range-diff.c

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