range-diff.c

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