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 "repository.h"
  17 #include "userdiff.h"
  18 #include "apply.h"
  19 #include "revision.h"
  20
  21 struct patch_util {
  22         /* For the search for an exact match */
  23         struct hashmap_entry e;
  24         const char *diff, *patch;
  25
  26         int i, shown;
  27         int diffsize;
  28         size_t diff_offset;
  29         /* the index of the matching item in the other branch, or -1 */
  30         int matching;
  31         struct object_id oid;
  32 };
  33
  34 /*
  35  * Reads the patches into a string list, with the `util` field being populated
  36  * as struct object_id (will need to be free()d).
  37  */
  38 static int read_patches(const char *range, struct string_list *list,
  39                         const struct strvec *other_arg)
  40 {
  41         struct child_process cp = CHILD_PROCESS_INIT;
  42         struct strbuf buf = STRBUF_INIT, contents = STRBUF_INIT;
  43         struct patch_util *util = NULL;
  44         int in_header = 1;
  45         char *line, *current_filename = NULL;
  46         ssize_t len;
  47         size_t size;
  48         int ret = -1;
  49
  50         strvec_pushl(&cp.args, "log", "--no-color", "-p", "--no-merges",
  51                      "--reverse", "--date-order", "--decorate=no",
  52                      "--no-prefix", "--submodule=short",
  53                      /*
  54                       * Choose indicators that are not used anywhere
  55                       * else in diffs, but still look reasonable
  56                       * (e.g. will not be confusing when debugging)
  57                       */
  58                      "--output-indicator-new=>",
  59                      "--output-indicator-old=<",
  60                      "--output-indicator-context=#",
  61                      "--no-abbrev-commit",
  62                      "--pretty=medium",
  63                      "--show-notes-by-default",
  64                      NULL);
  65         strvec_push(&cp.args, range);
  66         if (other_arg)
  67                 strvec_pushv(&cp.args, other_arg->v);
  68         cp.out = -1;
  69         cp.no_stdin = 1;
  70         cp.git_cmd = 1;
  71
  72         if (start_command(&cp))
  73                 return error_errno(_("could not start `log`"));
  74         if (strbuf_read(&contents, cp.out, 0) < 0) {
  75                 error_errno(_("could not read `log` output"));
  76                 finish_command(&cp);
  77                 goto cleanup;
  78         }
  79         if (finish_command(&cp))
  80                 goto cleanup;
  81
  82         line = contents.buf;
  83         size = contents.len;
  84         for (; size > 0; size -= len, line += len) {
  85                 const char *p;
  86                 char *eol;
  87
  88                 eol = memchr(line, '\n', size);
  89                 if (eol) {
  90                         *eol = '\0';
  91                         len = eol + 1 - line;
  92                 } else {
  93                         len = size;
  94                 }
  95
  96                 if (skip_prefix(line, "commit ", &p)) {
  97                         if (util) {
  98                                 string_list_append(list, buf.buf)->util = util;
  99                                 strbuf_reset(&buf);
 100                         }
 101                         CALLOC_ARRAY(util, 1);
 102                         if (repo_get_oid(the_repository, p, &util->oid)) {
 103                                 error(_("could not parse commit '%s'"), p);
 104                                 FREE_AND_NULL(util);
 105                                 string_list_clear(list, 1);
 106                                 goto cleanup;
 107                         }
 108                         util->matching = -1;
 109                         in_header = 1;
 110                         continue;
 111                 }
 112
 113                 if (!util) {
 114                         error(_("could not parse first line of `log` output: "
 115                                 "did not start with 'commit ': '%s'"),
 116                               line);
 117                         string_list_clear(list, 1);
 118                         goto cleanup;
 119                 }
 120
 121                 if (starts_with(line, "diff --git")) {
 122                         struct patch patch = { 0 };
 123                         struct strbuf root = STRBUF_INIT;
 124                         int linenr = 0;
 125                         int orig_len;
 126
 127                         in_header = 0;
 128                         strbuf_addch(&buf, '\n');
 129                         if (!util->diff_offset)
 130                                 util->diff_offset = buf.len;
 131                         if (eol)
 132                                 *eol = '\n';
 133                         orig_len = len;
 134                         len = parse_git_diff_header(&root, &linenr, 0, line,
 135                                                     len, size, &patch);
 136                         if (len < 0) {
 137                                 error(_("could not parse git header '%.*s'"),
 138                                       orig_len, line);
 139                                 FREE_AND_NULL(util);
 140                                 string_list_clear(list, 1);
 141                                 goto cleanup;
 142                         }
 143                         strbuf_addstr(&buf, " ## ");
 144                         if (patch.is_new > 0)
 145                                 strbuf_addf(&buf, "%s (new)", patch.new_name);
 146                         else if (patch.is_delete > 0)
 147                                 strbuf_addf(&buf, "%s (deleted)", patch.old_name);
 148                         else if (patch.is_rename)
 149                                 strbuf_addf(&buf, "%s => %s", patch.old_name, patch.new_name);
 150                         else
 151                                 strbuf_addstr(&buf, patch.new_name);
 152
 153                         free(current_filename);
 154                         if (patch.is_delete > 0)
 155                                 current_filename = xstrdup(patch.old_name);
 156                         else
 157                                 current_filename = xstrdup(patch.new_name);
 158
 159                         if (patch.new_mode && patch.old_mode &&
 160                             patch.old_mode != patch.new_mode)
 161                                 strbuf_addf(&buf, " (mode change %06o => %06o)",
 162                                             patch.old_mode, patch.new_mode);
 163
 164                         strbuf_addstr(&buf, " ##");
 165                         release_patch(&patch);
 166                 } else if (in_header) {
 167                         if (starts_with(line, "Author: ")) {
 168                                 strbuf_addstr(&buf, " ## Metadata ##\n");
 169                                 strbuf_addstr(&buf, line);
 170                                 strbuf_addstr(&buf, "\n\n");
 171                                 strbuf_addstr(&buf, " ## Commit message ##\n");
 172                         } else if (starts_with(line, "Notes") &&
 173                                    line[strlen(line) - 1] == ':') {
 174                                 strbuf_addstr(&buf, "\n\n");
 175                                 /* strip the trailing colon */
 176                                 strbuf_addf(&buf, " ## %.*s ##\n",
 177                                             (int)(strlen(line) - 1), line);
 178                         } else if (starts_with(line, "    ")) {
 179                                 p = line + len - 2;
 180                                 while (isspace(*p) && p >= line)
 181                                         p--;
 182                                 strbuf_add(&buf, line, p - line + 1);
 183                                 strbuf_addch(&buf, '\n');
 184                         }
 185                         continue;
 186                 } else if (skip_prefix(line, "@@ ", &p)) {
 187                         p = strstr(p, "@@");
 188                         strbuf_addstr(&buf, "@@");
 189                         if (current_filename && p[2])
 190                                 strbuf_addf(&buf, " %s:", current_filename);
 191                         if (p)
 192                                 strbuf_addstr(&buf, p + 2);
 193                 } else if (!line[0])
 194                         /*
 195                          * A completely blank (not ' \n', which is context)
 196                          * line is not valid in a diff.  We skip it
 197                          * silently, because this neatly handles the blank
 198                          * separator line between commits in git-log
 199                          * output.
 200                          */
 201                         continue;
 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 if (line[0] == '#') {
 209                         strbuf_addch(&buf, ' ');
 210                         strbuf_addstr(&buf, line + 1);
 211                 } else {
 212                         strbuf_addch(&buf, ' ');
 213                         strbuf_addstr(&buf, line);
 214                 }
 215
 216                 strbuf_addch(&buf, '\n');
 217                 util->diffsize++;
 218         }
 219
 220         ret = 0;
 221 cleanup:
 222         strbuf_release(&contents);
 223
 224         if (util)
 225                 string_list_append(list, buf.buf)->util = util;
 226         strbuf_release(&buf);
 227         free(current_filename);
 228
 229         return ret;
 230 }
 231
 232 static int patch_util_cmp(const void *cmp_data UNUSED,
 233                           const struct hashmap_entry *ha,
 234                           const struct hashmap_entry *hb,
 235                           const void *keydata)
 236 {
 237         const struct patch_util
 238                 *a = container_of(ha, const struct patch_util, e),
 239                 *b = container_of(hb, const struct patch_util, e);
 240         return strcmp(a->diff, keydata ? keydata : b->diff);
 241 }
 242
 243 static void find_exact_matches(struct string_list *a, struct string_list *b)
 244 {
 245         struct hashmap map = HASHMAP_INIT(patch_util_cmp, NULL);
 246         int i;
 247
 248         /* First, add the patches of a to a hash map */
 249         for (i = 0; i < a->nr; i++) {
 250                 struct patch_util *util = a->items[i].util;
 251
 252                 util->i = i;
 253                 util->patch = a->items[i].string;
 254                 util->diff = util->patch + util->diff_offset;
 255                 hashmap_entry_init(&util->e, strhash(util->diff));
 256                 hashmap_add(&map, &util->e);
 257         }
 258
 259         /* Now try to find exact matches in b */
 260         for (i = 0; i < b->nr; i++) {
 261                 struct patch_util *util = b->items[i].util, *other;
 262
 263                 util->i = i;
 264                 util->patch = b->items[i].string;
 265                 util->diff = util->patch + util->diff_offset;
 266                 hashmap_entry_init(&util->e, strhash(util->diff));
 267                 other = hashmap_remove_entry(&map, util, e, NULL);
 268                 if (other) {
 269                         if (other->matching >= 0)
 270                                 BUG("already assigned!");
 271
 272                         other->matching = i;
 273                         util->matching = other->i;
 274                 }
 275         }
 276
 277         hashmap_clear(&map);
 278 }
 279
 280 static int diffsize_consume(void *data,
 281                              char *line UNUSED,
 282                              unsigned long len UNUSED)
 283 {
 284         (*(int *)data)++;
 285         return 0;
 286 }
 287
 288 static void diffsize_hunk(void *data,
 289                           long ob UNUSED, long on UNUSED,
 290                           long nb UNUSED, long nn UNUSED,
 291                           const char *func UNUSED, long funclen UNUSED)
 292 {
 293         diffsize_consume(data, NULL, 0);
 294 }
 295
 296 static int diffsize(const char *a, const char *b)
 297 {
 298         xpparam_t pp = { 0 };
 299         xdemitconf_t cfg = { 0 };
 300         mmfile_t mf1, mf2;
 301         int count = 0;
 302
 303         mf1.ptr = (char *)a;
 304         mf1.size = strlen(a);
 305         mf2.ptr = (char *)b;
 306         mf2.size = strlen(b);
 307
 308         cfg.ctxlen = 3;
 309         if (!xdi_diff_outf(&mf1, &mf2,
 310                            diffsize_hunk, diffsize_consume, &count,
 311                            &pp, &cfg))
 312                 return count;
 313
 314         error(_("failed to generate diff"));
 315         return COST_MAX;
 316 }
 317
 318 static void get_correspondences(struct string_list *a, struct string_list *b,
 319                                 int creation_factor)
 320 {
 321         int n = a->nr + b->nr;
 322         int *cost, c, *a2b, *b2a;
 323         int i, j;
 324
 325         ALLOC_ARRAY(cost, st_mult(n, n));
 326         ALLOC_ARRAY(a2b, n);
 327         ALLOC_ARRAY(b2a, n);
 328
 329         for (i = 0; i < a->nr; i++) {
 330                 struct patch_util *a_util = a->items[i].util;
 331
 332                 for (j = 0; j < b->nr; j++) {
 333                         struct patch_util *b_util = b->items[j].util;
 334
 335                         if (a_util->matching == j)
 336                                 c = 0;
 337                         else if (a_util->matching < 0 && b_util->matching < 0)
 338                                 c = diffsize(a_util->diff, b_util->diff);
 339                         else
 340                                 c = COST_MAX;
 341                         cost[i + n * j] = c;
 342                 }
 343
 344                 c = a_util->matching < 0 ?
 345                         a_util->diffsize * creation_factor / 100 : COST_MAX;
 346                 for (j = b->nr; j < n; j++)
 347                         cost[i + n * j] = c;
 348         }
 349
 350         for (j = 0; j < b->nr; j++) {
 351                 struct patch_util *util = b->items[j].util;
 352
 353                 c = util->matching < 0 ?
 354                         util->diffsize * creation_factor / 100 : COST_MAX;
 355                 for (i = a->nr; i < n; i++)
 356                         cost[i + n * j] = c;
 357         }
 358
 359         for (i = a->nr; i < n; i++)
 360                 for (j = b->nr; j < n; j++)
 361                         cost[i + n * j] = 0;
 362
 363         compute_assignment(n, n, cost, a2b, b2a);
 364
 365         for (i = 0; i < a->nr; i++)
 366                 if (a2b[i] >= 0 && a2b[i] < b->nr) {
 367                         struct patch_util *a_util = a->items[i].util;
 368                         struct patch_util *b_util = b->items[a2b[i]].util;
 369
 370                         a_util->matching = a2b[i];
 371                         b_util->matching = i;
 372                 }
 373
 374         free(cost);
 375         free(a2b);
 376         free(b2a);
 377 }
 378
 379 static void output_pair_header(struct diff_options *diffopt,
 380                                int patch_no_width,
 381                                struct strbuf *buf,
 382                                struct strbuf *dashes,
 383                                struct patch_util *a_util,
 384                                struct patch_util *b_util)
 385 {
 386         struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
 387         struct commit *commit;
 388         char status;
 389         const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
 390         const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
 391         const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
 392         const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
 393         const char *color;
 394         int abbrev = diffopt->abbrev;
 395
 396         if (abbrev < 0)
 397                 abbrev = DEFAULT_ABBREV;
 398
 399         if (!dashes->len)
 400                 strbuf_addchars(dashes, '-',
 401                                 strlen(repo_find_unique_abbrev(the_repository, oid, abbrev)));
 402
 403         if (!b_util) {
 404                 color = color_old;
 405                 status = '<';
 406         } else if (!a_util) {
 407                 color = color_new;
 408                 status = '>';
 409         } else if (strcmp(a_util->patch, b_util->patch)) {
 410                 color = color_commit;
 411                 status = '!';
 412         } else {
 413                 color = color_commit;
 414                 status = '=';
 415         }
 416
 417         strbuf_reset(buf);
 418         strbuf_addstr(buf, status == '!' ? color_old : color);
 419         if (!a_util)
 420                 strbuf_addf(buf, "%*s:  %s ", patch_no_width, "-", dashes->buf);
 421         else
 422                 strbuf_addf(buf, "%*d:  %s ", patch_no_width, a_util->i + 1,
 423                             repo_find_unique_abbrev(the_repository, &a_util->oid, abbrev));
 424
 425         if (status == '!')
 426                 strbuf_addf(buf, "%s%s", color_reset, color);
 427         strbuf_addch(buf, status);
 428         if (status == '!')
 429                 strbuf_addf(buf, "%s%s", color_reset, color_new);
 430
 431         if (!b_util)
 432                 strbuf_addf(buf, " %*s:  %s", patch_no_width, "-", dashes->buf);
 433         else
 434                 strbuf_addf(buf, " %*d:  %s", patch_no_width, b_util->i + 1,
 435                             repo_find_unique_abbrev(the_repository, &b_util->oid, abbrev));
 436
 437         commit = lookup_commit_reference(the_repository, oid);
 438         if (commit) {
 439                 if (status == '!')
 440                         strbuf_addf(buf, "%s%s", color_reset, color);
 441
 442                 strbuf_addch(buf, ' ');
 443                 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
 444         }
 445         strbuf_addf(buf, "%s\n", color_reset);
 446
 447         fwrite(buf->buf, buf->len, 1, diffopt->file);
 448 }
 449
 450 static struct userdiff_driver section_headers = {
 451         .funcname = { "^ ## (.*) ##$\n"
 452                       "^.?@@ (.*)$", REG_EXTENDED }
 453 };
 454
 455 static struct diff_filespec *get_filespec(const char *name, const char *p)
 456 {
 457         struct diff_filespec *spec = alloc_filespec(name);
 458
 459         fill_filespec(spec, null_oid(), 0, 0100644);
 460         spec->data = (char *)p;
 461         spec->size = strlen(p);
 462         spec->should_munmap = 0;
 463         spec->is_stdin = 1;
 464         spec->driver = &section_headers;
 465
 466         return spec;
 467 }
 468
 469 static void patch_diff(const char *a, const char *b,
 470                        struct diff_options *diffopt)
 471 {
 472         diff_queue(&diff_queued_diff,
 473                    get_filespec("a", a), get_filespec("b", b));
 474
 475         diffcore_std(diffopt);
 476         diff_flush(diffopt);
 477 }
 478
 479 static struct strbuf *output_prefix_cb(struct diff_options *opt UNUSED, void *data)
 480 {
 481         return data;
 482 }
 483
 484 static void output(struct string_list *a, struct string_list *b,
 485                    struct range_diff_options *range_diff_opts)
 486 {
 487         struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
 488         int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
 489         int i = 0, j = 0;
 490         struct diff_options opts;
 491         struct strbuf indent = STRBUF_INIT;
 492
 493         if (range_diff_opts->diffopt)
 494                 memcpy(&opts, range_diff_opts->diffopt, sizeof(opts));
 495         else
 496                 repo_diff_setup(the_repository, &opts);
 497
 498         opts.no_free = 1;
 499         if (!opts.output_format)
 500                 opts.output_format = DIFF_FORMAT_PATCH;
 501         opts.flags.suppress_diff_headers = 1;
 502         opts.flags.dual_color_diffed_diffs =
 503                 range_diff_opts->dual_color;
 504         opts.flags.suppress_hunk_header_line_count = 1;
 505         opts.output_prefix = output_prefix_cb;
 506         strbuf_addstr(&indent, "    ");
 507         opts.output_prefix_data = &indent;
 508         diff_setup_done(&opts);
 509
 510         /*
 511          * We assume the user is really more interested in the second argument
 512          * ("newer" version). To that end, we print the output in the order of
 513          * the RHS (the `b` parameter). To put the LHS (the `a` parameter)
 514          * commits that are no longer in the RHS into a good place, we place
 515          * them once we have shown all of their predecessors in the LHS.
 516          */
 517
 518         while (i < a->nr || j < b->nr) {
 519                 struct patch_util *a_util, *b_util;
 520                 a_util = i < a->nr ? a->items[i].util : NULL;
 521                 b_util = j < b->nr ? b->items[j].util : NULL;
 522
 523                 /* Skip all the already-shown commits from the LHS. */
 524                 while (i < a->nr && a_util->shown)
 525                         a_util = ++i < a->nr ? a->items[i].util : NULL;
 526
 527                 /* Show unmatched LHS commit whose predecessors were shown. */
 528                 if (i < a->nr && a_util->matching < 0) {
 529                         if (!range_diff_opts->right_only)
 530                                 output_pair_header(&opts, patch_no_width,
 531                                            &buf, &dashes, a_util, NULL);
 532                         i++;
 533                         continue;
 534                 }
 535
 536                 /* Show unmatched RHS commits. */
 537                 while (j < b->nr && b_util->matching < 0) {
 538                         if (!range_diff_opts->left_only)
 539                                 output_pair_header(&opts, patch_no_width,
 540                                            &buf, &dashes, NULL, b_util);
 541                         b_util = ++j < b->nr ? b->items[j].util : NULL;
 542                 }
 543
 544                 /* Show matching LHS/RHS pair. */
 545                 if (j < b->nr) {
 546                         a_util = a->items[b_util->matching].util;
 547                         output_pair_header(&opts, patch_no_width,
 548                                            &buf, &dashes, a_util, b_util);
 549                         if (!(opts.output_format & DIFF_FORMAT_NO_OUTPUT))
 550                                 patch_diff(a->items[b_util->matching].string,
 551                                            b->items[j].string, &opts);
 552                         a_util->shown = 1;
 553                         j++;
 554                 }
 555         }
 556         strbuf_release(&buf);
 557         strbuf_release(&dashes);
 558         strbuf_release(&indent);
 559         opts.no_free = 0;
 560         diff_free(&opts);
 561 }
 562
 563 int show_range_diff(const char *range1, const char *range2,
 564                     struct range_diff_options *range_diff_opts)
 565 {
 566         int res = 0;
 567
 568         struct string_list branch1 = STRING_LIST_INIT_DUP;
 569         struct string_list branch2 = STRING_LIST_INIT_DUP;
 570
 571         if (range_diff_opts->left_only && range_diff_opts->right_only)
 572                 res = error(_("options '%s' and '%s' cannot be used together"), "--left-only", "--right-only");
 573
 574         if (!res && read_patches(range1, &branch1, range_diff_opts->other_arg))
 575                 res = error(_("could not parse log for '%s'"), range1);
 576         if (!res && read_patches(range2, &branch2, range_diff_opts->other_arg))
 577                 res = error(_("could not parse log for '%s'"), range2);
 578
 579         if (!res) {
 580                 find_exact_matches(&branch1, &branch2);
 581                 get_correspondences(&branch1, &branch2,
 582                                     range_diff_opts->creation_factor);
 583                 output(&branch1, &branch2, range_diff_opts);
 584         }
 585
 586         string_list_clear(&branch1, 1);
 587         string_list_clear(&branch2, 1);
 588
 589         return res;
 590 }
 591
 592 int is_range_diff_range(const char *arg)
 593 {
 594         char *copy = xstrdup(arg); /* setup_revisions() modifies it */
 595         const char *argv[] = { "", copy, "--", NULL };
 596         int i, positive = 0, negative = 0;
 597         struct rev_info revs;
 598
 599         repo_init_revisions(the_repository, &revs, NULL);
 600         if (setup_revisions(3, argv, &revs, NULL) == 1) {
 601                 for (i = 0; i < revs.pending.nr; i++)
 602                         if (revs.pending.objects[i].item->flags & UNINTERESTING)
 603                                 negative++;
 604                         else
 605                                 positive++;
 606                 for (i = 0; i < revs.pending.nr; i++) {
 607                         struct object *obj = revs.pending.objects[i].item;
 608
 609                         if (obj->type == OBJ_COMMIT)
 610                                 clear_commit_marks((struct commit *)obj,
 611                                                    ALL_REV_FLAGS);
 612                 }
 613         }
 614
 615         free(copy);
 616         release_revisions(&revs);
 617         return negative > 0 && positive > 0;
 618 }