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
  14 struct patch_util {
  15         /* For the search for an exact match */
  16         struct hashmap_entry e;
  17         const char *diff, *patch;
  18
  19         int i, shown;
  20         int diffsize;
  21         size_t diff_offset;
  22         /* the index of the matching item in the other branch, or -1 */
  23         int matching;
  24         struct object_id oid;
  25 };
  26
  27 static size_t find_end_of_line(char *buffer, unsigned long size)
  28 {
  29         char *eol = memchr(buffer, '\n', size);
  30
  31         if (!eol)
  32                 return size;
  33
  34         *eol = '\0';
  35         return eol + 1 - buffer;
  36 }
  37
  38 /*
  39  * Reads the patches into a string list, with the `util` field being populated
  40  * as struct object_id (will need to be free()d).
  41  */
  42 static int read_patches(const char *range, struct string_list *list)
  43 {
  44         struct child_process cp = CHILD_PROCESS_INIT;
  45         struct strbuf buf = STRBUF_INIT, contents = STRBUF_INIT;
  46         struct patch_util *util = NULL;
  47         int in_header = 1;
  48         char *line;
  49         int offset, len;
  50         size_t size;
  51
  52         argv_array_pushl(&cp.args, "log", "--no-color", "-p", "--no-merges",
  53                         "--reverse", "--date-order", "--decorate=no",
  54                         /*
  55                          * Choose indicators that are not used anywhere
  56                          * else in diffs, but still look reasonable
  57                          * (e.g. will not be confusing when debugging)
  58                          */
  59                         "--output-indicator-new=>",
  60                         "--output-indicator-old=<",
  61                         "--output-indicator-context=#",
  62                         "--no-abbrev-commit", range,
  63                         NULL);
  64         cp.out = -1;
  65         cp.no_stdin = 1;
  66         cp.git_cmd = 1;
  67
  68         if (start_command(&cp))
  69                 return error_errno(_("could not start `log`"));
  70         if (strbuf_read(&contents, cp.out, 0) < 0) {
  71                 error_errno(_("could not read `log` output"));
  72                 finish_command(&cp);
  73                 return -1;
  74         }
  75
  76         line = contents.buf;
  77         size = contents.len;
  78         for (offset = 0; size > 0; offset += len, size -= len, line += len) {
  79                 const char *p;
  80
  81                 len = find_end_of_line(line, size);
  82                 line[len - 1] = '\0';
  83                 if (skip_prefix(line, "commit ", &p)) {
  84                         if (util) {
  85                                 string_list_append(list, buf.buf)->util = util;
  86                                 strbuf_reset(&buf);
  87                         }
  88                         util = xcalloc(sizeof(*util), 1);
  89                         if (get_oid(p, &util->oid)) {
  90                                 error(_("could not parse commit '%s'"), p);
  91                                 free(util);
  92                                 string_list_clear(list, 1);
  93                                 strbuf_release(&buf);
  94                                 strbuf_release(&contents);
  95                                 finish_command(&cp);
  96                                 return -1;
  97                         }
  98                         util->matching = -1;
  99                         in_header = 1;
 100                         continue;
 101                 }
 102
 103                 if (starts_with(line, "diff --git")) {
 104                         in_header = 0;
 105                         strbuf_addch(&buf, '\n');
 106                         if (!util->diff_offset)
 107                                 util->diff_offset = buf.len;
 108                         strbuf_addch(&buf, ' ');
 109                         strbuf_addstr(&buf, line);
 110                 } else if (in_header) {
 111                         if (starts_with(line, "Author: ")) {
 112                                 strbuf_addstr(&buf, line);
 113                                 strbuf_addstr(&buf, "\n\n");
 114                         } else if (starts_with(line, "    ")) {
 115                                 p = line + len - 2;
 116                                 while (isspace(*p) && p >= line)
 117                                         p--;
 118                                 strbuf_add(&buf, line, p - line + 1);
 119                                 strbuf_addch(&buf, '\n');
 120                         }
 121                         continue;
 122                 } else if (starts_with(line, "@@ "))
 123                         strbuf_addstr(&buf, "@@");
 124                 else if (!line[0] || starts_with(line, "index "))
 125                         /*
 126                          * A completely blank (not ' \n', which is context)
 127                          * line is not valid in a diff.  We skip it
 128                          * silently, because this neatly handles the blank
 129                          * separator line between commits in git-log
 130                          * output.
 131                          *
 132                          * We also want to ignore the diff's `index` lines
 133                          * because they contain exact blob hashes in which
 134                          * we are not interested.
 135                          */
 136                         continue;
 137                 else if (line[0] == '>') {
 138                         strbuf_addch(&buf, '+');
 139                         strbuf_addstr(&buf, line + 1);
 140                 } else if (line[0] == '<') {
 141                         strbuf_addch(&buf, '-');
 142                         strbuf_addstr(&buf, line + 1);
 143                 } else if (line[0] == '#') {
 144                         strbuf_addch(&buf, ' ');
 145                         strbuf_addstr(&buf, line + 1);
 146                 } else {
 147                         strbuf_addch(&buf, ' ');
 148                         strbuf_addstr(&buf, line);
 149                 }
 150
 151                 strbuf_addch(&buf, '\n');
 152                 util->diffsize++;
 153         }
 154         strbuf_release(&contents);
 155
 156         if (util)
 157                 string_list_append(list, buf.buf)->util = util;
 158         strbuf_release(&buf);
 159
 160         if (finish_command(&cp))
 161                 return -1;
 162
 163         return 0;
 164 }
 165
 166 static int patch_util_cmp(const void *dummy, const struct patch_util *a,
 167                           const struct patch_util *b, const char *keydata)
 168 {
 169         return strcmp(a->diff, keydata ? keydata : b->diff);
 170 }
 171
 172 static void find_exact_matches(struct string_list *a, struct string_list *b)
 173 {
 174         struct hashmap map;
 175         int i;
 176
 177         hashmap_init(&map, (hashmap_cmp_fn)patch_util_cmp, NULL, 0);
 178
 179         /* First, add the patches of a to a hash map */
 180         for (i = 0; i < a->nr; i++) {
 181                 struct patch_util *util = a->items[i].util;
 182
 183                 util->i = i;
 184                 util->patch = a->items[i].string;
 185                 util->diff = util->patch + util->diff_offset;
 186                 hashmap_entry_init(util, strhash(util->diff));
 187                 hashmap_add(&map, util);
 188         }
 189
 190         /* Now try to find exact matches in b */
 191         for (i = 0; i < b->nr; i++) {
 192                 struct patch_util *util = b->items[i].util, *other;
 193
 194                 util->i = i;
 195                 util->patch = b->items[i].string;
 196                 util->diff = util->patch + util->diff_offset;
 197                 hashmap_entry_init(util, strhash(util->diff));
 198                 other = hashmap_remove(&map, util, NULL);
 199                 if (other) {
 200                         if (other->matching >= 0)
 201                                 BUG("already assigned!");
 202
 203                         other->matching = i;
 204                         util->matching = other->i;
 205                 }
 206         }
 207
 208         hashmap_free(&map, 0);
 209 }
 210
 211 static void diffsize_consume(void *data, char *line, unsigned long len)
 212 {
 213         (*(int *)data)++;
 214 }
 215
 216 static void diffsize_hunk(void *data, long ob, long on, long nb, long nn,
 217                           const char *funcline, long funclen)
 218 {
 219         diffsize_consume(data, NULL, 0);
 220 }
 221
 222 static int diffsize(const char *a, const char *b)
 223 {
 224         xpparam_t pp = { 0 };
 225         xdemitconf_t cfg = { 0 };
 226         mmfile_t mf1, mf2;
 227         int count = 0;
 228
 229         mf1.ptr = (char *)a;
 230         mf1.size = strlen(a);
 231         mf2.ptr = (char *)b;
 232         mf2.size = strlen(b);
 233
 234         cfg.ctxlen = 3;
 235         if (!xdi_diff_outf(&mf1, &mf2,
 236                            diffsize_hunk, diffsize_consume, &count,
 237                            &pp, &cfg))
 238                 return count;
 239
 240         error(_("failed to generate diff"));
 241         return COST_MAX;
 242 }
 243
 244 static void get_correspondences(struct string_list *a, struct string_list *b,
 245                                 int creation_factor)
 246 {
 247         int n = a->nr + b->nr;
 248         int *cost, c, *a2b, *b2a;
 249         int i, j;
 250
 251         ALLOC_ARRAY(cost, st_mult(n, n));
 252         ALLOC_ARRAY(a2b, n);
 253         ALLOC_ARRAY(b2a, n);
 254
 255         for (i = 0; i < a->nr; i++) {
 256                 struct patch_util *a_util = a->items[i].util;
 257
 258                 for (j = 0; j < b->nr; j++) {
 259                         struct patch_util *b_util = b->items[j].util;
 260
 261                         if (a_util->matching == j)
 262                                 c = 0;
 263                         else if (a_util->matching < 0 && b_util->matching < 0)
 264                                 c = diffsize(a_util->diff, b_util->diff);
 265                         else
 266                                 c = COST_MAX;
 267                         cost[i + n * j] = c;
 268                 }
 269
 270                 c = a_util->matching < 0 ?
 271                         a_util->diffsize * creation_factor / 100 : COST_MAX;
 272                 for (j = b->nr; j < n; j++)
 273                         cost[i + n * j] = c;
 274         }
 275
 276         for (j = 0; j < b->nr; j++) {
 277                 struct patch_util *util = b->items[j].util;
 278
 279                 c = util->matching < 0 ?
 280                         util->diffsize * creation_factor / 100 : COST_MAX;
 281                 for (i = a->nr; i < n; i++)
 282                         cost[i + n * j] = c;
 283         }
 284
 285         for (i = a->nr; i < n; i++)
 286                 for (j = b->nr; j < n; j++)
 287                         cost[i + n * j] = 0;
 288
 289         compute_assignment(n, n, cost, a2b, b2a);
 290
 291         for (i = 0; i < a->nr; i++)
 292                 if (a2b[i] >= 0 && a2b[i] < b->nr) {
 293                         struct patch_util *a_util = a->items[i].util;
 294                         struct patch_util *b_util = b->items[a2b[i]].util;
 295
 296                         a_util->matching = a2b[i];
 297                         b_util->matching = i;
 298                 }
 299
 300         free(cost);
 301         free(a2b);
 302         free(b2a);
 303 }
 304
 305 static void output_pair_header(struct diff_options *diffopt,
 306                                int patch_no_width,
 307                                struct strbuf *buf,
 308                                struct strbuf *dashes,
 309                                struct patch_util *a_util,
 310                                struct patch_util *b_util)
 311 {
 312         struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
 313         struct commit *commit;
 314         char status;
 315         const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
 316         const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
 317         const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
 318         const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
 319         const char *color;
 320
 321         if (!dashes->len)
 322                 strbuf_addchars(dashes, '-',
 323                                 strlen(find_unique_abbrev(oid,
 324                                                           DEFAULT_ABBREV)));
 325
 326         if (!b_util) {
 327                 color = color_old;
 328                 status = '<';
 329         } else if (!a_util) {
 330                 color = color_new;
 331                 status = '>';
 332         } else if (strcmp(a_util->patch, b_util->patch)) {
 333                 color = color_commit;
 334                 status = '!';
 335         } else {
 336                 color = color_commit;
 337                 status = '=';
 338         }
 339
 340         strbuf_reset(buf);
 341         strbuf_addstr(buf, status == '!' ? color_old : color);
 342         if (!a_util)
 343                 strbuf_addf(buf, "%*s:  %s ", patch_no_width, "-", dashes->buf);
 344         else
 345                 strbuf_addf(buf, "%*d:  %s ", patch_no_width, a_util->i + 1,
 346                             find_unique_abbrev(&a_util->oid, DEFAULT_ABBREV));
 347
 348         if (status == '!')
 349                 strbuf_addf(buf, "%s%s", color_reset, color);
 350         strbuf_addch(buf, status);
 351         if (status == '!')
 352                 strbuf_addf(buf, "%s%s", color_reset, color_new);
 353
 354         if (!b_util)
 355                 strbuf_addf(buf, " %*s:  %s", patch_no_width, "-", dashes->buf);
 356         else
 357                 strbuf_addf(buf, " %*d:  %s", patch_no_width, b_util->i + 1,
 358                             find_unique_abbrev(&b_util->oid, DEFAULT_ABBREV));
 359
 360         commit = lookup_commit_reference(the_repository, oid);
 361         if (commit) {
 362                 if (status == '!')
 363                         strbuf_addf(buf, "%s%s", color_reset, color);
 364
 365                 strbuf_addch(buf, ' ');
 366                 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
 367         }
 368         strbuf_addf(buf, "%s\n", color_reset);
 369
 370         fwrite(buf->buf, buf->len, 1, diffopt->file);
 371 }
 372
 373 static struct userdiff_driver no_func_name = {
 374         .funcname = { "$^", 0 }
 375 };
 376
 377 static struct diff_filespec *get_filespec(const char *name, const char *p)
 378 {
 379         struct diff_filespec *spec = alloc_filespec(name);
 380
 381         fill_filespec(spec, &null_oid, 0, 0100644);
 382         spec->data = (char *)p;
 383         spec->size = strlen(p);
 384         spec->should_munmap = 0;
 385         spec->is_stdin = 1;
 386         spec->driver = &no_func_name;
 387
 388         return spec;
 389 }
 390
 391 static void patch_diff(const char *a, const char *b,
 392                        struct diff_options *diffopt)
 393 {
 394         diff_queue(&diff_queued_diff,
 395                    get_filespec("a", a), get_filespec("b", b));
 396
 397         diffcore_std(diffopt);
 398         diff_flush(diffopt);
 399 }
 400
 401 static void output(struct string_list *a, struct string_list *b,
 402                    struct diff_options *diffopt)
 403 {
 404         struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
 405         int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
 406         int i = 0, j = 0;
 407
 408         /*
 409          * We assume the user is really more interested in the second argument
 410          * ("newer" version). To that end, we print the output in the order of
 411          * the RHS (the `b` parameter). To put the LHS (the `a` parameter)
 412          * commits that are no longer in the RHS into a good place, we place
 413          * them once we have shown all of their predecessors in the LHS.
 414          */
 415
 416         while (i < a->nr || j < b->nr) {
 417                 struct patch_util *a_util, *b_util;
 418                 a_util = i < a->nr ? a->items[i].util : NULL;
 419                 b_util = j < b->nr ? b->items[j].util : NULL;
 420
 421                 /* Skip all the already-shown commits from the LHS. */
 422                 while (i < a->nr && a_util->shown)
 423                         a_util = ++i < a->nr ? a->items[i].util : NULL;
 424
 425                 /* Show unmatched LHS commit whose predecessors were shown. */
 426                 if (i < a->nr && a_util->matching < 0) {
 427                         output_pair_header(diffopt, patch_no_width,
 428                                            &buf, &dashes, a_util, NULL);
 429                         i++;
 430                         continue;
 431                 }
 432
 433                 /* Show unmatched RHS commits. */
 434                 while (j < b->nr && b_util->matching < 0) {
 435                         output_pair_header(diffopt, patch_no_width,
 436                                            &buf, &dashes, NULL, b_util);
 437                         b_util = ++j < b->nr ? b->items[j].util : NULL;
 438                 }
 439
 440                 /* Show matching LHS/RHS pair. */
 441                 if (j < b->nr) {
 442                         a_util = a->items[b_util->matching].util;
 443                         output_pair_header(diffopt, patch_no_width,
 444                                            &buf, &dashes, a_util, b_util);
 445                         if (!(diffopt->output_format & DIFF_FORMAT_NO_OUTPUT))
 446                                 patch_diff(a->items[b_util->matching].string,
 447                                            b->items[j].string, diffopt);
 448                         a_util->shown = 1;
 449                         j++;
 450                 }
 451         }
 452         strbuf_release(&buf);
 453         strbuf_release(&dashes);
 454 }
 455
 456 static struct strbuf *output_prefix_cb(struct diff_options *opt, void *data)
 457 {
 458         return data;
 459 }
 460
 461 int show_range_diff(const char *range1, const char *range2,
 462                     int creation_factor, int dual_color,
 463                     struct diff_options *diffopt)
 464 {
 465         int res = 0;
 466
 467         struct string_list branch1 = STRING_LIST_INIT_DUP;
 468         struct string_list branch2 = STRING_LIST_INIT_DUP;
 469
 470         if (read_patches(range1, &branch1))
 471                 res = error(_("could not parse log for '%s'"), range1);
 472         if (!res && read_patches(range2, &branch2))
 473                 res = error(_("could not parse log for '%s'"), range2);
 474
 475         if (!res) {
 476                 struct diff_options opts;
 477                 struct strbuf indent = STRBUF_INIT;
 478
 479                 if (diffopt)
 480                         memcpy(&opts, diffopt, sizeof(opts));
 481                 else
 482                         diff_setup(&opts);
 483
 484                 if (!opts.output_format)
 485                         opts.output_format = DIFF_FORMAT_PATCH;
 486                 opts.flags.suppress_diff_headers = 1;
 487                 opts.flags.dual_color_diffed_diffs = dual_color;
 488                 opts.output_prefix = output_prefix_cb;
 489                 strbuf_addstr(&indent, "    ");
 490                 opts.output_prefix_data = &indent;
 491                 diff_setup_done(&opts);
 492
 493                 find_exact_matches(&branch1, &branch2);
 494                 get_correspondences(&branch1, &branch2, creation_factor);
 495                 output(&branch1, &branch2, &opts);
 496
 497                 strbuf_release(&indent);
 498         }
 499
 500         string_list_clear(&branch1, 1);
 501         string_list_clear(&branch2, 1);
 502
 503         return res;
 504 }