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 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 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 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 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 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 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         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         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(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 commit *commit, struct diff_filespec *spec)
 500 {
 501         unsigned mode;
 502         unsigned char sha1[20];
 503
 504         if (get_tree_entry(commit->object.sha1, spec->path,
 505                            sha1, &mode))
 506                 die("There is no path %s in the commit", spec->path);
 507         fill_filespec(spec, sha1, 1, mode);
 508
 509         return;
 510 }
 511
 512 static void fill_line_ends(struct diff_filespec *spec, long *lines,
 513                            unsigned long **line_ends)
 514 {
 515         int num = 0, size = 50;
 516         long cur = 0;
 517         unsigned long *ends = NULL;
 518         char *data = NULL;
 519
 520         if (diff_populate_filespec(spec, 0))
 521                 die("Cannot read blob %s", sha1_to_hex(spec->sha1));
 522
 523         ends = xmalloc(size * sizeof(*ends));
 524         ends[cur++] = 0;
 525         data = spec->data;
 526         while (num < spec->size) {
 527                 if (data[num] == '\n' || num == spec->size - 1) {
 528                         ALLOC_GROW(ends, (cur + 1), size);
 529                         ends[cur++] = num;
 530                 }
 531                 num++;
 532         }
 533
 534         /* shrink the array to fit the elements */
 535         REALLOC_ARRAY(ends, cur);
 536         *lines = cur-1;
 537         *line_ends = ends;
 538 }
 539
 540 struct nth_line_cb {
 541         struct diff_filespec *spec;
 542         long lines;
 543         unsigned long *line_ends;
 544 };
 545
 546 static const char *nth_line(void *data, long line)
 547 {
 548         struct nth_line_cb *d = data;
 549         assert(d && line <= d->lines);
 550         assert(d->spec && d->spec->data);
 551
 552         if (line == 0)
 553                 return (char *)d->spec->data;
 554         else
 555                 return (char *)d->spec->data + d->line_ends[line] + 1;
 556 }
 557
 558 static struct line_log_data *
 559 parse_lines(struct commit *commit, const char *prefix, struct string_list *args)
 560 {
 561         long lines = 0;
 562         unsigned long *ends = NULL;
 563         struct nth_line_cb cb_data;
 564         struct string_list_item *item;
 565         struct line_log_data *ranges = NULL;
 566         struct line_log_data *p;
 567
 568         for_each_string_list_item(item, args) {
 569                 const char *name_part, *range_part;
 570                 char *full_name;
 571                 struct diff_filespec *spec;
 572                 long begin = 0, end = 0;
 573                 long anchor;
 574
 575                 name_part = skip_range_arg(item->string);
 576                 if (!name_part || *name_part != ':' || !name_part[1])
 577                         die("-L argument not 'start,end:file' or ':funcname:file': %s",
 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                 p = search_line_log_data(ranges, full_name, NULL);
 593                 if (p && p->ranges.nr)
 594                         anchor = p->ranges.ranges[p->ranges.nr - 1].end + 1;
 595                 else
 596                         anchor = 1;
 597
 598                 if (parse_range_arg(range_part, nth_line, &cb_data,
 599                                     lines, anchor, &begin, &end,
 600                                     full_name))
 601                         die("malformed -L argument '%s'", range_part);
 602                 if (lines < end || ((lines || begin) && lines < begin))
 603                         die("file %s has only %lu lines", name_part, lines);
 604                 if (begin < 1)
 605                         begin = 1;
 606                 if (end < 1)
 607                         end = lines;
 608                 begin--;
 609                 line_log_data_insert(&ranges, full_name, begin, end);
 610
 611                 free_filespec(spec);
 612                 free(ends);
 613                 ends = NULL;
 614         }
 615
 616         for (p = ranges; p; p = p->next)
 617                 sort_and_merge_range_set(&p->ranges);
 618
 619         return ranges;
 620 }
 621
 622 static struct line_log_data *line_log_data_copy_one(struct line_log_data *r)
 623 {
 624         struct line_log_data *ret = xmalloc(sizeof(*ret));
 625
 626         assert(r);
 627         line_log_data_init(ret);
 628         range_set_copy(&ret->ranges, &r->ranges);
 629
 630         ret->path = xstrdup(r->path);
 631
 632         return ret;
 633 }
 634
 635 static struct line_log_data *
 636 line_log_data_copy(struct line_log_data *r)
 637 {
 638         struct line_log_data *ret = NULL;
 639         struct line_log_data *tmp = NULL, *prev = NULL;
 640
 641         assert(r);
 642         ret = tmp = prev = line_log_data_copy_one(r);
 643         r = r->next;
 644         while (r) {
 645                 tmp = line_log_data_copy_one(r);
 646                 prev->next = tmp;
 647                 prev = tmp;
 648                 r = r->next;
 649         }
 650
 651         return ret;
 652 }
 653
 654 /* merge two range sets across files */
 655 static struct line_log_data *line_log_data_merge(struct line_log_data *a,
 656                                                  struct line_log_data *b)
 657 {
 658         struct line_log_data *head = NULL, **pp = &head;
 659
 660         while (a || b) {
 661                 struct line_log_data *src;
 662                 struct line_log_data *src2 = NULL;
 663                 struct line_log_data *d;
 664                 int cmp;
 665                 if (!a)
 666                         cmp = 1;
 667                 else if (!b)
 668                         cmp = -1;
 669                 else
 670                         cmp = strcmp(a->path, b->path);
 671                 if (cmp < 0) {
 672                         src = a;
 673                         a = a->next;
 674                 } else if (cmp == 0) {
 675                         src = a;
 676                         a = a->next;
 677                         src2 = b;
 678                         b = b->next;
 679                 } else {
 680                         src = b;
 681                         b = b->next;
 682                 }
 683                 d = xmalloc(sizeof(struct line_log_data));
 684                 line_log_data_init(d);
 685                 d->path = xstrdup(src->path);
 686                 *pp = d;
 687                 pp = &d->next;
 688                 if (src2)
 689                         range_set_union(&d->ranges, &src->ranges, &src2->ranges);
 690                 else
 691                         range_set_copy(&d->ranges, &src->ranges);
 692         }
 693
 694         return head;
 695 }
 696
 697 static void add_line_range(struct rev_info *revs, struct commit *commit,
 698                            struct line_log_data *range)
 699 {
 700         struct line_log_data *old = NULL;
 701         struct line_log_data *new = NULL;
 702
 703         old = lookup_decoration(&revs->line_log_data, &commit->object);
 704         if (old && range) {
 705                 new = line_log_data_merge(old, range);
 706                 free_line_log_data(old);
 707         } else if (range)
 708                 new = line_log_data_copy(range);
 709
 710         if (new)
 711                 add_decoration(&revs->line_log_data, &commit->object, new);
 712 }
 713
 714 static void clear_commit_line_range(struct rev_info *revs, struct commit *commit)
 715 {
 716         struct line_log_data *r;
 717         r = lookup_decoration(&revs->line_log_data, &commit->object);
 718         if (!r)
 719                 return;
 720         free_line_log_data(r);
 721         add_decoration(&revs->line_log_data, &commit->object, NULL);
 722 }
 723
 724 static struct line_log_data *lookup_line_range(struct rev_info *revs,
 725                                                struct commit *commit)
 726 {
 727         struct line_log_data *ret = NULL;
 728         struct line_log_data *d;
 729
 730         ret = lookup_decoration(&revs->line_log_data, &commit->object);
 731
 732         for (d = ret; d; d = d->next)
 733                 range_set_check_invariants(&d->ranges);
 734
 735         return ret;
 736 }
 737
 738 void line_log_init(struct rev_info *rev, const char *prefix, struct string_list *args)
 739 {
 740         struct commit *commit = NULL;
 741         struct line_log_data *range;
 742
 743         commit = check_single_commit(rev);
 744         range = parse_lines(commit, prefix, args);
 745         add_line_range(rev, commit, range);
 746
 747         if (!rev->diffopt.detect_rename) {
 748                 int i, count = 0;
 749                 struct line_log_data *r = range;
 750                 const char **paths;
 751                 while (r) {
 752                         count++;
 753                         r = r->next;
 754                 }
 755                 paths = xmalloc((count+1)*sizeof(char *));
 756                 r = range;
 757                 for (i = 0; i < count; i++) {
 758                         paths[i] = xstrdup(r->path);
 759                         r = r->next;
 760                 }
 761                 paths[count] = NULL;
 762                 parse_pathspec(&rev->diffopt.pathspec, 0,
 763                                PATHSPEC_PREFER_FULL, "", paths);
 764                 free(paths);
 765         }
 766 }
 767
 768 static void move_diff_queue(struct diff_queue_struct *dst,
 769                             struct diff_queue_struct *src)
 770 {
 771         assert(src != dst);
 772         memcpy(dst, src, sizeof(struct diff_queue_struct));
 773         DIFF_QUEUE_CLEAR(src);
 774 }
 775
 776 static void filter_diffs_for_paths(struct line_log_data *range, int keep_deletions)
 777 {
 778         int i;
 779         struct diff_queue_struct outq;
 780         DIFF_QUEUE_CLEAR(&outq);
 781
 782         for (i = 0; i < diff_queued_diff.nr; i++) {
 783                 struct diff_filepair *p = diff_queued_diff.queue[i];
 784                 struct line_log_data *rg = NULL;
 785
 786                 if (!DIFF_FILE_VALID(p->two)) {
 787                         if (keep_deletions)
 788                                 diff_q(&outq, p);
 789                         else
 790                                 diff_free_filepair(p);
 791                         continue;
 792                 }
 793                 for (rg = range; rg; rg = rg->next) {
 794                         if (!strcmp(rg->path, p->two->path))
 795                                 break;
 796                 }
 797                 if (rg)
 798                         diff_q(&outq, p);
 799                 else
 800                         diff_free_filepair(p);
 801         }
 802         free(diff_queued_diff.queue);
 803         diff_queued_diff = outq;
 804 }
 805
 806 static inline int diff_might_be_rename(void)
 807 {
 808         int i;
 809         for (i = 0; i < diff_queued_diff.nr; i++)
 810                 if (!DIFF_FILE_VALID(diff_queued_diff.queue[i]->one)) {
 811                         /* fprintf(stderr, "diff_might_be_rename found creation of: %s\n", */
 812                         /*      diff_queued_diff.queue[i]->two->path); */
 813                         return 1;
 814                 }
 815         return 0;
 816 }
 817
 818 static void queue_diffs(struct line_log_data *range,
 819                         struct diff_options *opt,
 820                         struct diff_queue_struct *queue,
 821                         struct commit *commit, struct commit *parent)
 822 {
 823         assert(commit);
 824
 825         DIFF_QUEUE_CLEAR(&diff_queued_diff);
 826         diff_tree_sha1(parent ? parent->tree->object.sha1 : NULL,
 827                         commit->tree->object.sha1, "", opt);
 828         if (opt->detect_rename) {
 829                 filter_diffs_for_paths(range, 1);
 830                 if (diff_might_be_rename())
 831                         diffcore_std(opt);
 832                 filter_diffs_for_paths(range, 0);
 833         }
 834         move_diff_queue(queue, &diff_queued_diff);
 835 }
 836
 837 static char *get_nth_line(long line, unsigned long *ends, void *data)
 838 {
 839         if (line == 0)
 840                 return (char *)data;
 841         else
 842                 return (char *)data + ends[line] + 1;
 843 }
 844
 845 static void print_line(const char *prefix, char first,
 846                        long line, unsigned long *ends, void *data,
 847                        const char *color, const char *reset)
 848 {
 849         char *begin = get_nth_line(line, ends, data);
 850         char *end = get_nth_line(line+1, ends, data);
 851         int had_nl = 0;
 852
 853         if (end > begin && end[-1] == '\n') {
 854                 end--;
 855                 had_nl = 1;
 856         }
 857
 858         fputs(prefix, stdout);
 859         fputs(color, stdout);
 860         putchar(first);
 861         fwrite(begin, 1, end-begin, stdout);
 862         fputs(reset, stdout);
 863         putchar('\n');
 864         if (!had_nl)
 865                 fputs("\\ No newline at end of file\n", stdout);
 866 }
 867
 868 static char *output_prefix(struct diff_options *opt)
 869 {
 870         char *prefix = "";
 871
 872         if (opt->output_prefix) {
 873                 struct strbuf *sb = opt->output_prefix(opt, opt->output_prefix_data);
 874                 prefix = sb->buf;
 875         }
 876
 877         return prefix;
 878 }
 879
 880 static void dump_diff_hacky_one(struct rev_info *rev, struct line_log_data *range)
 881 {
 882         int i, j = 0;
 883         long p_lines, t_lines;
 884         unsigned long *p_ends = NULL, *t_ends = NULL;
 885         struct diff_filepair *pair = range->pair;
 886         struct diff_ranges *diff = &range->diff;
 887
 888         struct diff_options *opt = &rev->diffopt;
 889         char *prefix = output_prefix(opt);
 890         const char *c_reset = diff_get_color(opt->use_color, DIFF_RESET);
 891         const char *c_frag = diff_get_color(opt->use_color, DIFF_FRAGINFO);
 892         const char *c_meta = diff_get_color(opt->use_color, DIFF_METAINFO);
 893         const char *c_old = diff_get_color(opt->use_color, DIFF_FILE_OLD);
 894         const char *c_new = diff_get_color(opt->use_color, DIFF_FILE_NEW);
 895         const char *c_context = diff_get_color(opt->use_color, DIFF_CONTEXT);
 896
 897         if (!pair || !diff)
 898                 return;
 899
 900         if (pair->one->sha1_valid)
 901                 fill_line_ends(pair->one, &p_lines, &p_ends);
 902         fill_line_ends(pair->two, &t_lines, &t_ends);
 903
 904         printf("%s%sdiff --git a/%s b/%s%s\n", prefix, c_meta, pair->one->path, pair->two->path, c_reset);
 905         printf("%s%s--- %s%s%s\n", prefix, c_meta,
 906                pair->one->sha1_valid ? "a/" : "",
 907                pair->one->sha1_valid ? pair->one->path : "/dev/null",
 908                c_reset);
 909         printf("%s%s+++ b/%s%s\n", prefix, c_meta, pair->two->path, c_reset);
 910         for (i = 0; i < range->ranges.nr; i++) {
 911                 long p_start, p_end;
 912                 long t_start = range->ranges.ranges[i].start;
 913                 long t_end = range->ranges.ranges[i].end;
 914                 long t_cur = t_start;
 915                 int j_last;
 916
 917                 while (j < diff->target.nr && diff->target.ranges[j].end < t_start)
 918                         j++;
 919                 if (j == diff->target.nr || diff->target.ranges[j].start > t_end)
 920                         continue;
 921
 922                 /* Scan ahead to determine the last diff that falls in this range */
 923                 j_last = j;
 924                 while (j_last < diff->target.nr && diff->target.ranges[j_last].start < t_end)
 925                         j_last++;
 926                 if (j_last > j)
 927                         j_last--;
 928
 929                 /*
 930                  * Compute parent hunk headers: we know that the diff
 931                  * has the correct line numbers (but not all hunks).
 932                  * So it suffices to shift the start/end according to
 933                  * the line numbers of the first/last hunk(s) that
 934                  * fall in this range.
 935                  */
 936                 if (t_start < diff->target.ranges[j].start)
 937                         p_start = diff->parent.ranges[j].start - (diff->target.ranges[j].start-t_start);
 938                 else
 939                         p_start = diff->parent.ranges[j].start;
 940                 if (t_end > diff->target.ranges[j_last].end)
 941                         p_end = diff->parent.ranges[j_last].end + (t_end-diff->target.ranges[j_last].end);
 942                 else
 943                         p_end = diff->parent.ranges[j_last].end;
 944
 945                 if (!p_start && !p_end) {
 946                         p_start = -1;
 947                         p_end = -1;
 948                 }
 949
 950                 /* Now output a diff hunk for this range */
 951                 printf("%s%s@@ -%ld,%ld +%ld,%ld @@%s\n",
 952                        prefix, c_frag,
 953                        p_start+1, p_end-p_start, t_start+1, t_end-t_start,
 954                        c_reset);
 955                 while (j < diff->target.nr && diff->target.ranges[j].start < t_end) {
 956                         int k;
 957                         for (; t_cur < diff->target.ranges[j].start; t_cur++)
 958                                 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
 959                                            c_context, c_reset);
 960                         for (k = diff->parent.ranges[j].start; k < diff->parent.ranges[j].end; k++)
 961                                 print_line(prefix, '-', k, p_ends, pair->one->data,
 962                                            c_old, c_reset);
 963                         for (; t_cur < diff->target.ranges[j].end && t_cur < t_end; t_cur++)
 964                                 print_line(prefix, '+', t_cur, t_ends, pair->two->data,
 965                                            c_new, c_reset);
 966                         j++;
 967                 }
 968                 for (; t_cur < t_end; t_cur++)
 969                         print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
 970                                    c_context, c_reset);
 971         }
 972
 973         free(p_ends);
 974         free(t_ends);
 975 }
 976
 977 /*
 978  * NEEDSWORK: manually building a diff here is not the Right
 979  * Thing(tm).  log -L should be built into the diff pipeline.
 980  */
 981 static void dump_diff_hacky(struct rev_info *rev, struct line_log_data *range)
 982 {
 983         puts(output_prefix(&rev->diffopt));
 984         while (range) {
 985                 dump_diff_hacky_one(rev, range);
 986                 range = range->next;
 987         }
 988 }
 989
 990 /*
 991  * Unlike most other functions, this destructively operates on
 992  * 'range'.
 993  */
 994 static int process_diff_filepair(struct rev_info *rev,
 995                                  struct diff_filepair *pair,
 996                                  struct line_log_data *range,
 997                                  struct diff_ranges **diff_out)
 998 {
 999         struct line_log_data *rg = range;
1000         struct range_set tmp;
1001         struct diff_ranges diff;
1002         mmfile_t file_parent, file_target;
1003
1004         assert(pair->two->path);
1005         while (rg) {
1006                 assert(rg->path);
1007                 if (!strcmp(rg->path, pair->two->path))
1008                         break;
1009                 rg = rg->next;
1010         }
1011
1012         if (!rg)
1013                 return 0;
1014         if (rg->ranges.nr == 0)
1015                 return 0;
1016
1017         assert(pair->two->sha1_valid);
1018         diff_populate_filespec(pair->two, 0);
1019         file_target.ptr = pair->two->data;
1020         file_target.size = pair->two->size;
1021
1022         if (pair->one->sha1_valid) {
1023                 diff_populate_filespec(pair->one, 0);
1024                 file_parent.ptr = pair->one->data;
1025                 file_parent.size = pair->one->size;
1026         } else {
1027                 file_parent.ptr = "";
1028                 file_parent.size = 0;
1029         }
1030
1031         diff_ranges_init(&diff);
1032         if (collect_diff(&file_parent, &file_target, &diff))
1033                 die("unable to generate diff for %s", pair->one->path);
1034
1035         /* NEEDSWORK should apply some heuristics to prevent mismatches */
1036         free(rg->path);
1037         rg->path = xstrdup(pair->one->path);
1038
1039         range_set_init(&tmp, 0);
1040         range_set_map_across_diff(&tmp, &rg->ranges, &diff, diff_out);
1041         range_set_release(&rg->ranges);
1042         range_set_move(&rg->ranges, &tmp);
1043
1044         diff_ranges_release(&diff);
1045
1046         return ((*diff_out)->parent.nr > 0);
1047 }
1048
1049 static struct diff_filepair *diff_filepair_dup(struct diff_filepair *pair)
1050 {
1051         struct diff_filepair *new = xmalloc(sizeof(struct diff_filepair));
1052         new->one = pair->one;
1053         new->two = pair->two;
1054         new->one->count++;
1055         new->two->count++;
1056         return new;
1057 }
1058
1059 static void free_diffqueues(int n, struct diff_queue_struct *dq)
1060 {
1061         int i, j;
1062         for (i = 0; i < n; i++)
1063                 for (j = 0; j < dq[i].nr; j++)
1064                         diff_free_filepair(dq[i].queue[j]);
1065         free(dq);
1066 }
1067
1068 static int process_all_files(struct line_log_data **range_out,
1069                              struct rev_info *rev,
1070                              struct diff_queue_struct *queue,
1071                              struct line_log_data *range)
1072 {
1073         int i, changed = 0;
1074
1075         *range_out = line_log_data_copy(range);
1076
1077         for (i = 0; i < queue->nr; i++) {
1078                 struct diff_ranges *pairdiff = NULL;
1079                 struct diff_filepair *pair = queue->queue[i];
1080                 if (process_diff_filepair(rev, pair, *range_out, &pairdiff)) {
1081                         /*
1082                          * Store away the diff for later output.  We
1083                          * tuck it in the ranges we got as _input_,
1084                          * since that's the commit that caused the
1085                          * diff.
1086                          *
1087                          * NEEDSWORK not enough when we get around to
1088                          * doing something interesting with merges;
1089                          * currently each invocation on a merge parent
1090                          * trashes the previous one's diff.
1091                          *
1092                          * NEEDSWORK tramples over data structures not owned here
1093                          */
1094                         struct line_log_data *rg = range;
1095                         changed++;
1096                         while (rg && strcmp(rg->path, pair->two->path))
1097                                 rg = rg->next;
1098                         assert(rg);
1099                         rg->pair = diff_filepair_dup(queue->queue[i]);
1100                         memcpy(&rg->diff, pairdiff, sizeof(struct diff_ranges));
1101                 }
1102                 free(pairdiff);
1103         }
1104
1105         return changed;
1106 }
1107
1108 int line_log_print(struct rev_info *rev, struct commit *commit)
1109 {
1110         struct line_log_data *range = lookup_line_range(rev, commit);
1111
1112         show_log(rev);
1113         dump_diff_hacky(rev, range);
1114         return 1;
1115 }
1116
1117 static int process_ranges_ordinary_commit(struct rev_info *rev, struct commit *commit,
1118                                           struct line_log_data *range)
1119 {
1120         struct commit *parent = NULL;
1121         struct diff_queue_struct queue;
1122         struct line_log_data *parent_range;
1123         int changed;
1124
1125         if (commit->parents)
1126                 parent = commit->parents->item;
1127
1128         queue_diffs(range, &rev->diffopt, &queue, commit, parent);
1129         changed = process_all_files(&parent_range, rev, &queue, range);
1130         if (parent)
1131                 add_line_range(rev, parent, parent_range);
1132         return changed;
1133 }
1134
1135 static int process_ranges_merge_commit(struct rev_info *rev, struct commit *commit,
1136                                        struct line_log_data *range)
1137 {
1138         struct diff_queue_struct *diffqueues;
1139         struct line_log_data **cand;
1140         struct commit **parents;
1141         struct commit_list *p;
1142         int i;
1143         int nparents = commit_list_count(commit->parents);
1144
1145         if (nparents > 1 && rev->first_parent_only)
1146                 nparents = 1;
1147
1148         diffqueues = xmalloc(nparents * sizeof(*diffqueues));
1149         cand = xmalloc(nparents * sizeof(*cand));
1150         parents = xmalloc(nparents * sizeof(*parents));
1151
1152         p = commit->parents;
1153         for (i = 0; i < nparents; i++) {
1154                 parents[i] = p->item;
1155                 p = p->next;
1156                 queue_diffs(range, &rev->diffopt, &diffqueues[i], commit, parents[i]);
1157         }
1158
1159         for (i = 0; i < nparents; i++) {
1160                 int changed;
1161                 cand[i] = NULL;
1162                 changed = process_all_files(&cand[i], rev, &diffqueues[i], range);
1163                 if (!changed) {
1164                         /*
1165                          * This parent can take all the blame, so we
1166                          * don't follow any other path in history
1167                          */
1168                         add_line_range(rev, parents[i], cand[i]);
1169                         clear_commit_line_range(rev, commit);
1170                         commit_list_append(parents[i], &commit->parents);
1171                         free(parents);
1172                         free(cand);
1173                         free_diffqueues(nparents, diffqueues);
1174                         /* NEEDSWORK leaking like a sieve */
1175                         return 0;
1176                 }
1177         }
1178
1179         /*
1180          * No single parent took the blame.  We add the candidates
1181          * from the above loop to the parents.
1182          */
1183         for (i = 0; i < nparents; i++) {
1184                 add_line_range(rev, parents[i], cand[i]);
1185         }
1186
1187         clear_commit_line_range(rev, commit);
1188         free(parents);
1189         free(cand);
1190         free_diffqueues(nparents, diffqueues);
1191         return 1;
1192
1193         /* NEEDSWORK evil merge detection stuff */
1194         /* NEEDSWORK leaking like a sieve */
1195 }
1196
1197 static int process_ranges_arbitrary_commit(struct rev_info *rev, struct commit *commit)
1198 {
1199         struct line_log_data *range = lookup_line_range(rev, commit);
1200         int changed = 0;
1201
1202         if (range) {
1203                 if (!commit->parents || !commit->parents->next)
1204                         changed = process_ranges_ordinary_commit(rev, commit, range);
1205                 else
1206                         changed = process_ranges_merge_commit(rev, commit, range);
1207         }
1208
1209         if (!changed)
1210                 commit->object.flags |= TREESAME;
1211
1212         return changed;
1213 }
1214
1215 static enum rewrite_result line_log_rewrite_one(struct rev_info *rev, struct commit **pp)
1216 {
1217         for (;;) {
1218                 struct commit *p = *pp;
1219                 if (p->parents && p->parents->next)
1220                         return rewrite_one_ok;
1221                 if (p->object.flags & UNINTERESTING)
1222                         return rewrite_one_ok;
1223                 if (!(p->object.flags & TREESAME))
1224                         return rewrite_one_ok;
1225                 if (!p->parents)
1226                         return rewrite_one_noparents;
1227                 *pp = p->parents->item;
1228         }
1229 }
1230
1231 int line_log_filter(struct rev_info *rev)
1232 {
1233         struct commit *commit;
1234         struct commit_list *list = rev->commits;
1235         struct commit_list *out = NULL, **pp = &out;
1236
1237         while (list) {
1238                 struct commit_list *to_free = NULL;
1239                 commit = list->item;
1240                 if (process_ranges_arbitrary_commit(rev, commit)) {
1241                         *pp = list;
1242                         pp = &list->next;
1243                 } else
1244                         to_free = list;
1245                 list = list->next;
1246                 free(to_free);
1247         }
1248         *pp = NULL;
1249
1250         for (list = out; list; list = list->next)
1251                 rewrite_parents(rev, list->item, line_log_rewrite_one);
1252
1253         rev->commits = out;
1254
1255         return 0;
1256 }