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
  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 strvec *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         strvec_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                 strvec_pushv(&cp.args, other_arg->v);
  71         strvec_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         if (finish_command(&cp))
  84                 return -1;
  85
  86         line = contents.buf;
  87         size = contents.len;
  88         for (offset = 0; size > 0; offset += len, size -= len, line += len) {
  89                 const char *p;
  90
  91                 len = find_end_of_line(line, size);
  92                 line[len - 1] = '\0';
  93                 if (skip_prefix(line, "commit ", &p)) {
  94                         if (util) {
  95                                 string_list_append(list, buf.buf)->util = util;
  96                                 strbuf_reset(&buf);
  97                         }
  98                         util = xcalloc(sizeof(*util), 1);
  99                         if (get_oid(p, &util->oid)) {
 100                                 error(_("could not parse commit '%s'"), p);
 101                                 free(util);
 102                                 free(current_filename);
 103                                 string_list_clear(list, 1);
 104                                 strbuf_release(&buf);
 105                                 strbuf_release(&contents);
 106                                 return -1;
 107                         }
 108                         util->matching = -1;
 109                         in_header = 1;
 110                         continue;
 111                 }
 112
 113                 if (!util) {
 114                         error(_("could not parse first line of `log` output: "
 115                                 "did not start with 'commit ': '%s'"),
 116                               line);
 117                         free(current_filename);
 118                         string_list_clear(list, 1);
 119                         strbuf_release(&buf);
 120                         strbuf_release(&contents);
 121                         return -1;
 122                 }
 123
 124                 if (starts_with(line, "diff --git")) {
 125                         struct patch patch = { 0 };
 126                         struct strbuf root = STRBUF_INIT;
 127                         int linenr = 0;
 128                         int orig_len;
 129
 130                         in_header = 0;
 131                         strbuf_addch(&buf, '\n');
 132                         if (!util->diff_offset)
 133                                 util->diff_offset = buf.len;
 134                         line[len - 1] = '\n';
 135                         orig_len = len;
 136                         len = parse_git_diff_header(&root, &linenr, 0, line,
 137                                                     len, size, &patch);
 138                         if (len < 0) {
 139                                 error(_("could not parse git header '%.*s'"),
 140                                       orig_len, line);
 141                                 free(util);
 142                                 free(current_filename);
 143                                 string_list_clear(list, 1);
 144                                 strbuf_release(&buf);
 145                                 strbuf_release(&contents);
 146                                 return -1;
 147                         }
 148                         strbuf_addstr(&buf, " ## ");
 149                         if (patch.is_new > 0)
 150                                 strbuf_addf(&buf, "%s (new)", patch.new_name);
 151                         else if (patch.is_delete > 0)
 152                                 strbuf_addf(&buf, "%s (deleted)", patch.old_name);
 153                         else if (patch.is_rename)
 154                                 strbuf_addf(&buf, "%s => %s", patch.old_name, patch.new_name);
 155                         else
 156                                 strbuf_addstr(&buf, patch.new_name);
 157
 158                         free(current_filename);
 159                         if (patch.is_delete > 0)
 160                                 current_filename = xstrdup(patch.old_name);
 161                         else
 162                                 current_filename = xstrdup(patch.new_name);
 163
 164                         if (patch.new_mode && patch.old_mode &&
 165                             patch.old_mode != patch.new_mode)
 166                                 strbuf_addf(&buf, " (mode change %06o => %06o)",
 167                                             patch.old_mode, patch.new_mode);
 168
 169                         strbuf_addstr(&buf, " ##");
 170                 } else if (in_header) {
 171                         if (starts_with(line, "Author: ")) {
 172                                 strbuf_addstr(&buf, " ## Metadata ##\n");
 173                                 strbuf_addstr(&buf, line);
 174                                 strbuf_addstr(&buf, "\n\n");
 175                                 strbuf_addstr(&buf, " ## Commit message ##\n");
 176                         } else if (starts_with(line, "Notes") &&
 177                                    line[strlen(line) - 1] == ':') {
 178                                 strbuf_addstr(&buf, "\n\n");
 179                                 /* strip the trailing colon */
 180                                 strbuf_addf(&buf, " ## %.*s ##\n",
 181                                             (int)(strlen(line) - 1), line);
 182                         } else if (starts_with(line, "    ")) {
 183                                 p = line + len - 2;
 184                                 while (isspace(*p) && p >= line)
 185                                         p--;
 186                                 strbuf_add(&buf, line, p - line + 1);
 187                                 strbuf_addch(&buf, '\n');
 188                         }
 189                         continue;
 190                 } else if (skip_prefix(line, "@@ ", &p)) {
 191                         p = strstr(p, "@@");
 192                         strbuf_addstr(&buf, "@@");
 193                         if (current_filename && p[2])
 194                                 strbuf_addf(&buf, " %s:", current_filename);
 195                         if (p)
 196                                 strbuf_addstr(&buf, p + 2);
 197                 } else if (!line[0])
 198                         /*
 199                          * A completely blank (not ' \n', which is context)
 200                          * line is not valid in a diff.  We skip it
 201                          * silently, because this neatly handles the blank
 202                          * separator line between commits in git-log
 203                          * output.
 204                          */
 205                         continue;
 206                 else if (line[0] == '>') {
 207                         strbuf_addch(&buf, '+');
 208                         strbuf_addstr(&buf, line + 1);
 209                 } else if (line[0] == '<') {
 210                         strbuf_addch(&buf, '-');
 211                         strbuf_addstr(&buf, line + 1);
 212                 } else if (line[0] == '#') {
 213                         strbuf_addch(&buf, ' ');
 214                         strbuf_addstr(&buf, line + 1);
 215                 } else {
 216                         strbuf_addch(&buf, ' ');
 217                         strbuf_addstr(&buf, line);
 218                 }
 219
 220                 strbuf_addch(&buf, '\n');
 221                 util->diffsize++;
 222         }
 223         strbuf_release(&contents);
 224
 225         if (util)
 226                 string_list_append(list, buf.buf)->util = util;
 227         strbuf_release(&buf);
 228         free(current_filename);
 229
 230         return 0;
 231 }
 232
 233 static int patch_util_cmp(const void *dummy, const struct patch_util *a,
 234                           const struct patch_util *b, const char *keydata)
 235 {
 236         return strcmp(a->diff, keydata ? keydata : b->diff);
 237 }
 238
 239 static void find_exact_matches(struct string_list *a, struct string_list *b)
 240 {
 241         struct hashmap map = HASHMAP_INIT((hashmap_cmp_fn)patch_util_cmp, NULL);
 242         int i;
 243
 244         /* First, add the patches of a to a hash map */
 245         for (i = 0; i < a->nr; i++) {
 246                 struct patch_util *util = a->items[i].util;
 247
 248                 util->i = i;
 249                 util->patch = a->items[i].string;
 250                 util->diff = util->patch + util->diff_offset;
 251                 hashmap_entry_init(&util->e, strhash(util->diff));
 252                 hashmap_add(&map, &util->e);
 253         }
 254
 255         /* Now try to find exact matches in b */
 256         for (i = 0; i < b->nr; i++) {
 257                 struct patch_util *util = b->items[i].util, *other;
 258
 259                 util->i = i;
 260                 util->patch = b->items[i].string;
 261                 util->diff = util->patch + util->diff_offset;
 262                 hashmap_entry_init(&util->e, strhash(util->diff));
 263                 other = hashmap_remove_entry(&map, util, e, NULL);
 264                 if (other) {
 265                         if (other->matching >= 0)
 266                                 BUG("already assigned!");
 267
 268                         other->matching = i;
 269                         util->matching = other->i;
 270                 }
 271         }
 272
 273         hashmap_clear(&map);
 274 }
 275
 276 static void diffsize_consume(void *data, char *line, unsigned long len)
 277 {
 278         (*(int *)data)++;
 279 }
 280
 281 static void diffsize_hunk(void *data, long ob, long on, long nb, long nn,
 282                           const char *funcline, long funclen)
 283 {
 284         diffsize_consume(data, NULL, 0);
 285 }
 286
 287 static int diffsize(const char *a, const char *b)
 288 {
 289         xpparam_t pp = { 0 };
 290         xdemitconf_t cfg = { 0 };
 291         mmfile_t mf1, mf2;
 292         int count = 0;
 293
 294         mf1.ptr = (char *)a;
 295         mf1.size = strlen(a);
 296         mf2.ptr = (char *)b;
 297         mf2.size = strlen(b);
 298
 299         cfg.ctxlen = 3;
 300         if (!xdi_diff_outf(&mf1, &mf2,
 301                            diffsize_hunk, diffsize_consume, &count,
 302                            &pp, &cfg))
 303                 return count;
 304
 305         error(_("failed to generate diff"));
 306         return COST_MAX;
 307 }
 308
 309 static void get_correspondences(struct string_list *a, struct string_list *b,
 310                                 int creation_factor)
 311 {
 312         int n = a->nr + b->nr;
 313         int *cost, c, *a2b, *b2a;
 314         int i, j;
 315
 316         ALLOC_ARRAY(cost, st_mult(n, n));
 317         ALLOC_ARRAY(a2b, n);
 318         ALLOC_ARRAY(b2a, n);
 319
 320         for (i = 0; i < a->nr; i++) {
 321                 struct patch_util *a_util = a->items[i].util;
 322
 323                 for (j = 0; j < b->nr; j++) {
 324                         struct patch_util *b_util = b->items[j].util;
 325
 326                         if (a_util->matching == j)
 327                                 c = 0;
 328                         else if (a_util->matching < 0 && b_util->matching < 0)
 329                                 c = diffsize(a_util->diff, b_util->diff);
 330                         else
 331                                 c = COST_MAX;
 332                         cost[i + n * j] = c;
 333                 }
 334
 335                 c = a_util->matching < 0 ?
 336                         a_util->diffsize * creation_factor / 100 : COST_MAX;
 337                 for (j = b->nr; j < n; j++)
 338                         cost[i + n * j] = c;
 339         }
 340
 341         for (j = 0; j < b->nr; j++) {
 342                 struct patch_util *util = b->items[j].util;
 343
 344                 c = util->matching < 0 ?
 345                         util->diffsize * creation_factor / 100 : COST_MAX;
 346                 for (i = a->nr; i < n; i++)
 347                         cost[i + n * j] = c;
 348         }
 349
 350         for (i = a->nr; i < n; i++)
 351                 for (j = b->nr; j < n; j++)
 352                         cost[i + n * j] = 0;
 353
 354         compute_assignment(n, n, cost, a2b, b2a);
 355
 356         for (i = 0; i < a->nr; i++)
 357                 if (a2b[i] >= 0 && a2b[i] < b->nr) {
 358                         struct patch_util *a_util = a->items[i].util;
 359                         struct patch_util *b_util = b->items[a2b[i]].util;
 360
 361                         a_util->matching = a2b[i];
 362                         b_util->matching = i;
 363                 }
 364
 365         free(cost);
 366         free(a2b);
 367         free(b2a);
 368 }
 369
 370 static void output_pair_header(struct diff_options *diffopt,
 371                                int patch_no_width,
 372                                struct strbuf *buf,
 373                                struct strbuf *dashes,
 374                                struct patch_util *a_util,
 375                                struct patch_util *b_util)
 376 {
 377         struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
 378         struct commit *commit;
 379         char status;
 380         const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
 381         const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
 382         const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
 383         const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
 384         const char *color;
 385
 386         if (!dashes->len)
 387                 strbuf_addchars(dashes, '-',
 388                                 strlen(find_unique_abbrev(oid,
 389                                                           DEFAULT_ABBREV)));
 390
 391         if (!b_util) {
 392                 color = color_old;
 393                 status = '<';
 394         } else if (!a_util) {
 395                 color = color_new;
 396                 status = '>';
 397         } else if (strcmp(a_util->patch, b_util->patch)) {
 398                 color = color_commit;
 399                 status = '!';
 400         } else {
 401                 color = color_commit;
 402                 status = '=';
 403         }
 404
 405         strbuf_reset(buf);
 406         strbuf_addstr(buf, status == '!' ? color_old : color);
 407         if (!a_util)
 408                 strbuf_addf(buf, "%*s:  %s ", patch_no_width, "-", dashes->buf);
 409         else
 410                 strbuf_addf(buf, "%*d:  %s ", patch_no_width, a_util->i + 1,
 411                             find_unique_abbrev(&a_util->oid, DEFAULT_ABBREV));
 412
 413         if (status == '!')
 414                 strbuf_addf(buf, "%s%s", color_reset, color);
 415         strbuf_addch(buf, status);
 416         if (status == '!')
 417                 strbuf_addf(buf, "%s%s", color_reset, color_new);
 418
 419         if (!b_util)
 420                 strbuf_addf(buf, " %*s:  %s", patch_no_width, "-", dashes->buf);
 421         else
 422                 strbuf_addf(buf, " %*d:  %s", patch_no_width, b_util->i + 1,
 423                             find_unique_abbrev(&b_util->oid, DEFAULT_ABBREV));
 424
 425         commit = lookup_commit_reference(the_repository, oid);
 426         if (commit) {
 427                 if (status == '!')
 428                         strbuf_addf(buf, "%s%s", color_reset, color);
 429
 430                 strbuf_addch(buf, ' ');
 431                 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
 432         }
 433         strbuf_addf(buf, "%s\n", color_reset);
 434
 435         fwrite(buf->buf, buf->len, 1, diffopt->file);
 436 }
 437
 438 static struct userdiff_driver section_headers = {
 439         .funcname = { "^ ## (.*) ##$\n"
 440                       "^.?@@ (.*)$", REG_EXTENDED }
 441 };
 442
 443 static struct diff_filespec *get_filespec(const char *name, const char *p)
 444 {
 445         struct diff_filespec *spec = alloc_filespec(name);
 446
 447         fill_filespec(spec, &null_oid, 0, 0100644);
 448         spec->data = (char *)p;
 449         spec->size = strlen(p);
 450         spec->should_munmap = 0;
 451         spec->is_stdin = 1;
 452         spec->driver = &section_headers;
 453
 454         return spec;
 455 }
 456
 457 static void patch_diff(const char *a, const char *b,
 458                        struct diff_options *diffopt)
 459 {
 460         diff_queue(&diff_queued_diff,
 461                    get_filespec("a", a), get_filespec("b", b));
 462
 463         diffcore_std(diffopt);
 464         diff_flush(diffopt);
 465 }
 466
 467 static struct strbuf *output_prefix_cb(struct diff_options *opt, void *data)
 468 {
 469         return data;
 470 }
 471
 472 static void output(struct string_list *a, struct string_list *b,
 473                    struct range_diff_options *range_diff_opts)
 474 {
 475         struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
 476         int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
 477         int i = 0, j = 0;
 478         struct diff_options opts;
 479         struct strbuf indent = STRBUF_INIT;
 480
 481         if (range_diff_opts->diffopt)
 482                 memcpy(&opts, range_diff_opts->diffopt, sizeof(opts));
 483         else
 484                 diff_setup(&opts);
 485
 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 }
 547
 548 int show_range_diff(const char *range1, const char *range2,
 549                     struct range_diff_options *range_diff_opts)
 550 {
 551         int res = 0;
 552
 553         struct string_list branch1 = STRING_LIST_INIT_DUP;
 554         struct string_list branch2 = STRING_LIST_INIT_DUP;
 555
 556         if (range_diff_opts->left_only && range_diff_opts->right_only)
 557                 res = error(_("--left-only and --right-only are mutually exclusive"));
 558
 559         if (!res && read_patches(range1, &branch1, range_diff_opts->other_arg))
 560                 res = error(_("could not parse log for '%s'"), range1);
 561         if (!res && read_patches(range2, &branch2, range_diff_opts->other_arg))
 562                 res = error(_("could not parse log for '%s'"), range2);
 563
 564         if (!res) {
 565                 find_exact_matches(&branch1, &branch2);
 566                 get_correspondences(&branch1, &branch2,
 567                                     range_diff_opts->creation_factor);
 568                 output(&branch1, &branch2, range_diff_opts);
 569         }
 570
 571         string_list_clear(&branch1, 1);
 572         string_list_clear(&branch2, 1);
 573
 574         return res;
 575 }