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 /*
  28  * Reads the patches into a string list, with the `util` field being populated
  29  * as struct object_id (will need to be free()d).
  30  */
  31 static int read_patches(const char *range, struct string_list *list)
  32 {
  33         struct child_process cp = CHILD_PROCESS_INIT;
  34         FILE *in;
  35         struct strbuf buf = STRBUF_INIT, line = STRBUF_INIT;
  36         struct patch_util *util = NULL;
  37         int in_header = 1;
  38
  39         argv_array_pushl(&cp.args, "log", "--no-color", "-p", "--no-merges",
  40                         "--reverse", "--date-order", "--decorate=no",
  41                         /*
  42                          * Choose indicators that are not used anywhere
  43                          * else in diffs, but still look reasonable
  44                          * (e.g. will not be confusing when debugging)
  45                          */
  46                         "--output-indicator-new=>",
  47                         "--output-indicator-old=<",
  48                         "--output-indicator-context=#",
  49                         "--no-abbrev-commit", range,
  50                         NULL);
  51         cp.out = -1;
  52         cp.no_stdin = 1;
  53         cp.git_cmd = 1;
  54
  55         if (start_command(&cp))
  56                 return error_errno(_("could not start `log`"));
  57         in = fdopen(cp.out, "r");
  58         if (!in) {
  59                 error_errno(_("could not read `log` output"));
  60                 finish_command(&cp);
  61                 return -1;
  62         }
  63
  64         while (strbuf_getline(&line, in) != EOF) {
  65                 const char *p;
  66
  67                 if (skip_prefix(line.buf, "commit ", &p)) {
  68                         if (util) {
  69                                 string_list_append(list, buf.buf)->util = util;
  70                                 strbuf_reset(&buf);
  71                         }
  72                         util = xcalloc(sizeof(*util), 1);
  73                         if (get_oid(p, &util->oid)) {
  74                                 error(_("could not parse commit '%s'"), p);
  75                                 free(util);
  76                                 string_list_clear(list, 1);
  77                                 strbuf_release(&buf);
  78                                 strbuf_release(&line);
  79                                 fclose(in);
  80                                 finish_command(&cp);
  81                                 return -1;
  82                         }
  83                         util->matching = -1;
  84                         in_header = 1;
  85                         continue;
  86                 }
  87
  88                 if (starts_with(line.buf, "diff --git")) {
  89                         in_header = 0;
  90                         strbuf_addch(&buf, '\n');
  91                         if (!util->diff_offset)
  92                                 util->diff_offset = buf.len;
  93                         strbuf_addch(&buf, ' ');
  94                         strbuf_addbuf(&buf, &line);
  95                 } else if (in_header) {
  96                         if (starts_with(line.buf, "Author: ")) {
  97                                 strbuf_addbuf(&buf, &line);
  98                                 strbuf_addstr(&buf, "\n\n");
  99                         } else if (starts_with(line.buf, "    ")) {
 100                                 strbuf_rtrim(&line);
 101                                 strbuf_addbuf(&buf, &line);
 102                                 strbuf_addch(&buf, '\n');
 103                         }
 104                         continue;
 105                 } else if (starts_with(line.buf, "@@ "))
 106                         strbuf_addstr(&buf, "@@");
 107                 else if (!line.buf[0] || starts_with(line.buf, "index "))
 108                         /*
 109                          * A completely blank (not ' \n', which is context)
 110                          * line is not valid in a diff.  We skip it
 111                          * silently, because this neatly handles the blank
 112                          * separator line between commits in git-log
 113                          * output.
 114                          *
 115                          * We also want to ignore the diff's `index` lines
 116                          * because they contain exact blob hashes in which
 117                          * we are not interested.
 118                          */
 119                         continue;
 120                 else if (line.buf[0] == '>') {
 121                         strbuf_addch(&buf, '+');
 122                         strbuf_add(&buf, line.buf + 1, line.len - 1);
 123                 } else if (line.buf[0] == '<') {
 124                         strbuf_addch(&buf, '-');
 125                         strbuf_add(&buf, line.buf + 1, line.len - 1);
 126                 } else if (line.buf[0] == '#') {
 127                         strbuf_addch(&buf, ' ');
 128                         strbuf_add(&buf, line.buf + 1, line.len - 1);
 129                 } else {
 130                         strbuf_addch(&buf, ' ');
 131                         strbuf_addbuf(&buf, &line);
 132                 }
 133
 134                 strbuf_addch(&buf, '\n');
 135                 util->diffsize++;
 136         }
 137         fclose(in);
 138         strbuf_release(&line);
 139
 140         if (util)
 141                 string_list_append(list, buf.buf)->util = util;
 142         strbuf_release(&buf);
 143
 144         if (finish_command(&cp))
 145                 return -1;
 146
 147         return 0;
 148 }
 149
 150 static int patch_util_cmp(const void *dummy, const struct patch_util *a,
 151                      const struct patch_util *b, const char *keydata)
 152 {
 153         return strcmp(a->diff, keydata ? keydata : b->diff);
 154 }
 155
 156 static void find_exact_matches(struct string_list *a, struct string_list *b)
 157 {
 158         struct hashmap map;
 159         int i;
 160
 161         hashmap_init(&map, (hashmap_cmp_fn)patch_util_cmp, NULL, 0);
 162
 163         /* First, add the patches of a to a hash map */
 164         for (i = 0; i < a->nr; i++) {
 165                 struct patch_util *util = a->items[i].util;
 166
 167                 util->i = i;
 168                 util->patch = a->items[i].string;
 169                 util->diff = util->patch + util->diff_offset;
 170                 hashmap_entry_init(util, strhash(util->diff));
 171                 hashmap_add(&map, util);
 172         }
 173
 174         /* Now try to find exact matches in b */
 175         for (i = 0; i < b->nr; i++) {
 176                 struct patch_util *util = b->items[i].util, *other;
 177
 178                 util->i = i;
 179                 util->patch = b->items[i].string;
 180                 util->diff = util->patch + util->diff_offset;
 181                 hashmap_entry_init(util, strhash(util->diff));
 182                 other = hashmap_remove(&map, util, NULL);
 183                 if (other) {
 184                         if (other->matching >= 0)
 185                                 BUG("already assigned!");
 186
 187                         other->matching = i;
 188                         util->matching = other->i;
 189                 }
 190         }
 191
 192         hashmap_free(&map, 0);
 193 }
 194
 195 static void diffsize_consume(void *data, char *line, unsigned long len)
 196 {
 197         (*(int *)data)++;
 198 }
 199
 200 static int diffsize(const char *a, const char *b)
 201 {
 202         xpparam_t pp = { 0 };
 203         xdemitconf_t cfg = { 0 };
 204         mmfile_t mf1, mf2;
 205         int count = 0;
 206
 207         mf1.ptr = (char *)a;
 208         mf1.size = strlen(a);
 209         mf2.ptr = (char *)b;
 210         mf2.size = strlen(b);
 211
 212         cfg.ctxlen = 3;
 213         if (!xdi_diff_outf(&mf1, &mf2, diffsize_consume, &count, &pp, &cfg))
 214                 return count;
 215
 216         error(_("failed to generate diff"));
 217         return COST_MAX;
 218 }
 219
 220 static void get_correspondences(struct string_list *a, struct string_list *b,
 221                                 int creation_factor)
 222 {
 223         int n = a->nr + b->nr;
 224         int *cost, c, *a2b, *b2a;
 225         int i, j;
 226
 227         ALLOC_ARRAY(cost, st_mult(n, n));
 228         ALLOC_ARRAY(a2b, n);
 229         ALLOC_ARRAY(b2a, n);
 230
 231         for (i = 0; i < a->nr; i++) {
 232                 struct patch_util *a_util = a->items[i].util;
 233
 234                 for (j = 0; j < b->nr; j++) {
 235                         struct patch_util *b_util = b->items[j].util;
 236
 237                         if (a_util->matching == j)
 238                                 c = 0;
 239                         else if (a_util->matching < 0 && b_util->matching < 0)
 240                                 c = diffsize(a_util->diff, b_util->diff);
 241                         else
 242                                 c = COST_MAX;
 243                         cost[i + n * j] = c;
 244                 }
 245
 246                 c = a_util->matching < 0 ?
 247                         a_util->diffsize * creation_factor / 100 : COST_MAX;
 248                 for (j = b->nr; j < n; j++)
 249                         cost[i + n * j] = c;
 250         }
 251
 252         for (j = 0; j < b->nr; j++) {
 253                 struct patch_util *util = b->items[j].util;
 254
 255                 c = util->matching < 0 ?
 256                         util->diffsize * creation_factor / 100 : COST_MAX;
 257                 for (i = a->nr; i < n; i++)
 258                         cost[i + n * j] = c;
 259         }
 260
 261         for (i = a->nr; i < n; i++)
 262                 for (j = b->nr; j < n; j++)
 263                         cost[i + n * j] = 0;
 264
 265         compute_assignment(n, n, cost, a2b, b2a);
 266
 267         for (i = 0; i < a->nr; i++)
 268                 if (a2b[i] >= 0 && a2b[i] < b->nr) {
 269                         struct patch_util *a_util = a->items[i].util;
 270                         struct patch_util *b_util = b->items[a2b[i]].util;
 271
 272                         a_util->matching = a2b[i];
 273                         b_util->matching = i;
 274                 }
 275
 276         free(cost);
 277         free(a2b);
 278         free(b2a);
 279 }
 280
 281 static void output_pair_header(struct diff_options *diffopt,
 282                                int patch_no_width,
 283                                struct strbuf *buf,
 284                                struct strbuf *dashes,
 285                                struct patch_util *a_util,
 286                                struct patch_util *b_util)
 287 {
 288         struct object_id *oid = a_util ? &a_util->oid : &b_util->oid;
 289         struct commit *commit;
 290         char status;
 291         const char *color_reset = diff_get_color_opt(diffopt, DIFF_RESET);
 292         const char *color_old = diff_get_color_opt(diffopt, DIFF_FILE_OLD);
 293         const char *color_new = diff_get_color_opt(diffopt, DIFF_FILE_NEW);
 294         const char *color_commit = diff_get_color_opt(diffopt, DIFF_COMMIT);
 295         const char *color;
 296
 297         if (!dashes->len)
 298                 strbuf_addchars(dashes, '-',
 299                                 strlen(find_unique_abbrev(oid,
 300                                                           DEFAULT_ABBREV)));
 301
 302         if (!b_util) {
 303                 color = color_old;
 304                 status = '<';
 305         } else if (!a_util) {
 306                 color = color_new;
 307                 status = '>';
 308         } else if (strcmp(a_util->patch, b_util->patch)) {
 309                 color = color_commit;
 310                 status = '!';
 311         } else {
 312                 color = color_commit;
 313                 status = '=';
 314         }
 315
 316         strbuf_reset(buf);
 317         strbuf_addstr(buf, status == '!' ? color_old : color);
 318         if (!a_util)
 319                 strbuf_addf(buf, "%*s:  %s ", patch_no_width, "-", dashes->buf);
 320         else
 321                 strbuf_addf(buf, "%*d:  %s ", patch_no_width, a_util->i + 1,
 322                             find_unique_abbrev(&a_util->oid, DEFAULT_ABBREV));
 323
 324         if (status == '!')
 325                 strbuf_addf(buf, "%s%s", color_reset, color);
 326         strbuf_addch(buf, status);
 327         if (status == '!')
 328                 strbuf_addf(buf, "%s%s", color_reset, color_new);
 329
 330         if (!b_util)
 331                 strbuf_addf(buf, " %*s:  %s", patch_no_width, "-", dashes->buf);
 332         else
 333                 strbuf_addf(buf, " %*d:  %s", patch_no_width, b_util->i + 1,
 334                             find_unique_abbrev(&b_util->oid, DEFAULT_ABBREV));
 335
 336         commit = lookup_commit_reference(the_repository, oid);
 337         if (commit) {
 338                 if (status == '!')
 339                         strbuf_addf(buf, "%s%s", color_reset, color);
 340
 341                 strbuf_addch(buf, ' ');
 342                 pp_commit_easy(CMIT_FMT_ONELINE, commit, buf);
 343         }
 344         strbuf_addf(buf, "%s\n", color_reset);
 345
 346         fwrite(buf->buf, buf->len, 1, stdout);
 347 }
 348
 349 static struct userdiff_driver no_func_name = {
 350         .funcname = { "$^", 0 }
 351 };
 352
 353 static struct diff_filespec *get_filespec(const char *name, const char *p)
 354 {
 355         struct diff_filespec *spec = alloc_filespec(name);
 356
 357         fill_filespec(spec, &null_oid, 0, 0644);
 358         spec->data = (char *)p;
 359         spec->size = strlen(p);
 360         spec->should_munmap = 0;
 361         spec->is_stdin = 1;
 362         spec->driver = &no_func_name;
 363
 364         return spec;
 365 }
 366
 367 static void patch_diff(const char *a, const char *b,
 368                               struct diff_options *diffopt)
 369 {
 370         diff_queue(&diff_queued_diff,
 371                    get_filespec("a", a), get_filespec("b", b));
 372
 373         diffcore_std(diffopt);
 374         diff_flush(diffopt);
 375 }
 376
 377 static void output(struct string_list *a, struct string_list *b,
 378                    struct diff_options *diffopt)
 379 {
 380         struct strbuf buf = STRBUF_INIT, dashes = STRBUF_INIT;
 381         int patch_no_width = decimal_width(1 + (a->nr > b->nr ? a->nr : b->nr));
 382         int i = 0, j = 0;
 383
 384         /*
 385          * We assume the user is really more interested in the second argument
 386          * ("newer" version). To that end, we print the output in the order of
 387          * the RHS (the `b` parameter). To put the LHS (the `a` parameter)
 388          * commits that are no longer in the RHS into a good place, we place
 389          * them once we have shown all of their predecessors in the LHS.
 390          */
 391
 392         while (i < a->nr || j < b->nr) {
 393                 struct patch_util *a_util, *b_util;
 394                 a_util = i < a->nr ? a->items[i].util : NULL;
 395                 b_util = j < b->nr ? b->items[j].util : NULL;
 396
 397                 /* Skip all the already-shown commits from the LHS. */
 398                 while (i < a->nr && a_util->shown)
 399                         a_util = ++i < a->nr ? a->items[i].util : NULL;
 400
 401                 /* Show unmatched LHS commit whose predecessors were shown. */
 402                 if (i < a->nr && a_util->matching < 0) {
 403                         output_pair_header(diffopt, patch_no_width,
 404                                            &buf, &dashes, a_util, NULL);
 405                         i++;
 406                         continue;
 407                 }
 408
 409                 /* Show unmatched RHS commits. */
 410                 while (j < b->nr && b_util->matching < 0) {
 411                         output_pair_header(diffopt, patch_no_width,
 412                                            &buf, &dashes, NULL, b_util);
 413                         b_util = ++j < b->nr ? b->items[j].util : NULL;
 414                 }
 415
 416                 /* Show matching LHS/RHS pair. */
 417                 if (j < b->nr) {
 418                         a_util = a->items[b_util->matching].util;
 419                         output_pair_header(diffopt, patch_no_width,
 420                                            &buf, &dashes, a_util, b_util);
 421                         if (!(diffopt->output_format & DIFF_FORMAT_NO_OUTPUT))
 422                                 patch_diff(a->items[b_util->matching].string,
 423                                            b->items[j].string, diffopt);
 424                         a_util->shown = 1;
 425                         j++;
 426                 }
 427         }
 428         strbuf_release(&buf);
 429         strbuf_release(&dashes);
 430 }
 431
 432 int show_range_diff(const char *range1, const char *range2,
 433                     int creation_factor, struct diff_options *diffopt)
 434 {
 435         int res = 0;
 436
 437         struct string_list branch1 = STRING_LIST_INIT_DUP;
 438         struct string_list branch2 = STRING_LIST_INIT_DUP;
 439
 440         if (read_patches(range1, &branch1))
 441                 res = error(_("could not parse log for '%s'"), range1);
 442         if (!res && read_patches(range2, &branch2))
 443                 res = error(_("could not parse log for '%s'"), range2);
 444
 445         if (!res) {
 446                 find_exact_matches(&branch1, &branch2);
 447                 get_correspondences(&branch1, &branch2, creation_factor);
 448                 output(&branch1, &branch2, diffopt);
 449         }
 450
 451         string_list_clear(&branch1, 1);
 452         string_list_clear(&branch2, 1);
 453
 454         return res;
 455 }