range-diff.c

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