range-diff.c

   1 #define USE_THE_REPOSITORY_VARIABLE
   2
   3 #include "git-compat-util.h"
   4 #include "environment.h"
   5 #include "gettext.h"
   6 #include "range-diff.h"
   7 #include "object-name.h"
   8 #include "string-list.h"
   9 #include "run-command.h"
  10 #include "strvec.h"
  11 #include "hashmap.h"
  12 #include "xdiff-interface.h"
  13 #include "linear-assignment.h"
  14 #include "diffcore.h"
  15 #include "commit.h"
  16 #include "pager.h"
  17 #include "pretty.h"
  18 #include "repository.h"
  19 #include "userdiff.h"
  20 #include "apply.h"
  21 #include "revision.h"
  22
  23 struct patch_util {
  24         /* For the search for an exact match */
  25         struct hashmap_entry e;
  26         const char *diff, *patch;
  27
  28         int i, shown;
  29         int diffsize;
  30         size_t diff_offset;
  31         /* the index of the matching item in the other branch, or -1 */
  32         int matching;
  33         struct object_id oid;
  34 };
  35
  36 /*
  37  * Reads the patches into a string list, with the `util` field being populated
  38  * as struct object_id (will need to be free()d).
  39  */
  40 static int read_patches(const char *range, struct string_list *list,
  41                         const struct strvec *other_arg)
  42 {
  43         struct child_process cp = CHILD_PROCESS_INIT;
  44         struct strbuf buf = STRBUF_INIT, contents = STRBUF_INIT;
  45         struct patch_util *util = NULL;
  46         int in_header = 1;
  47         char *line, *current_filename = NULL;
  48         ssize_t len;
  49         size_t size;
  50         int ret = -1;
  51
  52         strvec_pushl(&cp.args, "log", "--no-color", "-p", "--no-merges",
  53                      "--reverse", "--date-order", "--decorate=no",
  54                      "--no-prefix", "--submodule=short",
  55                      /*
  56                       * Choose indicators that are not used anywhere
  57                       * else in diffs, but still look reasonable
  58                       * (e.g. will not be confusing when debugging)
  59                       */
  60                      "--output-indicator-new=>",
  61                      "--output-indicator-old=<",
  62                      "--output-indicator-context=#",
  63                      "--no-abbrev-commit",
  64                      "--pretty=medium",
  65                      "--show-notes-by-default",
  66                      NULL);
  67         strvec_push(&cp.args, range);
  68         if (other_arg)
  69                 strvec_pushv(&cp.args, other_arg->v);
  70         cp.out = -1;
  71         cp.no_stdin = 1;
  72         cp.git_cmd = 1;
  73
  74         if (start_command(&cp))
  75                 return error_errno(_("could not start `log`"));
  76         if (strbuf_read(&contents, cp.out, 0) < 0) {
  77                 error_errno(_("could not read `log` output"));
  78                 finish_command(&cp);
  79                 goto cleanup;
  80         }
  81         if (finish_command(&cp))
  82                 goto cleanup;
  83
  84         line = contents.buf;
  85         size = contents.len;
  86         for (; size > 0; size -= len, line += len) {
  87                 const char *p;
  88                 char *eol;
  89
  90                 eol = memchr(line, '\n', size);
  91                 if (eol) {
  92                         *eol = '\0';
  93                         len = eol + 1 - line;
  94                 } else {
  95                         len = size;
  96                 }
  97
  98                 if (skip_prefix(line, "commit ", &p)) {
  99                         if (util) {
 100                                 string_list_append(list, buf.buf)->util = util;
 101                                 strbuf_reset(&buf);
 102                         }
 103                         CALLOC_ARRAY(util, 1);
 104                         if (repo_get_oid(the_repository, p, &util->oid)) {
 105                                 error(_("could not parse commit '%s'"), p);
 106                                 FREE_AND_NULL(util);
 107                                 string_list_clear(list, 1);
 108                                 goto cleanup;
 109                         }
 110                         util->matching = -1;
 111                         in_header = 1;
 112                         continue;
 113                 }
 114
 115                 if (!util) {
 116                         error(_("could not parse first line of `log` output: "
 117                                 "did not start with 'commit ': '%s'"),
 118                               line);
 119                         string_list_clear(list, 1);
 120                         goto cleanup;
 121                 }
 122
 123                 if (starts_with(line, "diff --git")) {
 124                         struct patch patch = { 0 };
 125                         struct strbuf root = STRBUF_INIT;
 126                         int linenr = 0;
 127                         int orig_len;
 128
 129                         in_header = 0;
 130                         strbuf_addch(&buf, '\n');
 131                         if (!util->diff_offset)
 132                                 util->diff_offset = buf.len;
 133                         if (eol)
 134                                 *eol = '\n';
 135                         orig_len = len;
 136                         len = parse_git_diff_header(&root, &linenr, 0, line,
 137                                                     len, size, &patch);
 138                         if (len < 0) {
 139                                 error(_("could not parse git header '%.*s'"),
 140                                       orig_len, line);
 141                                 FREE_AND_NULL(util);
 142                                 string_list_clear(list, 1);
 143                                 goto cleanup;
 144                         }
 145                         strbuf_addstr(&buf, " ## ");
 146                         if (patch.is_new > 0)
 147                                 strbuf_addf(&buf, "%s (new)", patch.new_name);
 148                         else if (patch.is_delete > 0)
 149                                 strbuf_addf(&buf, "%s (deleted)", patch.old_name);
 150                         else if (patch.is_rename)
 151                                 strbuf_addf(&buf, "%s => %s", patch.old_name, patch.new_name);
 152                         else
 153                                 strbuf_addstr(&buf, patch.new_name);
 154
 155                         free(current_filename);
 156                         if (patch.is_delete > 0)
 157                                 current_filename = xstrdup(patch.old_name);
 158                         else
 159                                 current_filename = xstrdup(patch.new_name);
 160
 161                         if (patch.new_mode && patch.old_mode &&
 162                             patch.old_mode != patch.new_mode)
 163                                 strbuf_addf(&buf, " (mode change %06o => %06o)",
 164                                             patch.old_mode, patch.new_mode);
 165
 166                         strbuf_addstr(&buf, " ##");
 167                         release_patch(&patch);
 168                 } else if (in_header) {
 169                         if (starts_with(line, "Author: ")) {
 170                                 strbuf_addstr(&buf, " ## Metadata ##\n");
 171                                 strbuf_addstr(&buf, line);
 172                                 strbuf_addstr(&buf, "\n\n");
 173                                 strbuf_addstr(&buf, " ## Commit message ##\n");
 174                         } else if (starts_with(line, "Notes") &&
 175                                    line[strlen(line) - 1] == ':') {
 176                                 strbuf_addstr(&buf, "\n\n");
 177                                 /* strip the trailing colon */
 178                                 strbuf_addf(&buf, " ## %.*s ##\n",
 179                                             (int)(strlen(line) - 1), line);
 180                         } else if (starts_with(line, "    ")) {
 181                                 p = line + len - 2;
 182                                 while (isspace(*p) && p >= line)
 183                                         p--;
 184                                 strbuf_add(&buf, line, p - line + 1);
 185                                 strbuf_addch(&buf, '\n');
 186                         }
 187                         continue;
 188                 } else if (skip_prefix(line, "@@ ", &p)) {
 189                         p = strstr(p, "@@");
 190                         strbuf_addstr(&buf, "@@");
 191                         if (current_filename && p[2])
 192                                 strbuf_addf(&buf, " %s:", current_filename);
 193                         if (p)
 194                                 strbuf_addstr(&buf, p + 2);
 195                 } else if (!line[0])
 196                         /*
 197                          * A completely blank (not ' \n', which is context)
 198                          * line is not valid in a diff.  We skip it
 199                          * silently, because this neatly handles the blank
 200                          * separator line between commits in git-log
 201                          * output.
 202                          */
 203                         continue;
 204                 else if (line[0] == '>') {
 205                         strbuf_addch(&buf, '+');
 206                         strbuf_addstr(&buf, line + 1);
 207                 } else if (line[0] == '<') {
 208                         strbuf_addch(&buf, '-');
 209                         strbuf_addstr(&buf, line + 1);
 210                 } else if (line[0] == '#') {
 211                         strbuf_addch(&buf, ' ');
 212                         strbuf_addstr(&buf, line + 1);
 213                 } else {
 214                         strbuf_addch(&buf, ' ');
 215                         strbuf_addstr(&buf, line);
 216                 }
 217
 218                 strbuf_addch(&buf, '\n');
 219                 util->diffsize++;
 220         }
 221
 222         ret = 0;
 223 cleanup:
 224         strbuf_release(&contents);
 225
 226         if (util)
 227                 string_list_append(list, buf.buf)->util = util;
 228         strbuf_release(&buf);
 229         free(current_filename);
 230
 231         return ret;
 232 }
 233
 234 static int patch_util_cmp(const void *cmp_data UNUSED,
 235                           const struct hashmap_entry *ha,
 236                           const struct hashmap_entry *hb,
 237                           const void *keydata)
 238 {
 239         const struct patch_util
 240                 *a = container_of(ha, const struct patch_util, e),
 241                 *b = container_of(hb, const struct patch_util, e);
 242         return strcmp(a->diff, keydata ? keydata : b->diff);
 243 }
 244
 245 static void find_exact_matches(struct string_list *a, struct string_list *b)
 246 {
 247         struct hashmap map = HASHMAP_INIT(patch_util_cmp, NULL);
 248         int i;
 249
 250         /* First, add the patches of a to a hash map */
 251         for (i = 0; i < a->nr; i++) {
 252                 struct patch_util *util = a->items[i].util;
 253
 254                 util->i = i;
 255                 util->patch = a->items[i].string;
 256                 util->diff = util->patch + util->diff_offset;
 257                 hashmap_entry_init(&util->e, strhash(util->diff));
 258                 hashmap_add(&map, &util->e);
 259         }
 260
 261         /* Now try to find exact matches in b */
 262         for (i = 0; i < b->nr; i++) {
 263                 struct patch_util *util = b->items[i].util, *other;
 264
 265                 util->i = i;
 266                 util->patch = b->items[i].string;
 267                 util->diff = util->patch + util->diff_offset;
 268                 hashmap_entry_init(&util->e, strhash(util->diff));
 269                 other = hashmap_remove_entry(&map, util, e, NULL);
 270                 if (other) {
 271                         if (other->matching >= 0)
 272                                 BUG("already assigned!");
 273
 274                         other->matching = i;
 275                         util->matching = other->i;
 276                 }
 277         }
 278
 279         hashmap_clear(&map);
 280 }
 281
 282 static int diffsize_consume(void *data,
 283                              char *line UNUSED,
 284                              unsigned long len UNUSED)
 285 {
 286         (*(int *)data)++;
 287         return 0;
 288 }
 289
 290 static void diffsize_hunk(void *data,
 291                           long ob UNUSED, long on UNUSED,
 292                           long nb UNUSED, long nn UNUSED,
 293                           const char *func UNUSED, long funclen UNUSED)
 294 {
 295         diffsize_consume(data, NULL, 0);
 296 }
 297
 298 static int diffsize(const char *a, const char *b)
 299 {
 300         xpparam_t pp = { 0 };
 301         xdemitconf_t cfg = { 0 };
 302         mmfile_t mf1, mf2;
 303         int count = 0;
 304
 305         mf1.ptr = (char *)a;
 306         mf1.size = strlen(a);
 307         mf2.ptr = (char *)b;
 308         mf2.size = strlen(b);
 309
 310         cfg.ctxlen = 3;
 311         if (!xdi_diff_outf(&mf1, &mf2,
 312                            diffsize_hunk, diffsize_consume, &count,
 313                            &pp, &cfg))
 314                 return count;
 315
 316         error(_("failed to generate diff"));
 317         return COST_MAX;
 318 }
 319
 320 static void get_correspondences(struct string_list *a, struct string_list *b,
 321                                 int creation_factor)
 322 {
 323         int n = a->nr + b->nr;
 324         int *cost, c, *a2b, *b2a;
 325         int i, j;
 326
 327         ALLOC_ARRAY(cost, st_mult(n, n));
 328         ALLOC_ARRAY(a2b, n);
 329         ALLOC_ARRAY(b2a, n);
 330
 331         for (i = 0; i < a->nr; i++) {
 332                 struct patch_util *a_util = a->items[i].util;
 333
 334                 for (j = 0; j < b->nr; j++) {
 335                         struct patch_util *b_util = b->items[j].util;
 336
 337                         if (a_util->matching == j)
 338                                 c = 0;
 339                         else if (a_util->matching < 0 && b_util->matching < 0)
 340                                 c = diffsize(a_util->diff, b_util->diff);
 341                         else
 342                                 c = COST_MAX;
 343                         cost[i + n * j] = c;
 344                 }
 345
 346                 c = a_util->matching < 0 ?
 347                         a_util->diffsize * creation_factor / 100 : COST_MAX;
 348                 for (j = b->nr; j < n; j++)
 349                         cost[i + n * j] = c;
 350         }
 351
 352         for (j = 0; j < b->nr; j++) {
 353                 struct patch_util *util = b->items[j].util;
 354
 355                 c = util->matching < 0 ?
 356                         util->diffsize * creation_factor / 100 : COST_MAX;
 357                 for (i = a->nr; i < n; i++)
 358                         cost[i + n * j] = c;
 359         }
 360
 361         for (i = a->nr; i < n; i++)
 362                 for (j = b->nr; j < n; j++)
 363                         cost[i + n * j] = 0;
 364
 365         compute_assignment(n, n, cost, a2b, b2a);
 366
 367         for (i = 0; i < a->nr; i++)
 368                 if (a2b[i] >= 0 && a2b[i] < b->nr) {
 369                         struct patch_util *a_util = a->items[i].util;
 370                         struct patch_util *b_util = b->items[a2b[i]].util;
 371
 372                         a_util->matching = a2b[i];
 373                         b_util->matching = i;
 374                 }
 375
 376         free(cost);
 377         free(a2b);
 378         free(b2a);
 379 }
 380
 381 static void output_pair_header(struct diff_options *diffopt,
 382                                int patch_no_width,
 383                                struct strbuf *buf,
 384                                struct strbuf *dashes,
 385                                struct patch_util *a_util,
 386                                struct patch_util *b_util)
 387 {
 388         struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
 389         struct commit *commit;
 390         char status;
 391         const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
 392         const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
 393         const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
 394         const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
 395         const char *color;
 396         int abbrev = diffopt->abbrev;
 397
 398         if (abbrev < 0)
 399                 abbrev = DEFAULT_ABBREV;
 400
 401         if (!dashes->len)
 402                 strbuf_addchars(dashes, '-',
 403                                 strlen(repo_find_unique_abbrev(the_repository, oid, abbrev)));
 404
 405         if (!b_util) {
 406                 color = color_old;
 407                 status = '<';
 408         } else if (!a_util) {
 409                 color = color_new;
 410                 status = '>';
 411         } else if (strcmp(a_util->patch, b_util->patch)) {
 412                 color = color_commit;
 413                 status = '!';
 414         } else {
 415                 color = color_commit;
 416                 status = '=';
 417         }
 418
 419         strbuf_reset(buf);
 420         strbuf_addstr(buf, status == '!' ? color_old : color);
 421         if (!a_util)
 422                 strbuf_addf(buf, "%*s:  %s ", patch_no_width, "-", dashes->buf);
 423         else
 424                 strbuf_addf(buf, "%*d:  %s ", patch_no_width, a_util->i + 1,
 425                             repo_find_unique_abbrev(the_repository, &a_util->oid, abbrev));
 426
 427         if (status == '!')
 428                 strbuf_addf(buf, "%s%s", color_reset, color);
 429         strbuf_addch(buf, status);
 430         if (status == '!')
 431                 strbuf_addf(buf, "%s%s", color_reset, color_new);
 432
 433         if (!b_util)
 434                 strbuf_addf(buf, " %*s:  %s", patch_no_width, "-", dashes->buf);
 435         else
 436                 strbuf_addf(buf, " %*d:  %s", patch_no_width, b_util->i + 1,
 437                             repo_find_unique_abbrev(the_repository, &b_util->oid, abbrev));
 438
 439         commit = lookup_commit_reference(the_repository, oid);
 440         if (commit) {
 441                 if (status == '!')
 442                         strbuf_addf(buf, "%s%s", color_reset, color);
 443
 444                 strbuf_addch(buf, ' ');
 445                 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
 446         }
 447         strbuf_addf(buf, "%s\n", color_reset);
 448
 449         fwrite(buf->buf, buf->len, 1, diffopt->file);
 450 }
 451
 452 static struct userdiff_driver section_headers = {
 453         .funcname = { "^ ## (.*) ##$\n"
 454                       "^.?@@ (.*)$", REG_EXTENDED }
 455 };
 456
 457 static struct diff_filespec *get_filespec(const char *name, const char *p)
 458 {
 459         struct diff_filespec *spec = alloc_filespec(name);
 460
 461         fill_filespec(spec, null_oid(), 0, 0100644);
 462         spec->data = (char *)p;
 463         spec->size = strlen(p);
 464         spec->should_munmap = 0;
 465         spec->is_stdin = 1;
 466         spec->driver = &section_headers;
 467
 468         return spec;
 469 }
 470
 471 static void patch_diff(const char *a, const char *b,
 472                        struct diff_options *diffopt)
 473 {
 474         diff_queue(&diff_queued_diff,
 475                    get_filespec("a", a), get_filespec("b", b));
 476
 477         diffcore_std(diffopt);
 478         diff_flush(diffopt);
 479 }
 480
 481 static struct strbuf *output_prefix_cb(struct diff_options *opt UNUSED, void *data)
 482 {
 483         return data;
 484 }
 485
 486 static void output(struct string_list *a, struct string_list *b,
 487                    struct range_diff_options *range_diff_opts)
 488 {
 489         struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
 490         int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
 491         int i = 0, j = 0;
 492         struct diff_options opts;
 493         struct strbuf indent = STRBUF_INIT;
 494
 495         if (range_diff_opts->diffopt)
 496                 memcpy(&opts, range_diff_opts->diffopt, sizeof(opts));
 497         else
 498                 repo_diff_setup(the_repository, &opts);
 499
 500         opts.no_free = 1;
 501         if (!opts.output_format)
 502                 opts.output_format = DIFF_FORMAT_PATCH;
 503         opts.flags.suppress_diff_headers = 1;
 504         opts.flags.dual_color_diffed_diffs =
 505                 range_diff_opts->dual_color;
 506         opts.flags.suppress_hunk_header_line_count = 1;
 507         opts.output_prefix = output_prefix_cb;
 508         strbuf_addstr(&indent, "    ");
 509         opts.output_prefix_data = &indent;
 510         diff_setup_done(&opts);
 511
 512         /*
 513          * We assume the user is really more interested in the second argument
 514          * ("newer" version). To that end, we print the output in the order of
 515          * the RHS (the `b` parameter). To put the LHS (the `a` parameter)
 516          * commits that are no longer in the RHS into a good place, we place
 517          * them once we have shown all of their predecessors in the LHS.
 518          */
 519
 520         while (i < a->nr || j < b->nr) {
 521                 struct patch_util *a_util, *b_util;
 522                 a_util = i < a->nr ? a->items[i].util : NULL;
 523                 b_util = j < b->nr ? b->items[j].util : NULL;
 524
 525                 /* Skip all the already-shown commits from the LHS. */
 526                 while (i < a->nr && a_util->shown)
 527                         a_util = ++i < a->nr ? a->items[i].util : NULL;
 528
 529                 /* Show unmatched LHS commit whose predecessors were shown. */
 530                 if (i < a->nr && a_util->matching < 0) {
 531                         if (!range_diff_opts->right_only)
 532                                 output_pair_header(&opts, patch_no_width,
 533                                            &buf, &dashes, a_util, NULL);
 534                         i++;
 535                         continue;
 536                 }
 537
 538                 /* Show unmatched RHS commits. */
 539                 while (j < b->nr && b_util->matching < 0) {
 540                         if (!range_diff_opts->left_only)
 541                                 output_pair_header(&opts, patch_no_width,
 542                                            &buf, &dashes, NULL, b_util);
 543                         b_util = ++j < b->nr ? b->items[j].util : NULL;
 544                 }
 545
 546                 /* Show matching LHS/RHS pair. */
 547                 if (j < b->nr) {
 548                         a_util = a->items[b_util->matching].util;
 549                         output_pair_header(&opts, patch_no_width,
 550                                            &buf, &dashes, a_util, b_util);
 551                         if (!(opts.output_format & DIFF_FORMAT_NO_OUTPUT))
 552                                 patch_diff(a->items[b_util->matching].string,
 553                                            b->items[j].string, &opts);
 554                         a_util->shown = 1;
 555                         j++;
 556                 }
 557         }
 558         strbuf_release(&buf);
 559         strbuf_release(&dashes);
 560         strbuf_release(&indent);
 561         opts.no_free = 0;
 562         diff_free(&opts);
 563 }
 564
 565 int show_range_diff(const char *range1, const char *range2,
 566                     struct range_diff_options *range_diff_opts)
 567 {
 568         int res = 0;
 569
 570         struct string_list branch1 = STRING_LIST_INIT_DUP;
 571         struct string_list branch2 = STRING_LIST_INIT_DUP;
 572
 573         if (range_diff_opts->left_only && range_diff_opts->right_only)
 574                 res = error(_("options '%s' and '%s' cannot be used together"), "--left-only", "--right-only");
 575
 576         if (!res && read_patches(range1, &branch1, range_diff_opts->other_arg))
 577                 res = error(_("could not parse log for '%s'"), range1);
 578         if (!res && read_patches(range2, &branch2, range_diff_opts->other_arg))
 579                 res = error(_("could not parse log for '%s'"), range2);
 580
 581         if (!res) {
 582                 find_exact_matches(&branch1, &branch2);
 583                 get_correspondences(&branch1, &branch2,
 584                                     range_diff_opts->creation_factor);
 585                 output(&branch1, &branch2, range_diff_opts);
 586         }
 587
 588         string_list_clear(&branch1, 1);
 589         string_list_clear(&branch2, 1);
 590
 591         return res;
 592 }
 593
 594 int is_range_diff_range(const char *arg)
 595 {
 596         char *copy = xstrdup(arg); /* setup_revisions() modifies it */
 597         const char *argv[] = { "", copy, "--", NULL };
 598         int i, positive = 0, negative = 0;
 599         struct rev_info revs;
 600
 601         repo_init_revisions(the_repository, &revs, NULL);
 602         if (setup_revisions(3, argv, &revs, NULL) == 1) {
 603                 for (i = 0; i < revs.pending.nr; i++)
 604                         if (revs.pending.objects[i].item->flags & UNINTERESTING)
 605                                 negative++;
 606                         else
 607                                 positive++;
 608                 for (i = 0; i < revs.pending.nr; i++) {
 609                         struct object *obj = revs.pending.objects[i].item;
 610
 611                         if (obj->type == OBJ_COMMIT)
 612                                 clear_commit_marks((struct commit *)obj,
 613                                                    ALL_REV_FLAGS);
 614                 }
 615         }
 616
 617         free(copy);
 618         release_revisions(&revs);
 619         return negative > 0 && positive > 0;
 620 }