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(rs->ranges);
  38         rs->alloc = rs->nr = 0;
  39         rs->ranges = NULL;
  40 }
  41
  42 /* dst must be uninitialized! */
  43 static void range_set_copy(struct range_set *dst, struct range_set *src)
  44 {
  45         range_set_init(dst, src->nr);
  46         memcpy(dst->ranges, src->ranges, src->nr*sizeof(struct range_set));
  47         dst->nr = src->nr;
  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         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         int i;
 114         int o = 0; /* output cursor */
 115
 116         qsort(rs->ranges, rs->nr, sizeof(struct range), 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         int i = 0, j = 0, o = 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;
 155                 if (i < a->nr && j < b->nr) {
 156                         if (ra[i].start < rb[j].start)
 157                                 new = &ra[i++];
 158                         else if (ra[i].start > rb[j].start)
 159                                 new = &rb[j++];
 160                         else if (ra[i].end < rb[j].end)
 161                                 new = &ra[i++];
 162                         else
 163                                 new = &rb[j++];
 164                 } else if (i < a->nr)      /* b exhausted */
 165                         new = &ra[i++];
 166                 else                       /* a exhausted */
 167                         new = &rb[j++];
 168                 if (new->start == new->end)
 169                         ; /* empty range */
 170                 else if (!o || out->ranges[o-1].end < new->start) {
 171                         range_set_grow(out, 1);
 172                         out->ranges[o].start = new->start;
 173                         out->ranges[o].end = new->end;
 174                         o++;
 175                 } else if (out->ranges[o-1].end < new->end) {
 176                         out->ranges[o-1].end = new->end;
 177                 }
 178         }
 179         out->nr = o;
 180 }
 181
 182 /*
 183  * Difference of range sets (out = a \ b).  Pass the "interesting"
 184  * ranges as 'a' and the target side of the diff as 'b': it removes
 185  * the ranges for which the commit is responsible.
 186  */
 187 static void range_set_difference(struct range_set *out,
 188                                   struct range_set *a, struct range_set *b)
 189 {
 190         int i, j =  0;
 191         for (i = 0; i < a->nr; i++) {
 192                 long start = a->ranges[i].start;
 193                 long end = a->ranges[i].end;
 194                 while (start < end) {
 195                         while (j < b->nr && start >= b->ranges[j].end)
 196                                 /*
 197                                  * a:         |-------
 198                                  * b: ------|
 199                                  */
 200                                 j++;
 201                         if (j >= b->nr || end < b->ranges[j].start) {
 202                                 /*
 203                                  * b exhausted, or
 204                                  * a:  ----|
 205                                  * b:         |----
 206                                  */
 207                                 range_set_append(out, start, end);
 208                                 break;
 209                         }
 210                         if (start >= b->ranges[j].start) {
 211                                 /*
 212                                  * a:     |--????
 213                                  * b: |------|
 214                                  */
 215                                 start = b->ranges[j].end;
 216                         } else if (end > b->ranges[j].start) {
 217                                 /*
 218                                  * a: |-----|
 219                                  * b:    |--?????
 220                                  */
 221                                 if (start < b->ranges[j].start)
 222                                         range_set_append(out, start, b->ranges[j].start);
 223                                 start = b->ranges[j].end;
 224                         }
 225                 }
 226         }
 227 }
 228
 229 static void diff_ranges_init(struct diff_ranges *diff)
 230 {
 231         range_set_init(&diff->parent, 0);
 232         range_set_init(&diff->target, 0);
 233 }
 234
 235 static void diff_ranges_release(struct diff_ranges *diff)
 236 {
 237         range_set_release(&diff->parent);
 238         range_set_release(&diff->target);
 239 }
 240
 241 static void line_log_data_init(struct line_log_data *r)
 242 {
 243         memset(r, 0, sizeof(struct line_log_data));
 244         range_set_init(&r->ranges, 0);
 245 }
 246
 247 static void line_log_data_clear(struct line_log_data *r)
 248 {
 249         range_set_release(&r->ranges);
 250         if (r->pair)
 251                 diff_free_filepair(r->pair);
 252 }
 253
 254 static void free_line_log_data(struct line_log_data *r)
 255 {
 256         while (r) {
 257                 struct line_log_data *next = r->next;
 258                 line_log_data_clear(r);
 259                 free(r);
 260                 r = next;
 261         }
 262 }
 263
 264 static struct line_log_data *
 265 search_line_log_data(struct line_log_data *list, const char *path,
 266                      struct line_log_data **insertion_point)
 267 {
 268         struct line_log_data *p = list;
 269         if (insertion_point)
 270                 *insertion_point = NULL;
 271         while (p) {
 272                 int cmp = strcmp(p->path, path);
 273                 if (!cmp)
 274                         return p;
 275                 if (insertion_point && cmp < 0)
 276                         *insertion_point = p;
 277                 p = p->next;
 278         }
 279         return NULL;
 280 }
 281
 282 /*
 283  * Note: takes ownership of 'path', which happens to be what the only
 284  * caller needs.
 285  */
 286 static void line_log_data_insert(struct line_log_data **list,
 287                                  char *path,
 288                                  long begin, long end)
 289 {
 290         struct line_log_data *ip;
 291         struct line_log_data *p = search_line_log_data(*list, path, &ip);
 292
 293         if (p) {
 294                 range_set_append_unsafe(&p->ranges, begin, end);
 295                 free(path);
 296                 return;
 297         }
 298
 299         p = xcalloc(1, sizeof(struct line_log_data));
 300         p->path = path;
 301         range_set_append(&p->ranges, begin, end);
 302         if (ip) {
 303                 p->next = ip->next;
 304                 ip->next = p;
 305         } else {
 306                 p->next = *list;
 307                 *list = p;
 308         }
 309 }
 310
 311 struct collect_diff_cbdata {
 312         struct diff_ranges *diff;
 313 };
 314
 315 static int collect_diff_cb(long start_a, long count_a,
 316                            long start_b, long count_b,
 317                            void *data)
 318 {
 319         struct collect_diff_cbdata *d = data;
 320
 321         if (count_a >= 0)
 322                 range_set_append(&d->diff->parent, start_a, start_a + count_a);
 323         if (count_b >= 0)
 324                 range_set_append(&d->diff->target, start_b, start_b + count_b);
 325
 326         return 0;
 327 }
 328
 329 static int collect_diff(mmfile_t *parent, mmfile_t *target, struct diff_ranges *out)
 330 {
 331         struct collect_diff_cbdata cbdata = {NULL};
 332         xpparam_t xpp;
 333         xdemitconf_t xecfg;
 334         xdemitcb_t ecb;
 335
 336         memset(&xpp, 0, sizeof(xpp));
 337         memset(&xecfg, 0, sizeof(xecfg));
 338         xecfg.ctxlen = xecfg.interhunkctxlen = 0;
 339
 340         cbdata.diff = out;
 341         xecfg.hunk_func = collect_diff_cb;
 342         memset(&ecb, 0, sizeof(ecb));
 343         ecb.priv = &cbdata;
 344         return xdi_diff(parent, target, &xpp, &xecfg, &ecb);
 345 }
 346
 347 /*
 348  * These are handy for debugging.  Removing them with #if 0 silences
 349  * the "unused function" warning.
 350  */
 351 #if 0
 352 static void dump_range_set(struct range_set *rs, const char *desc)
 353 {
 354         int i;
 355         printf("range set %s (%d items):\n", desc, rs->nr);
 356         for (i = 0; i < rs->nr; i++)
 357                 printf("\t[%ld,%ld]\n", rs->ranges[i].start, rs->ranges[i].end);
 358 }
 359
 360 static void dump_line_log_data(struct line_log_data *r)
 361 {
 362         char buf[4096];
 363         while (r) {
 364                 snprintf(buf, 4096, "file %s\n", r->path);
 365                 dump_range_set(&r->ranges, buf);
 366                 r = r->next;
 367         }
 368 }
 369
 370 static void dump_diff_ranges(struct diff_ranges *diff, const char *desc)
 371 {
 372         int i;
 373         assert(diff->parent.nr == diff->target.nr);
 374         printf("diff ranges %s (%d items):\n", desc, diff->parent.nr);
 375         printf("\tparent\ttarget\n");
 376         for (i = 0; i < diff->parent.nr; i++) {
 377                 printf("\t[%ld,%ld]\t[%ld,%ld]\n",
 378                        diff->parent.ranges[i].start,
 379                        diff->parent.ranges[i].end,
 380                        diff->target.ranges[i].start,
 381                        diff->target.ranges[i].end);
 382         }
 383 }
 384 #endif
 385
 386
 387 static int ranges_overlap(struct range *a, struct range *b)
 388 {
 389         return !(a->end <= b->start || b->end <= a->start);
 390 }
 391
 392 /*
 393  * Given a diff and the set of interesting ranges, determine all hunks
 394  * of the diff which touch (overlap) at least one of the interesting
 395  * ranges in the target.
 396  */
 397 static void diff_ranges_filter_touched(struct diff_ranges *out,
 398                                        struct diff_ranges *diff,
 399                                        struct range_set *rs)
 400 {
 401         int i, j = 0;
 402
 403         assert(out->target.nr == 0);
 404
 405         for (i = 0; i < diff->target.nr; i++) {
 406                 while (diff->target.ranges[i].start > rs->ranges[j].end) {
 407                         j++;
 408                         if (j == rs->nr)
 409                                 return;
 410                 }
 411                 if (ranges_overlap(&diff->target.ranges[i], &rs->ranges[j])) {
 412                         range_set_append(&out->parent,
 413                                          diff->parent.ranges[i].start,
 414                                          diff->parent.ranges[i].end);
 415                         range_set_append(&out->target,
 416                                          diff->target.ranges[i].start,
 417                                          diff->target.ranges[i].end);
 418                 }
 419         }
 420 }
 421
 422 /*
 423  * Adjust the line counts in 'rs' to account for the lines
 424  * added/removed in the diff.
 425  */
 426 static void range_set_shift_diff(struct range_set *out,
 427                                  struct range_set *rs,
 428                                  struct diff_ranges *diff)
 429 {
 430         int i, j = 0;
 431         long offset = 0;
 432         struct range *src = rs->ranges;
 433         struct range *target = diff->target.ranges;
 434         struct range *parent = diff->parent.ranges;
 435
 436         for (i = 0; i < rs->nr; i++) {
 437                 while (j < diff->target.nr && src[i].start >= target[j].start) {
 438                         offset += (parent[j].end-parent[j].start)
 439                                 - (target[j].end-target[j].start);
 440                         j++;
 441                 }
 442                 range_set_append(out, src[i].start+offset, src[i].end+offset);
 443         }
 444 }
 445
 446 /*
 447  * Given a diff and the set of interesting ranges, map the ranges
 448  * across the diff.  That is: observe that the target commit takes
 449  * blame for all the + (target-side) ranges.  So for every pair of
 450  * ranges in the diff that was touched, we remove the latter and add
 451  * its parent side.
 452  */
 453 static void range_set_map_across_diff(struct range_set *out,
 454                                       struct range_set *rs,
 455                                       struct diff_ranges *diff,
 456                                       struct diff_ranges **touched_out)
 457 {
 458         struct diff_ranges *touched = xmalloc(sizeof(*touched));
 459         struct range_set tmp1 = RANGE_SET_INIT;
 460         struct range_set tmp2 = RANGE_SET_INIT;
 461
 462         diff_ranges_init(touched);
 463         diff_ranges_filter_touched(touched, diff, rs);
 464         range_set_difference(&tmp1, rs, &touched->target);
 465         range_set_shift_diff(&tmp2, &tmp1, diff);
 466         range_set_union(out, &tmp2, &touched->parent);
 467         range_set_release(&tmp1);
 468         range_set_release(&tmp2);
 469
 470         *touched_out = touched;
 471 }
 472
 473 static struct commit *check_single_commit(struct rev_info *revs)
 474 {
 475         struct object *commit = NULL;
 476         int found = -1;
 477         int i;
 478
 479         for (i = 0; i < revs->pending.nr; i++) {
 480                 struct object *obj = revs->pending.objects[i].item;
 481                 if (obj->flags & UNINTERESTING)
 482                         continue;
 483                 obj = deref_tag(obj, NULL, 0);
 484                 if (obj->type != OBJ_COMMIT)
 485                         die("Non commit %s?", revs->pending.objects[i].name);
 486                 if (commit)
 487                         die("More than one commit to dig from: %s and %s?",
 488                             revs->pending.objects[i].name,
 489                             revs->pending.objects[found].name);
 490                 commit = obj;
 491                 found = i;
 492         }
 493
 494         if (!commit)
 495                 die("No commit specified?");
 496
 497         return (struct commit *) commit;
 498 }
 499
 500 static void fill_blob_sha1(struct commit *commit, struct diff_filespec *spec)
 501 {
 502         unsigned mode;
 503         unsigned char sha1[20];
 504
 505         if (get_tree_entry(commit->object.oid.hash, spec->path,
 506                            sha1, &mode))
 507                 die("There is no path %s in the commit", spec->path);
 508         fill_filespec(spec, sha1, 1, mode);
 509
 510         return;
 511 }
 512
 513 static void fill_line_ends(struct diff_filespec *spec, long *lines,
 514                            unsigned long **line_ends)
 515 {
 516         int num = 0, size = 50;
 517         long cur = 0;
 518         unsigned long *ends = NULL;
 519         char *data = NULL;
 520
 521         if (diff_populate_filespec(spec, 0))
 522                 die("Cannot read blob %s", oid_to_hex(&spec->oid));
 523
 524         ALLOC_ARRAY(ends, size);
 525         ends[cur++] = 0;
 526         data = spec->data;
 527         while (num < spec->size) {
 528                 if (data[num] == '\n' || num == spec->size - 1) {
 529                         ALLOC_GROW(ends, (cur + 1), size);
 530                         ends[cur++] = num;
 531                 }
 532                 num++;
 533         }
 534
 535         /* shrink the array to fit the elements */
 536         REALLOC_ARRAY(ends, cur);
 537         *lines = cur-1;
 538         *line_ends = ends;
 539 }
 540
 541 struct nth_line_cb {
 542         struct diff_filespec *spec;
 543         long lines;
 544         unsigned long *line_ends;
 545 };
 546
 547 static const char *nth_line(void *data, long line)
 548 {
 549         struct nth_line_cb *d = data;
 550         assert(d && line <= d->lines);
 551         assert(d->spec && d->spec->data);
 552
 553         if (line == 0)
 554                 return (char *)d->spec->data;
 555         else
 556                 return (char *)d->spec->data + d->line_ends[line] + 1;
 557 }
 558
 559 static struct line_log_data *
 560 parse_lines(struct commit *commit, const char *prefix, struct string_list *args)
 561 {
 562         long lines = 0;
 563         unsigned long *ends = NULL;
 564         struct nth_line_cb cb_data;
 565         struct string_list_item *item;
 566         struct line_log_data *ranges = NULL;
 567         struct line_log_data *p;
 568
 569         for_each_string_list_item(item, args) {
 570                 const char *name_part, *range_part;
 571                 char *full_name;
 572                 struct diff_filespec *spec;
 573                 long begin = 0, end = 0;
 574                 long anchor;
 575
 576                 name_part = skip_range_arg(item->string);
 577                 if (!name_part || *name_part != ':' || !name_part[1])
 578                         die("-L argument not 'start,end:file' or ':funcname:file': %s",
 579                             item->string);
 580                 range_part = xstrndup(item->string, name_part - item->string);
 581                 name_part++;
 582
 583                 full_name = prefix_path(prefix, prefix ? strlen(prefix) : 0,
 584                                         name_part);
 585
 586                 spec = alloc_filespec(full_name);
 587                 fill_blob_sha1(commit, spec);
 588                 fill_line_ends(spec, &lines, &ends);
 589                 cb_data.spec = spec;
 590                 cb_data.lines = lines;
 591                 cb_data.line_ends = ends;
 592
 593                 p = search_line_log_data(ranges, full_name, NULL);
 594                 if (p && p->ranges.nr)
 595                         anchor = p->ranges.ranges[p->ranges.nr - 1].end + 1;
 596                 else
 597                         anchor = 1;
 598
 599                 if (parse_range_arg(range_part, nth_line, &cb_data,
 600                                     lines, anchor, &begin, &end,
 601                                     full_name))
 602                         die("malformed -L argument '%s'", range_part);
 603                 if (lines < end || ((lines || begin) && lines < begin))
 604                         die("file %s has only %lu lines", name_part, lines);
 605                 if (begin < 1)
 606                         begin = 1;
 607                 if (end < 1)
 608                         end = lines;
 609                 begin--;
 610                 line_log_data_insert(&ranges, full_name, begin, end);
 611
 612                 free_filespec(spec);
 613                 free(ends);
 614                 ends = NULL;
 615         }
 616
 617         for (p = ranges; p; p = p->next)
 618                 sort_and_merge_range_set(&p->ranges);
 619
 620         return ranges;
 621 }
 622
 623 static struct line_log_data *line_log_data_copy_one(struct line_log_data *r)
 624 {
 625         struct line_log_data *ret = xmalloc(sizeof(*ret));
 626
 627         assert(r);
 628         line_log_data_init(ret);
 629         range_set_copy(&ret->ranges, &r->ranges);
 630
 631         ret->path = xstrdup(r->path);
 632
 633         return ret;
 634 }
 635
 636 static struct line_log_data *
 637 line_log_data_copy(struct line_log_data *r)
 638 {
 639         struct line_log_data *ret = NULL;
 640         struct line_log_data *tmp = NULL, *prev = NULL;
 641
 642         assert(r);
 643         ret = tmp = prev = line_log_data_copy_one(r);
 644         r = r->next;
 645         while (r) {
 646                 tmp = line_log_data_copy_one(r);
 647                 prev->next = tmp;
 648                 prev = tmp;
 649                 r = r->next;
 650         }
 651
 652         return ret;
 653 }
 654
 655 /* merge two range sets across files */
 656 static struct line_log_data *line_log_data_merge(struct line_log_data *a,
 657                                                  struct line_log_data *b)
 658 {
 659         struct line_log_data *head = NULL, **pp = &head;
 660
 661         while (a || b) {
 662                 struct line_log_data *src;
 663                 struct line_log_data *src2 = NULL;
 664                 struct line_log_data *d;
 665                 int cmp;
 666                 if (!a)
 667                         cmp = 1;
 668                 else if (!b)
 669                         cmp = -1;
 670                 else
 671                         cmp = strcmp(a->path, b->path);
 672                 if (cmp < 0) {
 673                         src = a;
 674                         a = a->next;
 675                 } else if (cmp == 0) {
 676                         src = a;
 677                         a = a->next;
 678                         src2 = b;
 679                         b = b->next;
 680                 } else {
 681                         src = b;
 682                         b = b->next;
 683                 }
 684                 d = xmalloc(sizeof(struct line_log_data));
 685                 line_log_data_init(d);
 686                 d->path = xstrdup(src->path);
 687                 *pp = d;
 688                 pp = &d->next;
 689                 if (src2)
 690                         range_set_union(&d->ranges, &src->ranges, &src2->ranges);
 691                 else
 692                         range_set_copy(&d->ranges, &src->ranges);
 693         }
 694
 695         return head;
 696 }
 697
 698 static void add_line_range(struct rev_info *revs, struct commit *commit,
 699                            struct line_log_data *range)
 700 {
 701         struct line_log_data *old = NULL;
 702         struct line_log_data *new = NULL;
 703
 704         old = lookup_decoration(&revs->line_log_data, &commit->object);
 705         if (old && range) {
 706                 new = line_log_data_merge(old, range);
 707                 free_line_log_data(old);
 708         } else if (range)
 709                 new = line_log_data_copy(range);
 710
 711         if (new)
 712                 add_decoration(&revs->line_log_data, &commit->object, new);
 713 }
 714
 715 static void clear_commit_line_range(struct rev_info *revs, struct commit *commit)
 716 {
 717         struct line_log_data *r;
 718         r = lookup_decoration(&revs->line_log_data, &commit->object);
 719         if (!r)
 720                 return;
 721         free_line_log_data(r);
 722         add_decoration(&revs->line_log_data, &commit->object, NULL);
 723 }
 724
 725 static struct line_log_data *lookup_line_range(struct rev_info *revs,
 726                                                struct commit *commit)
 727 {
 728         struct line_log_data *ret = NULL;
 729         struct line_log_data *d;
 730
 731         ret = lookup_decoration(&revs->line_log_data, &commit->object);
 732
 733         for (d = ret; d; d = d->next)
 734                 range_set_check_invariants(&d->ranges);
 735
 736         return ret;
 737 }
 738
 739 void line_log_init(struct rev_info *rev, const char *prefix, struct string_list *args)
 740 {
 741         struct commit *commit = NULL;
 742         struct line_log_data *range;
 743
 744         commit = check_single_commit(rev);
 745         range = parse_lines(commit, prefix, args);
 746         add_line_range(rev, commit, range);
 747
 748         if (!rev->diffopt.detect_rename) {
 749                 struct line_log_data *r;
 750                 struct argv_array array = ARGV_ARRAY_INIT;
 751                 const char **paths;
 752
 753                 for (r = range; r; r = r->next)
 754                         argv_array_push(&array, r->path);
 755                 paths = argv_array_detach(&array);
 756
 757                 parse_pathspec(&rev->diffopt.pathspec, 0,
 758                                PATHSPEC_PREFER_FULL, "", paths);
 759                 /* strings are now owned by pathspec */
 760                 free(paths);
 761         }
 762 }
 763
 764 static void move_diff_queue(struct diff_queue_struct *dst,
 765                             struct diff_queue_struct *src)
 766 {
 767         assert(src != dst);
 768         memcpy(dst, src, sizeof(struct diff_queue_struct));
 769         DIFF_QUEUE_CLEAR(src);
 770 }
 771
 772 static void filter_diffs_for_paths(struct line_log_data *range, int keep_deletions)
 773 {
 774         int i;
 775         struct diff_queue_struct outq;
 776         DIFF_QUEUE_CLEAR(&outq);
 777
 778         for (i = 0; i < diff_queued_diff.nr; i++) {
 779                 struct diff_filepair *p = diff_queued_diff.queue[i];
 780                 struct line_log_data *rg = NULL;
 781
 782                 if (!DIFF_FILE_VALID(p->two)) {
 783                         if (keep_deletions)
 784                                 diff_q(&outq, p);
 785                         else
 786                                 diff_free_filepair(p);
 787                         continue;
 788                 }
 789                 for (rg = range; rg; rg = rg->next) {
 790                         if (!strcmp(rg->path, p->two->path))
 791                                 break;
 792                 }
 793                 if (rg)
 794                         diff_q(&outq, p);
 795                 else
 796                         diff_free_filepair(p);
 797         }
 798         free(diff_queued_diff.queue);
 799         diff_queued_diff = outq;
 800 }
 801
 802 static inline int diff_might_be_rename(void)
 803 {
 804         int i;
 805         for (i = 0; i < diff_queued_diff.nr; i++)
 806                 if (!DIFF_FILE_VALID(diff_queued_diff.queue[i]->one)) {
 807                         /* fprintf(stderr, "diff_might_be_rename found creation of: %s\n", */
 808                         /*      diff_queued_diff.queue[i]->two->path); */
 809                         return 1;
 810                 }
 811         return 0;
 812 }
 813
 814 static void queue_diffs(struct line_log_data *range,
 815                         struct diff_options *opt,
 816                         struct diff_queue_struct *queue,
 817                         struct commit *commit, struct commit *parent)
 818 {
 819         assert(commit);
 820
 821         DIFF_QUEUE_CLEAR(&diff_queued_diff);
 822         diff_tree_sha1(parent ? parent->tree->object.oid.hash : NULL,
 823                         commit->tree->object.oid.hash, "", opt);
 824         if (opt->detect_rename) {
 825                 filter_diffs_for_paths(range, 1);
 826                 if (diff_might_be_rename())
 827                         diffcore_std(opt);
 828                 filter_diffs_for_paths(range, 0);
 829         }
 830         move_diff_queue(queue, &diff_queued_diff);
 831 }
 832
 833 static char *get_nth_line(long line, unsigned long *ends, void *data)
 834 {
 835         if (line == 0)
 836                 return (char *)data;
 837         else
 838                 return (char *)data + ends[line] + 1;
 839 }
 840
 841 static void print_line(const char *prefix, char first,
 842                        long line, unsigned long *ends, void *data,
 843                        const char *color, const char *reset, FILE *file)
 844 {
 845         char *begin = get_nth_line(line, ends, data);
 846         char *end = get_nth_line(line+1, ends, data);
 847         int had_nl = 0;
 848
 849         if (end > begin && end[-1] == '\n') {
 850                 end--;
 851                 had_nl = 1;
 852         }
 853
 854         fputs(prefix, file);
 855         fputs(color, file);
 856         putc(first, file);
 857         fwrite(begin, 1, end-begin, file);
 858         fputs(reset, file);
 859         putc('\n', file);
 860         if (!had_nl)
 861                 fputs("\\ No newline at end of file\n", file);
 862 }
 863
 864 static char *output_prefix(struct diff_options *opt)
 865 {
 866         char *prefix = "";
 867
 868         if (opt->output_prefix) {
 869                 struct strbuf *sb = opt->output_prefix(opt, opt->output_prefix_data);
 870                 prefix = sb->buf;
 871         }
 872
 873         return prefix;
 874 }
 875
 876 static void dump_diff_hacky_one(struct rev_info *rev, struct line_log_data *range)
 877 {
 878         int i, j = 0;
 879         long p_lines, t_lines;
 880         unsigned long *p_ends = NULL, *t_ends = NULL;
 881         struct diff_filepair *pair = range->pair;
 882         struct diff_ranges *diff = &range->diff;
 883
 884         struct diff_options *opt = &rev->diffopt;
 885         char *prefix = output_prefix(opt);
 886         const char *c_reset = diff_get_color(opt->use_color, DIFF_RESET);
 887         const char *c_frag = diff_get_color(opt->use_color, DIFF_FRAGINFO);
 888         const char *c_meta = diff_get_color(opt->use_color, DIFF_METAINFO);
 889         const char *c_old = diff_get_color(opt->use_color, DIFF_FILE_OLD);
 890         const char *c_new = diff_get_color(opt->use_color, DIFF_FILE_NEW);
 891         const char *c_context = diff_get_color(opt->use_color, DIFF_CONTEXT);
 892
 893         if (!pair || !diff)
 894                 return;
 895
 896         if (pair->one->oid_valid)
 897                 fill_line_ends(pair->one, &p_lines, &p_ends);
 898         fill_line_ends(pair->two, &t_lines, &t_ends);
 899
 900         fprintf(opt->file, "%s%sdiff --git a/%s b/%s%s\n", prefix, c_meta, pair->one->path, pair->two->path, c_reset);
 901         fprintf(opt->file, "%s%s--- %s%s%s\n", prefix, c_meta,
 902                pair->one->oid_valid ? "a/" : "",
 903                pair->one->oid_valid ? pair->one->path : "/dev/null",
 904                c_reset);
 905         fprintf(opt->file, "%s%s+++ b/%s%s\n", prefix, c_meta, pair->two->path, c_reset);
 906         for (i = 0; i < range->ranges.nr; i++) {
 907                 long p_start, p_end;
 908                 long t_start = range->ranges.ranges[i].start;
 909                 long t_end = range->ranges.ranges[i].end;
 910                 long t_cur = t_start;
 911                 int j_last;
 912
 913                 while (j < diff->target.nr && diff->target.ranges[j].end < t_start)
 914                         j++;
 915                 if (j == diff->target.nr || diff->target.ranges[j].start > t_end)
 916                         continue;
 917
 918                 /* Scan ahead to determine the last diff that falls in this range */
 919                 j_last = j;
 920                 while (j_last < diff->target.nr && diff->target.ranges[j_last].start < t_end)
 921                         j_last++;
 922                 if (j_last > j)
 923                         j_last--;
 924
 925                 /*
 926                  * Compute parent hunk headers: we know that the diff
 927                  * has the correct line numbers (but not all hunks).
 928                  * So it suffices to shift the start/end according to
 929                  * the line numbers of the first/last hunk(s) that
 930                  * fall in this range.
 931                  */
 932                 if (t_start < diff->target.ranges[j].start)
 933                         p_start = diff->parent.ranges[j].start - (diff->target.ranges[j].start-t_start);
 934                 else
 935                         p_start = diff->parent.ranges[j].start;
 936                 if (t_end > diff->target.ranges[j_last].end)
 937                         p_end = diff->parent.ranges[j_last].end + (t_end-diff->target.ranges[j_last].end);
 938                 else
 939                         p_end = diff->parent.ranges[j_last].end;
 940
 941                 if (!p_start && !p_end) {
 942                         p_start = -1;
 943                         p_end = -1;
 944                 }
 945
 946                 /* Now output a diff hunk for this range */
 947                 fprintf(opt->file, "%s%s@@ -%ld,%ld +%ld,%ld @@%s\n",
 948                        prefix, c_frag,
 949                        p_start+1, p_end-p_start, t_start+1, t_end-t_start,
 950                        c_reset);
 951                 while (j < diff->target.nr && diff->target.ranges[j].start < t_end) {
 952                         int k;
 953                         for (; t_cur < diff->target.ranges[j].start; t_cur++)
 954                                 print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
 955                                            c_context, c_reset, opt->file);
 956                         for (k = diff->parent.ranges[j].start; k < diff->parent.ranges[j].end; k++)
 957                                 print_line(prefix, '-', k, p_ends, pair->one->data,
 958                                            c_old, c_reset, opt->file);
 959                         for (; t_cur < diff->target.ranges[j].end && t_cur < t_end; t_cur++)
 960                                 print_line(prefix, '+', t_cur, t_ends, pair->two->data,
 961                                            c_new, c_reset, opt->file);
 962                         j++;
 963                 }
 964                 for (; t_cur < t_end; t_cur++)
 965                         print_line(prefix, ' ', t_cur, t_ends, pair->two->data,
 966                                    c_context, c_reset, opt->file);
 967         }
 968
 969         free(p_ends);
 970         free(t_ends);
 971 }
 972
 973 /*
 974  * NEEDSWORK: manually building a diff here is not the Right
 975  * Thing(tm).  log -L should be built into the diff pipeline.
 976  */
 977 static void dump_diff_hacky(struct rev_info *rev, struct line_log_data *range)
 978 {
 979         fprintf(rev->diffopt.file, "%s\n", output_prefix(&rev->diffopt));
 980         while (range) {
 981                 dump_diff_hacky_one(rev, range);
 982                 range = range->next;
 983         }
 984 }
 985
 986 /*
 987  * Unlike most other functions, this destructively operates on
 988  * 'range'.
 989  */
 990 static int process_diff_filepair(struct rev_info *rev,
 991                                  struct diff_filepair *pair,
 992                                  struct line_log_data *range,
 993                                  struct diff_ranges **diff_out)
 994 {
 995         struct line_log_data *rg = range;
 996         struct range_set tmp;
 997         struct diff_ranges diff;
 998         mmfile_t file_parent, file_target;
 999
1000         assert(pair->two->path);
1001         while (rg) {
1002                 assert(rg->path);
1003                 if (!strcmp(rg->path, pair->two->path))
1004                         break;
1005                 rg = rg->next;
1006         }
1007
1008         if (!rg)
1009                 return 0;
1010         if (rg->ranges.nr == 0)
1011                 return 0;
1012
1013         assert(pair->two->oid_valid);
1014         diff_populate_filespec(pair->two, 0);
1015         file_target.ptr = pair->two->data;
1016         file_target.size = pair->two->size;
1017
1018         if (pair->one->oid_valid) {
1019                 diff_populate_filespec(pair->one, 0);
1020                 file_parent.ptr = pair->one->data;
1021                 file_parent.size = pair->one->size;
1022         } else {
1023                 file_parent.ptr = "";
1024                 file_parent.size = 0;
1025         }
1026
1027         diff_ranges_init(&diff);
1028         if (collect_diff(&file_parent, &file_target, &diff))
1029                 die("unable to generate diff for %s", pair->one->path);
1030
1031         /* NEEDSWORK should apply some heuristics to prevent mismatches */
1032         free(rg->path);
1033         rg->path = xstrdup(pair->one->path);
1034
1035         range_set_init(&tmp, 0);
1036         range_set_map_across_diff(&tmp, &rg->ranges, &diff, diff_out);
1037         range_set_release(&rg->ranges);
1038         range_set_move(&rg->ranges, &tmp);
1039
1040         diff_ranges_release(&diff);
1041
1042         return ((*diff_out)->parent.nr > 0);
1043 }
1044
1045 static struct diff_filepair *diff_filepair_dup(struct diff_filepair *pair)
1046 {
1047         struct diff_filepair *new = xmalloc(sizeof(struct diff_filepair));
1048         new->one = pair->one;
1049         new->two = pair->two;
1050         new->one->count++;
1051         new->two->count++;
1052         return new;
1053 }
1054
1055 static void free_diffqueues(int n, struct diff_queue_struct *dq)
1056 {
1057         int i, j;
1058         for (i = 0; i < n; i++)
1059                 for (j = 0; j < dq[i].nr; j++)
1060                         diff_free_filepair(dq[i].queue[j]);
1061         free(dq);
1062 }
1063
1064 static int process_all_files(struct line_log_data **range_out,
1065                              struct rev_info *rev,
1066                              struct diff_queue_struct *queue,
1067                              struct line_log_data *range)
1068 {
1069         int i, changed = 0;
1070
1071         *range_out = line_log_data_copy(range);
1072
1073         for (i = 0; i < queue->nr; i++) {
1074                 struct diff_ranges *pairdiff = NULL;
1075                 struct diff_filepair *pair = queue->queue[i];
1076                 if (process_diff_filepair(rev, pair, *range_out, &pairdiff)) {
1077                         /*
1078                          * Store away the diff for later output.  We
1079                          * tuck it in the ranges we got as _input_,
1080                          * since that's the commit that caused the
1081                          * diff.
1082                          *
1083                          * NEEDSWORK not enough when we get around to
1084                          * doing something interesting with merges;
1085                          * currently each invocation on a merge parent
1086                          * trashes the previous one's diff.
1087                          *
1088                          * NEEDSWORK tramples over data structures not owned here
1089                          */
1090                         struct line_log_data *rg = range;
1091                         changed++;
1092                         while (rg && strcmp(rg->path, pair->two->path))
1093                                 rg = rg->next;
1094                         assert(rg);
1095                         rg->pair = diff_filepair_dup(queue->queue[i]);
1096                         memcpy(&rg->diff, pairdiff, sizeof(struct diff_ranges));
1097                 }
1098                 free(pairdiff);
1099         }
1100
1101         return changed;
1102 }
1103
1104 int line_log_print(struct rev_info *rev, struct commit *commit)
1105 {
1106         struct line_log_data *range = lookup_line_range(rev, commit);
1107
1108         show_log(rev);
1109         dump_diff_hacky(rev, range);
1110         return 1;
1111 }
1112
1113 static int process_ranges_ordinary_commit(struct rev_info *rev, struct commit *commit,
1114                                           struct line_log_data *range)
1115 {
1116         struct commit *parent = NULL;
1117         struct diff_queue_struct queue;
1118         struct line_log_data *parent_range;
1119         int changed;
1120
1121         if (commit->parents)
1122                 parent = commit->parents->item;
1123
1124         queue_diffs(range, &rev->diffopt, &queue, commit, parent);
1125         changed = process_all_files(&parent_range, rev, &queue, range);
1126         if (parent)
1127                 add_line_range(rev, parent, parent_range);
1128         return changed;
1129 }
1130
1131 static int process_ranges_merge_commit(struct rev_info *rev, struct commit *commit,
1132                                        struct line_log_data *range)
1133 {
1134         struct diff_queue_struct *diffqueues;
1135         struct line_log_data **cand;
1136         struct commit **parents;
1137         struct commit_list *p;
1138         int i;
1139         int nparents = commit_list_count(commit->parents);
1140
1141         if (nparents > 1 && rev->first_parent_only)
1142                 nparents = 1;
1143
1144         ALLOC_ARRAY(diffqueues, nparents);
1145         ALLOC_ARRAY(cand, nparents);
1146         ALLOC_ARRAY(parents, nparents);
1147
1148         p = commit->parents;
1149         for (i = 0; i < nparents; i++) {
1150                 parents[i] = p->item;
1151                 p = p->next;
1152                 queue_diffs(range, &rev->diffopt, &diffqueues[i], commit, parents[i]);
1153         }
1154
1155         for (i = 0; i < nparents; i++) {
1156                 int changed;
1157                 cand[i] = NULL;
1158                 changed = process_all_files(&cand[i], rev, &diffqueues[i], range);
1159                 if (!changed) {
1160                         /*
1161                          * This parent can take all the blame, so we
1162                          * don't follow any other path in history
1163                          */
1164                         add_line_range(rev, parents[i], cand[i]);
1165                         clear_commit_line_range(rev, commit);
1166                         commit_list_append(parents[i], &commit->parents);
1167                         free(parents);
1168                         free(cand);
1169                         free_diffqueues(nparents, diffqueues);
1170                         /* NEEDSWORK leaking like a sieve */
1171                         return 0;
1172                 }
1173         }
1174
1175         /*
1176          * No single parent took the blame.  We add the candidates
1177          * from the above loop to the parents.
1178          */
1179         for (i = 0; i < nparents; i++) {
1180                 add_line_range(rev, parents[i], cand[i]);
1181         }
1182
1183         clear_commit_line_range(rev, commit);
1184         free(parents);
1185         free(cand);
1186         free_diffqueues(nparents, diffqueues);
1187         return 1;
1188
1189         /* NEEDSWORK evil merge detection stuff */
1190         /* NEEDSWORK leaking like a sieve */
1191 }
1192
1193 static int process_ranges_arbitrary_commit(struct rev_info *rev, struct commit *commit)
1194 {
1195         struct line_log_data *range = lookup_line_range(rev, commit);
1196         int changed = 0;
1197
1198         if (range) {
1199                 if (!commit->parents || !commit->parents->next)
1200                         changed = process_ranges_ordinary_commit(rev, commit, range);
1201                 else
1202                         changed = process_ranges_merge_commit(rev, commit, range);
1203         }
1204
1205         if (!changed)
1206                 commit->object.flags |= TREESAME;
1207
1208         return changed;
1209 }
1210
1211 static enum rewrite_result line_log_rewrite_one(struct rev_info *rev, struct commit **pp)
1212 {
1213         for (;;) {
1214                 struct commit *p = *pp;
1215                 if (p->parents && p->parents->next)
1216                         return rewrite_one_ok;
1217                 if (p->object.flags & UNINTERESTING)
1218                         return rewrite_one_ok;
1219                 if (!(p->object.flags & TREESAME))
1220                         return rewrite_one_ok;
1221                 if (!p->parents)
1222                         return rewrite_one_noparents;
1223                 *pp = p->parents->item;
1224         }
1225 }
1226
1227 int line_log_filter(struct rev_info *rev)
1228 {
1229         struct commit *commit;
1230         struct commit_list *list = rev->commits;
1231         struct commit_list *out = NULL, **pp = &out;
1232
1233         while (list) {
1234                 struct commit_list *to_free = NULL;
1235                 commit = list->item;
1236                 if (process_ranges_arbitrary_commit(rev, commit)) {
1237                         *pp = list;
1238                         pp = &list->next;
1239                 } else
1240                         to_free = list;
1241                 list = list->next;
1242                 free(to_free);
1243         }
1244         *pp = NULL;
1245
1246         for (list = out; list; list = list->next)
1247                 rewrite_parents(rev, list->item, line_log_rewrite_one);
1248
1249         rev->commits = out;
1250
1251         return 0;
1252 }