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