git.oblomov.eu Git - git/blob - line-log.c

   1 #include "git-compat-util.h"
   2 #include "line-range.h"
   3 #include "cache.h"
   4 #include "tag.h"
   5 #include "blob.h"
   6 #include "tree.h"
   7 #include "diff.h"
   8 #include "commit.h"
   9 #include "decorate.h"
  10 #include "revision.h"
  11 #include "xdiff-interface.h"
  12 #include "strbuf.h"
  13 #include "log-tree.h"
  14 #include "graph.h"
  15 #include "userdiff.h"
  16 #include "line-log.h"
  17
  18 static void range_set_grow(struct range_set *rs, size_t extra)
  19 {
  20         ALLOC_GROW(rs->ranges, rs->nr + extra, rs->alloc);
  21 }
  22
  23 /* Either initialization would be fine */
  24 #define RANGE_SET_INIT {0}
  25
  26 static void range_set_init(struct range_set *rs, size_t prealloc)
  27 {
  28         rs->alloc = rs->nr = 0;
  29         rs->ranges = NULL;
  30         if (prealloc)
  31                 range_set_grow(rs, prealloc);
  32 }
  33
  34 static void range_set_release(struct range_set *rs)
  35 {
  36         free(rs->ranges);
  37         rs->alloc = rs->nr = 0;
  38         rs->ranges = NULL;
  39 }
  40
  41 /* dst must be uninitialized! */
  42 static void range_set_copy(struct range_set *dst, struct range_set *src)
  43 {
  44         range_set_init(dst, src->nr);
  45         memcpy(dst->ranges, src->ranges, src->nr*sizeof(struct range_set));
  46         dst->nr = src->nr;
  47 }
  48 static void range_set_move(struct range_set *dst, struct range_set *src)
  49 {
  50         range_set_release(dst);
  51         dst->ranges = src->ranges;
  52         dst->nr = src->nr;
  53         dst->alloc = src->alloc;
  54         src->ranges = NULL;
  55         src->alloc = src->nr = 0;
  56 }
  57
  58 /* tack on a _new_ range _at the end_ */
  59 static void range_set_append_unsafe(struct range_set *rs, long a, long b)
  60 {
  61         assert(a <= b);
  62         range_set_grow(rs, 1);
  63         rs->ranges[rs->nr].start = a;
  64         rs->ranges[rs->nr].end = b;
  65         rs->nr++;
  66 }
  67
  68 static void range_set_append(struct range_set *rs, long a, long b)
  69 {
  70         assert(rs->nr == 0 || rs->ranges[rs->nr-1].end <= a);
  71         range_set_append_unsafe(rs, a, b);
  72 }
  73
  74 static int range_cmp(const void *_r, const void *_s)
  75 {
  76         const struct range *r = _r;
  77         const struct range *s = _s;
  78
  79         /* this could be simply 'return r.start-s.start', but for the types */
  80         if (r->start == s->start)
  81                 return 0;
  82         if (r->start < s->start)
  83                 return -1;
  84         return 1;
  85 }
  86
  87 /*
  88  * Check that the ranges are non-empty, sorted and non-overlapping
  89  */
  90 static void range_set_check_invariants(struct range_set *rs)
  91 {
  92         int i;
  93
  94         if (!rs)
  95                 return;
  96
  97         if (rs->nr)
  98                 assert(rs->ranges[0].start < rs->ranges[0].end);
  99
 100         for (i = 1; i < rs->nr; i++) {
 101                 assert(rs->ranges[i-1].end < rs->ranges[i].start);
 102                 assert(rs->ranges[i].start < rs->ranges[i].end);
 103         }
 104 }
 105
 106 /*
 107  * In-place pass of sorting and merging the ranges in the range set,
 108  * to establish the invariants when we get the ranges from the user
 109  */
 110 static void sort_and_merge_range_set(struct range_set *rs)
 111 {
 112         int i;
 113         int o = 0; /* output cursor */
 114
 115         qsort(rs->ranges, rs->nr, sizeof(struct range), range_cmp);
 116
 117         for (i = 0; i < rs->nr; i++) {
 118                 if (rs->ranges[i].start == rs->ranges[i].end)
 119                         continue;
 120                 if (o > 0 && rs->ranges[i].start <= rs->ranges[o-1].end) {
 121                         if (rs->ranges[o-1].end < rs->ranges[i].end)
 122                                 rs->ranges[o-1].end = rs->ranges[i].end;
 123                 } else {
 124                         rs->ranges[o].start = rs->ranges[i].start;
 125                         rs->ranges[o].end = rs->ranges[i].end;
 126                         o++;
 127                 }
 128         }
 129         assert(o <= rs->nr);
 130         rs->nr = o;
 131
 132         range_set_check_invariants(rs);
 133 }
 134
 135 /*
 136  * Union of range sets (i.e., sets of line numbers).  Used to merge
 137  * them when searches meet at a common ancestor.
 138  *
 139  * This is also where the ranges are consolidated into canonical form:
 140  * overlapping and adjacent ranges are merged, and empty ranges are
 141  * removed.
 142  */
 143 static void range_set_union(struct range_set *out,
 144                              struct range_set *a, struct range_set *b)
 145 {
 146         int i = 0, j = 0, o = 0;
 147         struct range *ra = a->ranges;
 148         struct range *rb = b->ranges;
 149         /* cannot make an alias of out->ranges: it may change during grow */
 150
 151         assert(out->nr == 0);
 152         while (i < a->nr || j < b->nr) {
 153                 struct range *new;
 154                 if (i < a->nr && j < b->nr) {
 155                         if (ra[i].start < rb[j].start)
 156                                 new = &ra[i++];
 157                         else if (ra[i].start > rb[j].start)
 158                                 new = &rb[j++];
 159                         else if (ra[i].end < rb[j].end)
 160                                 new = &ra[i++];
 161                         else
 162                                 new = &rb[j++];
 163                 } else if (i < a->nr)      /* b exhausted */
 164                         new = &ra[i++];
 165                 else                       /* a exhausted */
 166                         new = &rb[j++];
 167                 if (new->start == new->end)
 168                         ; /* empty range */
 169                 else if (!o || out->ranges[o-1].end < new->start) {
 170                         range_set_grow(out, 1);
 171                         out->ranges[o].start = new->start;
 172                         out->ranges[o].end = new->end;
 173                         o++;
 174                 } else if (out->ranges[o-1].end < new->end) {
 175                         out->ranges[o-1].end = new->end;
 176                 }
 177         }
 178         out->nr = o;
 179 }
 180
 181 /*
 182  * Difference of range sets (out = a \ b).  Pass the "interesting"
 183  * ranges as 'a' and the target side of the diff as 'b': it removes
 184  * the ranges for which the commit is responsible.
 185  */
 186 static void range_set_difference(struct range_set *out,
 187                                   struct range_set *a, struct range_set *b)
 188 {
 189         int i, j =  0;
 190         for (i = 0; i < a->nr; i++) {
 191                 long start = a->ranges[i].start;
 192                 long end = a->ranges[i].end;
 193                 while (start < end) {
 194                         while (j < b->nr && start >= b->ranges[j].end)
 195                                 /*
 196                                  * a:         |-------
 197                                  * b: ------|
 198                                  */
 199                                 j++;
 200                         if (j >= b->nr || end < b->ranges[j].start) {
 201                                 /*
 202                                  * b exhausted, or
 203                                  * a:  ----|
 204                                  * b:         |----
 205                                  */
 206                                 range_set_append(out, start, end);
 207                                 break;
 208                         }
 209                         if (start >= b->ranges[j].start) {
 210                                 /*
 211                                  * a:     |--????
 212                                  * b: |------|
 213                                  */
 214                                 start = b->ranges[j].end;
 215                         } else if (end > b->ranges[j].start) {
 216                                 /*
 217                                  * a: |-----|
 218                                  * b:    |--?????
 219                                  */
 220                                 if (start < b->ranges[j].start)
 221                                         range_set_append(out, start, b->ranges[j].start);
 222                                 start = b->ranges[j].end;
 223                         }
 224                 }
 225         }
 226 }
 227
 228 static void diff_ranges_init(struct diff_ranges *diff)
 229 {
 230         range_set_init(&diff->parent, 0);
 231         range_set_init(&diff->target, 0);
 232 }
 233
 234 static void diff_ranges_release(struct diff_ranges *diff)
 235 {
 236         range_set_release(&diff->parent);
 237         range_set_release(&diff->target);
 238 }
 239
 240 void line_log_data_init(struct line_log_data *r)
 241 {
 242         memset(r, 0, sizeof(struct line_log_data));
 243         range_set_init(&r->ranges, 0);
 244 }
 245
 246 static void line_log_data_clear(struct line_log_data *r)
 247 {
 248         range_set_release(&r->ranges);
 249         if (r->pair)
 250                 diff_free_filepair(r->pair);
 251 }
 252
 253 static void free_line_log_data(struct line_log_data *r)
 254 {
 255         while (r) {
 256                 struct line_log_data *next = r->next;
 257                 line_log_data_clear(r);
 258                 free(r);
 259                 r = next;
 260         }
 261 }
 262
 263 static struct line_log_data *
 264 search_line_log_data(struct line_log_data *list, const char *path,
 265                      struct line_log_data **insertion_point)
 266 {
 267         struct line_log_data *p = list;
 268         if (insertion_point)
 269                 *insertion_point = NULL;
 270         while (p) {
 271                 int cmp = strcmp(p->path, path);
 272                 if (!cmp)
 273                         return p;
 274                 if (insertion_point && cmp < 0)
 275                         *insertion_point = p;
 276                 p = p->next;
 277         }
 278         return NULL;
 279 }
 280
 281 /*
 282  * Note: takes ownership of 'path', which happens to be what the only
 283  * caller needs.
 284  */
 285 static void line_log_data_insert(struct line_log_data **list,
 286                                  char *path,
 287                                  long begin, long end)
 288 {
 289         struct line_log_data *ip;
 290         struct line_log_data *p = search_line_log_data(*list, path, &ip);
 291
 292         if (p) {
 293                 range_set_append_unsafe(&p->ranges, begin, end);
 294                 sort_and_merge_range_set(&p->ranges);
 295                 free(path);
 296                 return;
 297         }
 298
 299         p = xcalloc(1, sizeof(struct line_log_data));
 300         p->path = path;
 301         range_set_append(&p->ranges, begin, end);
 302         if (ip) {
 303                 p->next = ip->next;
 304                 ip->next = p;
 305         } else {
 306                 p->next = *list;
 307                 *list = p;
 308         }
 309 }
 310
 311 struct collect_diff_cbdata {
 312         struct diff_ranges *diff;
 313 };
 314
 315 static int collect_diff_cb(long start_a, long count_a,
 316                            long start_b, long count_b,
 317                            void *data)
 318 {
 319         struct collect_diff_cbdata *d = data;
 320
 321         if (count_a >= 0)
 322                 range_set_append(&d->diff->parent, start_a, start_a + count_a);
 323         if (count_b >= 0)
 324                 range_set_append(&d->diff->target, start_b, start_b + count_b);
 325
 326         return 0;
 327 }
 328
 329 static void collect_diff(mmfile_t *parent, mmfile_t *target, struct diff_ranges *out)
 330 {
 331         struct collect_diff_cbdata cbdata = {NULL};
 332         xpparam_t xpp;
 333         xdemitconf_t xecfg;
 334         xdemitcb_t ecb;
 335
 336         memset(&xpp, 0, sizeof(xpp));
 337         memset(&xecfg, 0, sizeof(xecfg));
 338         xecfg.ctxlen = xecfg.interhunkctxlen = 0;
 339
 340         cbdata.diff = out;
 341         xecfg.hunk_func = collect_diff_cb;
 342         memset(&ecb, 0, sizeof(ecb));
 343         ecb.priv = &cbdata;
 344         xdi_diff(parent, target, &xpp, &xecfg, &ecb);
 345 }
 346
 347 /*
 348  * These are handy for debugging.  Removing them with #if 0 silences
 349  * the "unused function" warning.
 350  */
 351 #if 0
 352 static void dump_range_set(struct range_set *rs, const char *desc)
 353 {
 354         int i;
 355         printf("range set %s (%d items):\n", desc, rs->nr);
 356         for (i = 0; i < rs->nr; i++)
 357                 printf("\t[%ld,%ld]\n", rs->ranges[i].start, rs->ranges[i].end);
 358 }
 359
 360 static void dump_line_log_data(struct line_log_data *r)
 361 {
 362         char buf[4096];
 363         while (r) {
 364                 snprintf(buf, 4096, "file %s\n", r->path);
 365                 dump_range_set(&r->ranges, buf);
 366                 r = r->next;
 367         }
 368 }
 369
 370 static void dump_diff_ranges(struct diff_ranges *diff, const char *desc)
 371 {
 372         int i;
 373         assert(diff->parent.nr == diff->target.nr);
 374         printf("diff ranges %s (%d items):\n", desc, diff->parent.nr);
 375         printf("\tparent\ttarget\n");
 376         for (i = 0; i < diff->parent.nr; i++) {
 377                 printf("\t[%ld,%ld]\t[%ld,%ld]\n",
 378                        diff->parent.ranges[i].start,
 379                        diff->parent.ranges[i].end,
 380                        diff->target.ranges[i].start,
 381                        diff->target.ranges[i].end);
 382         }
 383 }
 384 #endif
 385
 386
 387 static int ranges_overlap(struct range *a, struct range *b)
 388 {
 389         return !(a->end <= b->start || b->end <= a->start);
 390 }
 391
 392 /*
 393  * Given a diff and the set of interesting ranges, determine all hunks
 394  * of the diff which touch (overlap) at least one of the interesting
 395  * ranges in the target.
 396  */
 397 static void diff_ranges_filter_touched(struct diff_ranges *out,
 398                                        struct diff_ranges *diff,
 399                                        struct range_set *rs)
 400 {
 401         int i, j = 0;
 402
 403         assert(out->target.nr == 0);
 404
 405         for (i = 0; i < diff->target.nr; i++) {
 406                 while (diff->target.ranges[i].start > rs->ranges[j].end) {
 407                         j++;
 408                         if (j == rs->nr)
 409                                 return;
 410                 }
 411                 if (ranges_overlap(&diff->target.ranges[i], &rs->ranges[j])) {
 412                         range_set_append(&out->parent,
 413                                          diff->parent.ranges[i].start,
 414                                          diff->parent.ranges[i].end);
 415                         range_set_append(&out->target,
 416                                          diff->target.ranges[i].start,
 417                                          diff->target.ranges[i].end);
 418                 }
 419         }
 420 }
 421
 422 /*
 423  * Adjust the line counts in 'rs' to account for the lines
 424  * added/removed in the diff.
 425  */
 426 static void range_set_shift_diff(struct range_set *out,
 427                                  struct range_set *rs,
 428                                  struct diff_ranges *diff)
 429 {
 430         int i, j = 0;
 431         long offset = 0;
 432         struct range *src = rs->ranges;
 433         struct range *target = diff->target.ranges;
 434         struct range *parent = diff->parent.ranges;
 435
 436         for (i = 0; i < rs->nr; i++) {
 437                 while (j < diff->target.nr && src[i].start >= target[j].start) {
 438                         offset += (parent[j].end-parent[j].start)
 439                                 - (target[j].end-target[j].start);
 440                         j++;
 441                 }
 442                 range_set_append(out, src[i].start+offset, src[i].end+offset);
 443         }
 444 }
 445
 446 /*
 447  * Given a diff and the set of interesting ranges, map the ranges
 448  * across the diff.  That is: observe that the target commit takes
 449  * blame for all the + (target-side) ranges.  So for every pair of
 450  * ranges in the diff that was touched, we remove the latter and add
 451  * its parent side.
 452  */
 453 static void range_set_map_across_diff(struct range_set *out,
 454                                       struct range_set *rs,
 455                                       struct diff_ranges *diff,
 456                                       struct diff_ranges **touched_out)
 457 {
 458         struct diff_ranges *touched = xmalloc(sizeof(*touched));
 459         struct range_set tmp1 = RANGE_SET_INIT;
 460         struct range_set tmp2 = RANGE_SET_INIT;
 461
 462         diff_ranges_init(touched);
 463         diff_ranges_filter_touched(touched, diff, rs);
 464         range_set_difference(&tmp1, rs, &touched->target);
 465         range_set_shift_diff(&tmp2, &tmp1, diff);
 466         range_set_union(out, &tmp2, &touched->parent);
 467         range_set_release(&tmp1);
 468         range_set_release(&tmp2);
 469
 470         *touched_out = touched;
 471 }
 472
 473 static struct commit *check_single_commit(struct rev_info *revs)
 474 {
 475         struct object *commit = NULL;
 476         int found = -1;
 477         int i;
 478
 479         for (i = 0; i < revs->pending.nr; i++) {
 480                 struct object *obj = revs->pending.objects[i].item;
 481                 if (obj->flags & UNINTERESTING)
 482                         continue;
 483                 while (obj->type == OBJ_TAG)
 484                         obj = deref_tag(obj, NULL, 0);
 485                 if (obj->type != OBJ_COMMIT)
 486                         die("Non commit %s?", revs->pending.objects[i].name);
 487                 if (commit)
 488                         die("More than one commit to dig from: %s and %s?",
 489                             revs->pending.objects[i].name,
 490                             revs->pending.objects[found].name);
 491                 commit = obj;
 492                 found = i;
 493         }
 494
 495         if (!commit)
 496                 die("No commit specified?");
 497
 498         return (struct commit *) commit;
 499 }
 500
 501 static void fill_blob_sha1(struct commit *commit, struct diff_filespec *spec)
 502 {
 503         unsigned mode;
 504         unsigned char sha1[20];
 505
 506         if (get_tree_entry(commit->object.sha1, spec->path,
 507                            sha1, &mode))
 508                 die("There is no path %s in the commit", spec->path);
 509         fill_filespec(spec, sha1, 1, mode);
 510
 511         return;
 512 }
 513
 514 static void fill_line_ends(struct diff_filespec *spec, long *lines,
 515                            unsigned long **line_ends)
 516 {
 517         int num = 0, size = 50;
 518         long cur = 0;
 519         unsigned long *ends = NULL;
 520         char *data = NULL;
 521
 522         if (diff_populate_filespec(spec, 0))
 523                 die("Cannot read blob %s", sha1_to_hex(spec->sha1));
 524
 525         ends = xmalloc(size * sizeof(*ends));
 526         ends[cur++] = 0;
 527         data = spec->data;
 528         while (num < spec->size) {
 529                 if (data[num] == '\n' || num == spec->size - 1) {
 530                         ALLOC_GROW(ends, (cur + 1), size);
 531                         ends[cur++] = num;
 532                 }
 533                 num++;
 534         }
 535
 536         /* shrink the array to fit the elements */
 537         ends = xrealloc(ends, cur * sizeof(*ends));
 538         *lines = cur-1;
 539         *line_ends = ends;
 540 }
 541
 542 struct nth_line_cb {
 543         struct diff_filespec *spec;
 544         long lines;
 545         unsigned long *line_ends;
 546 };
 547
 548 static const char *nth_line(void *data, long line)
 549 {
 550         struct nth_line_cb *d = data;
 551         assert(d && line <= d->lines);
 552         assert(d->spec && d->spec->data);
 553
 554         if (line == 0)
 555                 return (char *)d->spec->data;
 556         else
 557                 return (char *)d->spec->data + d->line_ends[line] + 1;
 558 }
 559
 560 static struct line_log_data *
 561 parse_lines(struct commit *commit, const char *prefix, struct string_list *args)
 562 {
 563         long lines = 0;
 564         unsigned long *ends = NULL;
 565         struct nth_line_cb cb_data;
 566         struct string_list_item *item;
 567         struct line_log_data *ranges = NULL;
 568
 569         for_each_string_list_item(item, args) {
 570                 const char *name_part, *range_part;
 571                 char *full_name;
 572                 struct diff_filespec *spec;
 573                 long begin = 0, end = 0;
 574
 575                 name_part = skip_range_arg(item->string);
 576                 if (!name_part || *name_part != ':' || !name_part[1])
 577                         die("-L argument '%s' not of the form start,end:file",
 578                             item->string);
 579                 range_part = xstrndup(item->string, name_part - item->string);
 580                 name_part++;
 581
 582                 full_name = prefix_path(prefix, prefix ? strlen(prefix) : 0,
 583                                         name_part);
 584
 585                 spec = alloc_filespec(full_name);
 586                 fill_blob_sha1(commit, spec);
 587                 fill_line_ends(spec, &lines, &ends);
 588                 cb_data.spec = spec;
 589                 cb_data.lines = lines;
 590                 cb_data.line_ends = ends;
 591
 592                 if (parse_range_arg(range_part, nth_line, &cb_data,
 593                                     lines, &begin, &end,
 594                                     full_name))
 595                         die("malformed -L argument '%s'", range_part);
 596                 if (begin < 1)
 597                         begin = 1;
 598                 if (end < 1)
 599                         end = lines;
 600                 begin--;
 601                 if (lines < end || lines < begin)
 602                         die("file %s has only %ld lines", name_part, lines);
 603                 line_log_data_insert(&ranges, full_name, begin, end);
 604
 605                 free_filespec(spec);
 606                 free(ends);
 607                 ends = NULL;
 608         }
 609
 610         return ranges;
 611 }
 612
 613 static struct line_log_data *line_log_data_copy_one(struct line_log_data *r)
 614 {
 615         struct line_log_data *ret = xmalloc(sizeof(*ret));
 616
 617         assert(r);
 618         line_log_data_init(ret);
 619         range_set_copy(&ret->ranges, &r->ranges);
 620
 621         ret->path = xstrdup(r->path);
 622
 623         return ret;
 624 }
 625
 626 static struct line_log_data *
 627 line_log_data_copy(struct line_log_data *r)
 628 {
 629         struct line_log_data *ret = NULL;
 630         struct line_log_data *tmp = NULL, *prev = NULL;
 631
 632         assert(r);
 633         ret = tmp = prev = line_log_data_copy_one(r);
 634         r = r->next;
 635         while (r) {
 636                 tmp = line_log_data_copy_one(r);
 637                 prev->next = tmp;
 638                 prev = tmp;
 639                 r = r->next;
 640         }
 641
 642         return ret;
 643 }
 644
 645 /* merge two range sets across files */
 646 static struct line_log_data *line_log_data_merge(struct line_log_data *a,
 647                                                  struct line_log_data *b)
 648 {
 649         struct line_log_data *head = NULL, **pp = &head;
 650
 651         while (a || b) {
 652                 struct line_log_data *src;
 653                 struct line_log_data *src2 = NULL;
 654                 struct line_log_data *d;
 655                 int cmp;
 656                 if (!a)
 657                         cmp = 1;
 658                 else if (!b)
 659                         cmp = -1;
 660                 else
 661                         cmp = strcmp(a->path, b->path);
 662                 if (cmp < 0) {
 663                         src = a;
 664                         a = a->next;
 665                 } else if (cmp == 0) {
 666                         src = a;
 667                         a = a->next;
 668                         src2 = b;
 669                         b = b->next;
 670                 } else {
 671                         src = b;
 672                         b = b->next;
 673                 }
 674                 d = xmalloc(sizeof(struct line_log_data));
 675                 line_log_data_init(d);
 676                 d->path = xstrdup(src->path);
 677                 *pp = d;
 678                 pp = &d->next;
 679                 if (src2)
 680                         range_set_union(&d->ranges, &src->ranges, &src2->ranges);
 681                 else
 682                         range_set_copy(&d->ranges, &src->ranges);
 683         }
 684
 685         return head;
 686 }
 687
 688 static void add_line_range(struct rev_info *revs, struct commit *commit,
 689                            struct line_log_data *range)
 690 {
 691         struct line_log_data *old = NULL;
 692         struct line_log_data *new = NULL;
 693
 694         old = lookup_decoration(&revs->line_log_data, &commit->object);
 695         if (old && range) {
 696                 new = line_log_data_merge(old, range);
 697                 free_line_log_data(old);
 698         } else if (range)
 699                 new = line_log_data_copy(range);
 700
 701         if (new)
 702                 add_decoration(&revs->line_log_data, &commit->object, new);
 703 }
 704
 705 static void clear_commit_line_range(struct rev_info *revs, struct commit *commit)
 706 {
 707         struct line_log_data *r;
 708         r = lookup_decoration(&revs->line_log_data, &commit->object);
 709         if (!r)
 710                 return;
 711         free_line_log_data(r);
 712         add_decoration(&revs->line_log_data, &commit->object, NULL);
 713 }
 714
 715 static struct line_log_data *lookup_line_range(struct rev_info *revs,
 716                                                struct commit *commit)
 717 {
 718         struct line_log_data *ret = NULL;
 719         struct line_log_data *d;
 720
 721         ret = lookup_decoration(&revs->line_log_data, &commit->object);
 722
 723         for (d = ret; d; d = d->next)
 724                 range_set_check_invariants(&d->ranges);
 725
 726         return ret;
 727 }
 728
 729 void line_log_init(struct rev_info *rev, const char *prefix, struct string_list *args)
 730 {
 731         struct commit *commit = NULL;
 732         struct line_log_data *range;
 733
 734         commit = check_single_commit(rev);
 735         range = parse_lines(commit, prefix, args);
 736         add_line_range(rev, commit, range);
 737
 738         if (!rev->diffopt.detect_rename) {
 739                 int i, count = 0;
 740                 struct line_log_data *r = range;
 741                 const char **paths;
 742                 while (r) {
 743                         count++;
 744                         r = r->next;
 745                 }
 746                 paths = xmalloc((count+1)*sizeof(char *));
 747                 r = range;
 748                 for (i = 0; i < count; i++) {
 749                         paths[i] = xstrdup(r->path);
 750                         r = r->next;
 751                 }
 752                 paths[count] = NULL;
 753                 init_pathspec(&rev->diffopt.pathspec, paths);
 754                 free(paths);
 755         }
 756 }
 757
 758 static void load_tree_desc(struct tree_desc *desc, void **tree,
 759                            const unsigned char *sha1)
 760 {
 761         unsigned long size;
 762         *tree = read_object_with_reference(sha1, tree_type, &size, NULL);
 763         if (!*tree)
 764                 die("Unable to read tree (%s)", sha1_to_hex(sha1));
 765         init_tree_desc(desc, *tree, size);
 766 }
 767
 768 static int count_parents(struct commit *commit)
 769 {
 770         struct commit_list *parents = commit->parents;
 771         int count = 0;
 772         while (parents) {
 773                 count++;
 774                 parents = parents->next;
 775         }
 776         return count;
 777 }
 778
 779 static void move_diff_queue(struct diff_queue_struct *dst,
 780                             struct diff_queue_struct *src)
 781 {
 782         assert(src != dst);
 783         memcpy(dst, src, sizeof(struct diff_queue_struct));
 784         DIFF_QUEUE_CLEAR(src);
 785 }
 786
 787 static void filter_diffs_for_paths(struct line_log_data *range, int keep_deletions)
 788 {
 789         int i;
 790         struct diff_queue_struct outq;
 791         DIFF_QUEUE_CLEAR(&outq);
 792
 793         for (i = 0; i < diff_queued_diff.nr; i++) {
 794                 struct diff_filepair *p = diff_queued_diff.queue[i];
 795                 struct line_log_data *rg = NULL;
 796
 797                 if (!DIFF_FILE_VALID(p->two)) {
 798                         if (keep_deletions)
 799                                 diff_q(&outq, p);
 800                         else
 801                                 diff_free_filepair(p);
 802                         continue;
 803                 }
 804                 for (rg = range; rg; rg = rg->next) {
 805                         if (!strcmp(rg->path, p->two->path))
 806                                 break;
 807                 }
 808                 if (rg)
 809                         diff_q(&outq, p);
 810                 else
 811                         diff_free_filepair(p);
 812         }
 813         free(diff_queued_diff.queue);
 814         diff_queued_diff = outq;
 815 }
 816
 817 static inline int diff_might_be_rename(void)
 818 {
 819         int i;
 820         for (i = 0; i < diff_queued_diff.nr; i++)
 821                 if (!DIFF_FILE_VALID(diff_queued_diff.queue[i]->one)) {
 822                         /* fprintf(stderr, "diff_might_be_rename found creation of: %s\n", */
 823                         /*      diff_queued_diff.queue[i]->two->path); */
 824                         return 1;
 825                 }
 826         return 0;
 827 }
 828
 829 static void queue_diffs(struct line_log_data *range,
 830                         struct diff_options *opt,
 831                         struct diff_queue_struct *queue,
 832                         struct commit *commit, struct commit *parent)
 833 {
 834         void *tree1 = NULL, *tree2 = NULL;
 835         struct tree_desc desc1, desc2;
 836
 837         assert(commit);
 838         load_tree_desc(&desc2, &tree2, commit->tree->object.sha1);
 839         if (parent)
 840                 load_tree_desc(&desc1, &tree1, parent->tree->object.sha1);
 841         else
 842                 init_tree_desc(&desc1, "", 0);
 843
 844         DIFF_QUEUE_CLEAR(&diff_queued_diff);
 845         diff_tree(&desc1, &desc2, "", opt);
 846         if (opt->detect_rename) {
 847                 filter_diffs_for_paths(range, 1);
 848                 if (diff_might_be_rename())
 849                         diffcore_std(opt);
 850                 filter_diffs_for_paths(range, 0);
 851         }
 852         move_diff_queue(queue, &diff_queued_diff);
 853
 854         if (tree1)
 855                 free(tree1);
 856         if (tree2)
 857                 free(tree2);
 858 }
 859
 860 static char *get_nth_line(long line, unsigned long *ends, void *data)
 861 {
 862         if (line == 0)
 863                 return (char *)data;
 864         else
 865                 return (char *)data + ends[line] + 1;
 866 }
 867
 868 static void print_line(const char *prefix, char first,
 869                        long line, unsigned long *ends, void *data,
 870                        const char *color, const char *reset)
 871 {
 872         char *begin = get_nth_line(line, ends, data);
 873         char *end = get_nth_line(line+1, ends, data);
 874         int had_nl = 0;
 875
 876         if (end > begin && end[-1] == '\n') {
 877                 end--;
 878                 had_nl = 1;
 879         }
 880
 881         fputs(prefix, stdout);
 882         fputs(color, stdout);
 883         putchar(first);
 884         fwrite(begin, 1, end-begin, stdout);
 885         fputs(reset, stdout);
 886         putchar('\n');
 887         if (!had_nl)
 888                 fputs("\\ No newline at end of file\n", stdout);
 889 }
 890
 891 static char *output_prefix(struct diff_options *opt)
 892 {
 893         char *prefix = "";
 894
 895         if (opt->output_prefix) {
 896                 struct strbuf *sb = opt->output_prefix(opt, opt->output_prefix_data);
 897                 prefix = sb->buf;
 898         }
 899
 900         return prefix;
 901 }
 902
 903 static void dump_diff_hacky_one(struct rev_info *rev, struct line_log_data *range)
 904 {
 905         int i, j = 0;
 906         long p_lines, t_lines;
 907         unsigned long *p_ends = NULL, *t_ends = NULL;
 908         struct diff_filepair *pair = range->pair;
 909         struct diff_ranges *diff = &range->diff;
 910
 911         struct diff_options *opt = &rev->diffopt;
 912         char *prefix = output_prefix(opt);
 913         const char *c_reset = diff_get_color(opt->use_color, DIFF_RESET);
 914         const char *c_frag = diff_get_color(opt->use_color, DIFF_FRAGINFO);
 915         const char *c_meta = diff_get_color(opt->use_color, DIFF_METAINFO);
 916         const char *c_old = diff_get_color(opt->use_color, DIFF_FILE_OLD);
 917         const char *c_new = diff_get_color(opt->use_color, DIFF_FILE_NEW);
 918         const char *c_plain = diff_get_color(opt->use_color, DIFF_PLAIN);
 919
 920         if (!pair || !diff)
 921                 return;
 922
 923         if (pair->one->sha1_valid)
 924                 fill_line_ends(pair->one, &p_lines, &p_ends);
 925         fill_line_ends(pair->two, &t_lines, &t_ends);
 926
 927         printf("%s%sdiff --git a/%s b/%s%s\n", prefix, c_meta, pair->one->path, pair->two->path, c_reset);
 928         printf("%s%s--- %s%s%s\n", prefix, c_meta,
 929                pair->one->sha1_valid ? "a/" : "",
 930                pair->one->sha1_valid ? pair->one->path : "/dev/null",
 931                c_reset);
 932         printf("%s%s+++ b/%s%s\n", prefix, c_meta, pair->two->path, c_reset);
 933         for (i = 0; i < range->ranges.nr; i++) {
 934                 long p_start, p_end;
 935                 long t_start = range->ranges.ranges[i].start;
 936                 long t_end = range->ranges.ranges[i].end;
 937                 long t_cur = t_start;
 938                 int j_last;
 939
 940                 while (j < diff->target.nr && diff->target.ranges[j].end < t_start)
 941                         j++;
 942                 if (j == diff->target.nr || diff->target.ranges[j].start > t_end)
 943                         continue;
 944
 945                 /* Scan ahead to determine the last diff that falls in this range */
 946                 j_last = j;
 947                 while (j_last < diff->target.nr && diff->target.ranges[j_last].start < t_end)
 948                         j_last++;
 949                 if (j_last > j)
 950                         j_last--;
 951
 952                 /*
 953                  * Compute parent hunk headers: we know that the diff
 954                  * has the correct line numbers (but not all hunks).
 955                  * So it suffices to shift the start/end according to
 956                  * the line numbers of the first/last hunk(s) that
 957                  * fall in this range.
 958                  */
 959                 if (t_start < diff->target.ranges[j].start)
 960                         p_start = diff->parent.ranges[j].start - (diff->target.ranges[j].start-t_start);
 961                 else
 962                         p_start = diff->parent.ranges[j].start;
 963                 if (t_end > diff->target.ranges[j_last].end)
 964                         p_end = diff->parent.ranges[j_last].end + (t_end-diff->target.ranges[j_last].end);
 965                 else
 966                         p_end = diff->parent.ranges[j_last].end;
 967
 968                 if (!p_start && !p_end) {
 969                         p_start = -1;
 970                         p_end = -1;
 971                 }
 972
 973                 /* Now output a diff hunk for this range */
 974                 printf("%s%s@@ -%ld,%ld +%ld,%ld @@%s\n",
 975                        prefix, c_frag,
 976                        p_start+1, p_end-p_start, t_start+1, t_end-t_start,
 977                        c_reset);
 978                 while (j < diff->target.nr && diff->target.ranges[j].start < t_end) {
 979                         int k;
 980                         for (; t_cur < diff->target.ranges[j].start; t_cur++)
 981                                 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
 982                                            c_plain, c_reset);
 983                         for (k = diff->parent.ranges[j].start; k < diff->parent.ranges[j].end; k++)
 984                                 print_line(prefix, '-', k, p_ends, pair->one->data,
 985                                            c_old, c_reset);
 986                         for (; t_cur < diff->target.ranges[j].end && t_cur < t_end; t_cur++)
 987                                 print_line(prefix, '+', t_cur, t_ends, pair->two->data,
 988                                            c_new, c_reset);
 989                         j++;
 990                 }
 991                 for (; t_cur < t_end; t_cur++)
 992                         print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
 993                                    c_plain, c_reset);
 994         }
 995
 996         free(p_ends);
 997         free(t_ends);
 998 }
 999
1000 /*
1001  * NEEDSWORK: manually building a diff here is not the Right
1002  * Thing(tm).  log -L should be built into the diff pipeline.
1003  */
1004 static void dump_diff_hacky(struct rev_info *rev, struct line_log_data *range)
1005 {
1006         puts(output_prefix(&rev->diffopt));
1007         while (range) {
1008                 dump_diff_hacky_one(rev, range);
1009                 range = range->next;
1010         }
1011 }
1012
1013 /*
1014  * Unlike most other functions, this destructively operates on
1015  * 'range'.
1016  */
1017 static int process_diff_filepair(struct rev_info *rev,
1018                                  struct diff_filepair *pair,
1019                                  struct line_log_data *range,
1020                                  struct diff_ranges **diff_out)
1021 {
1022         struct line_log_data *rg = range;
1023         struct range_set tmp;
1024         struct diff_ranges diff;
1025         mmfile_t file_parent, file_target;
1026
1027         assert(pair->two->path);
1028         while (rg) {
1029                 assert(rg->path);
1030                 if (!strcmp(rg->path, pair->two->path))
1031                         break;
1032                 rg = rg->next;
1033         }
1034
1035         if (!rg)
1036                 return 0;
1037         if (rg->ranges.nr == 0)
1038                 return 0;
1039
1040         assert(pair->two->sha1_valid);
1041         diff_populate_filespec(pair->two, 0);
1042         file_target.ptr = pair->two->data;
1043         file_target.size = pair->two->size;
1044
1045         if (pair->one->sha1_valid) {
1046                 diff_populate_filespec(pair->one, 0);
1047                 file_parent.ptr = pair->one->data;
1048                 file_parent.size = pair->one->size;
1049         } else {
1050                 file_parent.ptr = "";
1051                 file_parent.size = 0;
1052         }
1053
1054         diff_ranges_init(&diff);
1055         collect_diff(&file_parent, &file_target, &diff);
1056
1057         /* NEEDSWORK should apply some heuristics to prevent mismatches */
1058         free(rg->path);
1059         rg->path = xstrdup(pair->one->path);
1060
1061         range_set_init(&tmp, 0);
1062         range_set_map_across_diff(&tmp, &rg->ranges, &diff, diff_out);
1063         range_set_release(&rg->ranges);
1064         range_set_move(&rg->ranges, &tmp);
1065
1066         diff_ranges_release(&diff);
1067
1068         return ((*diff_out)->parent.nr > 0);
1069 }
1070
1071 static struct diff_filepair *diff_filepair_dup(struct diff_filepair *pair)
1072 {
1073         struct diff_filepair *new = xmalloc(sizeof(struct diff_filepair));
1074         new->one = pair->one;
1075         new->two = pair->two;
1076         new->one->count++;
1077         new->two->count++;
1078         return new;
1079 }
1080
1081 static void free_diffqueues(int n, struct diff_queue_struct *dq)
1082 {
1083         int i, j;
1084         for (i = 0; i < n; i++)
1085                 for (j = 0; j < dq[i].nr; j++)
1086                         diff_free_filepair(dq[i].queue[j]);
1087         free(dq);
1088 }
1089
1090 static int process_all_files(struct line_log_data **range_out,
1091                              struct rev_info *rev,
1092                              struct diff_queue_struct *queue,
1093                              struct line_log_data *range)
1094 {
1095         int i, changed = 0;
1096
1097         *range_out = line_log_data_copy(range);
1098
1099         for (i = 0; i < queue->nr; i++) {
1100                 struct diff_ranges *pairdiff = NULL;
1101                 struct diff_filepair *pair = queue->queue[i];
1102                 if (process_diff_filepair(rev, pair, *range_out, &pairdiff)) {
1103                         /*
1104                          * Store away the diff for later output.  We
1105                          * tuck it in the ranges we got as _input_,
1106                          * since that's the commit that caused the
1107                          * diff.
1108                          *
1109                          * NEEDSWORK not enough when we get around to
1110                          * doing something interesting with merges;
1111                          * currently each invocation on a merge parent
1112                          * trashes the previous one's diff.
1113                          *
1114                          * NEEDSWORK tramples over data structures not owned here
1115                          */
1116                         struct line_log_data *rg = range;
1117                         changed++;
1118                         while (rg && strcmp(rg->path, pair->two->path))
1119                                 rg = rg->next;
1120                         assert(rg);
1121                         rg->pair = diff_filepair_dup(queue->queue[i]);
1122                         memcpy(&rg->diff, pairdiff, sizeof(struct diff_ranges));
1123                 }
1124         }
1125
1126         return changed;
1127 }
1128
1129 int line_log_print(struct rev_info *rev, struct commit *commit)
1130 {
1131         struct line_log_data *range = lookup_line_range(rev, commit);
1132
1133         show_log(rev);
1134         dump_diff_hacky(rev, range);
1135         return 1;
1136 }
1137
1138 static int process_ranges_ordinary_commit(struct rev_info *rev, struct commit *commit,
1139                                           struct line_log_data *range)
1140 {
1141         struct commit *parent = NULL;
1142         struct diff_queue_struct queue;
1143         struct line_log_data *parent_range;
1144         int changed;
1145
1146         if (commit->parents)
1147                 parent = commit->parents->item;
1148
1149         queue_diffs(range, &rev->diffopt, &queue, commit, parent);
1150         changed = process_all_files(&parent_range, rev, &queue, range);
1151         if (parent)
1152                 add_line_range(rev, parent, parent_range);
1153         return changed;
1154 }
1155
1156 static int process_ranges_merge_commit(struct rev_info *rev, struct commit *commit,
1157                                        struct line_log_data *range)
1158 {
1159         struct diff_queue_struct *diffqueues;
1160         struct line_log_data **cand;
1161         struct commit **parents;
1162         struct commit_list *p;
1163         int i;
1164         int nparents = count_parents(commit);
1165
1166         diffqueues = xmalloc(nparents * sizeof(*diffqueues));
1167         cand = xmalloc(nparents * sizeof(*cand));
1168         parents = xmalloc(nparents * sizeof(*parents));
1169
1170         p = commit->parents;
1171         for (i = 0; i < nparents; i++) {
1172                 parents[i] = p->item;
1173                 p = p->next;
1174                 queue_diffs(range, &rev->diffopt, &diffqueues[i], commit, parents[i]);
1175         }
1176
1177         for (i = 0; i < nparents; i++) {
1178                 int changed;
1179                 cand[i] = NULL;
1180                 changed = process_all_files(&cand[i], rev, &diffqueues[i], range);
1181                 if (!changed) {
1182                         /*
1183                          * This parent can take all the blame, so we
1184                          * don't follow any other path in history
1185                          */
1186                         add_line_range(rev, parents[i], cand[i]);
1187                         clear_commit_line_range(rev, commit);
1188                         commit->parents = xmalloc(sizeof(struct commit_list));
1189                         commit->parents->item = parents[i];
1190                         commit->parents->next = NULL;
1191                         free(parents);
1192                         free(cand);
1193                         free_diffqueues(nparents, diffqueues);
1194                         /* NEEDSWORK leaking like a sieve */
1195                         return 0;
1196                 }
1197         }
1198
1199         /*
1200          * No single parent took the blame.  We add the candidates
1201          * from the above loop to the parents.
1202          */
1203         for (i = 0; i < nparents; i++) {
1204                 add_line_range(rev, parents[i], cand[i]);
1205         }
1206
1207         clear_commit_line_range(rev, commit);
1208         free(parents);
1209         free(cand);
1210         free_diffqueues(nparents, diffqueues);
1211         return 1;
1212
1213         /* NEEDSWORK evil merge detection stuff */
1214         /* NEEDSWORK leaking like a sieve */
1215 }
1216
1217 static int process_ranges_arbitrary_commit(struct rev_info *rev, struct commit *commit)
1218 {
1219         struct line_log_data *range = lookup_line_range(rev, commit);
1220         int changed = 0;
1221
1222         if (range) {
1223                 if (!commit->parents || !commit->parents->next)
1224                         changed = process_ranges_ordinary_commit(rev, commit, range);
1225                 else
1226                         changed = process_ranges_merge_commit(rev, commit, range);
1227         }
1228
1229         if (!changed)
1230                 commit->object.flags |= TREESAME;
1231
1232         return changed;
1233 }
1234
1235 static enum rewrite_result line_log_rewrite_one(struct rev_info *rev, struct commit **pp)
1236 {
1237         for (;;) {
1238                 struct commit *p = *pp;
1239                 if (p->parents && p->parents->next)
1240                         return rewrite_one_ok;
1241                 if (p->object.flags & UNINTERESTING)
1242                         return rewrite_one_ok;
1243                 if (!(p->object.flags & TREESAME))
1244                         return rewrite_one_ok;
1245                 if (!p->parents)
1246                         return rewrite_one_noparents;
1247                 *pp = p->parents->item;
1248         }
1249 }
1250
1251 int line_log_filter(struct rev_info *rev)
1252 {
1253         struct commit *commit;
1254         struct commit_list *list = rev->commits;
1255         struct commit_list *out = NULL, **pp = &out;
1256
1257         while (list) {
1258                 struct commit_list *to_free = NULL;
1259                 commit = list->item;
1260                 if (process_ranges_arbitrary_commit(rev, commit)) {
1261                         *pp = list;
1262                         pp = &list->next;
1263                 } else
1264                         to_free = list;
1265                 list = list->next;
1266                 free(to_free);
1267         }
1268         *pp = NULL;
1269
1270         for (list = out; list; list = list->next)
1271                 rewrite_parents(rev, list->item, line_log_rewrite_one);
1272
1273         rev->commits = out;
1274
1275         return 0;
1276 }