range-diff.c

   1 #include "git-compat-util.h"
   2 #include "environment.h"
   3 #include "gettext.h"
   4 #include "range-diff.h"
   5 #include "object-name.h"
   6 #include "string-list.h"
   7 #include "run-command.h"
   8 #include "strvec.h"
   9 #include "hashmap.h"
  10 #include "xdiff-interface.h"
  11 #include "linear-assignment.h"
  12 #include "diffcore.h"
  13 #include "commit.h"
  14 #include "pager.h"
  15 #include "pretty.h"
  16 #include "userdiff.h"
  17 #include "apply.h"
  18 #include "revision.h"
  19
  20 struct patch_util {
  21         /* For the search for an exact match */
  22         struct hashmap_entry e;
  23         const char *diff, *patch;
  24
  25         int i, shown;
  26         int diffsize;
  27         size_t diff_offset;
  28         /* the index of the matching item in the other branch, or -1 */
  29         int matching;
  30         struct object_id oid;
  31 };
  32
  33 /*
  34  * Reads the patches into a string list, with the `util` field being populated
  35  * as struct object_id (will need to be free()d).
  36  */
  37 static int read_patches(const char *range, struct string_list *list,
  38                         const struct strvec *other_arg)
  39 {
  40         struct child_process cp = CHILD_PROCESS_INIT;
  41         struct strbuf buf = STRBUF_INIT, contents = STRBUF_INIT;
  42         struct patch_util *util = NULL;
  43         int in_header = 1;
  44         char *line, *current_filename = NULL;
  45         ssize_t len;
  46         size_t size;
  47         int ret = -1;
  48
  49         strvec_pushl(&cp.args, "log", "--no-color", "-p", "--no-merges",
  50                      "--reverse", "--date-order", "--decorate=no",
  51                      "--no-prefix", "--submodule=short",
  52                      /*
  53                       * Choose indicators that are not used anywhere
  54                       * else in diffs, but still look reasonable
  55                       * (e.g. will not be confusing when debugging)
  56                       */
  57                      "--output-indicator-new=>",
  58                      "--output-indicator-old=<",
  59                      "--output-indicator-context=#",
  60                      "--no-abbrev-commit",
  61                      "--pretty=medium",
  62                      "--notes",
  63                      NULL);
  64         strvec_push(&cp.args, range);
  65         if (other_arg)
  66                 strvec_pushv(&cp.args, other_arg->v);
  67         cp.out = -1;
  68         cp.no_stdin = 1;
  69         cp.git_cmd = 1;
  70
  71         if (start_command(&cp))
  72                 return error_errno(_("could not start `log`"));
  73         if (strbuf_read(&contents, cp.out, 0) < 0) {
  74                 error_errno(_("could not read `log` output"));
  75                 finish_command(&cp);
  76                 goto cleanup;
  77         }
  78         if (finish_command(&cp))
  79                 goto cleanup;
  80
  81         line = contents.buf;
  82         size = contents.len;
  83         for (; size > 0; size -= len, line += len) {
  84                 const char *p;
  85                 char *eol;
  86
  87                 eol = memchr(line, '\n', size);
  88                 if (eol) {
  89                         *eol = '\0';
  90                         len = eol + 1 - line;
  91                 } else {
  92                         len = size;
  93                 }
  94
  95                 if (skip_prefix(line, "commit ", &p)) {
  96                         if (util) {
  97                                 string_list_append(list, buf.buf)->util = util;
  98                                 strbuf_reset(&buf);
  99                         }
 100                         CALLOC_ARRAY(util, 1);
 101                         if (repo_get_oid(the_repository, p, &util->oid)) {
 102                                 error(_("could not parse commit '%s'"), p);
 103                                 FREE_AND_NULL(util);
 104                                 string_list_clear(list, 1);
 105                                 goto cleanup;
 106                         }
 107                         util->matching = -1;
 108                         in_header = 1;
 109                         continue;
 110                 }
 111
 112                 if (!util) {
 113                         error(_("could not parse first line of `log` output: "
 114                                 "did not start with 'commit ': '%s'"),
 115                               line);
 116                         string_list_clear(list, 1);
 117                         goto cleanup;
 118                 }
 119
 120                 if (starts_with(line, "diff --git")) {
 121                         struct patch patch = { 0 };
 122                         struct strbuf root = STRBUF_INIT;
 123                         int linenr = 0;
 124                         int orig_len;
 125
 126                         in_header = 0;
 127                         strbuf_addch(&buf, '\n');
 128                         if (!util->diff_offset)
 129                                 util->diff_offset = buf.len;
 130                         if (eol)
 131                                 *eol = '\n';
 132                         orig_len = len;
 133                         len = parse_git_diff_header(&root, &linenr, 0, line,
 134                                                     len, size, &patch);
 135                         if (len < 0) {
 136                                 error(_("could not parse git header '%.*s'"),
 137                                       orig_len, line);
 138                                 FREE_AND_NULL(util);
 139                                 string_list_clear(list, 1);
 140                                 goto cleanup;
 141                         }
 142                         strbuf_addstr(&buf, " ## ");
 143                         if (patch.is_new > 0)
 144                                 strbuf_addf(&buf, "%s (new)", patch.new_name);
 145                         else if (patch.is_delete > 0)
 146                                 strbuf_addf(&buf, "%s (deleted)", patch.old_name);
 147                         else if (patch.is_rename)
 148                                 strbuf_addf(&buf, "%s => %s", patch.old_name, patch.new_name);
 149                         else
 150                                 strbuf_addstr(&buf, patch.new_name);
 151
 152                         free(current_filename);
 153                         if (patch.is_delete > 0)
 154                                 current_filename = xstrdup(patch.old_name);
 155                         else
 156                                 current_filename = xstrdup(patch.new_name);
 157
 158                         if (patch.new_mode && patch.old_mode &&
 159                             patch.old_mode != patch.new_mode)
 160                                 strbuf_addf(&buf, " (mode change %06o => %06o)",
 161                                             patch.old_mode, patch.new_mode);
 162
 163                         strbuf_addstr(&buf, " ##");
 164                         release_patch(&patch);
 165                 } else if (in_header) {
 166                         if (starts_with(line, "Author: ")) {
 167                                 strbuf_addstr(&buf, " ## Metadata ##\n");
 168                                 strbuf_addstr(&buf, line);
 169                                 strbuf_addstr(&buf, "\n\n");
 170                                 strbuf_addstr(&buf, " ## Commit message ##\n");
 171                         } else if (starts_with(line, "Notes") &&
 172                                    line[strlen(line) - 1] == ':') {
 173                                 strbuf_addstr(&buf, "\n\n");
 174                                 /* strip the trailing colon */
 175                                 strbuf_addf(&buf, " ## %.*s ##\n",
 176                                             (int)(strlen(line) - 1), line);
 177                         } else if (starts_with(line, "    ")) {
 178                                 p = line + len - 2;
 179                                 while (isspace(*p) && p >= line)
 180                                         p--;
 181                                 strbuf_add(&buf, line, p - line + 1);
 182                                 strbuf_addch(&buf, '\n');
 183                         }
 184                         continue;
 185                 } else if (skip_prefix(line, "@@ ", &p)) {
 186                         p = strstr(p, "@@");
 187                         strbuf_addstr(&buf, "@@");
 188                         if (current_filename && p[2])
 189                                 strbuf_addf(&buf, " %s:", current_filename);
 190                         if (p)
 191                                 strbuf_addstr(&buf, p + 2);
 192                 } else if (!line[0])
 193                         /*
 194                          * A completely blank (not ' \n', which is context)
 195                          * line is not valid in a diff.  We skip it
 196                          * silently, because this neatly handles the blank
 197                          * separator line between commits in git-log
 198                          * output.
 199                          */
 200                         continue;
 201                 else if (line[0] == '>') {
 202                         strbuf_addch(&buf, '+');
 203                         strbuf_addstr(&buf, line + 1);
 204                 } else if (line[0] == '<') {
 205                         strbuf_addch(&buf, '-');
 206                         strbuf_addstr(&buf, line + 1);
 207                 } else if (line[0] == '#') {
 208                         strbuf_addch(&buf, ' ');
 209                         strbuf_addstr(&buf, line + 1);
 210                 } else {
 211                         strbuf_addch(&buf, ' ');
 212                         strbuf_addstr(&buf, line);
 213                 }
 214
 215                 strbuf_addch(&buf, '\n');
 216                 util->diffsize++;
 217         }
 218
 219         ret = 0;
 220 cleanup:
 221         strbuf_release(&contents);
 222
 223         if (util)
 224                 string_list_append(list, buf.buf)->util = util;
 225         strbuf_release(&buf);
 226         free(current_filename);
 227
 228         return ret;
 229 }
 230
 231 static int patch_util_cmp(const void *cmp_data UNUSED,
 232                           const struct patch_util *a,
 233                           const struct patch_util *b,
 234                           const char *keydata)
 235 {
 236         return strcmp(a->diff, keydata ? keydata : b->diff);
 237 }
 238
 239 static void find_exact_matches(struct string_list *a, struct string_list *b)
 240 {
 241         struct hashmap map = HASHMAP_INIT((hashmap_cmp_fn)patch_util_cmp, NULL);
 242         int i;
 243
 244         /* First, add the patches of a to a hash map */
 245         for (i = 0; i < a->nr; i++) {
 246                 struct patch_util *util = a->items[i].util;
 247
 248                 util->i = i;
 249                 util->patch = a->items[i].string;
 250                 util->diff = util->patch + util->diff_offset;
 251                 hashmap_entry_init(&util->e, strhash(util->diff));
 252                 hashmap_add(&map, &util->e);
 253         }
 254
 255         /* Now try to find exact matches in b */
 256         for (i = 0; i < b->nr; i++) {
 257                 struct patch_util *util = b->items[i].util, *other;
 258
 259                 util->i = i;
 260                 util->patch = b->items[i].string;
 261                 util->diff = util->patch + util->diff_offset;
 262                 hashmap_entry_init(&util->e, strhash(util->diff));
 263                 other = hashmap_remove_entry(&map, util, e, NULL);
 264                 if (other) {
 265                         if (other->matching >= 0)
 266                                 BUG("already assigned!");
 267
 268                         other->matching = i;
 269                         util->matching = other->i;
 270                 }
 271         }
 272
 273         hashmap_clear(&map);
 274 }
 275
 276 static int diffsize_consume(void *data,
 277                              char *line UNUSED,
 278                              unsigned long len UNUSED)
 279 {
 280         (*(int *)data)++;
 281         return 0;
 282 }
 283
 284 static void diffsize_hunk(void *data,
 285                           long ob UNUSED, long on UNUSED,
 286                           long nb UNUSED, long nn UNUSED,
 287                           const char *func UNUSED, long funclen UNUSED)
 288 {
 289         diffsize_consume(data, NULL, 0);
 290 }
 291
 292 static int diffsize(const char *a, const char *b)
 293 {
 294         xpparam_t pp = { 0 };
 295         xdemitconf_t cfg = { 0 };
 296         mmfile_t mf1, mf2;
 297         int count = 0;
 298
 299         mf1.ptr = (char *)a;
 300         mf1.size = strlen(a);
 301         mf2.ptr = (char *)b;
 302         mf2.size = strlen(b);
 303
 304         cfg.ctxlen = 3;
 305         if (!xdi_diff_outf(&mf1, &mf2,
 306                            diffsize_hunk, diffsize_consume, &count,
 307                            &pp, &cfg))
 308                 return count;
 309
 310         error(_("failed to generate diff"));
 311         return COST_MAX;
 312 }
 313
 314 static void get_correspondences(struct string_list *a, struct string_list *b,
 315                                 int creation_factor)
 316 {
 317         int n = a->nr + b->nr;
 318         int *cost, c, *a2b, *b2a;
 319         int i, j;
 320
 321         ALLOC_ARRAY(cost, st_mult(n, n));
 322         ALLOC_ARRAY(a2b, n);
 323         ALLOC_ARRAY(b2a, n);
 324
 325         for (i = 0; i < a->nr; i++) {
 326                 struct patch_util *a_util = a->items[i].util;
 327
 328                 for (j = 0; j < b->nr; j++) {
 329                         struct patch_util *b_util = b->items[j].util;
 330
 331                         if (a_util->matching == j)
 332                                 c = 0;
 333                         else if (a_util->matching < 0 && b_util->matching < 0)
 334                                 c = diffsize(a_util->diff, b_util->diff);
 335                         else
 336                                 c = COST_MAX;
 337                         cost[i + n * j] = c;
 338                 }
 339
 340                 c = a_util->matching < 0 ?
 341                         a_util->diffsize * creation_factor / 100 : COST_MAX;
 342                 for (j = b->nr; j < n; j++)
 343                         cost[i + n * j] = c;
 344         }
 345
 346         for (j = 0; j < b->nr; j++) {
 347                 struct patch_util *util = b->items[j].util;
 348
 349                 c = util->matching < 0 ?
 350                         util->diffsize * creation_factor / 100 : COST_MAX;
 351                 for (i = a->nr; i < n; i++)
 352                         cost[i + n * j] = c;
 353         }
 354
 355         for (i = a->nr; i < n; i++)
 356                 for (j = b->nr; j < n; j++)
 357                         cost[i + n * j] = 0;
 358
 359         compute_assignment(n, n, cost, a2b, b2a);
 360
 361         for (i = 0; i < a->nr; i++)
 362                 if (a2b[i] >= 0 && a2b[i] < b->nr) {
 363                         struct patch_util *a_util = a->items[i].util;
 364                         struct patch_util *b_util = b->items[a2b[i]].util;
 365
 366                         a_util->matching = a2b[i];
 367                         b_util->matching = i;
 368                 }
 369
 370         free(cost);
 371         free(a2b);
 372         free(b2a);
 373 }
 374
 375 static void output_pair_header(struct diff_options *diffopt,
 376                                int patch_no_width,
 377                                struct strbuf *buf,
 378                                struct strbuf *dashes,
 379                                struct patch_util *a_util,
 380                                struct patch_util *b_util)
 381 {
 382         struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
 383         struct commit *commit;
 384         char status;
 385         const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
 386         const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
 387         const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
 388         const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
 389         const char *color;
 390         int abbrev = diffopt->abbrev;
 391
 392         if (abbrev < 0)
 393                 abbrev = DEFAULT_ABBREV;
 394
 395         if (!dashes->len)
 396                 strbuf_addchars(dashes, '-',
 397                                 strlen(repo_find_unique_abbrev(the_repository, oid, abbrev)));
 398
 399         if (!b_util) {
 400                 color = color_old;
 401                 status = '<';
 402         } else if (!a_util) {
 403                 color = color_new;
 404                 status = '>';
 405         } else if (strcmp(a_util->patch, b_util->patch)) {
 406                 color = color_commit;
 407                 status = '!';
 408         } else {
 409                 color = color_commit;
 410                 status = '=';
 411         }
 412
 413         strbuf_reset(buf);
 414         strbuf_addstr(buf, status == '!' ? color_old : color);
 415         if (!a_util)
 416                 strbuf_addf(buf, "%*s:  %s ", patch_no_width, "-", dashes->buf);
 417         else
 418                 strbuf_addf(buf, "%*d:  %s ", patch_no_width, a_util->i + 1,
 419                             repo_find_unique_abbrev(the_repository, &a_util->oid, abbrev));
 420
 421         if (status == '!')
 422                 strbuf_addf(buf, "%s%s", color_reset, color);
 423         strbuf_addch(buf, status);
 424         if (status == '!')
 425                 strbuf_addf(buf, "%s%s", color_reset, color_new);
 426
 427         if (!b_util)
 428                 strbuf_addf(buf, " %*s:  %s", patch_no_width, "-", dashes->buf);
 429         else
 430                 strbuf_addf(buf, " %*d:  %s", patch_no_width, b_util->i + 1,
 431                             repo_find_unique_abbrev(the_repository, &b_util->oid, abbrev));
 432
 433         commit = lookup_commit_reference(the_repository, oid);
 434         if (commit) {
 435                 if (status == '!')
 436                         strbuf_addf(buf, "%s%s", color_reset, color);
 437
 438                 strbuf_addch(buf, ' ');
 439                 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
 440         }
 441         strbuf_addf(buf, "%s\n", color_reset);
 442
 443         fwrite(buf->buf, buf->len, 1, diffopt->file);
 444 }
 445
 446 static struct userdiff_driver section_headers = {
 447         .funcname = { "^ ## (.*) ##$\n"
 448                       "^.?@@ (.*)$", REG_EXTENDED }
 449 };
 450
 451 static struct diff_filespec *get_filespec(const char *name, const char *p)
 452 {
 453         struct diff_filespec *spec = alloc_filespec(name);
 454
 455         fill_filespec(spec, null_oid(), 0, 0100644);
 456         spec->data = (char *)p;
 457         spec->size = strlen(p);
 458         spec->should_munmap = 0;
 459         spec->is_stdin = 1;
 460         spec->driver = &section_headers;
 461
 462         return spec;
 463 }
 464
 465 static void patch_diff(const char *a, const char *b,
 466                        struct diff_options *diffopt)
 467 {
 468         diff_queue(&diff_queued_diff,
 469                    get_filespec("a", a), get_filespec("b", b));
 470
 471         diffcore_std(diffopt);
 472         diff_flush(diffopt);
 473 }
 474
 475 static struct strbuf *output_prefix_cb(struct diff_options *opt UNUSED, void *data)
 476 {
 477         return data;
 478 }
 479
 480 static void output(struct string_list *a, struct string_list *b,
 481                    struct range_diff_options *range_diff_opts)
 482 {
 483         struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
 484         int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
 485         int i = 0, j = 0;
 486         struct diff_options opts;
 487         struct strbuf indent = STRBUF_INIT;
 488
 489         if (range_diff_opts->diffopt)
 490                 memcpy(&opts, range_diff_opts->diffopt, sizeof(opts));
 491         else
 492                 repo_diff_setup(the_repository, &opts);
 493
 494         opts.no_free = 1;
 495         if (!opts.output_format)
 496                 opts.output_format = DIFF_FORMAT_PATCH;
 497         opts.flags.suppress_diff_headers = 1;
 498         opts.flags.dual_color_diffed_diffs =
 499                 range_diff_opts->dual_color;
 500         opts.flags.suppress_hunk_header_line_count = 1;
 501         opts.output_prefix = output_prefix_cb;
 502         strbuf_addstr(&indent, "    ");
 503         opts.output_prefix_data = &indent;
 504         diff_setup_done(&opts);
 505
 506         /*
 507          * We assume the user is really more interested in the second argument
 508          * ("newer" version). To that end, we print the output in the order of
 509          * the RHS (the `b` parameter). To put the LHS (the `a` parameter)
 510          * commits that are no longer in the RHS into a good place, we place
 511          * them once we have shown all of their predecessors in the LHS.
 512          */
 513
 514         while (i < a->nr || j < b->nr) {
 515                 struct patch_util *a_util, *b_util;
 516                 a_util = i < a->nr ? a->items[i].util : NULL;
 517                 b_util = j < b->nr ? b->items[j].util : NULL;
 518
 519                 /* Skip all the already-shown commits from the LHS. */
 520                 while (i < a->nr && a_util->shown)
 521                         a_util = ++i < a->nr ? a->items[i].util : NULL;
 522
 523                 /* Show unmatched LHS commit whose predecessors were shown. */
 524                 if (i < a->nr && a_util->matching < 0) {
 525                         if (!range_diff_opts->right_only)
 526                                 output_pair_header(&opts, patch_no_width,
 527                                            &buf, &dashes, a_util, NULL);
 528                         i++;
 529                         continue;
 530                 }
 531
 532                 /* Show unmatched RHS commits. */
 533                 while (j < b->nr && b_util->matching < 0) {
 534                         if (!range_diff_opts->left_only)
 535                                 output_pair_header(&opts, patch_no_width,
 536                                            &buf, &dashes, NULL, b_util);
 537                         b_util = ++j < b->nr ? b->items[j].util : NULL;
 538                 }
 539
 540                 /* Show matching LHS/RHS pair. */
 541                 if (j < b->nr) {
 542                         a_util = a->items[b_util->matching].util;
 543                         output_pair_header(&opts, patch_no_width,
 544                                            &buf, &dashes, a_util, b_util);
 545                         if (!(opts.output_format & DIFF_FORMAT_NO_OUTPUT))
 546                                 patch_diff(a->items[b_util->matching].string,
 547                                            b->items[j].string, &opts);
 548                         a_util->shown = 1;
 549                         j++;
 550                 }
 551         }
 552         strbuf_release(&buf);
 553         strbuf_release(&dashes);
 554         strbuf_release(&indent);
 555         opts.no_free = 0;
 556         diff_free(&opts);
 557 }
 558
 559 int show_range_diff(const char *range1, const char *range2,
 560                     struct range_diff_options *range_diff_opts)
 561 {
 562         int res = 0;
 563
 564         struct string_list branch1 = STRING_LIST_INIT_DUP;
 565         struct string_list branch2 = STRING_LIST_INIT_DUP;
 566
 567         if (range_diff_opts->left_only && range_diff_opts->right_only)
 568                 res = error(_("options '%s' and '%s' cannot be used together"), "--left-only", "--right-only");
 569
 570         if (!res && read_patches(range1, &branch1, range_diff_opts->other_arg))
 571                 res = error(_("could not parse log for '%s'"), range1);
 572         if (!res && read_patches(range2, &branch2, range_diff_opts->other_arg))
 573                 res = error(_("could not parse log for '%s'"), range2);
 574
 575         if (!res) {
 576                 find_exact_matches(&branch1, &branch2);
 577                 get_correspondences(&branch1, &branch2,
 578                                     range_diff_opts->creation_factor);
 579                 output(&branch1, &branch2, range_diff_opts);
 580         }
 581
 582         string_list_clear(&branch1, 1);
 583         string_list_clear(&branch2, 1);
 584
 585         return res;
 586 }
 587
 588 int is_range_diff_range(const char *arg)
 589 {
 590         char *copy = xstrdup(arg); /* setup_revisions() modifies it */
 591         const char *argv[] = { "", copy, "--", NULL };
 592         int i, positive = 0, negative = 0;
 593         struct rev_info revs;
 594
 595         repo_init_revisions(the_repository, &revs, NULL);
 596         if (setup_revisions(3, argv, &revs, NULL) == 1) {
 597                 for (i = 0; i < revs.pending.nr; i++)
 598                         if (revs.pending.objects[i].item->flags & UNINTERESTING)
 599                                 negative++;
 600                         else
 601                                 positive++;
 602                 for (i = 0; i < revs.pending.nr; i++) {
 603                         struct object *obj = revs.pending.objects[i].item;
 604
 605                         if (obj->type == OBJ_COMMIT)
 606                                 clear_commit_marks((struct commit *)obj,
 607                                                    ALL_REV_FLAGS);
 608                 }
 609         }
 610
 611         free(copy);
 612         release_revisions(&revs);
 613         return negative > 0 && positive > 0;
 614 }