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