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 /*
  29  * Reads the patches into a string list, with the `util` field being populated
  30  * as struct object_id (will need to be free()d).
  31  */
  32 static int read_patches(const char *range, struct string_list *list,
  33                         const struct strvec *other_arg)
  34 {
  35         struct child_process cp = CHILD_PROCESS_INIT;
  36         struct strbuf buf = STRBUF_INIT, contents = STRBUF_INIT;
  37         struct patch_util *util = NULL;
  38         int in_header = 1;
  39         char *line, *current_filename = NULL;
  40         ssize_t len;
  41         size_t size;
  42
  43         strvec_pushl(&cp.args, "log", "--no-color", "-p", "--no-merges",
  44                      "--reverse", "--date-order", "--decorate=no",
  45                      "--no-prefix",
  46                      /*
  47                       * Choose indicators that are not used anywhere
  48                       * else in diffs, but still look reasonable
  49                       * (e.g. will not be confusing when debugging)
  50                       */
  51                      "--output-indicator-new=>",
  52                      "--output-indicator-old=<",
  53                      "--output-indicator-context=#",
  54                      "--no-abbrev-commit",
  55                      "--pretty=medium",
  56                      "--notes",
  57                      NULL);
  58         if (other_arg)
  59                 strvec_pushv(&cp.args, other_arg->v);
  60         strvec_push(&cp.args, range);
  61         cp.out = -1;
  62         cp.no_stdin = 1;
  63         cp.git_cmd = 1;
  64
  65         if (start_command(&cp))
  66                 return error_errno(_("could not start `log`"));
  67         if (strbuf_read(&contents, cp.out, 0) < 0) {
  68                 error_errno(_("could not read `log` output"));
  69                 finish_command(&cp);
  70                 return -1;
  71         }
  72
  73         line = contents.buf;
  74         size = contents.len;
  75         for (; size > 0; size -= len, line += len) {
  76                 const char *p;
  77                 char *eol;
  78
  79                 eol = memchr(line, '\n', size);
  80                 if (eol) {
  81                         *eol = '\0';
  82                         len = eol + 1 - line;
  83                 } else {
  84                         len = size;
  85                 }
  86
  87                 if (skip_prefix(line, "commit ", &p)) {
  88                         if (util) {
  89                                 string_list_append(list, buf.buf)->util = util;
  90                                 strbuf_reset(&buf);
  91                         }
  92                         util = xcalloc(sizeof(*util), 1);
  93                         if (get_oid(p, &util->oid)) {
  94                                 error(_("could not parse commit '%s'"), p);
  95                                 free(util);
  96                                 string_list_clear(list, 1);
  97                                 strbuf_release(&buf);
  98                                 strbuf_release(&contents);
  99                                 finish_command(&cp);
 100                                 return -1;
 101                         }
 102                         util->matching = -1;
 103                         in_header = 1;
 104                         continue;
 105                 }
 106
 107                 if (!util) {
 108                         error(_("could not parse first line of `log` output: "
 109                                 "did not start with 'commit ': '%s'"),
 110                               line);
 111                         string_list_clear(list, 1);
 112                         strbuf_release(&buf);
 113                         strbuf_release(&contents);
 114                         finish_command(&cp);
 115                         return -1;
 116                 }
 117
 118                 if (starts_with(line, "diff --git")) {
 119                         struct patch patch = { 0 };
 120                         struct strbuf root = STRBUF_INIT;
 121                         int linenr = 0;
 122                         int orig_len;
 123
 124                         in_header = 0;
 125                         strbuf_addch(&buf, '\n');
 126                         if (!util->diff_offset)
 127                                 util->diff_offset = buf.len;
 128                         if (eol)
 129                                 *eol = '\n';
 130                         orig_len = len;
 131                         len = parse_git_diff_header(&root, &linenr, 0, line,
 132                                                     len, size, &patch);
 133                         if (len < 0)
 134                                 die(_("could not parse git header '%.*s'"),
 135                                     orig_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 = HASHMAP_INIT((hashmap_cmp_fn)patch_util_cmp, NULL);
 233         int i;
 234
 235         /* First, add the patches of a to a hash map */
 236         for (i = 0; i < a->nr; i++) {
 237                 struct patch_util *util = a->items[i].util;
 238
 239                 util->i = i;
 240                 util->patch = a->items[i].string;
 241                 util->diff = util->patch + util->diff_offset;
 242                 hashmap_entry_init(&util->e, strhash(util->diff));
 243                 hashmap_add(&map, &util->e);
 244         }
 245
 246         /* Now try to find exact matches in b */
 247         for (i = 0; i < b->nr; i++) {
 248                 struct patch_util *util = b->items[i].util, *other;
 249
 250                 util->i = i;
 251                 util->patch = b->items[i].string;
 252                 util->diff = util->patch + util->diff_offset;
 253                 hashmap_entry_init(&util->e, strhash(util->diff));
 254                 other = hashmap_remove_entry(&map, util, e, NULL);
 255                 if (other) {
 256                         if (other->matching >= 0)
 257                                 BUG("already assigned!");
 258
 259                         other->matching = i;
 260                         util->matching = other->i;
 261                 }
 262         }
 263
 264         hashmap_clear(&map);
 265 }
 266
 267 static void diffsize_consume(void *data, char *line, unsigned long len)
 268 {
 269         (*(int *)data)++;
 270 }
 271
 272 static void diffsize_hunk(void *data, long ob, long on, long nb, long nn,
 273                           const char *funcline, long funclen)
 274 {
 275         diffsize_consume(data, NULL, 0);
 276 }
 277
 278 static int diffsize(const char *a, const char *b)
 279 {
 280         xpparam_t pp = { 0 };
 281         xdemitconf_t cfg = { 0 };
 282         mmfile_t mf1, mf2;
 283         int count = 0;
 284
 285         mf1.ptr = (char *)a;
 286         mf1.size = strlen(a);
 287         mf2.ptr = (char *)b;
 288         mf2.size = strlen(b);
 289
 290         cfg.ctxlen = 3;
 291         if (!xdi_diff_outf(&mf1, &mf2,
 292                            diffsize_hunk, diffsize_consume, &count,
 293                            &pp, &cfg))
 294                 return count;
 295
 296         error(_("failed to generate diff"));
 297         return COST_MAX;
 298 }
 299
 300 static void get_correspondences(struct string_list *a, struct string_list *b,
 301                                 int creation_factor)
 302 {
 303         int n = a->nr + b->nr;
 304         int *cost, c, *a2b, *b2a;
 305         int i, j;
 306
 307         ALLOC_ARRAY(cost, st_mult(n, n));
 308         ALLOC_ARRAY(a2b, n);
 309         ALLOC_ARRAY(b2a, n);
 310
 311         for (i = 0; i < a->nr; i++) {
 312                 struct patch_util *a_util = a->items[i].util;
 313
 314                 for (j = 0; j < b->nr; j++) {
 315                         struct patch_util *b_util = b->items[j].util;
 316
 317                         if (a_util->matching == j)
 318                                 c = 0;
 319                         else if (a_util->matching < 0 && b_util->matching < 0)
 320                                 c = diffsize(a_util->diff, b_util->diff);
 321                         else
 322                                 c = COST_MAX;
 323                         cost[i + n * j] = c;
 324                 }
 325
 326                 c = a_util->matching < 0 ?
 327                         a_util->diffsize * creation_factor / 100 : COST_MAX;
 328                 for (j = b->nr; j < n; j++)
 329                         cost[i + n * j] = c;
 330         }
 331
 332         for (j = 0; j < b->nr; j++) {
 333                 struct patch_util *util = b->items[j].util;
 334
 335                 c = util->matching < 0 ?
 336                         util->diffsize * creation_factor / 100 : COST_MAX;
 337                 for (i = a->nr; i < n; i++)
 338                         cost[i + n * j] = c;
 339         }
 340
 341         for (i = a->nr; i < n; i++)
 342                 for (j = b->nr; j < n; j++)
 343                         cost[i + n * j] = 0;
 344
 345         compute_assignment(n, n, cost, a2b, b2a);
 346
 347         for (i = 0; i < a->nr; i++)
 348                 if (a2b[i] >= 0 && a2b[i] < b->nr) {
 349                         struct patch_util *a_util = a->items[i].util;
 350                         struct patch_util *b_util = b->items[a2b[i]].util;
 351
 352                         a_util->matching = a2b[i];
 353                         b_util->matching = i;
 354                 }
 355
 356         free(cost);
 357         free(a2b);
 358         free(b2a);
 359 }
 360
 361 static void output_pair_header(struct diff_options *diffopt,
 362                                int patch_no_width,
 363                                struct strbuf *buf,
 364                                struct strbuf *dashes,
 365                                struct patch_util *a_util,
 366                                struct patch_util *b_util)
 367 {
 368         struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
 369         struct commit *commit;
 370         char status;
 371         const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
 372         const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
 373         const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
 374         const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
 375         const char *color;
 376
 377         if (!dashes->len)
 378                 strbuf_addchars(dashes, '-',
 379                                 strlen(find_unique_abbrev(oid,
 380                                                           DEFAULT_ABBREV)));
 381
 382         if (!b_util) {
 383                 color = color_old;
 384                 status = '<';
 385         } else if (!a_util) {
 386                 color = color_new;
 387                 status = '>';
 388         } else if (strcmp(a_util->patch, b_util->patch)) {
 389                 color = color_commit;
 390                 status = '!';
 391         } else {
 392                 color = color_commit;
 393                 status = '=';
 394         }
 395
 396         strbuf_reset(buf);
 397         strbuf_addstr(buf, status == '!' ? color_old : color);
 398         if (!a_util)
 399                 strbuf_addf(buf, "%*s:  %s ", patch_no_width, "-", dashes->buf);
 400         else
 401                 strbuf_addf(buf, "%*d:  %s ", patch_no_width, a_util->i + 1,
 402                             find_unique_abbrev(&a_util->oid, DEFAULT_ABBREV));
 403
 404         if (status == '!')
 405                 strbuf_addf(buf, "%s%s", color_reset, color);
 406         strbuf_addch(buf, status);
 407         if (status == '!')
 408                 strbuf_addf(buf, "%s%s", color_reset, color_new);
 409
 410         if (!b_util)
 411                 strbuf_addf(buf, " %*s:  %s", patch_no_width, "-", dashes->buf);
 412         else
 413                 strbuf_addf(buf, " %*d:  %s", patch_no_width, b_util->i + 1,
 414                             find_unique_abbrev(&b_util->oid, DEFAULT_ABBREV));
 415
 416         commit = lookup_commit_reference(the_repository, oid);
 417         if (commit) {
 418                 if (status == '!')
 419                         strbuf_addf(buf, "%s%s", color_reset, color);
 420
 421                 strbuf_addch(buf, ' ');
 422                 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
 423         }
 424         strbuf_addf(buf, "%s\n", color_reset);
 425
 426         fwrite(buf->buf, buf->len, 1, diffopt->file);
 427 }
 428
 429 static struct userdiff_driver section_headers = {
 430         .funcname = { "^ ## (.*) ##$\n"
 431                       "^.?@@ (.*)$", REG_EXTENDED }
 432 };
 433
 434 static struct diff_filespec *get_filespec(const char *name, const char *p)
 435 {
 436         struct diff_filespec *spec = alloc_filespec(name);
 437
 438         fill_filespec(spec, &null_oid, 0, 0100644);
 439         spec->data = (char *)p;
 440         spec->size = strlen(p);
 441         spec->should_munmap = 0;
 442         spec->is_stdin = 1;
 443         spec->driver = &section_headers;
 444
 445         return spec;
 446 }
 447
 448 static void patch_diff(const char *a, const char *b,
 449                        struct diff_options *diffopt)
 450 {
 451         diff_queue(&diff_queued_diff,
 452                    get_filespec("a", a), get_filespec("b", b));
 453
 454         diffcore_std(diffopt);
 455         diff_flush(diffopt);
 456 }
 457
 458 static void output(struct string_list *a, struct string_list *b,
 459                    struct diff_options *diffopt)
 460 {
 461         struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
 462         int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
 463         int i = 0, j = 0;
 464
 465         /*
 466          * We assume the user is really more interested in the second argument
 467          * ("newer" version). To that end, we print the output in the order of
 468          * the RHS (the `b` parameter). To put the LHS (the `a` parameter)
 469          * commits that are no longer in the RHS into a good place, we place
 470          * them once we have shown all of their predecessors in the LHS.
 471          */
 472
 473         while (i < a->nr || j < b->nr) {
 474                 struct patch_util *a_util, *b_util;
 475                 a_util = i < a->nr ? a->items[i].util : NULL;
 476                 b_util = j < b->nr ? b->items[j].util : NULL;
 477
 478                 /* Skip all the already-shown commits from the LHS. */
 479                 while (i < a->nr && a_util->shown)
 480                         a_util = ++i < a->nr ? a->items[i].util : NULL;
 481
 482                 /* Show unmatched LHS commit whose predecessors were shown. */
 483                 if (i < a->nr && a_util->matching < 0) {
 484                         output_pair_header(diffopt, patch_no_width,
 485                                            &buf, &dashes, a_util, NULL);
 486                         i++;
 487                         continue;
 488                 }
 489
 490                 /* Show unmatched RHS commits. */
 491                 while (j < b->nr && b_util->matching < 0) {
 492                         output_pair_header(diffopt, patch_no_width,
 493                                            &buf, &dashes, NULL, b_util);
 494                         b_util = ++j < b->nr ? b->items[j].util : NULL;
 495                 }
 496
 497                 /* Show matching LHS/RHS pair. */
 498                 if (j < b->nr) {
 499                         a_util = a->items[b_util->matching].util;
 500                         output_pair_header(diffopt, patch_no_width,
 501                                            &buf, &dashes, a_util, b_util);
 502                         if (!(diffopt->output_format & DIFF_FORMAT_NO_OUTPUT))
 503                                 patch_diff(a->items[b_util->matching].string,
 504                                            b->items[j].string, diffopt);
 505                         a_util->shown = 1;
 506                         j++;
 507                 }
 508         }
 509         strbuf_release(&buf);
 510         strbuf_release(&dashes);
 511 }
 512
 513 static struct strbuf *output_prefix_cb(struct diff_options *opt, void *data)
 514 {
 515         return data;
 516 }
 517
 518 int show_range_diff(const char *range1, const char *range2,
 519                     int creation_factor, int dual_color,
 520                     const struct diff_options *diffopt,
 521                     const struct strvec *other_arg)
 522 {
 523         int res = 0;
 524
 525         struct string_list branch1 = STRING_LIST_INIT_DUP;
 526         struct string_list branch2 = STRING_LIST_INIT_DUP;
 527
 528         if (read_patches(range1, &branch1, other_arg))
 529                 res = error(_("could not parse log for '%s'"), range1);
 530         if (!res && read_patches(range2, &branch2, other_arg))
 531                 res = error(_("could not parse log for '%s'"), range2);
 532
 533         if (!res) {
 534                 struct diff_options opts;
 535                 struct strbuf indent = STRBUF_INIT;
 536
 537                 if (diffopt)
 538                         memcpy(&opts, diffopt, sizeof(opts));
 539                 else
 540                         diff_setup(&opts);
 541
 542                 if (!opts.output_format)
 543                         opts.output_format = DIFF_FORMAT_PATCH;
 544                 opts.flags.suppress_diff_headers = 1;
 545                 opts.flags.dual_color_diffed_diffs = dual_color;
 546                 opts.flags.suppress_hunk_header_line_count = 1;
 547                 opts.output_prefix = output_prefix_cb;
 548                 strbuf_addstr(&indent, "    ");
 549                 opts.output_prefix_data = &indent;
 550                 diff_setup_done(&opts);
 551
 552                 find_exact_matches(&branch1, &branch2);
 553                 get_correspondences(&branch1, &branch2, creation_factor);
 554                 output(&branch1, &branch2, &opts);
 555
 556                 strbuf_release(&indent);
 557         }
 558
 559         string_list_clear(&branch1, 1);
 560         string_list_clear(&branch2, 1);
 561
 562         return res;
 563 }