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 {
  45         struct child_process cp = CHILD_PROCESS_INIT;
  46         struct strbuf buf = STRBUF_INIT, contents = STRBUF_INIT;
  47         struct patch_util *util = NULL;
  48         int in_header = 1;
  49         char *line, *current_filename = NULL;
  50         int offset, len;
  51         size_t size;
  52
  53         argv_array_pushl(&cp.args, "log", "--no-color", "-p", "--no-merges",
  54                         "--reverse", "--date-order", "--decorate=no",
  55                         /*
  56                          * Choose indicators that are not used anywhere
  57                          * else in diffs, but still look reasonable
  58                          * (e.g. will not be confusing when debugging)
  59                          */
  60                         "--output-indicator-new=>",
  61                         "--output-indicator-old=<",
  62                         "--output-indicator-context=#",
  63                         "--no-abbrev-commit", range,
  64                         NULL);
  65         cp.out = -1;
  66         cp.no_stdin = 1;
  67         cp.git_cmd = 1;
  68
  69         if (start_command(&cp))
  70                 return error_errno(_("could not start `log`"));
  71         if (strbuf_read(&contents, cp.out, 0) < 0) {
  72                 error_errno(_("could not read `log` output"));
  73                 finish_command(&cp);
  74                 return -1;
  75         }
  76
  77         line = contents.buf;
  78         size = contents.len;
  79         for (offset = 0; size > 0; offset += len, size -= len, line += len) {
  80                 const char *p;
  81
  82                 len = find_end_of_line(line, size);
  83                 line[len - 1] = '\0';
  84                 if (skip_prefix(line, "commit ", &p)) {
  85                         if (util) {
  86                                 string_list_append(list, buf.buf)->util = util;
  87                                 strbuf_reset(&buf);
  88                         }
  89                         util = xcalloc(sizeof(*util), 1);
  90                         if (get_oid(p, &util->oid)) {
  91                                 error(_("could not parse commit '%s'"), p);
  92                                 free(util);
  93                                 string_list_clear(list, 1);
  94                                 strbuf_release(&buf);
  95                                 strbuf_release(&contents);
  96                                 finish_command(&cp);
  97                                 return -1;
  98                         }
  99                         util->matching = -1;
 100                         in_header = 1;
 101                         continue;
 102                 }
 103
 104                 if (starts_with(line, "diff --git")) {
 105                         struct patch patch = { 0 };
 106                         struct strbuf root = STRBUF_INIT;
 107                         int linenr = 0;
 108
 109                         in_header = 0;
 110                         strbuf_addch(&buf, '\n');
 111                         if (!util->diff_offset)
 112                                 util->diff_offset = buf.len;
 113                         line[len - 1] = '\n';
 114                         len = parse_git_diff_header(&root, &linenr, 1, line,
 115                                                     len, size, &patch);
 116                         if (len < 0)
 117                                 die(_("could not parse git header '%.*s'"), (int)len, line);
 118                         strbuf_addstr(&buf, " ## ");
 119                         if (patch.is_new > 0)
 120                                 strbuf_addf(&buf, "%s (new)", patch.new_name);
 121                         else if (patch.is_delete > 0)
 122                                 strbuf_addf(&buf, "%s (deleted)", patch.old_name);
 123                         else if (patch.is_rename)
 124                                 strbuf_addf(&buf, "%s => %s", patch.old_name, patch.new_name);
 125                         else
 126                                 strbuf_addstr(&buf, patch.new_name);
 127
 128                         free(current_filename);
 129                         if (patch.is_delete > 0)
 130                                 current_filename = xstrdup(patch.old_name);
 131                         else
 132                                 current_filename = xstrdup(patch.new_name);
 133
 134                         if (patch.new_mode && patch.old_mode &&
 135                             patch.old_mode != patch.new_mode)
 136                                 strbuf_addf(&buf, " (mode change %06o => %06o)",
 137                                             patch.old_mode, patch.new_mode);
 138
 139                         strbuf_addstr(&buf, " ##");
 140                 } else if (in_header) {
 141                         if (starts_with(line, "Author: ")) {
 142                                 strbuf_addstr(&buf, " ## Metadata ##\n");
 143                                 strbuf_addstr(&buf, line);
 144                                 strbuf_addstr(&buf, "\n\n");
 145                                 strbuf_addstr(&buf, " ## Commit message ##\n");
 146                         } else if (starts_with(line, "    ")) {
 147                                 p = line + len - 2;
 148                                 while (isspace(*p) && p >= line)
 149                                         p--;
 150                                 strbuf_add(&buf, line, p - line + 1);
 151                                 strbuf_addch(&buf, '\n');
 152                         }
 153                         continue;
 154                 } else if (skip_prefix(line, "@@ ", &p)) {
 155                         p = strstr(p, "@@");
 156                         strbuf_addstr(&buf, "@@");
 157                         if (current_filename && p[2])
 158                                 strbuf_addf(&buf, " %s:", current_filename);
 159                         if (p)
 160                                 strbuf_addstr(&buf, p + 2);
 161                 } else if (!line[0])
 162                         /*
 163                          * A completely blank (not ' \n', which is context)
 164                          * line is not valid in a diff.  We skip it
 165                          * silently, because this neatly handles the blank
 166                          * separator line between commits in git-log
 167                          * output.
 168                          */
 169                         continue;
 170                 else if (line[0] == '>') {
 171                         strbuf_addch(&buf, '+');
 172                         strbuf_addstr(&buf, line + 1);
 173                 } else if (line[0] == '<') {
 174                         strbuf_addch(&buf, '-');
 175                         strbuf_addstr(&buf, line + 1);
 176                 } else if (line[0] == '#') {
 177                         strbuf_addch(&buf, ' ');
 178                         strbuf_addstr(&buf, line + 1);
 179                 } else {
 180                         strbuf_addch(&buf, ' ');
 181                         strbuf_addstr(&buf, line);
 182                 }
 183
 184                 strbuf_addch(&buf, '\n');
 185                 util->diffsize++;
 186         }
 187         strbuf_release(&contents);
 188
 189         if (util)
 190                 string_list_append(list, buf.buf)->util = util;
 191         strbuf_release(&buf);
 192         free(current_filename);
 193
 194         if (finish_command(&cp))
 195                 return -1;
 196
 197         return 0;
 198 }
 199
 200 static int patch_util_cmp(const void *dummy, const struct patch_util *a,
 201                           const struct patch_util *b, const char *keydata)
 202 {
 203         return strcmp(a->diff, keydata ? keydata : b->diff);
 204 }
 205
 206 static void find_exact_matches(struct string_list *a, struct string_list *b)
 207 {
 208         struct hashmap map;
 209         int i;
 210
 211         hashmap_init(&map, (hashmap_cmp_fn)patch_util_cmp, NULL, 0);
 212
 213         /* First, add the patches of a to a hash map */
 214         for (i = 0; i < a->nr; i++) {
 215                 struct patch_util *util = a->items[i].util;
 216
 217                 util->i = i;
 218                 util->patch = a->items[i].string;
 219                 util->diff = util->patch + util->diff_offset;
 220                 hashmap_entry_init(&util->e, strhash(util->diff));
 221                 hashmap_add(&map, &util->e);
 222         }
 223
 224         /* Now try to find exact matches in b */
 225         for (i = 0; i < b->nr; i++) {
 226                 struct patch_util *util = b->items[i].util, *other;
 227
 228                 util->i = i;
 229                 util->patch = b->items[i].string;
 230                 util->diff = util->patch + util->diff_offset;
 231                 hashmap_entry_init(&util->e, strhash(util->diff));
 232                 other = hashmap_remove_entry(&map, util, NULL,
 233                                         struct patch_util,
 234                                         e /* member name */);
 235                 if (other) {
 236                         if (other->matching >= 0)
 237                                 BUG("already assigned!");
 238
 239                         other->matching = i;
 240                         util->matching = other->i;
 241                 }
 242         }
 243
 244         hashmap_free(&map);
 245 }
 246
 247 static void diffsize_consume(void *data, char *line, unsigned long len)
 248 {
 249         (*(int *)data)++;
 250 }
 251
 252 static void diffsize_hunk(void *data, long ob, long on, long nb, long nn,
 253                           const char *funcline, long funclen)
 254 {
 255         diffsize_consume(data, NULL, 0);
 256 }
 257
 258 static int diffsize(const char *a, const char *b)
 259 {
 260         xpparam_t pp = { 0 };
 261         xdemitconf_t cfg = { 0 };
 262         mmfile_t mf1, mf2;
 263         int count = 0;
 264
 265         mf1.ptr = (char *)a;
 266         mf1.size = strlen(a);
 267         mf2.ptr = (char *)b;
 268         mf2.size = strlen(b);
 269
 270         cfg.ctxlen = 3;
 271         if (!xdi_diff_outf(&mf1, &mf2,
 272                            diffsize_hunk, diffsize_consume, &count,
 273                            &pp, &cfg))
 274                 return count;
 275
 276         error(_("failed to generate diff"));
 277         return COST_MAX;
 278 }
 279
 280 static void get_correspondences(struct string_list *a, struct string_list *b,
 281                                 int creation_factor)
 282 {
 283         int n = a->nr + b->nr;
 284         int *cost, c, *a2b, *b2a;
 285         int i, j;
 286
 287         ALLOC_ARRAY(cost, st_mult(n, n));
 288         ALLOC_ARRAY(a2b, n);
 289         ALLOC_ARRAY(b2a, n);
 290
 291         for (i = 0; i < a->nr; i++) {
 292                 struct patch_util *a_util = a->items[i].util;
 293
 294                 for (j = 0; j < b->nr; j++) {
 295                         struct patch_util *b_util = b->items[j].util;
 296
 297                         if (a_util->matching == j)
 298                                 c = 0;
 299                         else if (a_util->matching < 0 && b_util->matching < 0)
 300                                 c = diffsize(a_util->diff, b_util->diff);
 301                         else
 302                                 c = COST_MAX;
 303                         cost[i + n * j] = c;
 304                 }
 305
 306                 c = a_util->matching < 0 ?
 307                         a_util->diffsize * creation_factor / 100 : COST_MAX;
 308                 for (j = b->nr; j < n; j++)
 309                         cost[i + n * j] = c;
 310         }
 311
 312         for (j = 0; j < b->nr; j++) {
 313                 struct patch_util *util = b->items[j].util;
 314
 315                 c = util->matching < 0 ?
 316                         util->diffsize * creation_factor / 100 : COST_MAX;
 317                 for (i = a->nr; i < n; i++)
 318                         cost[i + n * j] = c;
 319         }
 320
 321         for (i = a->nr; i < n; i++)
 322                 for (j = b->nr; j < n; j++)
 323                         cost[i + n * j] = 0;
 324
 325         compute_assignment(n, n, cost, a2b, b2a);
 326
 327         for (i = 0; i < a->nr; i++)
 328                 if (a2b[i] >= 0 && a2b[i] < b->nr) {
 329                         struct patch_util *a_util = a->items[i].util;
 330                         struct patch_util *b_util = b->items[a2b[i]].util;
 331
 332                         a_util->matching = a2b[i];
 333                         b_util->matching = i;
 334                 }
 335
 336         free(cost);
 337         free(a2b);
 338         free(b2a);
 339 }
 340
 341 static void output_pair_header(struct diff_options *diffopt,
 342                                int patch_no_width,
 343                                struct strbuf *buf,
 344                                struct strbuf *dashes,
 345                                struct patch_util *a_util,
 346                                struct patch_util *b_util)
 347 {
 348         struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
 349         struct commit *commit;
 350         char status;
 351         const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
 352         const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
 353         const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
 354         const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
 355         const char *color;
 356
 357         if (!dashes->len)
 358                 strbuf_addchars(dashes, '-',
 359                                 strlen(find_unique_abbrev(oid,
 360                                                           DEFAULT_ABBREV)));
 361
 362         if (!b_util) {
 363                 color = color_old;
 364                 status = '<';
 365         } else if (!a_util) {
 366                 color = color_new;
 367                 status = '>';
 368         } else if (strcmp(a_util->patch, b_util->patch)) {
 369                 color = color_commit;
 370                 status = '!';
 371         } else {
 372                 color = color_commit;
 373                 status = '=';
 374         }
 375
 376         strbuf_reset(buf);
 377         strbuf_addstr(buf, status == '!' ? color_old : color);
 378         if (!a_util)
 379                 strbuf_addf(buf, "%*s:  %s ", patch_no_width, "-", dashes->buf);
 380         else
 381                 strbuf_addf(buf, "%*d:  %s ", patch_no_width, a_util->i + 1,
 382                             find_unique_abbrev(&a_util->oid, DEFAULT_ABBREV));
 383
 384         if (status == '!')
 385                 strbuf_addf(buf, "%s%s", color_reset, color);
 386         strbuf_addch(buf, status);
 387         if (status == '!')
 388                 strbuf_addf(buf, "%s%s", color_reset, color_new);
 389
 390         if (!b_util)
 391                 strbuf_addf(buf, " %*s:  %s", patch_no_width, "-", dashes->buf);
 392         else
 393                 strbuf_addf(buf, " %*d:  %s", patch_no_width, b_util->i + 1,
 394                             find_unique_abbrev(&b_util->oid, DEFAULT_ABBREV));
 395
 396         commit = lookup_commit_reference(the_repository, oid);
 397         if (commit) {
 398                 if (status == '!')
 399                         strbuf_addf(buf, "%s%s", color_reset, color);
 400
 401                 strbuf_addch(buf, ' ');
 402                 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
 403         }
 404         strbuf_addf(buf, "%s\n", color_reset);
 405
 406         fwrite(buf->buf, buf->len, 1, diffopt->file);
 407 }
 408
 409 static struct userdiff_driver section_headers = {
 410         .funcname = { "^ ## (.*) ##$\n"
 411                       "^.?@@ (.*)$", REG_EXTENDED }
 412 };
 413
 414 static struct diff_filespec *get_filespec(const char *name, const char *p)
 415 {
 416         struct diff_filespec *spec = alloc_filespec(name);
 417
 418         fill_filespec(spec, &null_oid, 0, 0100644);
 419         spec->data = (char *)p;
 420         spec->size = strlen(p);
 421         spec->should_munmap = 0;
 422         spec->is_stdin = 1;
 423         spec->driver = &section_headers;
 424
 425         return spec;
 426 }
 427
 428 static void patch_diff(const char *a, const char *b,
 429                        struct diff_options *diffopt)
 430 {
 431         diff_queue(&diff_queued_diff,
 432                    get_filespec("a", a), get_filespec("b", b));
 433
 434         diffcore_std(diffopt);
 435         diff_flush(diffopt);
 436 }
 437
 438 static void output(struct string_list *a, struct string_list *b,
 439                    struct diff_options *diffopt)
 440 {
 441         struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
 442         int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
 443         int i = 0, j = 0;
 444
 445         /*
 446          * We assume the user is really more interested in the second argument
 447          * ("newer" version). To that end, we print the output in the order of
 448          * the RHS (the `b` parameter). To put the LHS (the `a` parameter)
 449          * commits that are no longer in the RHS into a good place, we place
 450          * them once we have shown all of their predecessors in the LHS.
 451          */
 452
 453         while (i < a->nr || j < b->nr) {
 454                 struct patch_util *a_util, *b_util;
 455                 a_util = i < a->nr ? a->items[i].util : NULL;
 456                 b_util = j < b->nr ? b->items[j].util : NULL;
 457
 458                 /* Skip all the already-shown commits from the LHS. */
 459                 while (i < a->nr && a_util->shown)
 460                         a_util = ++i < a->nr ? a->items[i].util : NULL;
 461
 462                 /* Show unmatched LHS commit whose predecessors were shown. */
 463                 if (i < a->nr && a_util->matching < 0) {
 464                         output_pair_header(diffopt, patch_no_width,
 465                                            &buf, &dashes, a_util, NULL);
 466                         i++;
 467                         continue;
 468                 }
 469
 470                 /* Show unmatched RHS commits. */
 471                 while (j < b->nr && b_util->matching < 0) {
 472                         output_pair_header(diffopt, patch_no_width,
 473                                            &buf, &dashes, NULL, b_util);
 474                         b_util = ++j < b->nr ? b->items[j].util : NULL;
 475                 }
 476
 477                 /* Show matching LHS/RHS pair. */
 478                 if (j < b->nr) {
 479                         a_util = a->items[b_util->matching].util;
 480                         output_pair_header(diffopt, patch_no_width,
 481                                            &buf, &dashes, a_util, b_util);
 482                         if (!(diffopt->output_format & DIFF_FORMAT_NO_OUTPUT))
 483                                 patch_diff(a->items[b_util->matching].string,
 484                                            b->items[j].string, diffopt);
 485                         a_util->shown = 1;
 486                         j++;
 487                 }
 488         }
 489         strbuf_release(&buf);
 490         strbuf_release(&dashes);
 491 }
 492
 493 static struct strbuf *output_prefix_cb(struct diff_options *opt, void *data)
 494 {
 495         return data;
 496 }
 497
 498 int show_range_diff(const char *range1, const char *range2,
 499                     int creation_factor, int dual_color,
 500                     struct diff_options *diffopt)
 501 {
 502         int res = 0;
 503
 504         struct string_list branch1 = STRING_LIST_INIT_DUP;
 505         struct string_list branch2 = STRING_LIST_INIT_DUP;
 506
 507         if (read_patches(range1, &branch1))
 508                 res = error(_("could not parse log for '%s'"), range1);
 509         if (!res && read_patches(range2, &branch2))
 510                 res = error(_("could not parse log for '%s'"), range2);
 511
 512         if (!res) {
 513                 struct diff_options opts;
 514                 struct strbuf indent = STRBUF_INIT;
 515
 516                 if (diffopt)
 517                         memcpy(&opts, diffopt, sizeof(opts));
 518                 else
 519                         diff_setup(&opts);
 520
 521                 if (!opts.output_format)
 522                         opts.output_format = DIFF_FORMAT_PATCH;
 523                 opts.flags.suppress_diff_headers = 1;
 524                 opts.flags.dual_color_diffed_diffs = dual_color;
 525                 opts.flags.suppress_hunk_header_line_count = 1;
 526                 opts.output_prefix = output_prefix_cb;
 527                 strbuf_addstr(&indent, "    ");
 528                 opts.output_prefix_data = &indent;
 529                 diff_setup_done(&opts);
 530
 531                 find_exact_matches(&branch1, &branch2);
 532                 get_correspondences(&branch1, &branch2, creation_factor);
 533                 output(&branch1, &branch2, &opts);
 534
 535                 strbuf_release(&indent);
 536         }
 537
 538         string_list_clear(&branch1, 1);
 539         string_list_clear(&branch2, 1);
 540
 541         return res;
 542 }