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                      "--notes",
  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 patch_util *a,
 234                           const struct patch_util *b,
 235                           const char *keydata)
 236 {
 237         return strcmp(a->diff, keydata ? keydata : b->diff);
 238 }
 239
 240 static void find_exact_matches(struct string_list *a, struct string_list *b)
 241 {
 242         struct hashmap map = HASHMAP_INIT((hashmap_cmp_fn)patch_util_cmp, NULL);
 243         int i;
 244
 245         /* First, add the patches of a to a hash map */
 246         for (i = 0; i < a->nr; i++) {
 247                 struct patch_util *util = a->items[i].util;
 248
 249                 util->i = i;
 250                 util->patch = a->items[i].string;
 251                 util->diff = util->patch + util->diff_offset;
 252                 hashmap_entry_init(&util->e, strhash(util->diff));
 253                 hashmap_add(&map, &util->e);
 254         }
 255
 256         /* Now try to find exact matches in b */
 257         for (i = 0; i < b->nr; i++) {
 258                 struct patch_util *util = b->items[i].util, *other;
 259
 260                 util->i = i;
 261                 util->patch = b->items[i].string;
 262                 util->diff = util->patch + util->diff_offset;
 263                 hashmap_entry_init(&util->e, strhash(util->diff));
 264                 other = hashmap_remove_entry(&map, util, e, NULL);
 265                 if (other) {
 266                         if (other->matching >= 0)
 267                                 BUG("already assigned!");
 268
 269                         other->matching = i;
 270                         util->matching = other->i;
 271                 }
 272         }
 273
 274         hashmap_clear(&map);
 275 }
 276
 277 static int diffsize_consume(void *data,
 278                              char *line UNUSED,
 279                              unsigned long len UNUSED)
 280 {
 281         (*(int *)data)++;
 282         return 0;
 283 }
 284
 285 static void diffsize_hunk(void *data,
 286                           long ob UNUSED, long on UNUSED,
 287                           long nb UNUSED, long nn UNUSED,
 288                           const char *func UNUSED, long funclen UNUSED)
 289 {
 290         diffsize_consume(data, NULL, 0);
 291 }
 292
 293 static int diffsize(const char *a, const char *b)
 294 {
 295         xpparam_t pp = { 0 };
 296         xdemitconf_t cfg = { 0 };
 297         mmfile_t mf1, mf2;
 298         int count = 0;
 299
 300         mf1.ptr = (char *)a;
 301         mf1.size = strlen(a);
 302         mf2.ptr = (char *)b;
 303         mf2.size = strlen(b);
 304
 305         cfg.ctxlen = 3;
 306         if (!xdi_diff_outf(&mf1, &mf2,
 307                            diffsize_hunk, diffsize_consume, &count,
 308                            &pp, &cfg))
 309                 return count;
 310
 311         error(_("failed to generate diff"));
 312         return COST_MAX;
 313 }
 314
 315 static void get_correspondences(struct string_list *a, struct string_list *b,
 316                                 int creation_factor)
 317 {
 318         int n = a->nr + b->nr;
 319         int *cost, c, *a2b, *b2a;
 320         int i, j;
 321
 322         ALLOC_ARRAY(cost, st_mult(n, n));
 323         ALLOC_ARRAY(a2b, n);
 324         ALLOC_ARRAY(b2a, n);
 325
 326         for (i = 0; i < a->nr; i++) {
 327                 struct patch_util *a_util = a->items[i].util;
 328
 329                 for (j = 0; j < b->nr; j++) {
 330                         struct patch_util *b_util = b->items[j].util;
 331
 332                         if (a_util->matching == j)
 333                                 c = 0;
 334                         else if (a_util->matching < 0 && b_util->matching < 0)
 335                                 c = diffsize(a_util->diff, b_util->diff);
 336                         else
 337                                 c = COST_MAX;
 338                         cost[i + n * j] = c;
 339                 }
 340
 341                 c = a_util->matching < 0 ?
 342                         a_util->diffsize * creation_factor / 100 : COST_MAX;
 343                 for (j = b->nr; j < n; j++)
 344                         cost[i + n * j] = c;
 345         }
 346
 347         for (j = 0; j < b->nr; j++) {
 348                 struct patch_util *util = b->items[j].util;
 349
 350                 c = util->matching < 0 ?
 351                         util->diffsize * creation_factor / 100 : COST_MAX;
 352                 for (i = a->nr; i < n; i++)
 353                         cost[i + n * j] = c;
 354         }
 355
 356         for (i = a->nr; i < n; i++)
 357                 for (j = b->nr; j < n; j++)
 358                         cost[i + n * j] = 0;
 359
 360         compute_assignment(n, n, cost, a2b, b2a);
 361
 362         for (i = 0; i < a->nr; i++)
 363                 if (a2b[i] >= 0 && a2b[i] < b->nr) {
 364                         struct patch_util *a_util = a->items[i].util;
 365                         struct patch_util *b_util = b->items[a2b[i]].util;
 366
 367                         a_util->matching = a2b[i];
 368                         b_util->matching = i;
 369                 }
 370
 371         free(cost);
 372         free(a2b);
 373         free(b2a);
 374 }
 375
 376 static void output_pair_header(struct diff_options *diffopt,
 377                                int patch_no_width,
 378                                struct strbuf *buf,
 379                                struct strbuf *dashes,
 380                                struct patch_util *a_util,
 381                                struct patch_util *b_util)
 382 {
 383         struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
 384         struct commit *commit;
 385         char status;
 386         const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
 387         const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
 388         const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
 389         const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
 390         const char *color;
 391         int abbrev = diffopt->abbrev;
 392
 393         if (abbrev < 0)
 394                 abbrev = DEFAULT_ABBREV;
 395
 396         if (!dashes->len)
 397                 strbuf_addchars(dashes, '-',
 398                                 strlen(repo_find_unique_abbrev(the_repository, oid, abbrev)));
 399
 400         if (!b_util) {
 401                 color = color_old;
 402                 status = '<';
 403         } else if (!a_util) {
 404                 color = color_new;
 405                 status = '>';
 406         } else if (strcmp(a_util->patch, b_util->patch)) {
 407                 color = color_commit;
 408                 status = '!';
 409         } else {
 410                 color = color_commit;
 411                 status = '=';
 412         }
 413
 414         strbuf_reset(buf);
 415         strbuf_addstr(buf, status == '!' ? color_old : color);
 416         if (!a_util)
 417                 strbuf_addf(buf, "%*s:  %s ", patch_no_width, "-", dashes->buf);
 418         else
 419                 strbuf_addf(buf, "%*d:  %s ", patch_no_width, a_util->i + 1,
 420                             repo_find_unique_abbrev(the_repository, &a_util->oid, abbrev));
 421
 422         if (status == '!')
 423                 strbuf_addf(buf, "%s%s", color_reset, color);
 424         strbuf_addch(buf, status);
 425         if (status == '!')
 426                 strbuf_addf(buf, "%s%s", color_reset, color_new);
 427
 428         if (!b_util)
 429                 strbuf_addf(buf, " %*s:  %s", patch_no_width, "-", dashes->buf);
 430         else
 431                 strbuf_addf(buf, " %*d:  %s", patch_no_width, b_util->i + 1,
 432                             repo_find_unique_abbrev(the_repository, &b_util->oid, abbrev));
 433
 434         commit = lookup_commit_reference(the_repository, oid);
 435         if (commit) {
 436                 if (status == '!')
 437                         strbuf_addf(buf, "%s%s", color_reset, color);
 438
 439                 strbuf_addch(buf, ' ');
 440                 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
 441         }
 442         strbuf_addf(buf, "%s\n", color_reset);
 443
 444         fwrite(buf->buf, buf->len, 1, diffopt->file);
 445 }
 446
 447 static struct userdiff_driver section_headers = {
 448         .funcname = { "^ ## (.*) ##$\n"
 449                       "^.?@@ (.*)$", REG_EXTENDED }
 450 };
 451
 452 static struct diff_filespec *get_filespec(const char *name, const char *p)
 453 {
 454         struct diff_filespec *spec = alloc_filespec(name);
 455
 456         fill_filespec(spec, null_oid(), 0, 0100644);
 457         spec->data = (char *)p;
 458         spec->size = strlen(p);
 459         spec->should_munmap = 0;
 460         spec->is_stdin = 1;
 461         spec->driver = &section_headers;
 462
 463         return spec;
 464 }
 465
 466 static void patch_diff(const char *a, const char *b,
 467                        struct diff_options *diffopt)
 468 {
 469         diff_queue(&diff_queued_diff,
 470                    get_filespec("a", a), get_filespec("b", b));
 471
 472         diffcore_std(diffopt);
 473         diff_flush(diffopt);
 474 }
 475
 476 static struct strbuf *output_prefix_cb(struct diff_options *opt UNUSED, void *data)
 477 {
 478         return data;
 479 }
 480
 481 static void output(struct string_list *a, struct string_list *b,
 482                    struct range_diff_options *range_diff_opts)
 483 {
 484         struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
 485         int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
 486         int i = 0, j = 0;
 487         struct diff_options opts;
 488         struct strbuf indent = STRBUF_INIT;
 489
 490         if (range_diff_opts->diffopt)
 491                 memcpy(&opts, range_diff_opts->diffopt, sizeof(opts));
 492         else
 493                 repo_diff_setup(the_repository, &opts);
 494
 495         opts.no_free = 1;
 496         if (!opts.output_format)
 497                 opts.output_format = DIFF_FORMAT_PATCH;
 498         opts.flags.suppress_diff_headers = 1;
 499         opts.flags.dual_color_diffed_diffs =
 500                 range_diff_opts->dual_color;
 501         opts.flags.suppress_hunk_header_line_count = 1;
 502         opts.output_prefix = output_prefix_cb;
 503         strbuf_addstr(&indent, "    ");
 504         opts.output_prefix_data = &indent;
 505         diff_setup_done(&opts);
 506
 507         /*
 508          * We assume the user is really more interested in the second argument
 509          * ("newer" version). To that end, we print the output in the order of
 510          * the RHS (the `b` parameter). To put the LHS (the `a` parameter)
 511          * commits that are no longer in the RHS into a good place, we place
 512          * them once we have shown all of their predecessors in the LHS.
 513          */
 514
 515         while (i < a->nr || j < b->nr) {
 516                 struct patch_util *a_util, *b_util;
 517                 a_util = i < a->nr ? a->items[i].util : NULL;
 518                 b_util = j < b->nr ? b->items[j].util : NULL;
 519
 520                 /* Skip all the already-shown commits from the LHS. */
 521                 while (i < a->nr && a_util->shown)
 522                         a_util = ++i < a->nr ? a->items[i].util : NULL;
 523
 524                 /* Show unmatched LHS commit whose predecessors were shown. */
 525                 if (i < a->nr && a_util->matching < 0) {
 526                         if (!range_diff_opts->right_only)
 527                                 output_pair_header(&opts, patch_no_width,
 528                                            &buf, &dashes, a_util, NULL);
 529                         i++;
 530                         continue;
 531                 }
 532
 533                 /* Show unmatched RHS commits. */
 534                 while (j < b->nr && b_util->matching < 0) {
 535                         if (!range_diff_opts->left_only)
 536                                 output_pair_header(&opts, patch_no_width,
 537                                            &buf, &dashes, NULL, b_util);
 538                         b_util = ++j < b->nr ? b->items[j].util : NULL;
 539                 }
 540
 541                 /* Show matching LHS/RHS pair. */
 542                 if (j < b->nr) {
 543                         a_util = a->items[b_util->matching].util;
 544                         output_pair_header(&opts, patch_no_width,
 545                                            &buf, &dashes, a_util, b_util);
 546                         if (!(opts.output_format & DIFF_FORMAT_NO_OUTPUT))
 547                                 patch_diff(a->items[b_util->matching].string,
 548                                            b->items[j].string, &opts);
 549                         a_util->shown = 1;
 550                         j++;
 551                 }
 552         }
 553         strbuf_release(&buf);
 554         strbuf_release(&dashes);
 555         strbuf_release(&indent);
 556         opts.no_free = 0;
 557         diff_free(&opts);
 558 }
 559
 560 int show_range_diff(const char *range1, const char *range2,
 561                     struct range_diff_options *range_diff_opts)
 562 {
 563         int res = 0;
 564
 565         struct string_list branch1 = STRING_LIST_INIT_DUP;
 566         struct string_list branch2 = STRING_LIST_INIT_DUP;
 567
 568         if (range_diff_opts->left_only && range_diff_opts->right_only)
 569                 res = error(_("options '%s' and '%s' cannot be used together"), "--left-only", "--right-only");
 570
 571         if (!res && read_patches(range1, &branch1, range_diff_opts->other_arg))
 572                 res = error(_("could not parse log for '%s'"), range1);
 573         if (!res && read_patches(range2, &branch2, range_diff_opts->other_arg))
 574                 res = error(_("could not parse log for '%s'"), range2);
 575
 576         if (!res) {
 577                 find_exact_matches(&branch1, &branch2);
 578                 get_correspondences(&branch1, &branch2,
 579                                     range_diff_opts->creation_factor);
 580                 output(&branch1, &branch2, range_diff_opts);
 581         }
 582
 583         string_list_clear(&branch1, 1);
 584         string_list_clear(&branch2, 1);
 585
 586         return res;
 587 }
 588
 589 int is_range_diff_range(const char *arg)
 590 {
 591         char *copy = xstrdup(arg); /* setup_revisions() modifies it */
 592         const char *argv[] = { "", copy, "--", NULL };
 593         int i, positive = 0, negative = 0;
 594         struct rev_info revs;
 595
 596         repo_init_revisions(the_repository, &revs, NULL);
 597         if (setup_revisions(3, argv, &revs, NULL) == 1) {
 598                 for (i = 0; i < revs.pending.nr; i++)
 599                         if (revs.pending.objects[i].item->flags & UNINTERESTING)
 600                                 negative++;
 601                         else
 602                                 positive++;
 603                 for (i = 0; i < revs.pending.nr; i++) {
 604                         struct object *obj = revs.pending.objects[i].item;
 605
 606                         if (obj->type == OBJ_COMMIT)
 607                                 clear_commit_marks((struct commit *)obj,
 608                                                    ALL_REV_FLAGS);
 609                 }
 610         }
 611
 612         free(copy);
 613         release_revisions(&revs);
 614         return negative > 0 && positive > 0;
 615 }