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