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 #include "argv-array.h"
  18
  19 static void range_set_grow(struct range_set *rs, size_t extra)
  20 {
  21         ALLOC_GROW(rs->ranges, rs->nr + extra, rs->alloc);
  22 }
  23
  24 /* Either initialization would be fine */
  25 #define RANGE_SET_INIT {0}
  26
  27 void range_set_init(struct range_set *rs, size_t prealloc)
  28 {
  29         rs->alloc = rs->nr = 0;
  30         rs->ranges = NULL;
  31         if (prealloc)
  32                 range_set_grow(rs, prealloc);
  33 }
  34
  35 void range_set_release(struct range_set *rs)
  36 {
  37         FREE_AND_NULL(rs->ranges);
  38         rs->alloc = rs->nr = 0;
  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         COPY_ARRAY(dst->ranges, src->ranges, src->nr);
  46         dst->nr = src->nr;
  47 }
  48
  49 static void range_set_move(struct range_set *dst, struct range_set *src)
  50 {
  51         range_set_release(dst);
  52         dst->ranges = src->ranges;
  53         dst->nr = src->nr;
  54         dst->alloc = src->alloc;
  55         src->ranges = NULL;
  56         src->alloc = src->nr = 0;
  57 }
  58
  59 /* tack on a _new_ range _at the end_ */
  60 void range_set_append_unsafe(struct range_set *rs, long a, long b)
  61 {
  62         assert(a <= b);
  63         range_set_grow(rs, 1);
  64         rs->ranges[rs->nr].start = a;
  65         rs->ranges[rs->nr].end = b;
  66         rs->nr++;
  67 }
  68
  69 void range_set_append(struct range_set *rs, long a, long b)
  70 {
  71         assert(rs->nr == 0 || rs->ranges[rs->nr-1].end <= a);
  72         range_set_append_unsafe(rs, a, b);
  73 }
  74
  75 static int range_cmp(const void *_r, const void *_s)
  76 {
  77         const struct range *r = _r;
  78         const struct range *s = _s;
  79
  80         /* this could be simply 'return r.start-s.start', but for the types */
  81         if (r->start == s->start)
  82                 return 0;
  83         if (r->start < s->start)
  84                 return -1;
  85         return 1;
  86 }
  87
  88 /*
  89  * Check that the ranges are non-empty, sorted and non-overlapping
  90  */
  91 static void range_set_check_invariants(struct range_set *rs)
  92 {
  93         unsigned int i;
  94
  95         if (!rs)
  96                 return;
  97
  98         if (rs->nr)
  99                 assert(rs->ranges[0].start < rs->ranges[0].end);
 100
 101         for (i = 1; i < rs->nr; i++) {
 102                 assert(rs->ranges[i-1].end < rs->ranges[i].start);
 103                 assert(rs->ranges[i].start < rs->ranges[i].end);
 104         }
 105 }
 106
 107 /*
 108  * In-place pass of sorting and merging the ranges in the range set,
 109  * to establish the invariants when we get the ranges from the user
 110  */
 111 void sort_and_merge_range_set(struct range_set *rs)
 112 {
 113         unsigned int i;
 114         unsigned int o = 0; /* output cursor */
 115
 116         QSORT(rs->ranges, rs->nr, range_cmp);
 117
 118         for (i = 0; i < rs->nr; i++) {
 119                 if (rs->ranges[i].start == rs->ranges[i].end)
 120                         continue;
 121                 if (o > 0 && rs->ranges[i].start <= rs->ranges[o-1].end) {
 122                         if (rs->ranges[o-1].end < rs->ranges[i].end)
 123                                 rs->ranges[o-1].end = rs->ranges[i].end;
 124                 } else {
 125                         rs->ranges[o].start = rs->ranges[i].start;
 126                         rs->ranges[o].end = rs->ranges[i].end;
 127                         o++;
 128                 }
 129         }
 130         assert(o <= rs->nr);
 131         rs->nr = o;
 132
 133         range_set_check_invariants(rs);
 134 }
 135
 136 /*
 137  * Union of range sets (i.e., sets of line numbers).  Used to merge
 138  * them when searches meet at a common ancestor.
 139  *
 140  * This is also where the ranges are consolidated into canonical form:
 141  * overlapping and adjacent ranges are merged, and empty ranges are
 142  * removed.
 143  */
 144 static void range_set_union(struct range_set *out,
 145                              struct range_set *a, struct range_set *b)
 146 {
 147         unsigned int i = 0, j = 0;
 148         struct range *ra = a->ranges;
 149         struct range *rb = b->ranges;
 150         /* cannot make an alias of out->ranges: it may change during grow */
 151
 152         assert(out->nr == 0);
 153         while (i < a->nr || j < b->nr) {
 154                 struct range *new_range;
 155                 if (i < a->nr && j < b->nr) {
 156                         if (ra[i].start < rb[j].start)
 157                                 new_range = &ra[i++];
 158                         else if (ra[i].start > rb[j].start)
 159                                 new_range = &rb[j++];
 160                         else if (ra[i].end < rb[j].end)
 161                                 new_range = &ra[i++];
 162                         else
 163                                 new_range = &rb[j++];
 164                 } else if (i < a->nr)      /* b exhausted */
 165                         new_range = &ra[i++];
 166                 else                       /* a exhausted */
 167                         new_range = &rb[j++];
 168                 if (new_range->start == new_range->end)
 169                         ; /* empty range */
 170                 else if (!out->nr || out->ranges[out->nr-1].end < new_range->start) {
 171                         range_set_grow(out, 1);
 172                         out->ranges[out->nr].start = new_range->start;
 173                         out->ranges[out->nr].end = new_range->end;
 174                         out->nr++;
 175                 } else if (out->ranges[out->nr-1].end < new_range->end) {
 176                         out->ranges[out->nr-1].end = new_range->end;
 177                 }
 178         }
 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         unsigned 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 static 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                 free(path);
 295                 return;
 296         }
 297
 298         p = xcalloc(1, sizeof(struct line_log_data));
 299         p->path = path;
 300         range_set_append(&p->ranges, begin, end);
 301         if (ip) {
 302                 p->next = ip->next;
 303                 ip->next = p;
 304         } else {
 305                 p->next = *list;
 306                 *list = p;
 307         }
 308 }
 309
 310 struct collect_diff_cbdata {
 311         struct diff_ranges *diff;
 312 };
 313
 314 static int collect_diff_cb(long start_a, long count_a,
 315                            long start_b, long count_b,
 316                            void *data)
 317 {
 318         struct collect_diff_cbdata *d = data;
 319
 320         if (count_a >= 0)
 321                 range_set_append(&d->diff->parent, start_a, start_a + count_a);
 322         if (count_b >= 0)
 323                 range_set_append(&d->diff->target, start_b, start_b + count_b);
 324
 325         return 0;
 326 }
 327
 328 static int collect_diff(mmfile_t *parent, mmfile_t *target, struct diff_ranges *out)
 329 {
 330         struct collect_diff_cbdata cbdata = {NULL};
 331         xpparam_t xpp;
 332         xdemitconf_t xecfg;
 333         xdemitcb_t ecb;
 334
 335         memset(&xpp, 0, sizeof(xpp));
 336         memset(&xecfg, 0, sizeof(xecfg));
 337         xecfg.ctxlen = xecfg.interhunkctxlen = 0;
 338
 339         cbdata.diff = out;
 340         xecfg.hunk_func = collect_diff_cb;
 341         memset(&ecb, 0, sizeof(ecb));
 342         ecb.priv = &cbdata;
 343         return xdi_diff(parent, target, &xpp, &xecfg, &ecb);
 344 }
 345
 346 /*
 347  * These are handy for debugging.  Removing them with #if 0 silences
 348  * the "unused function" warning.
 349  */
 350 #if 0
 351 static void dump_range_set(struct range_set *rs, const char *desc)
 352 {
 353         int i;
 354         printf("range set %s (%d items):\n", desc, rs->nr);
 355         for (i = 0; i < rs->nr; i++)
 356                 printf("\t[%ld,%ld]\n", rs->ranges[i].start, rs->ranges[i].end);
 357 }
 358
 359 static void dump_line_log_data(struct line_log_data *r)
 360 {
 361         char buf[4096];
 362         while (r) {
 363                 snprintf(buf, 4096, "file %s\n", r->path);
 364                 dump_range_set(&r->ranges, buf);
 365                 r = r->next;
 366         }
 367 }
 368
 369 static void dump_diff_ranges(struct diff_ranges *diff, const char *desc)
 370 {
 371         int i;
 372         assert(diff->parent.nr == diff->target.nr);
 373         printf("diff ranges %s (%d items):\n", desc, diff->parent.nr);
 374         printf("\tparent\ttarget\n");
 375         for (i = 0; i < diff->parent.nr; i++) {
 376                 printf("\t[%ld,%ld]\t[%ld,%ld]\n",
 377                        diff->parent.ranges[i].start,
 378                        diff->parent.ranges[i].end,
 379                        diff->target.ranges[i].start,
 380                        diff->target.ranges[i].end);
 381         }
 382 }
 383 #endif
 384
 385
 386 static int ranges_overlap(struct range *a, struct range *b)
 387 {
 388         return !(a->end <= b->start || b->end <= a->start);
 389 }
 390
 391 /*
 392  * Given a diff and the set of interesting ranges, determine all hunks
 393  * of the diff which touch (overlap) at least one of the interesting
 394  * ranges in the target.
 395  */
 396 static void diff_ranges_filter_touched(struct diff_ranges *out,
 397                                        struct diff_ranges *diff,
 398                                        struct range_set *rs)
 399 {
 400         unsigned int i, j = 0;
 401
 402         assert(out->target.nr == 0);
 403
 404         for (i = 0; i < diff->target.nr; i++) {
 405                 while (diff->target.ranges[i].start > rs->ranges[j].end) {
 406                         j++;
 407                         if (j == rs->nr)
 408                                 return;
 409                 }
 410                 if (ranges_overlap(&diff->target.ranges[i], &rs->ranges[j])) {
 411                         range_set_append(&out->parent,
 412                                          diff->parent.ranges[i].start,
 413                                          diff->parent.ranges[i].end);
 414                         range_set_append(&out->target,
 415                                          diff->target.ranges[i].start,
 416                                          diff->target.ranges[i].end);
 417                 }
 418         }
 419 }
 420
 421 /*
 422  * Adjust the line counts in 'rs' to account for the lines
 423  * added/removed in the diff.
 424  */
 425 static void range_set_shift_diff(struct range_set *out,
 426                                  struct range_set *rs,
 427                                  struct diff_ranges *diff)
 428 {
 429         unsigned int i, j = 0;
 430         long offset = 0;
 431         struct range *src = rs->ranges;
 432         struct range *target = diff->target.ranges;
 433         struct range *parent = diff->parent.ranges;
 434
 435         for (i = 0; i < rs->nr; i++) {
 436                 while (j < diff->target.nr && src[i].start >= target[j].start) {
 437                         offset += (parent[j].end-parent[j].start)
 438                                 - (target[j].end-target[j].start);
 439                         j++;
 440                 }
 441                 range_set_append(out, src[i].start+offset, src[i].end+offset);
 442         }
 443 }
 444
 445 /*
 446  * Given a diff and the set of interesting ranges, map the ranges
 447  * across the diff.  That is: observe that the target commit takes
 448  * blame for all the + (target-side) ranges.  So for every pair of
 449  * ranges in the diff that was touched, we remove the latter and add
 450  * its parent side.
 451  */
 452 static void range_set_map_across_diff(struct range_set *out,
 453                                       struct range_set *rs,
 454                                       struct diff_ranges *diff,
 455                                       struct diff_ranges **touched_out)
 456 {
 457         struct diff_ranges *touched = xmalloc(sizeof(*touched));
 458         struct range_set tmp1 = RANGE_SET_INIT;
 459         struct range_set tmp2 = RANGE_SET_INIT;
 460
 461         diff_ranges_init(touched);
 462         diff_ranges_filter_touched(touched, diff, rs);
 463         range_set_difference(&tmp1, rs, &touched->target);
 464         range_set_shift_diff(&tmp2, &tmp1, diff);
 465         range_set_union(out, &tmp2, &touched->parent);
 466         range_set_release(&tmp1);
 467         range_set_release(&tmp2);
 468
 469         *touched_out = touched;
 470 }
 471
 472 static struct commit *check_single_commit(struct rev_info *revs)
 473 {
 474         struct object *commit = NULL;
 475         int found = -1;
 476         int i;
 477
 478         for (i = 0; i < revs->pending.nr; i++) {
 479                 struct object *obj = revs->pending.objects[i].item;
 480                 if (obj->flags & UNINTERESTING)
 481                         continue;
 482                 obj = deref_tag(revs->repo, obj, NULL, 0);
 483                 if (obj->type != OBJ_COMMIT)
 484                         die("Non commit %s?", revs->pending.objects[i].name);
 485                 if (commit)
 486                         die("More than one commit to dig from: %s and %s?",
 487                             revs->pending.objects[i].name,
 488                             revs->pending.objects[found].name);
 489                 commit = obj;
 490                 found = i;
 491         }
 492
 493         if (!commit)
 494                 die("No commit specified?");
 495
 496         return (struct commit *) commit;
 497 }
 498
 499 static void fill_blob_sha1(struct repository *r, struct commit *commit,
 500                            struct diff_filespec *spec)
 501 {
 502         unsigned short mode;
 503         struct object_id oid;
 504
 505         if (get_tree_entry(r, &commit->object.oid, spec->path, &oid, &mode))
 506                 die("There is no path %s in the commit", spec->path);
 507         fill_filespec(spec, &oid, 1, mode);
 508
 509         return;
 510 }
 511
 512 static void fill_line_ends(struct repository *r,
 513                            struct diff_filespec *spec,
 514                            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(r, spec, NULL))
 523                 die("Cannot read blob %s", oid_to_hex(&spec->oid));
 524
 525         ALLOC_ARRAY(ends, size);
 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         REALLOC_ARRAY(ends, cur);
 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 repository *r, struct commit *commit,
 562             const char *prefix, struct string_list *args)
 563 {
 564         long lines = 0;
 565         unsigned long *ends = NULL;
 566         struct nth_line_cb cb_data;
 567         struct string_list_item *item;
 568         struct line_log_data *ranges = NULL;
 569         struct line_log_data *p;
 570
 571         for_each_string_list_item(item, args) {
 572                 const char *name_part, *range_part;
 573                 char *full_name;
 574                 struct diff_filespec *spec;
 575                 long begin = 0, end = 0;
 576                 long anchor;
 577
 578                 name_part = skip_range_arg(item->string, r->index);
 579                 if (!name_part || *name_part != ':' || !name_part[1])
 580                         die("-L argument not 'start,end:file' or ':funcname:file': %s",
 581                             item->string);
 582                 range_part = xstrndup(item->string, name_part - item->string);
 583                 name_part++;
 584
 585                 full_name = prefix_path(prefix, prefix ? strlen(prefix) : 0,
 586                                         name_part);
 587
 588                 spec = alloc_filespec(full_name);
 589                 fill_blob_sha1(r, commit, spec);
 590                 fill_line_ends(r, spec, &lines, &ends);
 591                 cb_data.spec = spec;
 592                 cb_data.lines = lines;
 593                 cb_data.line_ends = ends;
 594
 595                 p = search_line_log_data(ranges, full_name, NULL);
 596                 if (p && p->ranges.nr)
 597                         anchor = p->ranges.ranges[p->ranges.nr - 1].end + 1;
 598                 else
 599                         anchor = 1;
 600
 601                 if (parse_range_arg(range_part, nth_line, &cb_data,
 602                                     lines, anchor, &begin, &end,
 603                                     full_name, r->index))
 604                         die("malformed -L argument '%s'", range_part);
 605                 if ((!lines && (begin || end)) || lines < begin)
 606                         die("file %s has only %lu lines", name_part, lines);
 607                 if (begin < 1)
 608                         begin = 1;
 609                 if (end < 1 || lines < end)
 610                         end = lines;
 611                 begin--;
 612                 line_log_data_insert(&ranges, full_name, begin, end);
 613
 614                 free_filespec(spec);
 615                 FREE_AND_NULL(ends);
 616         }
 617
 618         for (p = ranges; p; p = p->next)
 619                 sort_and_merge_range_set(&p->ranges);
 620
 621         return ranges;
 622 }
 623
 624 static struct line_log_data *line_log_data_copy_one(struct line_log_data *r)
 625 {
 626         struct line_log_data *ret = xmalloc(sizeof(*ret));
 627
 628         assert(r);
 629         line_log_data_init(ret);
 630         range_set_copy(&ret->ranges, &r->ranges);
 631
 632         ret->path = xstrdup(r->path);
 633
 634         return ret;
 635 }
 636
 637 static struct line_log_data *
 638 line_log_data_copy(struct line_log_data *r)
 639 {
 640         struct line_log_data *ret = NULL;
 641         struct line_log_data *tmp = NULL, *prev = NULL;
 642
 643         assert(r);
 644         ret = tmp = prev = line_log_data_copy_one(r);
 645         r = r->next;
 646         while (r) {
 647                 tmp = line_log_data_copy_one(r);
 648                 prev->next = tmp;
 649                 prev = tmp;
 650                 r = r->next;
 651         }
 652
 653         return ret;
 654 }
 655
 656 /* merge two range sets across files */
 657 static struct line_log_data *line_log_data_merge(struct line_log_data *a,
 658                                                  struct line_log_data *b)
 659 {
 660         struct line_log_data *head = NULL, **pp = &head;
 661
 662         while (a || b) {
 663                 struct line_log_data *src;
 664                 struct line_log_data *src2 = NULL;
 665                 struct line_log_data *d;
 666                 int cmp;
 667                 if (!a)
 668                         cmp = 1;
 669                 else if (!b)
 670                         cmp = -1;
 671                 else
 672                         cmp = strcmp(a->path, b->path);
 673                 if (cmp < 0) {
 674                         src = a;
 675                         a = a->next;
 676                 } else if (cmp == 0) {
 677                         src = a;
 678                         a = a->next;
 679                         src2 = b;
 680                         b = b->next;
 681                 } else {
 682                         src = b;
 683                         b = b->next;
 684                 }
 685                 d = xmalloc(sizeof(struct line_log_data));
 686                 line_log_data_init(d);
 687                 d->path = xstrdup(src->path);
 688                 *pp = d;
 689                 pp = &d->next;
 690                 if (src2)
 691                         range_set_union(&d->ranges, &src->ranges, &src2->ranges);
 692                 else
 693                         range_set_copy(&d->ranges, &src->ranges);
 694         }
 695
 696         return head;
 697 }
 698
 699 static void add_line_range(struct rev_info *revs, struct commit *commit,
 700                            struct line_log_data *range)
 701 {
 702         struct line_log_data *old_line = NULL;
 703         struct line_log_data *new_line = NULL;
 704
 705         old_line = lookup_decoration(&revs->line_log_data, &commit->object);
 706         if (old_line && range) {
 707                 new_line = line_log_data_merge(old_line, range);
 708                 free_line_log_data(old_line);
 709         } else if (range)
 710                 new_line = line_log_data_copy(range);
 711
 712         if (new_line)
 713                 add_decoration(&revs->line_log_data, &commit->object, new_line);
 714 }
 715
 716 static void clear_commit_line_range(struct rev_info *revs, struct commit *commit)
 717 {
 718         struct line_log_data *r;
 719         r = lookup_decoration(&revs->line_log_data, &commit->object);
 720         if (!r)
 721                 return;
 722         free_line_log_data(r);
 723         add_decoration(&revs->line_log_data, &commit->object, NULL);
 724 }
 725
 726 static struct line_log_data *lookup_line_range(struct rev_info *revs,
 727                                                struct commit *commit)
 728 {
 729         struct line_log_data *ret = NULL;
 730         struct line_log_data *d;
 731
 732         ret = lookup_decoration(&revs->line_log_data, &commit->object);
 733
 734         for (d = ret; d; d = d->next)
 735                 range_set_check_invariants(&d->ranges);
 736
 737         return ret;
 738 }
 739
 740 static int same_paths_in_pathspec_and_range(struct pathspec *pathspec,
 741                                             struct line_log_data *range)
 742 {
 743         int i;
 744         struct line_log_data *r;
 745
 746         for (i = 0, r = range; i < pathspec->nr && r; i++, r = r->next)
 747                 if (strcmp(pathspec->items[i].match, r->path))
 748                         return 0;
 749         if (i < pathspec->nr || r)
 750                 /* different number of pathspec items and ranges */
 751                 return 0;
 752
 753         return 1;
 754 }
 755
 756 static void parse_pathspec_from_ranges(struct pathspec *pathspec,
 757                                        struct line_log_data *range)
 758 {
 759         struct line_log_data *r;
 760         struct argv_array array = ARGV_ARRAY_INIT;
 761         const char **paths;
 762
 763         for (r = range; r; r = r->next)
 764                 argv_array_push(&array, r->path);
 765         paths = argv_array_detach(&array);
 766
 767         parse_pathspec(pathspec, 0, PATHSPEC_PREFER_FULL, "", paths);
 768         /* strings are now owned by pathspec */
 769         free(paths);
 770 }
 771
 772 void line_log_init(struct rev_info *rev, const char *prefix, struct string_list *args)
 773 {
 774         struct commit *commit = NULL;
 775         struct line_log_data *range;
 776
 777         commit = check_single_commit(rev);
 778         range = parse_lines(rev->diffopt.repo, commit, prefix, args);
 779         add_line_range(rev, commit, range);
 780
 781         parse_pathspec_from_ranges(&rev->diffopt.pathspec, range);
 782 }
 783
 784 static void move_diff_queue(struct diff_queue_struct *dst,
 785                             struct diff_queue_struct *src)
 786 {
 787         assert(src != dst);
 788         memcpy(dst, src, sizeof(struct diff_queue_struct));
 789         DIFF_QUEUE_CLEAR(src);
 790 }
 791
 792 static void filter_diffs_for_paths(struct line_log_data *range, int keep_deletions)
 793 {
 794         int i;
 795         struct diff_queue_struct outq;
 796         DIFF_QUEUE_CLEAR(&outq);
 797
 798         for (i = 0; i < diff_queued_diff.nr; i++) {
 799                 struct diff_filepair *p = diff_queued_diff.queue[i];
 800                 struct line_log_data *rg = NULL;
 801
 802                 if (!DIFF_FILE_VALID(p->two)) {
 803                         if (keep_deletions)
 804                                 diff_q(&outq, p);
 805                         else
 806                                 diff_free_filepair(p);
 807                         continue;
 808                 }
 809                 for (rg = range; rg; rg = rg->next) {
 810                         if (!strcmp(rg->path, p->two->path))
 811                                 break;
 812                 }
 813                 if (rg)
 814                         diff_q(&outq, p);
 815                 else
 816                         diff_free_filepair(p);
 817         }
 818         free(diff_queued_diff.queue);
 819         diff_queued_diff = outq;
 820 }
 821
 822 static inline int diff_might_be_rename(void)
 823 {
 824         int i;
 825         for (i = 0; i < diff_queued_diff.nr; i++)
 826                 if (!DIFF_FILE_VALID(diff_queued_diff.queue[i]->one)) {
 827                         /* fprintf(stderr, "diff_might_be_rename found creation of: %s\n", */
 828                         /*      diff_queued_diff.queue[i]->two->path); */
 829                         return 1;
 830                 }
 831         return 0;
 832 }
 833
 834 static void queue_diffs(struct line_log_data *range,
 835                         struct diff_options *opt,
 836                         struct diff_queue_struct *queue,
 837                         struct commit *commit, struct commit *parent)
 838 {
 839         struct object_id *tree_oid, *parent_tree_oid;
 840
 841         assert(commit);
 842
 843         tree_oid = get_commit_tree_oid(commit);
 844         parent_tree_oid = parent ? get_commit_tree_oid(parent) : NULL;
 845
 846         if (opt->detect_rename &&
 847             !same_paths_in_pathspec_and_range(&opt->pathspec, range)) {
 848                 clear_pathspec(&opt->pathspec);
 849                 parse_pathspec_from_ranges(&opt->pathspec, range);
 850         }
 851         DIFF_QUEUE_CLEAR(&diff_queued_diff);
 852         diff_tree_oid(parent_tree_oid, tree_oid, "", opt);
 853         if (opt->detect_rename && diff_might_be_rename()) {
 854                 /* must look at the full tree diff to detect renames */
 855                 clear_pathspec(&opt->pathspec);
 856                 DIFF_QUEUE_CLEAR(&diff_queued_diff);
 857
 858                 diff_tree_oid(parent_tree_oid, tree_oid, "", opt);
 859
 860                 filter_diffs_for_paths(range, 1);
 861                 diffcore_std(opt);
 862                 filter_diffs_for_paths(range, 0);
 863         }
 864         move_diff_queue(queue, &diff_queued_diff);
 865 }
 866
 867 static char *get_nth_line(long line, unsigned long *ends, void *data)
 868 {
 869         if (line == 0)
 870                 return (char *)data;
 871         else
 872                 return (char *)data + ends[line] + 1;
 873 }
 874
 875 static void print_line(const char *prefix, char first,
 876                        long line, unsigned long *ends, void *data,
 877                        const char *color, const char *reset, FILE *file)
 878 {
 879         char *begin = get_nth_line(line, ends, data);
 880         char *end = get_nth_line(line+1, ends, data);
 881         int had_nl = 0;
 882
 883         if (end > begin && end[-1] == '\n') {
 884                 end--;
 885                 had_nl = 1;
 886         }
 887
 888         fputs(prefix, file);
 889         fputs(color, file);
 890         putc(first, file);
 891         fwrite(begin, 1, end-begin, file);
 892         fputs(reset, file);
 893         putc('\n', file);
 894         if (!had_nl)
 895                 fputs("\\ No newline at end of file\n", file);
 896 }
 897
 898 static char *output_prefix(struct diff_options *opt)
 899 {
 900         char *prefix = "";
 901
 902         if (opt->output_prefix) {
 903                 struct strbuf *sb = opt->output_prefix(opt, opt->output_prefix_data);
 904                 prefix = sb->buf;
 905         }
 906
 907         return prefix;
 908 }
 909
 910 static void dump_diff_hacky_one(struct rev_info *rev, struct line_log_data *range)
 911 {
 912         unsigned int i, j = 0;
 913         long p_lines, t_lines;
 914         unsigned long *p_ends = NULL, *t_ends = NULL;
 915         struct diff_filepair *pair = range->pair;
 916         struct diff_ranges *diff = &range->diff;
 917
 918         struct diff_options *opt = &rev->diffopt;
 919         char *prefix = output_prefix(opt);
 920         const char *c_reset = diff_get_color(opt->use_color, DIFF_RESET);
 921         const char *c_frag = diff_get_color(opt->use_color, DIFF_FRAGINFO);
 922         const char *c_meta = diff_get_color(opt->use_color, DIFF_METAINFO);
 923         const char *c_old = diff_get_color(opt->use_color, DIFF_FILE_OLD);
 924         const char *c_new = diff_get_color(opt->use_color, DIFF_FILE_NEW);
 925         const char *c_context = diff_get_color(opt->use_color, DIFF_CONTEXT);
 926
 927         if (!pair || !diff)
 928                 return;
 929
 930         if (pair->one->oid_valid)
 931                 fill_line_ends(rev->diffopt.repo, pair->one, &p_lines, &p_ends);
 932         fill_line_ends(rev->diffopt.repo, pair->two, &t_lines, &t_ends);
 933
 934         fprintf(opt->file, "%s%sdiff --git a/%s b/%s%s\n", prefix, c_meta, pair->one->path, pair->two->path, c_reset);
 935         fprintf(opt->file, "%s%s--- %s%s%s\n", prefix, c_meta,
 936                pair->one->oid_valid ? "a/" : "",
 937                pair->one->oid_valid ? pair->one->path : "/dev/null",
 938                c_reset);
 939         fprintf(opt->file, "%s%s+++ b/%s%s\n", prefix, c_meta, pair->two->path, c_reset);
 940         for (i = 0; i < range->ranges.nr; i++) {
 941                 long p_start, p_end;
 942                 long t_start = range->ranges.ranges[i].start;
 943                 long t_end = range->ranges.ranges[i].end;
 944                 long t_cur = t_start;
 945                 unsigned int j_last;
 946
 947                 while (j < diff->target.nr && diff->target.ranges[j].end < t_start)
 948                         j++;
 949                 if (j == diff->target.nr || diff->target.ranges[j].start > t_end)
 950                         continue;
 951
 952                 /* Scan ahead to determine the last diff that falls in this range */
 953                 j_last = j;
 954                 while (j_last < diff->target.nr && diff->target.ranges[j_last].start < t_end)
 955                         j_last++;
 956                 if (j_last > j)
 957                         j_last--;
 958
 959                 /*
 960                  * Compute parent hunk headers: we know that the diff
 961                  * has the correct line numbers (but not all hunks).
 962                  * So it suffices to shift the start/end according to
 963                  * the line numbers of the first/last hunk(s) that
 964                  * fall in this range.
 965                  */
 966                 if (t_start < diff->target.ranges[j].start)
 967                         p_start = diff->parent.ranges[j].start - (diff->target.ranges[j].start-t_start);
 968                 else
 969                         p_start = diff->parent.ranges[j].start;
 970                 if (t_end > diff->target.ranges[j_last].end)
 971                         p_end = diff->parent.ranges[j_last].end + (t_end-diff->target.ranges[j_last].end);
 972                 else
 973                         p_end = diff->parent.ranges[j_last].end;
 974
 975                 if (!p_start && !p_end) {
 976                         p_start = -1;
 977                         p_end = -1;
 978                 }
 979
 980                 /* Now output a diff hunk for this range */
 981                 fprintf(opt->file, "%s%s@@ -%ld,%ld +%ld,%ld @@%s\n",
 982                        prefix, c_frag,
 983                        p_start+1, p_end-p_start, t_start+1, t_end-t_start,
 984                        c_reset);
 985                 while (j < diff->target.nr && diff->target.ranges[j].start < t_end) {
 986                         int k;
 987                         for (; t_cur < diff->target.ranges[j].start; t_cur++)
 988                                 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
 989                                            c_context, c_reset, opt->file);
 990                         for (k = diff->parent.ranges[j].start; k < diff->parent.ranges[j].end; k++)
 991                                 print_line(prefix, '-', k, p_ends, pair->one->data,
 992                                            c_old, c_reset, opt->file);
 993                         for (; t_cur < diff->target.ranges[j].end && t_cur < t_end; t_cur++)
 994                                 print_line(prefix, '+', t_cur, t_ends, pair->two->data,
 995                                            c_new, c_reset, opt->file);
 996                         j++;
 997                 }
 998                 for (; t_cur < t_end; t_cur++)
 999                         print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
1000                                    c_context, c_reset, opt->file);
1001         }
1002
1003         free(p_ends);
1004         free(t_ends);
1005 }
1006
1007 /*
1008  * NEEDSWORK: manually building a diff here is not the Right
1009  * Thing(tm).  log -L should be built into the diff pipeline.
1010  */
1011 static void dump_diff_hacky(struct rev_info *rev, struct line_log_data *range)
1012 {
1013         fprintf(rev->diffopt.file, "%s\n", output_prefix(&rev->diffopt));
1014         while (range) {
1015                 dump_diff_hacky_one(rev, range);
1016                 range = range->next;
1017         }
1018 }
1019
1020 /*
1021  * Unlike most other functions, this destructively operates on
1022  * 'range'.
1023  */
1024 static int process_diff_filepair(struct rev_info *rev,
1025                                  struct diff_filepair *pair,
1026                                  struct line_log_data *range,
1027                                  struct diff_ranges **diff_out)
1028 {
1029         struct line_log_data *rg = range;
1030         struct range_set tmp;
1031         struct diff_ranges diff;
1032         mmfile_t file_parent, file_target;
1033
1034         assert(pair->two->path);
1035         while (rg) {
1036                 assert(rg->path);
1037                 if (!strcmp(rg->path, pair->two->path))
1038                         break;
1039                 rg = rg->next;
1040         }
1041
1042         if (!rg)
1043                 return 0;
1044         if (rg->ranges.nr == 0)
1045                 return 0;
1046
1047         assert(pair->two->oid_valid);
1048         diff_populate_filespec(rev->diffopt.repo, pair->two, NULL);
1049         file_target.ptr = pair->two->data;
1050         file_target.size = pair->two->size;
1051
1052         if (pair->one->oid_valid) {
1053                 diff_populate_filespec(rev->diffopt.repo, pair->one, NULL);
1054                 file_parent.ptr = pair->one->data;
1055                 file_parent.size = pair->one->size;
1056         } else {
1057                 file_parent.ptr = "";
1058                 file_parent.size = 0;
1059         }
1060
1061         diff_ranges_init(&diff);
1062         if (collect_diff(&file_parent, &file_target, &diff))
1063                 die("unable to generate diff for %s", pair->one->path);
1064
1065         /* NEEDSWORK should apply some heuristics to prevent mismatches */
1066         free(rg->path);
1067         rg->path = xstrdup(pair->one->path);
1068
1069         range_set_init(&tmp, 0);
1070         range_set_map_across_diff(&tmp, &rg->ranges, &diff, diff_out);
1071         range_set_release(&rg->ranges);
1072         range_set_move(&rg->ranges, &tmp);
1073
1074         diff_ranges_release(&diff);
1075
1076         return ((*diff_out)->parent.nr > 0);
1077 }
1078
1079 static struct diff_filepair *diff_filepair_dup(struct diff_filepair *pair)
1080 {
1081         struct diff_filepair *new_filepair = xmalloc(sizeof(struct diff_filepair));
1082         new_filepair->one = pair->one;
1083         new_filepair->two = pair->two;
1084         new_filepair->one->count++;
1085         new_filepair->two->count++;
1086         return new_filepair;
1087 }
1088
1089 static void free_diffqueues(int n, struct diff_queue_struct *dq)
1090 {
1091         int i, j;
1092         for (i = 0; i < n; i++)
1093                 for (j = 0; j < dq[i].nr; j++)
1094                         diff_free_filepair(dq[i].queue[j]);
1095         free(dq);
1096 }
1097
1098 static int process_all_files(struct line_log_data **range_out,
1099                              struct rev_info *rev,
1100                              struct diff_queue_struct *queue,
1101                              struct line_log_data *range)
1102 {
1103         int i, changed = 0;
1104
1105         *range_out = line_log_data_copy(range);
1106
1107         for (i = 0; i < queue->nr; i++) {
1108                 struct diff_ranges *pairdiff = NULL;
1109                 struct diff_filepair *pair = queue->queue[i];
1110                 if (process_diff_filepair(rev, pair, *range_out, &pairdiff)) {
1111                         /*
1112                          * Store away the diff for later output.  We
1113                          * tuck it in the ranges we got as _input_,
1114                          * since that's the commit that caused the
1115                          * diff.
1116                          *
1117                          * NEEDSWORK not enough when we get around to
1118                          * doing something interesting with merges;
1119                          * currently each invocation on a merge parent
1120                          * trashes the previous one's diff.
1121                          *
1122                          * NEEDSWORK tramples over data structures not owned here
1123                          */
1124                         struct line_log_data *rg = range;
1125                         changed++;
1126                         while (rg && strcmp(rg->path, pair->two->path))
1127                                 rg = rg->next;
1128                         assert(rg);
1129                         rg->pair = diff_filepair_dup(queue->queue[i]);
1130                         memcpy(&rg->diff, pairdiff, sizeof(struct diff_ranges));
1131                 }
1132                 free(pairdiff);
1133         }
1134
1135         return changed;
1136 }
1137
1138 int line_log_print(struct rev_info *rev, struct commit *commit)
1139 {
1140
1141         show_log(rev);
1142         if (!(rev->diffopt.output_format & DIFF_FORMAT_NO_OUTPUT)) {
1143                 struct line_log_data *range = lookup_line_range(rev, commit);
1144                 dump_diff_hacky(rev, range);
1145         }
1146         return 1;
1147 }
1148
1149 static int process_ranges_ordinary_commit(struct rev_info *rev, struct commit *commit,
1150                                           struct line_log_data *range)
1151 {
1152         struct commit *parent = NULL;
1153         struct diff_queue_struct queue;
1154         struct line_log_data *parent_range;
1155         int changed;
1156
1157         if (commit->parents)
1158                 parent = commit->parents->item;
1159
1160         queue_diffs(range, &rev->diffopt, &queue, commit, parent);
1161         changed = process_all_files(&parent_range, rev, &queue, range);
1162         if (parent)
1163                 add_line_range(rev, parent, parent_range);
1164         free_line_log_data(parent_range);
1165         return changed;
1166 }
1167
1168 static int process_ranges_merge_commit(struct rev_info *rev, struct commit *commit,
1169                                        struct line_log_data *range)
1170 {
1171         struct diff_queue_struct *diffqueues;
1172         struct line_log_data **cand;
1173         struct commit **parents;
1174         struct commit_list *p;
1175         int i;
1176         int nparents = commit_list_count(commit->parents);
1177
1178         if (nparents > 1 && rev->first_parent_only)
1179                 nparents = 1;
1180
1181         ALLOC_ARRAY(diffqueues, nparents);
1182         ALLOC_ARRAY(cand, nparents);
1183         ALLOC_ARRAY(parents, nparents);
1184
1185         p = commit->parents;
1186         for (i = 0; i < nparents; i++) {
1187                 parents[i] = p->item;
1188                 p = p->next;
1189                 queue_diffs(range, &rev->diffopt, &diffqueues[i], commit, parents[i]);
1190         }
1191
1192         for (i = 0; i < nparents; i++) {
1193                 int changed;
1194                 cand[i] = NULL;
1195                 changed = process_all_files(&cand[i], rev, &diffqueues[i], range);
1196                 if (!changed) {
1197                         /*
1198                          * This parent can take all the blame, so we
1199                          * don't follow any other path in history
1200                          */
1201                         add_line_range(rev, parents[i], cand[i]);
1202                         clear_commit_line_range(rev, commit);
1203                         commit_list_append(parents[i], &commit->parents);
1204                         free(parents);
1205                         free(cand);
1206                         free_diffqueues(nparents, diffqueues);
1207                         /* NEEDSWORK leaking like a sieve */
1208                         return 0;
1209                 }
1210         }
1211
1212         /*
1213          * No single parent took the blame.  We add the candidates
1214          * from the above loop to the parents.
1215          */
1216         for (i = 0; i < nparents; i++) {
1217                 add_line_range(rev, parents[i], cand[i]);
1218         }
1219
1220         clear_commit_line_range(rev, commit);
1221         free(parents);
1222         free(cand);
1223         free_diffqueues(nparents, diffqueues);
1224         return 1;
1225
1226         /* NEEDSWORK evil merge detection stuff */
1227         /* NEEDSWORK leaking like a sieve */
1228 }
1229
1230 static int process_ranges_arbitrary_commit(struct rev_info *rev, struct commit *commit)
1231 {
1232         struct line_log_data *range = lookup_line_range(rev, commit);
1233         int changed = 0;
1234
1235         if (range) {
1236                 if (!commit->parents || !commit->parents->next)
1237                         changed = process_ranges_ordinary_commit(rev, commit, range);
1238                 else
1239                         changed = process_ranges_merge_commit(rev, commit, range);
1240         }
1241
1242         if (!changed)
1243                 commit->object.flags |= TREESAME;
1244
1245         return changed;
1246 }
1247
1248 static enum rewrite_result line_log_rewrite_one(struct rev_info *rev, struct commit **pp)
1249 {
1250         for (;;) {
1251                 struct commit *p = *pp;
1252                 if (p->parents && p->parents->next)
1253                         return rewrite_one_ok;
1254                 if (p->object.flags & UNINTERESTING)
1255                         return rewrite_one_ok;
1256                 if (!(p->object.flags & TREESAME))
1257                         return rewrite_one_ok;
1258                 if (!p->parents)
1259                         return rewrite_one_noparents;
1260                 *pp = p->parents->item;
1261         }
1262 }
1263
1264 int line_log_filter(struct rev_info *rev)
1265 {
1266         struct commit *commit;
1267         struct commit_list *list = rev->commits;
1268         struct commit_list *out = NULL, **pp = &out;
1269
1270         while (list) {
1271                 struct commit_list *to_free = NULL;
1272                 commit = list->item;
1273                 if (process_ranges_arbitrary_commit(rev, commit)) {
1274                         *pp = list;
1275                         pp = &list->next;
1276                 } else
1277                         to_free = list;
1278                 list = list->next;
1279                 free(to_free);
1280         }
1281         *pp = NULL;
1282
1283         for (list = out; list; list = list->next)
1284                 rewrite_parents(rev, list->item, line_log_rewrite_one);
1285
1286         rev->commits = out;
1287
1288         return 0;
1289 }