[i] < b->nr) {
 361                         struct patch_util *a_util = a->items[i].util;
 362                         struct patch_util *b_util = b->items[a2b[i]].util;
 363
 364                         a_util->matching = a2b[i];
 365                         b_util->matching = i;
 366                 }
 367
 368         free(cost);
 369         free(a2b);
 370         free(b2a);
 371 }
 372
 373 static void output_pair_header(struct diff_options *diffopt,
 374                                int patch_no_width,
 375                                struct strbuf *buf,
 376                                struct strbuf *dashes,
 377                                struct patch_util *a_util,
 378                                struct patch_util *b_util)
 379 {
 380         struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
 381         struct commit *commit;
 382         char status;
 383         const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
 384         const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
 385         const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
 386         const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
 387         const char *color;
 388         int abbrev = diffopt->abbrev;
 389
 390         if (abbrev < 0)
 391                 abbrev = DEFAULT_ABBREV;
 392
 393         if (!dashes->len)
 394                 strbuf_addchars(dashes, '-',
 395                                 strlen(repo_find_unique_abbrev(the_repository, oid, abbrev)));
 396
 397         if (!b_util) {
 398                 color = color_old;
 399                 status = '<';
 400         } else if (!a_util) {
 401                 color = color_new;
 402                 status = '>';
 403         } else if (strcmp(a_util->patch, b_util->patch)) {
 404                 color = color_commit;
 405                 status = '!';
 406         } else {
 407                 color = color_commit;
 408                 status = '=';
 409         }
 410
 411         strbuf_reset(buf);
 412         strbuf_addstr(buf, status == '!' ? color_old : color);
 413         if (!a_util)
 414                 strbuf_addf(buf, "%*s:  %s ", patch_no_width, "-", dashes->buf);
 415         else
 416                 strbuf_addf(buf, "%*d:  %s ", patch_no_width, a_util->i + 1,
 417                             repo_find_unique_abbrev(the_repository, &a_util->oid, abbrev));
 418
 419         if (status == '!')
 420                 strbuf_addf(buf, "%s%s", color_reset, color);
 421         strbuf_addch(buf, status);
 422         if (status == '!')
 423                 strbuf_addf(buf, "%s%s", color_reset, color_new);
 424
 425         if (!b_util)
 426                 strbuf_addf(buf, " %*s:  %s", patch_no_width, "-", dashes->buf);
 427         else
 428                 strbuf_addf(buf, " %*d:  %s", patch_no_width, b_util->i + 1,
 429                             repo_find_unique_abbrev(the_repository, &b_util->oid, abbrev));
 430
 431         commit = lookup_commit_reference(the_repository, oid);
 432         if (commit) {
 433                 if (status == '!')
 434                         strbuf_addf(buf, "%s%s", color_reset, color);
 435
 436                 strbuf_addch(buf, ' ');
 437                 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
 438         }
 439         strbuf_addf(buf, "%s\n", color_reset);
 440
 441         fwrite(buf->buf, buf->len, 1, diffopt->file);
 442 }
 443
 444 static struct userdiff_driver section_headers = {
 445         .funcname = { "^ ## (.*) ##$\n"
 446                       "^.?@@ (.*)$", REG_EXTENDED }
 447 };
 448
 449 static struct diff_filespec *get_filespec(const char *name, const char *p)
 450 {
 451         struct diff_filespec *spec = alloc_filespec(name);
 452
 453         fill_filespec(spec, null_oid(), 0, 0100644);
 454         spec->data = (char *)p;
 455         spec->size = strlen(p);
 456         spec->should_munmap = 0;
 457         spec->is_stdin = 1;
 458         spec->driver = &section_headers;
 459
 460         return spec;
 461 }
 462
 463 static void patch_diff(const char *a, const char *b,
 464                        struct diff_options *diffopt)
 465 {
 466         diff_queue(&diff_queued_diff,
 467                    get_filespec("a", a), get_filespec("b", b));
 468
 469         diffcore_std(diffopt);
 470         diff_flush(diffopt);
 471 }
 472
 473 static struct strbuf *output_prefix_cb(struct diff_options *opt UNUSED, void *data)
 474 {
 475         return data;
 476 }
 477
 478 static void output(struct string_list *a, struct string_list *b,
 479                    struct range_diff_options *range_diff_opts)
 480 {
 481         struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
 482         int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
 483         int i = 0, j = 0;
 484         struct diff_options opts;
 485         struct strbuf indent = STRBUF_INIT;
 486
 487         if (range_diff_opts->diffopt)
 488                 memcpy(&opts, range_diff_opts->diffopt, sizeof(opts));
 489         else
 490                 repo_diff_setup(the_repository, &opts);
 491
 492         opts.no_free = 1;
 493         if (!opts.output_format)
 494                 opts.output_format = DIFF_FORMAT_PATCH;
 495         opts.flags.suppress_diff_headers = 1;
 496         opts.flags.dual_color_diffed_diffs =
 497                 range_diff_opts->dual_color;
 498         opts.flags.suppress_hunk_header_line_count = 1;
 499         opts.output_prefix = output_prefix_cb;
 500         strbuf_addstr(&indent, "    ");
 501         opts.output_prefix_data = &indent;
 502         diff_setup_done(&opts);
 503
 504         /*
 505          * We assume the user is really more interested in the second argument
 506          * ("newer" version). To that end, we print the output in the order of
 507          * the RHS (the `b` parameter). To put the LHS (the `a` parameter)
 508          * commits that are no longer in the RHS into a good place, we place
 509          * them once we have shown all of their predecessors in the LHS.
 510          */
 511
 512         while (i < a->nr || j < b->nr) {
 513                 struct patch_util *a_util, *b_util;
 514                 a_util = i < a->nr ? a->items[i].util : NULL;
 515                 b_util = j < b->nr ? b->items[j].util : NULL;
 516
 517                 /* Skip all the already-shown commits from the LHS. */
 518                 while (i < a->nr && a_util->shown)
 519                         a_util = ++i < a->nr ? a->items[i].util : NULL;
 520
 521                 /* Show unmatched LHS commit whose predecessors were shown. */
 522                 if (i < a->nr && a_util->matching < 0) {
 523                         if (!range_diff_opts->right_only)
 524                                 output_pair_header(&opts, patch_no_width,
 525                                            &buf, &dashes, a_util, NULL);
 526                         i++;
 527                         continue;
 528                 }
 529
 530                 /* Show unmatched RHS commits. */
 531                 while (j < b->nr && b_util->matching < 0) {
 532                         if (!range_diff_opts->left_only)
 533                                 output_pair_header(&opts, patch_no_width,
 534                                            &buf, &dashes, NULL, b_util);
 535                         b_util = ++j < b->nr ? b->items[j].util : NULL;
 536                 }
 537
 538                 /* Show matching LHS/RHS pair. */
 539                 if (j < b->nr) {
 540                         a_util = a->items[b_util->matching].util;
 541                         output_pair_header(&opts, patch_no_width,
 542                                            &buf, &dashes, a_util, b_util);
 543                         if (!(opts.output_format & DIFF_FORMAT_NO_OUTPUT))
 544                                 patch_diff(a->items[b_util->matching].string,
 545                                            b->items[j].string, &opts);
 546                         a_util->shown = 1;
 547                         j++;
 548                 }
 549         }
 550         strbuf_release(&buf);
 551         strbuf_release(&dashes);
 552         strbuf_release(&indent);
 553         opts.no_free = 0;
 554         diff_free(&opts);
 555 }
 556
 557 int show_range_diff(const char *range1, const char *range2,
 558                     struct range_diff_options *range_diff_opts)
 559 {
 560         int res = 0;
 561
 562         struct string_list branch1 = STRING_LIST_INIT_DUP;
 563         struct string_list branch2 = STRING_LIST_INIT_DUP;
 564
 565         if (range_diff_opts->left_only && range_diff_opts->right_only)
 566                 res = error(_("options '%s' and '%s' cannot be used together"), "--left-only", "--right-only");
 567
 568         if (!res && read_patches(range1, &branch1, range_diff_opts->other_arg))
 569                 res = error(_("could not parse log for '%s'"), range1);
 570         if (!res && read_patches(range2, &branch2, range_diff_opts->other_arg))
 571                 res = error(_("could not parse log for '%s'"), range2);
 572
 573         if (!res) {
 574                 find_exact_matches(&branch1, &branch2);
 575                 get_correspondences(&branch1, &branch2,
 576                                     range_diff_opts->creation_factor);
 577                 output(&branch1, &branch2, range_diff_opts);
 578         }
 579
 580         string_list_clear(&branch1, 1);
 581         string_list_clear(&branch2, 1);
 582
 583         return res;
 584 }
 585
 586 int is_range_diff_range(const char *arg)
 587 {
 588         char *copy = xstrdup(arg); /* setup_revisions() modifies it */
 589         const char *argv[] = { "", copy, "--", NULL };
 590         int i, positive = 0, negative = 0;
 591         struct rev_info revs;
 592
 593         repo_init_revisions(the_repository, &revs, NULL);
 594         if (setup_revisions(3, argv, &revs, NULL) == 1) {
 595                 for (i = 0; i < revs.pending.nr; i++)
 596                         if (revs.pending.objects[i].item->flags & UNINTERESTING)
 597                                 negative++;
 598                         else
 599                                 positive++;
 600                 for (i = 0; i < revs.pending.nr; i++) {
 601                         struct object *obj = revs.pending.objects[i].item;
 602
 603                         if (obj->type == OBJ_COMMIT)
 604                                 clear_commit_marks((struct commit *)obj,
 605                                                    ALL_REV_FLAGS);
 606                 }
 607         }
 608
 609         free(copy);
 610         release_revisions(&revs);
 611         return negative > 0 && positive > 0;
 612 }