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