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 *dummy, const struct patch_util *a,
 228                           const struct patch_util *b, const char *keydata)
 229 {
 230         return strcmp(a->diff, keydata ? keydata : b->diff);
 231 }
 232
 233 static void find_exact_matches(struct string_list *a, struct string_list *b)
 234 {
 235         struct hashmap map = HASHMAP_INIT((hashmap_cmp_fn)patch_util_cmp, NULL);
 236         int i;
 237
 238         /* First, add the patches of a to a hash map */
 239         for (i = 0; i < a->nr; i++) {
 240                 struct patch_util *util = a->items[i].util;
 241
 242                 util->i = i;
 243                 util->patch = a->items[i].string;
 244                 util->diff = util->patch + util->diff_offset;
 245                 hashmap_entry_init(&util->e, strhash(util->diff));
 246                 hashmap_add(&map, &util->e);
 247         }
 248
 249         /* Now try to find exact matches in b */
 250         for (i = 0; i < b->nr; i++) {
 251                 struct patch_util *util = b->items[i].util, *other;
 252
 253                 util->i = i;
 254                 util->patch = b->items[i].string;
 255                 util->diff = util->patch + util->diff_offset;
 256                 hashmap_entry_init(&util->e, strhash(util->diff));
 257                 other = hashmap_remove_entry(&map, util, e, NULL);
 258                 if (other) {
 259                         if (other->matching >= 0)
 260                                 BUG("already assigned!");
 261
 262                         other->matching = i;
 263                         util->matching = other->i;
 264                 }
 265         }
 266
 267         hashmap_clear(&map);
 268 }
 269
 270 static int diffsize_consume(void *data, char *line, unsigned long len)
 271 {
 272         (*(int *)data)++;
 273         return 0;
 274 }
 275
 276 static void diffsize_hunk(void *data, long ob, long on, long nb, long nn,
 277                           const char *funcline, long funclen)
 278 {
 279         diffsize_consume(data, NULL, 0);
 280 }
 281
 282 static int diffsize(const char *a, const char *b)
 283 {
 284         xpparam_t pp = { 0 };
 285         xdemitconf_t cfg = { 0 };
 286         mmfile_t mf1, mf2;
 287         int count = 0;
 288
 289         mf1.ptr = (char *)a;
 290         mf1.size = strlen(a);
 291         mf2.ptr = (char *)b;
 292         mf2.size = strlen(b);
 293
 294         cfg.ctxlen = 3;
 295         if (!xdi_diff_outf(&mf1, &mf2,
 296                            diffsize_hunk, diffsize_consume, &count,
 297                            &pp, &cfg))
 298                 return count;
 299
 300         error(_("failed to generate diff"));
 301         return COST_MAX;
 302 }
 303
 304 static void get_correspondences(struct string_list *a, struct string_list *b,
 305                                 int creation_factor)
 306 {
 307         int n = a->nr + b->nr;
 308         int *cost, c, *a2b, *b2a;
 309         int i, j;
 310
 311         ALLOC_ARRAY(cost, st_mult(n, n));
 312         ALLOC_ARRAY(a2b, n);
 313         ALLOC_ARRAY(b2a, n);
 314
 315         for (i = 0; i < a->nr; i++) {
 316                 struct patch_util *a_util = a->items[i].util;
 317
 318                 for (j = 0; j < b->nr; j++) {
 319                         struct patch_util *b_util = b->items[j].util;
 320
 321                         if (a_util->matching == j)
 322                                 c = 0;
 323                         else if (a_util->matching < 0 && b_util->matching < 0)
 324                                 c = diffsize(a_util->diff, b_util->diff);
 325                         else
 326                                 c = COST_MAX;
 327                         cost[i + n * j] = c;
 328                 }
 329
 330                 c = a_util->matching < 0 ?
 331                         a_util->diffsize * creation_factor / 100 : COST_MAX;
 332                 for (j = b->nr; j < n; j++)
 333                         cost[i + n * j] = c;
 334         }
 335
 336         for (j = 0; j < b->nr; j++) {
 337                 struct patch_util *util = b->items[j].util;
 338
 339                 c = util->matching < 0 ?
 340                         util->diffsize * creation_factor / 100 : COST_MAX;
 341                 for (i = a->nr; i < n; i++)
 342                         cost[i + n * j] = c;
 343         }
 344
 345         for (i = a->nr; i < n; i++)
 346                 for (j = b->nr; j < n; j++)
 347                         cost[i + n * j] = 0;
 348
 349         compute_assignment(n, n, cost, a2b, b2a);
 350
 351         for (i = 0; i < a->nr; i++)
 352                 if (a2b[i] >= 0 && a2b[i] < b->nr) {
 353                         struct patch_util *a_util = a->items[i].util;
 354                         struct patch_util *b_util = b->items[a2b[i]].util;
 355
 356                         a_util->matching = a2b[i];
 357                         b_util->matching = i;
 358                 }
 359
 360         free(cost);
 361         free(a2b);
 362         free(b2a);
 363 }
 364
 365 static void output_pair_header(struct diff_options *diffopt,
 366                                int patch_no_width,
 367                                struct strbuf *buf,
 368                                struct strbuf *dashes,
 369                                struct patch_util *a_util,
 370                                struct patch_util *b_util)
 371 {
 372         struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
 373         struct commit *commit;
 374         char status;
 375         const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
 376         const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
 377         const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
 378         const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
 379         const char *color;
 380
 381         if (!dashes->len)
 382                 strbuf_addchars(dashes, '-',
 383                                 strlen(find_unique_abbrev(oid,
 384                                                           DEFAULT_ABBREV)));
 385
 386         if (!b_util) {
 387                 color = color_old;
 388                 status = '<';
 389         } else if (!a_util) {
 390                 color = color_new;
 391                 status = '>';
 392         } else if (strcmp(a_util->patch, b_util->patch)) {
 393                 color = color_commit;
 394                 status = '!';
 395         } else {
 396                 color = color_commit;
 397                 status = '=';
 398         }
 399
 400         strbuf_reset(buf);
 401         strbuf_addstr(buf, status == '!' ? color_old : color);
 402         if (!a_util)
 403                 strbuf_addf(buf, "%*s:  %s ", patch_no_width, "-", dashes->buf);
 404         else
 405                 strbuf_addf(buf, "%*d:  %s ", patch_no_width, a_util->i + 1,
 406                             find_unique_abbrev(&a_util->oid, DEFAULT_ABBREV));
 407
 408         if (status == '!')
 409                 strbuf_addf(buf, "%s%s", color_reset, color);
 410         strbuf_addch(buf, status);
 411         if (status == '!')
 412                 strbuf_addf(buf, "%s%s", color_reset, color_new);
 413
 414         if (!b_util)
 415                 strbuf_addf(buf, " %*s:  %s", patch_no_width, "-", dashes->buf);
 416         else
 417                 strbuf_addf(buf, " %*d:  %s", patch_no_width, b_util->i + 1,
 418                             find_unique_abbrev(&b_util->oid, DEFAULT_ABBREV));
 419
 420         commit = lookup_commit_reference(the_repository, oid);
 421         if (commit) {
 422                 if (status == '!')
 423                         strbuf_addf(buf, "%s%s", color_reset, color);
 424
 425                 strbuf_addch(buf, ' ');
 426                 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
 427         }
 428         strbuf_addf(buf, "%s\n", color_reset);
 429
 430         fwrite(buf->buf, buf->len, 1, diffopt->file);
 431 }
 432
 433 static struct userdiff_driver section_headers = {
 434         .funcname = { "^ ## (.*) ##$\n"
 435                       "^.?@@ (.*)$", REG_EXTENDED }
 436 };
 437
 438 static struct diff_filespec *get_filespec(const char *name, const char *p)
 439 {
 440         struct diff_filespec *spec = alloc_filespec(name);
 441
 442         fill_filespec(spec, null_oid(), 0, 0100644);
 443         spec->data = (char *)p;
 444         spec->size = strlen(p);
 445         spec->should_munmap = 0;
 446         spec->is_stdin = 1;
 447         spec->driver = &section_headers;
 448
 449         return spec;
 450 }
 451
 452 static void patch_diff(const char *a, const char *b,
 453                        struct diff_options *diffopt)
 454 {
 455         diff_queue(&diff_queued_diff,
 456                    get_filespec("a", a), get_filespec("b", b));
 457
 458         diffcore_std(diffopt);
 459         diff_flush(diffopt);
 460 }
 461
 462 static struct strbuf *output_prefix_cb(struct diff_options *opt, void *data)
 463 {
 464         return data;
 465 }
 466
 467 static void output(struct string_list *a, struct string_list *b,
 468                    struct range_diff_options *range_diff_opts)
 469 {
 470         struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
 471         int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
 472         int i = 0, j = 0;
 473         struct diff_options opts;
 474         struct strbuf indent = STRBUF_INIT;
 475
 476         if (range_diff_opts->diffopt)
 477                 memcpy(&opts, range_diff_opts->diffopt, sizeof(opts));
 478         else
 479                 diff_setup(&opts);
 480
 481         opts.no_free = 1;
 482         if (!opts.output_format)
 483                 opts.output_format = DIFF_FORMAT_PATCH;
 484         opts.flags.suppress_diff_headers = 1;
 485         opts.flags.dual_color_diffed_diffs =
 486                 range_diff_opts->dual_color;
 487         opts.flags.suppress_hunk_header_line_count = 1;
 488         opts.output_prefix = output_prefix_cb;
 489         strbuf_addstr(&indent, "    ");
 490         opts.output_prefix_data = &indent;
 491         diff_setup_done(&opts);
 492
 493         /*
 494          * We assume the user is really more interested in the second argument
 495          * ("newer" version). To that end, we print the output in the order of
 496          * the RHS (the `b` parameter). To put the LHS (the `a` parameter)
 497          * commits that are no longer in the RHS into a good place, we place
 498          * them once we have shown all of their predecessors in the LHS.
 499          */
 500
 501         while (i < a->nr || j < b->nr) {
 502                 struct patch_util *a_util, *b_util;
 503                 a_util = i < a->nr ? a->items[i].util : NULL;
 504                 b_util = j < b->nr ? b->items[j].util : NULL;
 505
 506                 /* Skip all the already-shown commits from the LHS. */
 507                 while (i < a->nr && a_util->shown)
 508                         a_util = ++i < a->nr ? a->items[i].util : NULL;
 509
 510                 /* Show unmatched LHS commit whose predecessors were shown. */
 511                 if (i < a->nr && a_util->matching < 0) {
 512                         if (!range_diff_opts->right_only)
 513                                 output_pair_header(&opts, patch_no_width,
 514                                            &buf, &dashes, a_util, NULL);
 515                         i++;
 516                         continue;
 517                 }
 518
 519                 /* Show unmatched RHS commits. */
 520                 while (j < b->nr && b_util->matching < 0) {
 521                         if (!range_diff_opts->left_only)
 522                                 output_pair_header(&opts, patch_no_width,
 523                                            &buf, &dashes, NULL, b_util);
 524                         b_util = ++j < b->nr ? b->items[j].util : NULL;
 525                 }
 526
 527                 /* Show matching LHS/RHS pair. */
 528                 if (j < b->nr) {
 529                         a_util = a->items[b_util->matching].util;
 530                         output_pair_header(&opts, patch_no_width,
 531                                            &buf, &dashes, a_util, b_util);
 532                         if (!(opts.output_format & DIFF_FORMAT_NO_OUTPUT))
 533                                 patch_diff(a->items[b_util->matching].string,
 534                                            b->items[j].string, &opts);
 535                         a_util->shown = 1;
 536                         j++;
 537                 }
 538         }
 539         strbuf_release(&buf);
 540         strbuf_release(&dashes);
 541         strbuf_release(&indent);
 542         opts.no_free = 0;
 543         diff_free(&opts);
 544 }
 545
 546 int show_range_diff(const char *range1, const char *range2,
 547                     struct range_diff_options *range_diff_opts)
 548 {
 549         int res = 0;
 550
 551         struct string_list branch1 = STRING_LIST_INIT_DUP;
 552         struct string_list branch2 = STRING_LIST_INIT_DUP;
 553
 554         if (range_diff_opts->left_only && range_diff_opts->right_only)
 555                 res = error(_("options '%s' and '%s' cannot be used together"), "--left-only", "--right-only");
 556
 557         if (!res && read_patches(range1, &branch1, range_diff_opts->other_arg))
 558                 res = error(_("could not parse log for '%s'"), range1);
 559         if (!res && read_patches(range2, &branch2, range_diff_opts->other_arg))
 560                 res = error(_("could not parse log for '%s'"), range2);
 561
 562         if (!res) {
 563                 find_exact_matches(&branch1, &branch2);
 564                 get_correspondences(&branch1, &branch2,
 565                                     range_diff_opts->creation_factor);
 566                 output(&branch1, &branch2, range_diff_opts);
 567         }
 568
 569         string_list_clear(&branch1, 1);
 570         string_list_clear(&branch2, 1);
 571
 572         return res;
 573 }
 574
 575 int is_range_diff_range(const char *arg)
 576 {
 577         char *copy = xstrdup(arg); /* setup_revisions() modifies it */
 578         const char *argv[] = { "", copy, "--", NULL };
 579         int i, positive = 0, negative = 0;
 580         struct rev_info revs;
 581
 582         init_revisions(&revs, NULL);
 583         if (setup_revisions(3, argv, &revs, NULL) == 1) {
 584                 for (i = 0; i < revs.pending.nr; i++)
 585                         if (revs.pending.objects[i].item->flags & UNINTERESTING)
 586                                 negative++;
 587                         else
 588                                 positive++;
 589                 for (i = 0; i < revs.pending.nr; i++) {
 590                         struct object *obj = revs.pending.objects[i].item;
 591
 592                         if (obj->type == OBJ_COMMIT)
 593                                 clear_commit_marks((struct commit *)obj,
 594                                                    ALL_REV_FLAGS);
 595                 }
 596         }
 597
 598         free(copy);
 599         release_revisions(&revs);
 600         return negative > 0 && positive > 0;
 601 }