git.oblomov.eu Git - git/blob - 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                         "--no-prefix",
  56                         /*
  57                          * Choose indicators that are not used anywhere
  58                          * else in diffs, but still look reasonable
  59                          * (e.g. will not be confusing when debugging)
  60                          */
  61                         "--output-indicator-new=>",
  62                         "--output-indicator-old=<",
  63                         "--output-indicator-context=#",
  64                         "--no-abbrev-commit", range,
  65                         NULL);
  66         cp.out = -1;
  67         cp.no_stdin = 1;
  68         cp.git_cmd = 1;
  69
  70         if (start_command(&cp))
  71                 return error_errno(_("could not start `log`"));
  72         if (strbuf_read(&contents, cp.out, 0) < 0) {
  73                 error_errno(_("could not read `log` output"));
  74                 finish_command(&cp);
  75                 return -1;
  76         }
  77
  78         line = contents.buf;
  79         size = contents.len;
  80         for (offset = 0; size > 0; offset += len, size -= len, line += len) {
  81                 const char *p;
  82
  83                 len = find_end_of_line(line, size);
  84                 line[len - 1] = '\0';
  85                 if (skip_prefix(line, "commit ", &p)) {
  86                         if (util) {
  87                                 string_list_append(list, buf.buf)->util = util;
  88                                 strbuf_reset(&buf);
  89                         }
  90                         util = xcalloc(sizeof(*util), 1);
  91                         if (get_oid(p, &util->oid)) {
  92                                 error(_("could not parse commit '%s'"), p);
  93                                 free(util);
  94                                 string_list_clear(list, 1);
  95                                 strbuf_release(&buf);
  96                                 strbuf_release(&contents);
  97                                 finish_command(&cp);
  98                                 return -1;
  99                         }
 100                         util->matching = -1;
 101                         in_header = 1;
 102                         continue;
 103                 }
 104
 105                 if (starts_with(line, "diff --git")) {
 106                         struct patch patch = { 0 };
 107                         struct strbuf root = STRBUF_INIT;
 108                         int linenr = 0;
 109
 110                         in_header = 0;
 111                         strbuf_addch(&buf, '\n');
 112                         if (!util->diff_offset)
 113                                 util->diff_offset = buf.len;
 114                         line[len - 1] = '\n';
 115                         len = parse_git_diff_header(&root, &linenr, 0, line,
 116                                                     len, size, &patch);
 117                         if (len < 0)
 118                                 die(_("could not parse git header '%.*s'"), (int)len, line);
 119                         strbuf_addstr(&buf, " ## ");
 120                         if (patch.is_new > 0)
 121                                 strbuf_addf(&buf, "%s (new)", patch.new_name);
 122                         else if (patch.is_delete > 0)
 123                                 strbuf_addf(&buf, "%s (deleted)", patch.old_name);
 124                         else if (patch.is_rename)
 125                                 strbuf_addf(&buf, "%s => %s", patch.old_name, patch.new_name);
 126                         else
 127                                 strbuf_addstr(&buf, patch.new_name);
 128
 129                         free(current_filename);
 130                         if (patch.is_delete > 0)
 131                                 current_filename = xstrdup(patch.old_name);
 132                         else
 133                                 current_filename = xstrdup(patch.new_name);
 134
 135                         if (patch.new_mode && patch.old_mode &&
 136                             patch.old_mode != patch.new_mode)
 137                                 strbuf_addf(&buf, " (mode change %06o => %06o)",
 138                                             patch.old_mode, patch.new_mode);
 139
 140                         strbuf_addstr(&buf, " ##");
 141                 } else if (in_header) {
 142                         if (starts_with(line, "Author: ")) {
 143                                 strbuf_addstr(&buf, " ## Metadata ##\n");
 144                                 strbuf_addstr(&buf, line);
 145                                 strbuf_addstr(&buf, "\n\n");
 146                                 strbuf_addstr(&buf, " ## Commit message ##\n");
 147                         } else if (starts_with(line, "Notes") &&
 148                                    line[strlen(line) - 1] == ':') {
 149                                 strbuf_addstr(&buf, "\n\n");
 150                                 /* strip the trailing colon */
 151                                 strbuf_addf(&buf, " ## %.*s ##\n",
 152                                             (int)(strlen(line) - 1), line);
 153                         } else if (starts_with(line, "    ")) {
 154                                 p = line + len - 2;
 155                                 while (isspace(*p) && p >= line)
 156                                         p--;
 157                                 strbuf_add(&buf, line, p - line + 1);
 158                                 strbuf_addch(&buf, '\n');
 159                         }
 160                         continue;
 161                 } else if (skip_prefix(line, "@@ ", &p)) {
 162                         p = strstr(p, "@@");
 163                         strbuf_addstr(&buf, "@@");
 164                         if (current_filename && p[2])
 165                                 strbuf_addf(&buf, " %s:", current_filename);
 166                         if (p)
 167                                 strbuf_addstr(&buf, p + 2);
 168                 } else if (!line[0])
 169                         /*
 170                          * A completely blank (not ' \n', which is context)
 171                          * line is not valid in a diff.  We skip it
 172                          * silently, because this neatly handles the blank
 173                          * separator line between commits in git-log
 174                          * output.
 175                          */
 176                         continue;
 177                 else if (line[0] == '>') {
 178                         strbuf_addch(&buf, '+');
 179                         strbuf_addstr(&buf, line + 1);
 180                 } else if (line[0] == '<') {
 181                         strbuf_addch(&buf, '-');
 182                         strbuf_addstr(&buf, line + 1);
 183                 } else if (line[0] == '#') {
 184                         strbuf_addch(&buf, ' ');
 185                         strbuf_addstr(&buf, line + 1);
 186                 } else {
 187                         strbuf_addch(&buf, ' ');
 188                         strbuf_addstr(&buf, line);
 189                 }
 190
 191                 strbuf_addch(&buf, '\n');
 192                 util->diffsize++;
 193         }
 194         strbuf_release(&contents);
 195
 196         if (util)
 197                 string_list_append(list, buf.buf)->util = util;
 198         strbuf_release(&buf);
 199         free(current_filename);
 200
 201         if (finish_command(&cp))
 202                 return -1;
 203
 204         return 0;
 205 }
 206
 207 static int patch_util_cmp(const void *dummy, const struct patch_util *a,
 208                           const struct patch_util *b, const char *keydata)
 209 {
 210         return strcmp(a->diff, keydata ? keydata : b->diff);
 211 }
 212
 213 static void find_exact_matches(struct string_list *a, struct string_list *b)
 214 {
 215         struct hashmap map;
 216         int i;
 217
 218         hashmap_init(&map, (hashmap_cmp_fn)patch_util_cmp, NULL, 0);
 219
 220         /* First, add the patches of a to a hash map */
 221         for (i = 0; i < a->nr; i++) {
 222                 struct patch_util *util = a->items[i].util;
 223
 224                 util->i = i;
 225                 util->patch = a->items[i].string;
 226                 util->diff = util->patch + util->diff_offset;
 227                 hashmap_entry_init(&util->e, strhash(util->diff));
 228                 hashmap_add(&map, &util->e);
 229         }
 230
 231         /* Now try to find exact matches in b */
 232         for (i = 0; i < b->nr; i++) {
 233                 struct patch_util *util = b->items[i].util, *other;
 234
 235                 util->i = i;
 236                 util->patch = b->items[i].string;
 237                 util->diff = util->patch + util->diff_offset;
 238                 hashmap_entry_init(&util->e, strhash(util->diff));
 239                 other = hashmap_remove_entry(&map, util, e, NULL);
 240                 if (other) {
 241                         if (other->matching >= 0)
 242                                 BUG("already assigned!");
 243
 244                         other->matching = i;
 245                         util->matching = other->i;
 246                 }
 247         }
 248
 249         hashmap_free(&map);
 250 }
 251
 252 static void diffsize_consume(void *data, char *line, unsigned long len)
 253 {
 254         (*(int *)data)++;
 255 }
 256
 257 static void diffsize_hunk(void *data, long ob, long on, long nb, long nn,
 258                           const char *funcline, long funclen)
 259 {
 260         diffsize_consume(data, NULL, 0);
 261 }
 262
 263 static int diffsize(const char *a, const char *b)
 264 {
 265         xpparam_t pp = { 0 };
 266         xdemitconf_t cfg = { 0 };
 267         mmfile_t mf1, mf2;
 268         int count = 0;
 269
 270         mf1.ptr = (char *)a;
 271         mf1.size = strlen(a);
 272         mf2.ptr = (char *)b;
 273         mf2.size = strlen(b);
 274
 275         cfg.ctxlen = 3;
 276         if (!xdi_diff_outf(&mf1, &mf2,
 277                            diffsize_hunk, diffsize_consume, &count,
 278                            &pp, &cfg))
 279                 return count;
 280
 281         error(_("failed to generate diff"));
 282         return COST_MAX;
 283 }
 284
 285 static void get_correspondences(struct string_list *a, struct string_list *b,
 286                                 int creation_factor)
 287 {
 288         int n = a->nr + b->nr;
 289         int *cost, c, *a2b, *b2a;
 290         int i, j;
 291
 292         ALLOC_ARRAY(cost, st_mult(n, n));
 293         ALLOC_ARRAY(a2b, n);
 294         ALLOC_ARRAY(b2a, n);
 295
 296         for (i = 0; i < a->nr; i++) {
 297                 struct patch_util *a_util = a->items[i].util;
 298
 299                 for (j = 0; j < b->nr; j++) {
 300                         struct patch_util *b_util = b->items[j].util;
 301
 302                         if (a_util->matching == j)
 303                                 c = 0;
 304                         else if (a_util->matching < 0 && b_util->matching < 0)
 305                                 c = diffsize(a_util->diff, b_util->diff);
 306                         else
 307                                 c = COST_MAX;
 308                         cost[i + n * j] = c;
 309                 }
 310
 311                 c = a_util->matching < 0 ?
 312                         a_util->diffsize * creation_factor / 100 : COST_MAX;
 313                 for (j = b->nr; j < n; j++)
 314                         cost[i + n * j] = c;
 315         }
 316
 317         for (j = 0; j < b->nr; j++) {
 318                 struct patch_util *util = b->items[j].util;
 319
 320                 c = util->matching < 0 ?
 321                         util->diffsize * creation_factor / 100 : COST_MAX;
 322                 for (i = a->nr; i < n; i++)
 323                         cost[i + n * j] = c;
 324         }
 325
 326         for (i = a->nr; i < n; i++)
 327                 for (j = b->nr; j < n; j++)
 328                         cost[i + n * j] = 0;
 329
 330         compute_assignment(n, n, cost, a2b, b2a);
 331
 332         for (i = 0; i < a->nr; i++)
 333                 if (a2b[i] >= 0 && a2b[i] < b->nr) {
 334                         struct patch_util *a_util = a->items[i].util;
 335                         struct patch_util *b_util = b->items[a2b[i]].util;
 336
 337                         a_util->matching = a2b[i];
 338                         b_util->matching = i;
 339                 }
 340
 341         free(cost);
 342         free(a2b);
 343         free(b2a);
 344 }
 345
 346 static void output_pair_header(struct diff_options *diffopt,
 347                                int patch_no_width,
 348                                struct strbuf *buf,
 349                                struct strbuf *dashes,
 350                                struct patch_util *a_util,
 351                                struct patch_util *b_util)
 352 {
 353         struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
 354         struct commit *commit;
 355         char status;
 356         const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
 357         const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
 358         const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
 359         const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
 360         const char *color;
 361
 362         if (!dashes->len)
 363                 strbuf_addchars(dashes, '-',
 364                                 strlen(find_unique_abbrev(oid,
 365                                                           DEFAULT_ABBREV)));
 366
 367         if (!b_util) {
 368                 color = color_old;
 369                 status = '<';
 370         } else if (!a_util) {
 371                 color = color_new;
 372                 status = '>';
 373         } else if (strcmp(a_util->patch, b_util->patch)) {
 374                 color = color_commit;
 375                 status = '!';
 376         } else {
 377                 color = color_commit;
 378                 status = '=';
 379         }
 380
 381         strbuf_reset(buf);
 382         strbuf_addstr(buf, status == '!' ? color_old : color);
 383         if (!a_util)
 384                 strbuf_addf(buf, "%*s:  %s ", patch_no_width, "-", dashes->buf);
 385         else
 386                 strbuf_addf(buf, "%*d:  %s ", patch_no_width, a_util->i + 1,
 387                             find_unique_abbrev(&a_util->oid, DEFAULT_ABBREV));
 388
 389         if (status == '!')
 390                 strbuf_addf(buf, "%s%s", color_reset, color);
 391         strbuf_addch(buf, status);
 392         if (status == '!')
 393                 strbuf_addf(buf, "%s%s", color_reset, color_new);
 394
 395         if (!b_util)
 396                 strbuf_addf(buf, " %*s:  %s", patch_no_width, "-", dashes->buf);
 397         else
 398                 strbuf_addf(buf, " %*d:  %s", patch_no_width, b_util->i + 1,
 399                             find_unique_abbrev(&b_util->oid, DEFAULT_ABBREV));
 400
 401         commit = lookup_commit_reference(the_repository, oid);
 402         if (commit) {
 403                 if (status == '!')
 404                         strbuf_addf(buf, "%s%s", color_reset, color);
 405
 406                 strbuf_addch(buf, ' ');
 407                 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
 408         }
 409         strbuf_addf(buf, "%s\n", color_reset);
 410
 411         fwrite(buf->buf, buf->len, 1, diffopt->file);
 412 }
 413
 414 static struct userdiff_driver section_headers = {
 415         .funcname = { "^ ## (.*) ##$\n"
 416                       "^.?@@ (.*)$", REG_EXTENDED }
 417 };
 418
 419 static struct diff_filespec *get_filespec(const char *name, const char *p)
 420 {
 421         struct diff_filespec *spec = alloc_filespec(name);
 422
 423         fill_filespec(spec, &null_oid, 0, 0100644);
 424         spec->data = (char *)p;
 425         spec->size = strlen(p);
 426         spec->should_munmap = 0;
 427         spec->is_stdin = 1;
 428         spec->driver = &section_headers;
 429
 430         return spec;
 431 }
 432
 433 static void patch_diff(const char *a, const char *b,
 434                        struct diff_options *diffopt)
 435 {
 436         diff_queue(&diff_queued_diff,
 437                    get_filespec("a", a), get_filespec("b", b));
 438
 439         diffcore_std(diffopt);
 440         diff_flush(diffopt);
 441 }
 442
 443 static void output(struct string_list *a, struct string_list *b,
 444                    struct diff_options *diffopt)
 445 {
 446         struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
 447         int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
 448         int i = 0, j = 0;
 449
 450         /*
 451          * We assume the user is really more interested in the second argument
 452          * ("newer" version). To that end, we print the output in the order of
 453          * the RHS (the `b` parameter). To put the LHS (the `a` parameter)
 454          * commits that are no longer in the RHS into a good place, we place
 455          * them once we have shown all of their predecessors in the LHS.
 456          */
 457
 458         while (i < a->nr || j < b->nr) {
 459                 struct patch_util *a_util, *b_util;
 460                 a_util = i < a->nr ? a->items[i].util : NULL;
 461                 b_util = j < b->nr ? b->items[j].util : NULL;
 462
 463                 /* Skip all the already-shown commits from the LHS. */
 464                 while (i < a->nr && a_util->shown)
 465                         a_util = ++i < a->nr ? a->items[i].util : NULL;
 466
 467                 /* Show unmatched LHS commit whose predecessors were shown. */
 468                 if (i < a->nr && a_util->matching < 0) {
 469                         output_pair_header(diffopt, patch_no_width,
 470                                            &buf, &dashes, a_util, NULL);
 471                         i++;
 472                         continue;
 473                 }
 474
 475                 /* Show unmatched RHS commits. */
 476                 while (j < b->nr && b_util->matching < 0) {
 477                         output_pair_header(diffopt, patch_no_width,
 478                                            &buf, &dashes, NULL, b_util);
 479                         b_util = ++j < b->nr ? b->items[j].util : NULL;
 480                 }
 481
 482                 /* Show matching LHS/RHS pair. */
 483                 if (j < b->nr) {
 484                         a_util = a->items[b_util->matching].util;
 485                         output_pair_header(diffopt, patch_no_width,
 486                                            &buf, &dashes, a_util, b_util);
 487                         if (!(diffopt->output_format & DIFF_FORMAT_NO_OUTPUT))
 488                                 patch_diff(a->items[b_util->matching].string,
 489                                            b->items[j].string, diffopt);
 490                         a_util->shown = 1;
 491                         j++;
 492                 }
 493         }
 494         strbuf_release(&buf);
 495         strbuf_release(&dashes);
 496 }
 497
 498 static struct strbuf *output_prefix_cb(struct diff_options *opt, void *data)
 499 {
 500         return data;
 501 }
 502
 503 int show_range_diff(const char *range1, const char *range2,
 504                     int creation_factor, int dual_color,
 505                     struct diff_options *diffopt)
 506 {
 507         int res = 0;
 508
 509         struct string_list branch1 = STRING_LIST_INIT_DUP;
 510         struct string_list branch2 = STRING_LIST_INIT_DUP;
 511
 512         if (read_patches(range1, &branch1))
 513                 res = error(_("could not parse log for '%s'"), range1);
 514         if (!res && read_patches(range2, &branch2))
 515                 res = error(_("could not parse log for '%s'"), range2);
 516
 517         if (!res) {
 518                 struct diff_options opts;
 519                 struct strbuf indent = STRBUF_INIT;
 520
 521                 if (diffopt)
 522                         memcpy(&opts, diffopt, sizeof(opts));
 523                 else
 524                         diff_setup(&opts);
 525
 526                 if (!opts.output_format)
 527                         opts.output_format = DIFF_FORMAT_PATCH;
 528                 opts.flags.suppress_diff_headers = 1;
 529                 opts.flags.dual_color_diffed_diffs = dual_color;
 530                 opts.flags.suppress_hunk_header_line_count = 1;
 531                 opts.output_prefix = output_prefix_cb;
 532                 strbuf_addstr(&indent, "    ");
 533                 opts.output_prefix_data = &indent;
 534                 diff_setup_done(&opts);
 535
 536                 find_exact_matches(&branch1, &branch2);
 537                 get_correspondences(&branch1, &branch2, creation_factor);
 538                 output(&branch1, &branch2, &opts);
 539
 540                 strbuf_release(&indent);
 541         }
 542
 543         string_list_clear(&branch1, 1);
 544         string_list_clear(&branch2, 1);
 545
 546         return res;
 547 }