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 (skip_prefix(line, "@@ ", &p)) {
 123                         p = strstr(p, "@@");
 124                         strbuf_addstr(&buf, p ? p : "@@");
 125                 } else if (!line[0] || starts_with(line, "index "))
 126                         /*
 127                          * A completely blank (not ' \n', which is context)
 128                          * line is not valid in a diff.  We skip it
 129                          * silently, because this neatly handles the blank
 130                          * separator line between commits in git-log
 131                          * output.
 132                          *
 133                          * We also want to ignore the diff's `index` lines
 134                          * because they contain exact blob hashes in which
 135                          * we are not interested.
 136                          */
 137                         continue;
 138                 else if (line[0] == '>') {
 139                         strbuf_addch(&buf, '+');
 140                         strbuf_addstr(&buf, line + 1);
 141                 } else if (line[0] == '<') {
 142                         strbuf_addch(&buf, '-');
 143                         strbuf_addstr(&buf, line + 1);
 144                 } else if (line[0] == '#') {
 145                         strbuf_addch(&buf, ' ');
 146                         strbuf_addstr(&buf, line + 1);
 147                 } else {
 148                         strbuf_addch(&buf, ' ');
 149                         strbuf_addstr(&buf, line);
 150                 }
 151
 152                 strbuf_addch(&buf, '\n');
 153                 util->diffsize++;
 154         }
 155         strbuf_release(&contents);
 156
 157         if (util)
 158                 string_list_append(list, buf.buf)->util = util;
 159         strbuf_release(&buf);
 160
 161         if (finish_command(&cp))
 162                 return -1;
 163
 164         return 0;
 165 }
 166
 167 static int patch_util_cmp(const void *dummy, const struct patch_util *a,
 168                           const struct patch_util *b, const char *keydata)
 169 {
 170         return strcmp(a->diff, keydata ? keydata : b->diff);
 171 }
 172
 173 static void find_exact_matches(struct string_list *a, struct string_list *b)
 174 {
 175         struct hashmap map;
 176         int i;
 177
 178         hashmap_init(&map, (hashmap_cmp_fn)patch_util_cmp, NULL, 0);
 179
 180         /* First, add the patches of a to a hash map */
 181         for (i = 0; i < a->nr; i++) {
 182                 struct patch_util *util = a->items[i].util;
 183
 184                 util->i = i;
 185                 util->patch = a->items[i].string;
 186                 util->diff = util->patch + util->diff_offset;
 187                 hashmap_entry_init(util, strhash(util->diff));
 188                 hashmap_add(&map, util);
 189         }
 190
 191         /* Now try to find exact matches in b */
 192         for (i = 0; i < b->nr; i++) {
 193                 struct patch_util *util = b->items[i].util, *other;
 194
 195                 util->i = i;
 196                 util->patch = b->items[i].string;
 197                 util->diff = util->patch + util->diff_offset;
 198                 hashmap_entry_init(util, strhash(util->diff));
 199                 other = hashmap_remove(&map, util, NULL);
 200                 if (other) {
 201                         if (other->matching >= 0)
 202                                 BUG("already assigned!");
 203
 204                         other->matching = i;
 205                         util->matching = other->i;
 206                 }
 207         }
 208
 209         hashmap_free(&map, 0);
 210 }
 211
 212 static void diffsize_consume(void *data, char *line, unsigned long len)
 213 {
 214         (*(int *)data)++;
 215 }
 216
 217 static void diffsize_hunk(void *data, long ob, long on, long nb, long nn,
 218                           const char *funcline, long funclen)
 219 {
 220         diffsize_consume(data, NULL, 0);
 221 }
 222
 223 static int diffsize(const char *a, const char *b)
 224 {
 225         xpparam_t pp = { 0 };
 226         xdemitconf_t cfg = { 0 };
 227         mmfile_t mf1, mf2;
 228         int count = 0;
 229
 230         mf1.ptr = (char *)a;
 231         mf1.size = strlen(a);
 232         mf2.ptr = (char *)b;
 233         mf2.size = strlen(b);
 234
 235         cfg.ctxlen = 3;
 236         if (!xdi_diff_outf(&mf1, &mf2,
 237                            diffsize_hunk, diffsize_consume, &count,
 238                            &pp, &cfg))
 239                 return count;
 240
 241         error(_("failed to generate diff"));
 242         return COST_MAX;
 243 }
 244
 245 static void get_correspondences(struct string_list *a, struct string_list *b,
 246                                 int creation_factor)
 247 {
 248         int n = a->nr + b->nr;
 249         int *cost, c, *a2b, *b2a;
 250         int i, j;
 251
 252         ALLOC_ARRAY(cost, st_mult(n, n));
 253         ALLOC_ARRAY(a2b, n);
 254         ALLOC_ARRAY(b2a, n);
 255
 256         for (i = 0; i < a->nr; i++) {
 257                 struct patch_util *a_util = a->items[i].util;
 258
 259                 for (j = 0; j < b->nr; j++) {
 260                         struct patch_util *b_util = b->items[j].util;
 261
 262                         if (a_util->matching == j)
 263                                 c = 0;
 264                         else if (a_util->matching < 0 && b_util->matching < 0)
 265                                 c = diffsize(a_util->diff, b_util->diff);
 266                         else
 267                                 c = COST_MAX;
 268                         cost[i + n * j] = c;
 269                 }
 270
 271                 c = a_util->matching < 0 ?
 272                         a_util->diffsize * creation_factor / 100 : COST_MAX;
 273                 for (j = b->nr; j < n; j++)
 274                         cost[i + n * j] = c;
 275         }
 276
 277         for (j = 0; j < b->nr; j++) {
 278                 struct patch_util *util = b->items[j].util;
 279
 280                 c = util->matching < 0 ?
 281                         util->diffsize * creation_factor / 100 : COST_MAX;
 282                 for (i = a->nr; i < n; i++)
 283                         cost[i + n * j] = c;
 284         }
 285
 286         for (i = a->nr; i < n; i++)
 287                 for (j = b->nr; j < n; j++)
 288                         cost[i + n * j] = 0;
 289
 290         compute_assignment(n, n, cost, a2b, b2a);
 291
 292         for (i = 0; i < a->nr; i++)
 293                 if (a2b[i] >= 0 && a2b[i] < b->nr) {
 294                         struct patch_util *a_util = a->items[i].util;
 295                         struct patch_util *b_util = b->items[a2b[i]].util;
 296
 297                         a_util->matching = a2b[i];
 298                         b_util->matching = i;
 299                 }
 300
 301         free(cost);
 302         free(a2b);
 303         free(b2a);
 304 }
 305
 306 static void output_pair_header(struct diff_options *diffopt,
 307                                int patch_no_width,
 308                                struct strbuf *buf,
 309                                struct strbuf *dashes,
 310                                struct patch_util *a_util,
 311                                struct patch_util *b_util)
 312 {
 313         struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
 314         struct commit *commit;
 315         char status;
 316         const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
 317         const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
 318         const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
 319         const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
 320         const char *color;
 321
 322         if (!dashes->len)
 323                 strbuf_addchars(dashes, '-',
 324                                 strlen(find_unique_abbrev(oid,
 325                                                           DEFAULT_ABBREV)));
 326
 327         if (!b_util) {
 328                 color = color_old;
 329                 status = '<';
 330         } else if (!a_util) {
 331                 color = color_new;
 332                 status = '>';
 333         } else if (strcmp(a_util->patch, b_util->patch)) {
 334                 color = color_commit;
 335                 status = '!';
 336         } else {
 337                 color = color_commit;
 338                 status = '=';
 339         }
 340
 341         strbuf_reset(buf);
 342         strbuf_addstr(buf, status == '!' ? color_old : color);
 343         if (!a_util)
 344                 strbuf_addf(buf, "%*s:  %s ", patch_no_width, "-", dashes->buf);
 345         else
 346                 strbuf_addf(buf, "%*d:  %s ", patch_no_width, a_util->i + 1,
 347                             find_unique_abbrev(&a_util->oid, DEFAULT_ABBREV));
 348
 349         if (status == '!')
 350                 strbuf_addf(buf, "%s%s", color_reset, color);
 351         strbuf_addch(buf, status);
 352         if (status == '!')
 353                 strbuf_addf(buf, "%s%s", color_reset, color_new);
 354
 355         if (!b_util)
 356                 strbuf_addf(buf, " %*s:  %s", patch_no_width, "-", dashes->buf);
 357         else
 358                 strbuf_addf(buf, " %*d:  %s", patch_no_width, b_util->i + 1,
 359                             find_unique_abbrev(&b_util->oid, DEFAULT_ABBREV));
 360
 361         commit = lookup_commit_reference(the_repository, oid);
 362         if (commit) {
 363                 if (status == '!')
 364                         strbuf_addf(buf, "%s%s", color_reset, color);
 365
 366                 strbuf_addch(buf, ' ');
 367                 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
 368         }
 369         strbuf_addf(buf, "%s\n", color_reset);
 370
 371         fwrite(buf->buf, buf->len, 1, diffopt->file);
 372 }
 373
 374 static struct userdiff_driver no_func_name = {
 375         .funcname = { "$^", 0 }
 376 };
 377
 378 static struct diff_filespec *get_filespec(const char *name, const char *p)
 379 {
 380         struct diff_filespec *spec = alloc_filespec(name);
 381
 382         fill_filespec(spec, &null_oid, 0, 0100644);
 383         spec->data = (char *)p;
 384         spec->size = strlen(p);
 385         spec->should_munmap = 0;
 386         spec->is_stdin = 1;
 387         spec->driver = &no_func_name;
 388
 389         return spec;
 390 }
 391
 392 static void patch_diff(const char *a, const char *b,
 393                        struct diff_options *diffopt)
 394 {
 395         diff_queue(&diff_queued_diff,
 396                    get_filespec("a", a), get_filespec("b", b));
 397
 398         diffcore_std(diffopt);
 399         diff_flush(diffopt);
 400 }
 401
 402 static void output(struct string_list *a, struct string_list *b,
 403                    struct diff_options *diffopt)
 404 {
 405         struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
 406         int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
 407         int i = 0, j = 0;
 408
 409         /*
 410          * We assume the user is really more interested in the second argument
 411          * ("newer" version). To that end, we print the output in the order of
 412          * the RHS (the `b` parameter). To put the LHS (the `a` parameter)
 413          * commits that are no longer in the RHS into a good place, we place
 414          * them once we have shown all of their predecessors in the LHS.
 415          */
 416
 417         while (i < a->nr || j < b->nr) {
 418                 struct patch_util *a_util, *b_util;
 419                 a_util = i < a->nr ? a->items[i].util : NULL;
 420                 b_util = j < b->nr ? b->items[j].util : NULL;
 421
 422                 /* Skip all the already-shown commits from the LHS. */
 423                 while (i < a->nr && a_util->shown)
 424                         a_util = ++i < a->nr ? a->items[i].util : NULL;
 425
 426                 /* Show unmatched LHS commit whose predecessors were shown. */
 427                 if (i < a->nr && a_util->matching < 0) {
 428                         output_pair_header(diffopt, patch_no_width,
 429                                            &buf, &dashes, a_util, NULL);
 430                         i++;
 431                         continue;
 432                 }
 433
 434                 /* Show unmatched RHS commits. */
 435                 while (j < b->nr && b_util->matching < 0) {
 436                         output_pair_header(diffopt, patch_no_width,
 437                                            &buf, &dashes, NULL, b_util);
 438                         b_util = ++j < b->nr ? b->items[j].util : NULL;
 439                 }
 440
 441                 /* Show matching LHS/RHS pair. */
 442                 if (j < b->nr) {
 443                         a_util = a->items[b_util->matching].util;
 444                         output_pair_header(diffopt, patch_no_width,
 445                                            &buf, &dashes, a_util, b_util);
 446                         if (!(diffopt->output_format & DIFF_FORMAT_NO_OUTPUT))
 447                                 patch_diff(a->items[b_util->matching].string,
 448                                            b->items[j].string, diffopt);
 449                         a_util->shown = 1;
 450                         j++;
 451                 }
 452         }
 453         strbuf_release(&buf);
 454         strbuf_release(&dashes);
 455 }
 456
 457 static struct strbuf *output_prefix_cb(struct diff_options *opt, void *data)
 458 {
 459         return data;
 460 }
 461
 462 int show_range_diff(const char *range1, const char *range2,
 463                     int creation_factor, int dual_color,
 464                     struct diff_options *diffopt)
 465 {
 466         int res = 0;
 467
 468         struct string_list branch1 = STRING_LIST_INIT_DUP;
 469         struct string_list branch2 = STRING_LIST_INIT_DUP;
 470
 471         if (read_patches(range1, &branch1))
 472                 res = error(_("could not parse log for '%s'"), range1);
 473         if (!res && read_patches(range2, &branch2))
 474                 res = error(_("could not parse log for '%s'"), range2);
 475
 476         if (!res) {
 477                 struct diff_options opts;
 478                 struct strbuf indent = STRBUF_INIT;
 479
 480                 if (diffopt)
 481                         memcpy(&opts, diffopt, sizeof(opts));
 482                 else
 483                         diff_setup(&opts);
 484
 485                 if (!opts.output_format)
 486                         opts.output_format = DIFF_FORMAT_PATCH;
 487                 opts.flags.suppress_diff_headers = 1;
 488                 opts.flags.dual_color_diffed_diffs = dual_color;
 489                 opts.output_prefix = output_prefix_cb;
 490                 strbuf_addstr(&indent, "    ");
 491                 opts.output_prefix_data = &indent;
 492                 diff_setup_done(&opts);
 493
 494                 find_exact_matches(&branch1, &branch2);
 495                 get_correspondences(&branch1, &branch2, creation_factor);
 496                 output(&branch1, &branch2, &opts);
 497
 498                 strbuf_release(&indent);
 499         }
 500
 501         string_list_clear(&branch1, 1);
 502         string_list_clear(&branch2, 1);
 503
 504         return res;
 505 }