range-diff.c

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