builtin-blame.c

   1 /*
   2  * Pickaxe
   3  *
   4  * Copyright (c) 2006, Junio C Hamano
   5  */
   6
   7 #include "cache.h"
   8 #include "builtin.h"
   9 #include "blob.h"
  10 #include "commit.h"
  11 #include "tag.h"
  12 #include "tree-walk.h"
  13 #include "diff.h"
  14 #include "diffcore.h"
  15 #include "revision.h"
  16 #include "xdiff-interface.h"
  17
  18 #include <time.h>
  19 #include <sys/time.h>
  20 #include <regex.h>
  21
  22 static char blame_usage[] =
  23 "git-blame [-c] [-l] [-t] [-f] [-n] [-p] [-L n,m] [-S <revs-file>] [-M] [-C] [-C] [commit] [--] file\n"
  24 "  -c, --compatibility Use the same output mode as git-annotate (Default: off)\n"
  25 "  -b                  Show blank SHA-1 for boundary commits (Default: off)\n"
  26 "  -l, --long          Show long commit SHA1 (Default: off)\n"
  27 "  --root              Do not treat root commits as boundaries (Default: off)\n"
  28 "  -t, --time          Show raw timestamp (Default: off)\n"
  29 "  -f, --show-name     Show original filename (Default: auto)\n"
  30 "  -n, --show-number   Show original linenumber (Default: off)\n"
  31 "  -p, --porcelain     Show in a format designed for machine consumption\n"
  32 "  -L n,m              Process only line range n,m, counting from 1\n"
  33 "  -M, -C              Find line movements within and across files\n"
  34 "  -S revs-file        Use revisions from revs-file instead of calling git-rev-list\n";
  35
  36 static int longest_file;
  37 static int longest_author;
  38 static int max_orig_digits;
  39 static int max_digits;
  40 static int max_score_digits;
  41 static int show_root;
  42 static int blank_boundary;
  43
  44 #ifndef DEBUG
  45 #define DEBUG 0
  46 #endif
  47
  48 /* stats */
  49 static int num_read_blob;
  50 static int num_get_patch;
  51 static int num_commits;
  52
  53 #define PICKAXE_BLAME_MOVE              01
  54 #define PICKAXE_BLAME_COPY              02
  55 #define PICKAXE_BLAME_COPY_HARDER       04
  56
  57 /*
  58  * blame for a blame_entry with score lower than these thresholds
  59  * is not passed to the parent using move/copy logic.
  60  */
  61 static unsigned blame_move_score;
  62 static unsigned blame_copy_score;
  63 #define BLAME_DEFAULT_MOVE_SCORE        20
  64 #define BLAME_DEFAULT_COPY_SCORE        40
  65
  66 /* bits #0..7 in revision.h, #8..11 used for merge_bases() in commit.c */
  67 #define METAINFO_SHOWN          (1u<<12)
  68 #define MORE_THAN_ONE_PATH      (1u<<13)
  69
  70 /*
  71  * One blob in a commit that is being suspected
  72  */
  73 struct origin {
  74         int refcnt;
  75         struct commit *commit;
  76         mmfile_t file;
  77         unsigned char blob_sha1[20];
  78         char path[FLEX_ARRAY];
  79 };
  80
  81 static char *fill_origin_blob(struct origin *o, mmfile_t *file)
  82 {
  83         if (!o->file.ptr) {
  84                 char type[10];
  85                 num_read_blob++;
  86                 file->ptr = read_sha1_file(o->blob_sha1, type,
  87                                            (unsigned long *)(&(file->size)));
  88                 o->file = *file;
  89         }
  90         else
  91                 *file = o->file;
  92         return file->ptr;
  93 }
  94
  95 static inline struct origin *origin_incref(struct origin *o)
  96 {
  97         if (o)
  98                 o->refcnt++;
  99         return o;
 100 }
 101
 102 static void origin_decref(struct origin *o)
 103 {
 104         if (o && --o->refcnt <= 0) {
 105                 if (o->file.ptr)
 106                         free(o->file.ptr);
 107                 memset(o, 0, sizeof(*o));
 108                 free(o);
 109         }
 110 }
 111
 112 struct blame_entry {
 113         struct blame_entry *prev;
 114         struct blame_entry *next;
 115
 116         /* the first line of this group in the final image;
 117          * internally all line numbers are 0 based.
 118          */
 119         int lno;
 120
 121         /* how many lines this group has */
 122         int num_lines;
 123
 124         /* the commit that introduced this group into the final image */
 125         struct origin *suspect;
 126
 127         /* true if the suspect is truly guilty; false while we have not
 128          * checked if the group came from one of its parents.
 129          */
 130         char guilty;
 131
 132         /* the line number of the first line of this group in the
 133          * suspect's file; internally all line numbers are 0 based.
 134          */
 135         int s_lno;
 136
 137         /* how significant this entry is -- cached to avoid
 138          * scanning the lines over and over
 139          */
 140         unsigned score;
 141 };
 142
 143 struct scoreboard {
 144         /* the final commit (i.e. where we started digging from) */
 145         struct commit *final;
 146
 147         const char *path;
 148
 149         /* the contents in the final; pointed into by buf pointers of
 150          * blame_entries
 151          */
 152         const char *final_buf;
 153         unsigned long final_buf_size;
 154
 155         /* linked list of blames */
 156         struct blame_entry *ent;
 157
 158         /* look-up a line in the final buffer */
 159         int num_lines;
 160         int *lineno;
 161 };
 162
 163 static int cmp_suspect(struct origin *a, struct origin *b)
 164 {
 165         int cmp = hashcmp(a->commit->object.sha1, b->commit->object.sha1);
 166         if (cmp)
 167                 return cmp;
 168         return strcmp(a->path, b->path);
 169 }
 170
 171 #define cmp_suspect(a, b) ( ((a)==(b)) ? 0 : cmp_suspect(a,b) )
 172
 173 static void sanity_check_refcnt(struct scoreboard *);
 174
 175 static void coalesce(struct scoreboard *sb)
 176 {
 177         struct blame_entry *ent, *next;
 178
 179         for (ent = sb->ent; ent && (next = ent->next); ent = next) {
 180                 if (!cmp_suspect(ent->suspect, next->suspect) &&
 181                     ent->guilty == next->guilty &&
 182                     ent->s_lno + ent->num_lines == next->s_lno) {
 183                         ent->num_lines += next->num_lines;
 184                         ent->next = next->next;
 185                         if (ent->next)
 186                                 ent->next->prev = ent;
 187                         origin_decref(next->suspect);
 188                         free(next);
 189                         ent->score = 0;
 190                         next = ent; /* again */
 191                 }
 192         }
 193
 194         if (DEBUG) /* sanity */
 195                 sanity_check_refcnt(sb);
 196 }
 197
 198 static struct origin *make_origin(struct commit *commit, const char *path)
 199 {
 200         struct origin *o;
 201         o = xcalloc(1, sizeof(*o) + strlen(path) + 1);
 202         o->commit = commit;
 203         o->refcnt = 1;
 204         strcpy(o->path, path);
 205         return o;
 206 }
 207
 208 static struct origin *get_origin(struct scoreboard *sb,
 209                                  struct commit *commit,
 210                                  const char *path)
 211 {
 212         struct blame_entry *e;
 213
 214         for (e = sb->ent; e; e = e->next) {
 215                 if (e->suspect->commit == commit &&
 216                     !strcmp(e->suspect->path, path))
 217                         return origin_incref(e->suspect);
 218         }
 219         return make_origin(commit, path);
 220 }
 221
 222 static int fill_blob_sha1(struct origin *origin)
 223 {
 224         unsigned mode;
 225         char type[10];
 226
 227         if (!is_null_sha1(origin->blob_sha1))
 228                 return 0;
 229         if (get_tree_entry(origin->commit->object.sha1,
 230                            origin->path,
 231                            origin->blob_sha1, &mode))
 232                 goto error_out;
 233         if (sha1_object_info(origin->blob_sha1, type, NULL) ||
 234             strcmp(type, blob_type))
 235                 goto error_out;
 236         return 0;
 237  error_out:
 238         hashclr(origin->blob_sha1);
 239         return -1;
 240 }
 241
 242 static struct origin *find_origin(struct scoreboard *sb,
 243                                   struct commit *parent,
 244                                   struct origin *origin)
 245 {
 246         struct origin *porigin = NULL;
 247         struct diff_options diff_opts;
 248         const char *paths[2];
 249
 250         if (parent->util) {
 251                 /* This is a freestanding copy of origin and not
 252                  * refcounted.
 253                  */
 254                 struct origin *cached = parent->util;
 255                 if (!strcmp(cached->path, origin->path)) {
 256                         porigin = get_origin(sb, parent, cached->path);
 257                         if (porigin->refcnt == 1)
 258                                 hashcpy(porigin->blob_sha1, cached->blob_sha1);
 259                         return porigin;
 260                 }
 261                 /* otherwise it was not very useful; free it */
 262                 free(parent->util);
 263                 parent->util = NULL;
 264         }
 265
 266         /* See if the origin->path is different between parent
 267          * and origin first.  Most of the time they are the
 268          * same and diff-tree is fairly efficient about this.
 269          */
 270         diff_setup(&diff_opts);
 271         diff_opts.recursive = 1;
 272         diff_opts.detect_rename = 0;
 273         diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT;
 274         paths[0] = origin->path;
 275         paths[1] = NULL;
 276
 277         diff_tree_setup_paths(paths, &diff_opts);
 278         if (diff_setup_done(&diff_opts) < 0)
 279                 die("diff-setup");
 280         diff_tree_sha1(parent->tree->object.sha1,
 281                        origin->commit->tree->object.sha1,
 282                        "", &diff_opts);
 283         diffcore_std(&diff_opts);
 284
 285         /* It is either one entry that says "modified", or "created",
 286          * or nothing.
 287          */
 288         if (!diff_queued_diff.nr) {
 289                 /* The path is the same as parent */
 290                 porigin = get_origin(sb, parent, origin->path);
 291                 hashcpy(porigin->blob_sha1, origin->blob_sha1);
 292         }
 293         else if (diff_queued_diff.nr != 1)
 294                 die("internal error in blame::find_origin");
 295         else {
 296                 struct diff_filepair *p = diff_queued_diff.queue[0];
 297                 switch (p->status) {
 298                 default:
 299                         die("internal error in blame::find_origin (%c)",
 300                             p->status);
 301                 case 'M':
 302                         porigin = get_origin(sb, parent, origin->path);
 303                         hashcpy(porigin->blob_sha1, p->one->sha1);
 304                         break;
 305                 case 'A':
 306                 case 'T':
 307                         /* Did not exist in parent, or type changed */
 308                         break;
 309                 }
 310         }
 311         diff_flush(&diff_opts);
 312         if (porigin) {
 313                 struct origin *cached;
 314                 cached = make_origin(porigin->commit, porigin->path);
 315                 hashcpy(cached->blob_sha1, porigin->blob_sha1);
 316                 parent->util = cached;
 317         }
 318         return porigin;
 319 }
 320
 321 static struct origin *find_rename(struct scoreboard *sb,
 322                                   struct commit *parent,
 323                                   struct origin *origin)
 324 {
 325         struct origin *porigin = NULL;
 326         struct diff_options diff_opts;
 327         int i;
 328         const char *paths[2];
 329
 330         diff_setup(&diff_opts);
 331         diff_opts.recursive = 1;
 332         diff_opts.detect_rename = DIFF_DETECT_RENAME;
 333         diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT;
 334         diff_opts.single_follow = origin->path;
 335         paths[0] = NULL;
 336         diff_tree_setup_paths(paths, &diff_opts);
 337         if (diff_setup_done(&diff_opts) < 0)
 338                 die("diff-setup");
 339         diff_tree_sha1(parent->tree->object.sha1,
 340                        origin->commit->tree->object.sha1,
 341                        "", &diff_opts);
 342         diffcore_std(&diff_opts);
 343
 344         for (i = 0; i < diff_queued_diff.nr; i++) {
 345                 struct diff_filepair *p = diff_queued_diff.queue[i];
 346                 if ((p->status == 'R' || p->status == 'C') &&
 347                     !strcmp(p->two->path, origin->path)) {
 348                         porigin = get_origin(sb, parent, p->one->path);
 349                         hashcpy(porigin->blob_sha1, p->one->sha1);
 350                         break;
 351                 }
 352         }
 353         diff_flush(&diff_opts);
 354         return porigin;
 355 }
 356
 357 struct chunk {
 358         /* line number in postimage; up to but not including this
 359          * line is the same as preimage
 360          */
 361         int same;
 362
 363         /* preimage line number after this chunk */
 364         int p_next;
 365
 366         /* postimage line number after this chunk */
 367         int t_next;
 368 };
 369
 370 struct patch {
 371         struct chunk *chunks;
 372         int num;
 373 };
 374
 375 struct blame_diff_state {
 376         struct xdiff_emit_state xm;
 377         struct patch *ret;
 378         unsigned hunk_post_context;
 379         unsigned hunk_in_pre_context : 1;
 380 };
 381
 382 static void process_u_diff(void *state_, char *line, unsigned long len)
 383 {
 384         struct blame_diff_state *state = state_;
 385         struct chunk *chunk;
 386         int off1, off2, len1, len2, num;
 387
 388         num = state->ret->num;
 389         if (len < 4 || line[0] != '@' || line[1] != '@') {
 390                 if (state->hunk_in_pre_context && line[0] == ' ')
 391                         state->ret->chunks[num - 1].same++;
 392                 else {
 393                         state->hunk_in_pre_context = 0;
 394                         if (line[0] == ' ')
 395                                 state->hunk_post_context++;
 396                         else
 397                                 state->hunk_post_context = 0;
 398                 }
 399                 return;
 400         }
 401
 402         if (num && state->hunk_post_context) {
 403                 chunk = &state->ret->chunks[num - 1];
 404                 chunk->p_next -= state->hunk_post_context;
 405                 chunk->t_next -= state->hunk_post_context;
 406         }
 407         state->ret->num = ++num;
 408         state->ret->chunks = xrealloc(state->ret->chunks,
 409                                       sizeof(struct chunk) * num);
 410         chunk = &state->ret->chunks[num - 1];
 411         if (parse_hunk_header(line, len, &off1, &len1, &off2, &len2)) {
 412                 state->ret->num--;
 413                 return;
 414         }
 415
 416         /* Line numbers in patch output are one based. */
 417         off1--;
 418         off2--;
 419
 420         chunk->same = len2 ? off2 : (off2 + 1);
 421
 422         chunk->p_next = off1 + (len1 ? len1 : 1);
 423         chunk->t_next = chunk->same + len2;
 424         state->hunk_in_pre_context = 1;
 425         state->hunk_post_context = 0;
 426 }
 427
 428 static struct patch *compare_buffer(mmfile_t *file_p, mmfile_t *file_o,
 429                                     int context)
 430 {
 431         struct blame_diff_state state;
 432         xpparam_t xpp;
 433         xdemitconf_t xecfg;
 434         xdemitcb_t ecb;
 435
 436         xpp.flags = XDF_NEED_MINIMAL;
 437         xecfg.ctxlen = context;
 438         xecfg.flags = 0;
 439         ecb.outf = xdiff_outf;
 440         ecb.priv = &state;
 441         memset(&state, 0, sizeof(state));
 442         state.xm.consume = process_u_diff;
 443         state.ret = xmalloc(sizeof(struct patch));
 444         state.ret->chunks = NULL;
 445         state.ret->num = 0;
 446
 447         xdl_diff(file_p, file_o, &xpp, &xecfg, &ecb);
 448
 449         if (state.ret->num) {
 450                 struct chunk *chunk;
 451                 chunk = &state.ret->chunks[state.ret->num - 1];
 452                 chunk->p_next -= state.hunk_post_context;
 453                 chunk->t_next -= state.hunk_post_context;
 454         }
 455         return state.ret;
 456 }
 457
 458 static struct patch *get_patch(struct origin *parent, struct origin *origin)
 459 {
 460         mmfile_t file_p, file_o;
 461         struct patch *patch;
 462
 463         fill_origin_blob(parent, &file_p);
 464         fill_origin_blob(origin, &file_o);
 465         if (!file_p.ptr || !file_o.ptr)
 466                 return NULL;
 467         patch = compare_buffer(&file_p, &file_o, 0);
 468         num_get_patch++;
 469         return patch;
 470 }
 471
 472 static void free_patch(struct patch *p)
 473 {
 474         free(p->chunks);
 475         free(p);
 476 }
 477
 478 static void add_blame_entry(struct scoreboard *sb, struct blame_entry *e)
 479 {
 480         struct blame_entry *ent, *prev = NULL;
 481
 482         origin_incref(e->suspect);
 483
 484         for (ent = sb->ent; ent && ent->lno < e->lno; ent = ent->next)
 485                 prev = ent;
 486
 487         /* prev, if not NULL, is the last one that is below e */
 488         e->prev = prev;
 489         if (prev) {
 490                 e->next = prev->next;
 491                 prev->next = e;
 492         }
 493         else {
 494                 e->next = sb->ent;
 495                 sb->ent = e;
 496         }
 497         if (e->next)
 498                 e->next->prev = e;
 499 }
 500
 501 static void dup_entry(struct blame_entry *dst, struct blame_entry *src)
 502 {
 503         struct blame_entry *p, *n;
 504
 505         p = dst->prev;
 506         n = dst->next;
 507         origin_incref(src->suspect);
 508         origin_decref(dst->suspect);
 509         memcpy(dst, src, sizeof(*src));
 510         dst->prev = p;
 511         dst->next = n;
 512         dst->score = 0;
 513 }
 514
 515 static const char *nth_line(struct scoreboard *sb, int lno)
 516 {
 517         return sb->final_buf + sb->lineno[lno];
 518 }
 519
 520 static void split_overlap(struct blame_entry *split,
 521                           struct blame_entry *e,
 522                           int tlno, int plno, int same,
 523                           struct origin *parent)
 524 {
 525         /* it is known that lines between tlno to same came from
 526          * parent, and e has an overlap with that range.  it also is
 527          * known that parent's line plno corresponds to e's line tlno.
 528          *
 529          *                <---- e ----->
 530          *                   <------>
 531          *                   <------------>
 532          *             <------------>
 533          *             <------------------>
 534          *
 535          * Potentially we need to split e into three parts; before
 536          * this chunk, the chunk to be blamed for parent, and after
 537          * that portion.
 538          */
 539         int chunk_end_lno;
 540         memset(split, 0, sizeof(struct blame_entry [3]));
 541
 542         if (e->s_lno < tlno) {
 543                 /* there is a pre-chunk part not blamed on parent */
 544                 split[0].suspect = origin_incref(e->suspect);
 545                 split[0].lno = e->lno;
 546                 split[0].s_lno = e->s_lno;
 547                 split[0].num_lines = tlno - e->s_lno;
 548                 split[1].lno = e->lno + tlno - e->s_lno;
 549                 split[1].s_lno = plno;
 550         }
 551         else {
 552                 split[1].lno = e->lno;
 553                 split[1].s_lno = plno + (e->s_lno - tlno);
 554         }
 555
 556         if (same < e->s_lno + e->num_lines) {
 557                 /* there is a post-chunk part not blamed on parent */
 558                 split[2].suspect = origin_incref(e->suspect);
 559                 split[2].lno = e->lno + (same - e->s_lno);
 560                 split[2].s_lno = e->s_lno + (same - e->s_lno);
 561                 split[2].num_lines = e->s_lno + e->num_lines - same;
 562                 chunk_end_lno = split[2].lno;
 563         }
 564         else
 565                 chunk_end_lno = e->lno + e->num_lines;
 566         split[1].num_lines = chunk_end_lno - split[1].lno;
 567
 568         if (split[1].num_lines < 1)
 569                 return;
 570         split[1].suspect = origin_incref(parent);
 571 }
 572
 573 static void split_blame(struct scoreboard *sb,
 574                         struct blame_entry *split,
 575                         struct blame_entry *e)
 576 {
 577         struct blame_entry *new_entry;
 578
 579         if (split[0].suspect && split[2].suspect) {
 580                 /* we need to split e into two and add another for parent */
 581                 dup_entry(e, &split[0]);
 582
 583                 new_entry = xmalloc(sizeof(*new_entry));
 584                 memcpy(new_entry, &(split[2]), sizeof(struct blame_entry));
 585                 add_blame_entry(sb, new_entry);
 586
 587                 new_entry = xmalloc(sizeof(*new_entry));
 588                 memcpy(new_entry, &(split[1]), sizeof(struct blame_entry));
 589                 add_blame_entry(sb, new_entry);
 590         }
 591         else if (!split[0].suspect && !split[2].suspect)
 592                 /* parent covers the entire area */
 593                 dup_entry(e, &split[1]);
 594         else if (split[0].suspect) {
 595                 dup_entry(e, &split[0]);
 596
 597                 new_entry = xmalloc(sizeof(*new_entry));
 598                 memcpy(new_entry, &(split[1]), sizeof(struct blame_entry));
 599                 add_blame_entry(sb, new_entry);
 600         }
 601         else {
 602                 dup_entry(e, &split[1]);
 603
 604                 new_entry = xmalloc(sizeof(*new_entry));
 605                 memcpy(new_entry, &(split[2]), sizeof(struct blame_entry));
 606                 add_blame_entry(sb, new_entry);
 607         }
 608
 609         if (DEBUG) { /* sanity */
 610                 struct blame_entry *ent;
 611                 int lno = sb->ent->lno, corrupt = 0;
 612
 613                 for (ent = sb->ent; ent; ent = ent->next) {
 614                         if (lno != ent->lno)
 615                                 corrupt = 1;
 616                         if (ent->s_lno < 0)
 617                                 corrupt = 1;
 618                         lno += ent->num_lines;
 619                 }
 620                 if (corrupt) {
 621                         lno = sb->ent->lno;
 622                         for (ent = sb->ent; ent; ent = ent->next) {
 623                                 printf("L %8d l %8d n %8d\n",
 624                                        lno, ent->lno, ent->num_lines);
 625                                 lno = ent->lno + ent->num_lines;
 626                         }
 627                         die("oops");
 628                 }
 629         }
 630 }
 631
 632 static void decref_split(struct blame_entry *split)
 633 {
 634         int i;
 635
 636         for (i = 0; i < 3; i++)
 637                 origin_decref(split[i].suspect);
 638 }
 639
 640 static void blame_overlap(struct scoreboard *sb, struct blame_entry *e,
 641                           int tlno, int plno, int same,
 642                           struct origin *parent)
 643 {
 644         struct blame_entry split[3];
 645
 646         split_overlap(split, e, tlno, plno, same, parent);
 647         if (split[1].suspect)
 648                 split_blame(sb, split, e);
 649         decref_split(split);
 650 }
 651
 652 static int find_last_in_target(struct scoreboard *sb, struct origin *target)
 653 {
 654         struct blame_entry *e;
 655         int last_in_target = -1;
 656
 657         for (e = sb->ent; e; e = e->next) {
 658                 if (e->guilty || cmp_suspect(e->suspect, target))
 659                         continue;
 660                 if (last_in_target < e->s_lno + e->num_lines)
 661                         last_in_target = e->s_lno + e->num_lines;
 662         }
 663         return last_in_target;
 664 }
 665
 666 static void blame_chunk(struct scoreboard *sb,
 667                         int tlno, int plno, int same,
 668                         struct origin *target, struct origin *parent)
 669 {
 670         struct blame_entry *e;
 671
 672         for (e = sb->ent; e; e = e->next) {
 673                 if (e->guilty || cmp_suspect(e->suspect, target))
 674                         continue;
 675                 if (same <= e->s_lno)
 676                         continue;
 677                 if (tlno < e->s_lno + e->num_lines)
 678                         blame_overlap(sb, e, tlno, plno, same, parent);
 679         }
 680 }
 681
 682 static int pass_blame_to_parent(struct scoreboard *sb,
 683                                 struct origin *target,
 684                                 struct origin *parent)
 685 {
 686         int i, last_in_target, plno, tlno;
 687         struct patch *patch;
 688
 689         last_in_target = find_last_in_target(sb, target);
 690         if (last_in_target < 0)
 691                 return 1; /* nothing remains for this target */
 692
 693         patch = get_patch(parent, target);
 694         plno = tlno = 0;
 695         for (i = 0; i < patch->num; i++) {
 696                 struct chunk *chunk = &patch->chunks[i];
 697
 698                 blame_chunk(sb, tlno, plno, chunk->same, target, parent);
 699                 plno = chunk->p_next;
 700                 tlno = chunk->t_next;
 701         }
 702         /* rest (i.e. anything above tlno) are the same as parent */
 703         blame_chunk(sb, tlno, plno, last_in_target, target, parent);
 704
 705         free_patch(patch);
 706         return 0;
 707 }
 708
 709 static unsigned ent_score(struct scoreboard *sb, struct blame_entry *e)
 710 {
 711         unsigned score;
 712         const char *cp, *ep;
 713
 714         if (e->score)
 715                 return e->score;
 716
 717         score = 1;
 718         cp = nth_line(sb, e->lno);
 719         ep = nth_line(sb, e->lno + e->num_lines);
 720         while (cp < ep) {
 721                 unsigned ch = *((unsigned char *)cp);
 722                 if (isalnum(ch))
 723                         score++;
 724                 cp++;
 725         }
 726         e->score = score;
 727         return score;
 728 }
 729
 730 static void copy_split_if_better(struct scoreboard *sb,
 731                                  struct blame_entry *best_so_far,
 732                                  struct blame_entry *this)
 733 {
 734         int i;
 735
 736         if (!this[1].suspect)
 737                 return;
 738         if (best_so_far[1].suspect) {
 739                 if (ent_score(sb, &this[1]) < ent_score(sb, &best_so_far[1]))
 740                         return;
 741         }
 742
 743         for (i = 0; i < 3; i++)
 744                 origin_incref(this[i].suspect);
 745         decref_split(best_so_far);
 746         memcpy(best_so_far, this, sizeof(struct blame_entry [3]));
 747 }
 748
 749 static void find_copy_in_blob(struct scoreboard *sb,
 750                               struct blame_entry *ent,
 751                               struct origin *parent,
 752                               struct blame_entry *split,
 753                               mmfile_t *file_p)
 754 {
 755         const char *cp;
 756         int cnt;
 757         mmfile_t file_o;
 758         struct patch *patch;
 759         int i, plno, tlno;
 760
 761         cp = nth_line(sb, ent->lno);
 762         file_o.ptr = (char*) cp;
 763         cnt = ent->num_lines;
 764
 765         while (cnt && cp < sb->final_buf + sb->final_buf_size) {
 766                 if (*cp++ == '\n')
 767                         cnt--;
 768         }
 769         file_o.size = cp - file_o.ptr;
 770
 771         patch = compare_buffer(file_p, &file_o, 1);
 772
 773         memset(split, 0, sizeof(struct blame_entry [3]));
 774         plno = tlno = 0;
 775         for (i = 0; i < patch->num; i++) {
 776                 struct chunk *chunk = &patch->chunks[i];
 777
 778                 /* tlno to chunk->same are the same as ent */
 779                 if (ent->num_lines <= tlno)
 780                         break;
 781                 if (tlno < chunk->same) {
 782                         struct blame_entry this[3];
 783                         split_overlap(this, ent,
 784                                       tlno + ent->s_lno, plno,
 785                                       chunk->same + ent->s_lno,
 786                                       parent);
 787                         copy_split_if_better(sb, split, this);
 788                         decref_split(this);
 789                 }
 790                 plno = chunk->p_next;
 791                 tlno = chunk->t_next;
 792         }
 793         free_patch(patch);
 794 }
 795
 796 static int find_move_in_parent(struct scoreboard *sb,
 797                                struct origin *target,
 798                                struct origin *parent)
 799 {
 800         int last_in_target, made_progress;
 801         struct blame_entry *e, split[3];
 802         mmfile_t file_p;
 803
 804         last_in_target = find_last_in_target(sb, target);
 805         if (last_in_target < 0)
 806                 return 1; /* nothing remains for this target */
 807
 808         fill_origin_blob(parent, &file_p);
 809         if (!file_p.ptr)
 810                 return 0;
 811
 812         made_progress = 1;
 813         while (made_progress) {
 814                 made_progress = 0;
 815                 for (e = sb->ent; e; e = e->next) {
 816                         if (e->guilty || cmp_suspect(e->suspect, target))
 817                                 continue;
 818                         find_copy_in_blob(sb, e, parent, split, &file_p);
 819                         if (split[1].suspect &&
 820                             blame_move_score < ent_score(sb, &split[1])) {
 821                                 split_blame(sb, split, e);
 822                                 made_progress = 1;
 823                         }
 824                         decref_split(split);
 825                 }
 826         }
 827         return 0;
 828 }
 829
 830
 831 struct blame_list {
 832         struct blame_entry *ent;
 833         struct blame_entry split[3];
 834 };
 835
 836 static struct blame_list *setup_blame_list(struct scoreboard *sb,
 837                                            struct origin *target,
 838                                            int *num_ents_p)
 839 {
 840         struct blame_entry *e;
 841         int num_ents, i;
 842         struct blame_list *blame_list = NULL;
 843
 844         /* Count the number of entries the target is suspected for,
 845          * and prepare a list of entry and the best split.
 846          */
 847         for (e = sb->ent, num_ents = 0; e; e = e->next)
 848                 if (!e->guilty && !cmp_suspect(e->suspect, target))
 849                         num_ents++;
 850         if (num_ents) {
 851                 blame_list = xcalloc(num_ents, sizeof(struct blame_list));
 852                 for (e = sb->ent, i = 0; e; e = e->next)
 853                         if (!e->guilty && !cmp_suspect(e->suspect, target))
 854                                 blame_list[i++].ent = e;
 855         }
 856         *num_ents_p = num_ents;
 857         return blame_list;
 858 }
 859
 860 static int find_copy_in_parent(struct scoreboard *sb,
 861                                struct origin *target,
 862                                struct commit *parent,
 863                                struct origin *porigin,
 864                                int opt)
 865 {
 866         struct diff_options diff_opts;
 867         const char *paths[1];
 868         int i, j;
 869         int retval;
 870         struct blame_list *blame_list;
 871         int num_ents;
 872
 873         blame_list = setup_blame_list(sb, target, &num_ents);
 874         if (!blame_list)
 875                 return 1; /* nothing remains for this target */
 876
 877         diff_setup(&diff_opts);
 878         diff_opts.recursive = 1;
 879         diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT;
 880
 881         paths[0] = NULL;
 882         diff_tree_setup_paths(paths, &diff_opts);
 883         if (diff_setup_done(&diff_opts) < 0)
 884                 die("diff-setup");
 885
 886         /* Try "find copies harder" on new path if requested;
 887          * we do not want to use diffcore_rename() actually to
 888          * match things up; find_copies_harder is set only to
 889          * force diff_tree_sha1() to feed all filepairs to diff_queue,
 890          * and this code needs to be after diff_setup_done(), which
 891          * usually makes find-copies-harder imply copy detection.
 892          */
 893         if ((opt & PICKAXE_BLAME_COPY_HARDER) &&
 894             (!porigin || strcmp(target->path, porigin->path)))
 895                 diff_opts.find_copies_harder = 1;
 896
 897         diff_tree_sha1(parent->tree->object.sha1,
 898                        target->commit->tree->object.sha1,
 899                        "", &diff_opts);
 900
 901         if (!diff_opts.find_copies_harder)
 902                 diffcore_std(&diff_opts);
 903
 904         retval = 0;
 905         while (1) {
 906                 int made_progress = 0;
 907
 908                 for (i = 0; i < diff_queued_diff.nr; i++) {
 909                         struct diff_filepair *p = diff_queued_diff.queue[i];
 910                         struct origin *norigin;
 911                         mmfile_t file_p;
 912                         struct blame_entry this[3];
 913
 914                         if (!DIFF_FILE_VALID(p->one))
 915                                 continue; /* does not exist in parent */
 916                         if (porigin && !strcmp(p->one->path, porigin->path))
 917                                 /* find_move already dealt with this path */
 918                                 continue;
 919
 920                         norigin = get_origin(sb, parent, p->one->path);
 921                         hashcpy(norigin->blob_sha1, p->one->sha1);
 922                         fill_origin_blob(norigin, &file_p);
 923                         if (!file_p.ptr)
 924                                 continue;
 925
 926                         for (j = 0; j < num_ents; j++) {
 927                                 find_copy_in_blob(sb, blame_list[j].ent,
 928                                                   norigin, this, &file_p);
 929                                 copy_split_if_better(sb, blame_list[j].split,
 930                                                      this);
 931                                 decref_split(this);
 932                         }
 933                         origin_decref(norigin);
 934                 }
 935
 936                 for (j = 0; j < num_ents; j++) {
 937                         struct blame_entry *split = blame_list[j].split;
 938                         if (split[1].suspect &&
 939                             blame_copy_score < ent_score(sb, &split[1])) {
 940                                 split_blame(sb, split, blame_list[j].ent);
 941                                 made_progress = 1;
 942                         }
 943                         decref_split(split);
 944                 }
 945                 free(blame_list);
 946
 947                 if (!made_progress)
 948                         break;
 949                 blame_list = setup_blame_list(sb, target, &num_ents);
 950                 if (!blame_list) {
 951                         retval = 1;
 952                         break;
 953                 }
 954         }
 955         diff_flush(&diff_opts);
 956
 957         return retval;
 958 }
 959
 960 /* The blobs of origin and porigin exactly match, so everything
 961  * origin is suspected for can be blamed on the parent.
 962  */
 963 static void pass_whole_blame(struct scoreboard *sb,
 964                              struct origin *origin, struct origin *porigin)
 965 {
 966         struct blame_entry *e;
 967
 968         if (!porigin->file.ptr && origin->file.ptr) {
 969                 /* Steal its file */
 970                 porigin->file = origin->file;
 971                 origin->file.ptr = NULL;
 972         }
 973         for (e = sb->ent; e; e = e->next) {
 974                 if (cmp_suspect(e->suspect, origin))
 975                         continue;
 976                 origin_incref(porigin);
 977                 origin_decref(e->suspect);
 978                 e->suspect = porigin;
 979         }
 980 }
 981
 982 #define MAXPARENT 16
 983
 984 static void pass_blame(struct scoreboard *sb, struct origin *origin, int opt)
 985 {
 986         int i, pass;
 987         struct commit *commit = origin->commit;
 988         struct commit_list *parent;
 989         struct origin *parent_origin[MAXPARENT], *porigin;
 990
 991         memset(parent_origin, 0, sizeof(parent_origin));
 992
 993         /* The first pass looks for unrenamed path to optimize for
 994          * common cases, then we look for renames in the second pass.
 995          */
 996         for (pass = 0; pass < 2; pass++) {
 997                 struct origin *(*find)(struct scoreboard *,
 998                                        struct commit *, struct origin *);
 999                 find = pass ? find_rename : find_origin;
1000
1001                 for (i = 0, parent = commit->parents;
1002                      i < MAXPARENT && parent;
1003                      parent = parent->next, i++) {
1004                         struct commit *p = parent->item;
1005                         int j, same;
1006
1007                         if (parent_origin[i])
1008                                 continue;
1009                         if (parse_commit(p))
1010                                 continue;
1011                         porigin = find(sb, p, origin);
1012                         if (!porigin)
1013                                 continue;
1014                         if (!hashcmp(porigin->blob_sha1, origin->blob_sha1)) {
1015                                 pass_whole_blame(sb, origin, porigin);
1016                                 origin_decref(porigin);
1017                                 goto finish;
1018                         }
1019                         for (j = same = 0; j < i; j++)
1020                                 if (parent_origin[j] &&
1021                                     !hashcmp(parent_origin[j]->blob_sha1,
1022                                              porigin->blob_sha1)) {
1023                                         same = 1;
1024                                         break;
1025                                 }
1026                         if (!same)
1027                                 parent_origin[i] = porigin;
1028                         else
1029                                 origin_decref(porigin);
1030                 }
1031         }
1032
1033         num_commits++;
1034         for (i = 0, parent = commit->parents;
1035              i < MAXPARENT && parent;
1036              parent = parent->next, i++) {
1037                 struct origin *porigin = parent_origin[i];
1038                 if (!porigin)
1039                         continue;
1040                 if (pass_blame_to_parent(sb, origin, porigin))
1041                         goto finish;
1042         }
1043
1044         /*
1045          * Optionally run "miff" to find moves in parents' files here.
1046          */
1047         if (opt & PICKAXE_BLAME_MOVE)
1048                 for (i = 0, parent = commit->parents;
1049                      i < MAXPARENT && parent;
1050                      parent = parent->next, i++) {
1051                         struct origin *porigin = parent_origin[i];
1052                         if (!porigin)
1053                                 continue;
1054                         if (find_move_in_parent(sb, origin, porigin))
1055                                 goto finish;
1056                 }
1057
1058         /*
1059          * Optionally run "ciff" to find copies from parents' files here.
1060          */
1061         if (opt & PICKAXE_BLAME_COPY)
1062                 for (i = 0, parent = commit->parents;
1063                      i < MAXPARENT && parent;
1064                      parent = parent->next, i++) {
1065                         struct origin *porigin = parent_origin[i];
1066                         if (find_copy_in_parent(sb, origin, parent->item,
1067                                                 porigin, opt))
1068                                 goto finish;
1069                 }
1070
1071  finish:
1072         for (i = 0; i < MAXPARENT; i++)
1073                 origin_decref(parent_origin[i]);
1074 }
1075
1076 static void assign_blame(struct scoreboard *sb, struct rev_info *revs, int opt)
1077 {
1078         while (1) {
1079                 struct blame_entry *ent;
1080                 struct commit *commit;
1081                 struct origin *suspect = NULL;
1082
1083                 /* find one suspect to break down */
1084                 for (ent = sb->ent; !suspect && ent; ent = ent->next)
1085                         if (!ent->guilty)
1086                                 suspect = ent->suspect;
1087                 if (!suspect)
1088                         return; /* all done */
1089
1090                 origin_incref(suspect);
1091                 commit = suspect->commit;
1092                 if (!commit->object.parsed)
1093                         parse_commit(commit);
1094                 if (!(commit->object.flags & UNINTERESTING) &&
1095                     !(revs->max_age != -1 && commit->date  < revs->max_age))
1096                         pass_blame(sb, suspect, opt);
1097                 else {
1098                         commit->object.flags |= UNINTERESTING;
1099                         if (commit->object.parsed)
1100                                 mark_parents_uninteresting(commit);
1101                 }
1102                 /* treat root commit as boundary */
1103                 if (!commit->parents && !show_root)
1104                         commit->object.flags |= UNINTERESTING;
1105
1106                 /* Take responsibility for the remaining entries */
1107                 for (ent = sb->ent; ent; ent = ent->next)
1108                         if (!cmp_suspect(ent->suspect, suspect))
1109                                 ent->guilty = 1;
1110                 origin_decref(suspect);
1111
1112                 if (DEBUG) /* sanity */
1113                         sanity_check_refcnt(sb);
1114         }
1115 }
1116
1117 static const char *format_time(unsigned long time, const char *tz_str,
1118                                int show_raw_time)
1119 {
1120         static char time_buf[128];
1121         time_t t = time;
1122         int minutes, tz;
1123         struct tm *tm;
1124
1125         if (show_raw_time) {
1126                 sprintf(time_buf, "%lu %s", time, tz_str);
1127                 return time_buf;
1128         }
1129
1130         tz = atoi(tz_str);
1131         minutes = tz < 0 ? -tz : tz;
1132         minutes = (minutes / 100)*60 + (minutes % 100);
1133         minutes = tz < 0 ? -minutes : minutes;
1134         t = time + minutes * 60;
1135         tm = gmtime(&t);
1136
1137         strftime(time_buf, sizeof(time_buf), "%Y-%m-%d %H:%M:%S ", tm);
1138         strcat(time_buf, tz_str);
1139         return time_buf;
1140 }
1141
1142 struct commit_info
1143 {
1144         char *author;
1145         char *author_mail;
1146         unsigned long author_time;
1147         char *author_tz;
1148
1149         /* filled only when asked for details */
1150         char *committer;
1151         char *committer_mail;
1152         unsigned long committer_time;
1153         char *committer_tz;
1154
1155         char *summary;
1156 };
1157
1158 static void get_ac_line(const char *inbuf, const char *what,
1159                         int bufsz, char *person, char **mail,
1160                         unsigned long *time, char **tz)
1161 {
1162         int len;
1163         char *tmp, *endp;
1164
1165         tmp = strstr(inbuf, what);
1166         if (!tmp)
1167                 goto error_out;
1168         tmp += strlen(what);
1169         endp = strchr(tmp, '\n');
1170         if (!endp)
1171                 len = strlen(tmp);
1172         else
1173                 len = endp - tmp;
1174         if (bufsz <= len) {
1175         error_out:
1176                 /* Ugh */
1177                 person = *mail = *tz = "(unknown)";
1178                 *time = 0;
1179                 return;
1180         }
1181         memcpy(person, tmp, len);
1182
1183         tmp = person;
1184         tmp += len;
1185         *tmp = 0;
1186         while (*tmp != ' ')
1187                 tmp--;
1188         *tz = tmp+1;
1189
1190         *tmp = 0;
1191         while (*tmp != ' ')
1192                 tmp--;
1193         *time = strtoul(tmp, NULL, 10);
1194
1195         *tmp = 0;
1196         while (*tmp != ' ')
1197                 tmp--;
1198         *mail = tmp + 1;
1199         *tmp = 0;
1200 }
1201
1202 static void get_commit_info(struct commit *commit,
1203                             struct commit_info *ret,
1204                             int detailed)
1205 {
1206         int len;
1207         char *tmp, *endp;
1208         static char author_buf[1024];
1209         static char committer_buf[1024];
1210         static char summary_buf[1024];
1211
1212         /* We've operated without save_commit_buffer, so
1213          * we now need to populate them for output.
1214          */
1215         if (!commit->buffer) {
1216                 char type[20];
1217                 unsigned long size;
1218                 commit->buffer =
1219                         read_sha1_file(commit->object.sha1, type, &size);
1220         }
1221         ret->author = author_buf;
1222         get_ac_line(commit->buffer, "\nauthor ",
1223                     sizeof(author_buf), author_buf, &ret->author_mail,
1224                     &ret->author_time, &ret->author_tz);
1225
1226         if (!detailed)
1227                 return;
1228
1229         ret->committer = committer_buf;
1230         get_ac_line(commit->buffer, "\ncommitter ",
1231                     sizeof(committer_buf), committer_buf, &ret->committer_mail,
1232                     &ret->committer_time, &ret->committer_tz);
1233
1234         ret->summary = summary_buf;
1235         tmp = strstr(commit->buffer, "\n\n");
1236         if (!tmp) {
1237         error_out:
1238                 sprintf(summary_buf, "(%s)", sha1_to_hex(commit->object.sha1));
1239                 return;
1240         }
1241         tmp += 2;
1242         endp = strchr(tmp, '\n');
1243         if (!endp)
1244                 goto error_out;
1245         len = endp - tmp;
1246         if (len >= sizeof(summary_buf))
1247                 goto error_out;
1248         memcpy(summary_buf, tmp, len);
1249         summary_buf[len] = 0;
1250 }
1251
1252 #define OUTPUT_ANNOTATE_COMPAT  001
1253 #define OUTPUT_LONG_OBJECT_NAME 002
1254 #define OUTPUT_RAW_TIMESTAMP    004
1255 #define OUTPUT_PORCELAIN        010
1256 #define OUTPUT_SHOW_NAME        020
1257 #define OUTPUT_SHOW_NUMBER      040
1258 #define OUTPUT_SHOW_SCORE      0100
1259
1260 static void emit_porcelain(struct scoreboard *sb, struct blame_entry *ent)
1261 {
1262         int cnt;
1263         const char *cp;
1264         struct origin *suspect = ent->suspect;
1265         char hex[41];
1266
1267         strcpy(hex, sha1_to_hex(suspect->commit->object.sha1));
1268         printf("%s%c%d %d %d\n",
1269                hex,
1270                ent->guilty ? ' ' : '*', // purely for debugging
1271                ent->s_lno + 1,
1272                ent->lno + 1,
1273                ent->num_lines);
1274         if (!(suspect->commit->object.flags & METAINFO_SHOWN)) {
1275                 struct commit_info ci;
1276                 suspect->commit->object.flags |= METAINFO_SHOWN;
1277                 get_commit_info(suspect->commit, &ci, 1);
1278                 printf("author %s\n", ci.author);
1279                 printf("author-mail %s\n", ci.author_mail);
1280                 printf("author-time %lu\n", ci.author_time);
1281                 printf("author-tz %s\n", ci.author_tz);
1282                 printf("committer %s\n", ci.committer);
1283                 printf("committer-mail %s\n", ci.committer_mail);
1284                 printf("committer-time %lu\n", ci.committer_time);
1285                 printf("committer-tz %s\n", ci.committer_tz);
1286                 printf("filename %s\n", suspect->path);
1287                 printf("summary %s\n", ci.summary);
1288                 if (suspect->commit->object.flags & UNINTERESTING)
1289                         printf("boundary\n");
1290         }
1291         else if (suspect->commit->object.flags & MORE_THAN_ONE_PATH)
1292                 printf("filename %s\n", suspect->path);
1293
1294         cp = nth_line(sb, ent->lno);
1295         for (cnt = 0; cnt < ent->num_lines; cnt++) {
1296                 char ch;
1297                 if (cnt)
1298                         printf("%s %d %d\n", hex,
1299                                ent->s_lno + 1 + cnt,
1300                                ent->lno + 1 + cnt);
1301                 putchar('\t');
1302                 do {
1303                         ch = *cp++;
1304                         putchar(ch);
1305                 } while (ch != '\n' &&
1306                          cp < sb->final_buf + sb->final_buf_size);
1307         }
1308 }
1309
1310 static void emit_other(struct scoreboard *sb, struct blame_entry *ent, int opt)
1311 {
1312         int cnt;
1313         const char *cp;
1314         struct origin *suspect = ent->suspect;
1315         struct commit_info ci;
1316         char hex[41];
1317         int show_raw_time = !!(opt & OUTPUT_RAW_TIMESTAMP);
1318
1319         get_commit_info(suspect->commit, &ci, 1);
1320         strcpy(hex, sha1_to_hex(suspect->commit->object.sha1));
1321
1322         cp = nth_line(sb, ent->lno);
1323         for (cnt = 0; cnt < ent->num_lines; cnt++) {
1324                 char ch;
1325                 int length = (opt & OUTPUT_LONG_OBJECT_NAME) ? 40 : 8;
1326
1327                 if (suspect->commit->object.flags & UNINTERESTING) {
1328                         if (!blank_boundary) {
1329                                 length--;
1330                                 putchar('^');
1331                         }
1332                         else
1333                                 memset(hex, ' ', length);
1334                 }
1335
1336                 printf("%.*s", length, hex);
1337                 if (opt & OUTPUT_ANNOTATE_COMPAT)
1338                         printf("\t(%10s\t%10s\t%d)", ci.author,
1339                                format_time(ci.author_time, ci.author_tz,
1340                                            show_raw_time),
1341                                ent->lno + 1 + cnt);
1342                 else {
1343                         if (opt & OUTPUT_SHOW_SCORE)
1344                                 printf(" %*d %02d",
1345                                        max_score_digits, ent->score,
1346                                        ent->suspect->refcnt);
1347                         if (opt & OUTPUT_SHOW_NAME)
1348                                 printf(" %-*.*s", longest_file, longest_file,
1349                                        suspect->path);
1350                         if (opt & OUTPUT_SHOW_NUMBER)
1351                                 printf(" %*d", max_orig_digits,
1352                                        ent->s_lno + 1 + cnt);
1353                         printf(" (%-*.*s %10s %*d) ",
1354                                longest_author, longest_author, ci.author,
1355                                format_time(ci.author_time, ci.author_tz,
1356                                            show_raw_time),
1357                                max_digits, ent->lno + 1 + cnt);
1358                 }
1359                 do {
1360                         ch = *cp++;
1361                         putchar(ch);
1362                 } while (ch != '\n' &&
1363                          cp < sb->final_buf + sb->final_buf_size);
1364         }
1365 }
1366
1367 static void output(struct scoreboard *sb, int option)
1368 {
1369         struct blame_entry *ent;
1370
1371         if (option & OUTPUT_PORCELAIN) {
1372                 for (ent = sb->ent; ent; ent = ent->next) {
1373                         struct blame_entry *oth;
1374                         struct origin *suspect = ent->suspect;
1375                         struct commit *commit = suspect->commit;
1376                         if (commit->object.flags & MORE_THAN_ONE_PATH)
1377                                 continue;
1378                         for (oth = ent->next; oth; oth = oth->next) {
1379                                 if ((oth->suspect->commit != commit) ||
1380                                     !strcmp(oth->suspect->path, suspect->path))
1381                                         continue;
1382                                 commit->object.flags |= MORE_THAN_ONE_PATH;
1383                                 break;
1384                         }
1385                 }
1386         }
1387
1388         for (ent = sb->ent; ent; ent = ent->next) {
1389                 if (option & OUTPUT_PORCELAIN)
1390                         emit_porcelain(sb, ent);
1391                 else {
1392                         emit_other(sb, ent, option);
1393                 }
1394         }
1395 }
1396
1397 static int prepare_lines(struct scoreboard *sb)
1398 {
1399         const char *buf = sb->final_buf;
1400         unsigned long len = sb->final_buf_size;
1401         int num = 0, incomplete = 0, bol = 1;
1402
1403         if (len && buf[len-1] != '\n')
1404                 incomplete++; /* incomplete line at the end */
1405         while (len--) {
1406                 if (bol) {
1407                         sb->lineno = xrealloc(sb->lineno,
1408                                               sizeof(int* ) * (num + 1));
1409                         sb->lineno[num] = buf - sb->final_buf;
1410                         bol = 0;
1411                 }
1412                 if (*buf++ == '\n') {
1413                         num++;
1414                         bol = 1;
1415                 }
1416         }
1417         sb->lineno = xrealloc(sb->lineno,
1418                               sizeof(int* ) * (num + incomplete + 1));
1419         sb->lineno[num + incomplete] = buf - sb->final_buf;
1420         sb->num_lines = num + incomplete;
1421         return sb->num_lines;
1422 }
1423
1424 static int read_ancestry(const char *graft_file)
1425 {
1426         FILE *fp = fopen(graft_file, "r");
1427         char buf[1024];
1428         if (!fp)
1429                 return -1;
1430         while (fgets(buf, sizeof(buf), fp)) {
1431                 /* The format is just "Commit Parent1 Parent2 ...\n" */
1432                 int len = strlen(buf);
1433                 struct commit_graft *graft = read_graft_line(buf, len);
1434                 if (graft)
1435                         register_commit_graft(graft, 0);
1436         }
1437         fclose(fp);
1438         return 0;
1439 }
1440
1441 static int lineno_width(int lines)
1442 {
1443         int i, width;
1444
1445         for (width = 1, i = 10; i <= lines + 1; width++)
1446                 i *= 10;
1447         return width;
1448 }
1449
1450 static void find_alignment(struct scoreboard *sb, int *option)
1451 {
1452         int longest_src_lines = 0;
1453         int longest_dst_lines = 0;
1454         unsigned largest_score = 0;
1455         struct blame_entry *e;
1456
1457         for (e = sb->ent; e; e = e->next) {
1458                 struct origin *suspect = e->suspect;
1459                 struct commit_info ci;
1460                 int num;
1461
1462                 if (strcmp(suspect->path, sb->path))
1463                         *option |= OUTPUT_SHOW_NAME;
1464                 num = strlen(suspect->path);
1465                 if (longest_file < num)
1466                         longest_file = num;
1467                 if (!(suspect->commit->object.flags & METAINFO_SHOWN)) {
1468                         suspect->commit->object.flags |= METAINFO_SHOWN;
1469                         get_commit_info(suspect->commit, &ci, 1);
1470                         num = strlen(ci.author);
1471                         if (longest_author < num)
1472                                 longest_author = num;
1473                 }
1474                 num = e->s_lno + e->num_lines;
1475                 if (longest_src_lines < num)
1476                         longest_src_lines = num;
1477                 num = e->lno + e->num_lines;
1478                 if (longest_dst_lines < num)
1479                         longest_dst_lines = num;
1480                 if (largest_score < ent_score(sb, e))
1481                         largest_score = ent_score(sb, e);
1482         }
1483         max_orig_digits = lineno_width(longest_src_lines);
1484         max_digits = lineno_width(longest_dst_lines);
1485         max_score_digits = lineno_width(largest_score);
1486 }
1487
1488 static void sanity_check_refcnt(struct scoreboard *sb)
1489 {
1490         int baa = 0;
1491         struct blame_entry *ent;
1492
1493         for (ent = sb->ent; ent; ent = ent->next) {
1494                 /* Nobody should have zero or negative refcnt */
1495                 if (ent->suspect->refcnt <= 0) {
1496                         fprintf(stderr, "%s in %s has negative refcnt %d\n",
1497                                 ent->suspect->path,
1498                                 sha1_to_hex(ent->suspect->commit->object.sha1),
1499                                 ent->suspect->refcnt);
1500                         baa = 1;
1501                 }
1502         }
1503         for (ent = sb->ent; ent; ent = ent->next) {
1504                 /* Mark the ones that haven't been checked */
1505                 if (0 < ent->suspect->refcnt)
1506                         ent->suspect->refcnt = -ent->suspect->refcnt;
1507         }
1508         for (ent = sb->ent; ent; ent = ent->next) {
1509                 /* then pick each and see if they have the the correct
1510                  * refcnt.
1511                  */
1512                 int found;
1513                 struct blame_entry *e;
1514                 struct origin *suspect = ent->suspect;
1515
1516                 if (0 < suspect->refcnt)
1517                         continue;
1518                 suspect->refcnt = -suspect->refcnt; /* Unmark */
1519                 for (found = 0, e = sb->ent; e; e = e->next) {
1520                         if (e->suspect != suspect)
1521                                 continue;
1522                         found++;
1523                 }
1524                 if (suspect->refcnt != found) {
1525                         fprintf(stderr, "%s in %s has refcnt %d, not %d\n",
1526                                 ent->suspect->path,
1527                                 sha1_to_hex(ent->suspect->commit->object.sha1),
1528                                 ent->suspect->refcnt, found);
1529                         baa = 2;
1530                 }
1531         }
1532         if (baa) {
1533                 int opt = 0160;
1534                 find_alignment(sb, &opt);
1535                 output(sb, opt);
1536                 die("Baa %d!", baa);
1537         }
1538 }
1539
1540 static int has_path_in_work_tree(const char *path)
1541 {
1542         struct stat st;
1543         return !lstat(path, &st);
1544 }
1545
1546 static unsigned parse_score(const char *arg)
1547 {
1548         char *end;
1549         unsigned long score = strtoul(arg, &end, 10);
1550         if (*end)
1551                 return 0;
1552         return score;
1553 }
1554
1555 static const char *add_prefix(const char *prefix, const char *path)
1556 {
1557         if (!prefix || !prefix[0])
1558                 return path;
1559         return prefix_path(prefix, strlen(prefix), path);
1560 }
1561
1562 static const char *parse_loc(const char *spec,
1563                              struct scoreboard *sb, long lno,
1564                              long begin, long *ret)
1565 {
1566         char *term;
1567         const char *line;
1568         long num;
1569         int reg_error;
1570         regex_t regexp;
1571         regmatch_t match[1];
1572
1573         /* Allow "-L <something>,+20" to mean starting at <something>
1574          * for 20 lines, or "-L <something>,-5" for 5 lines ending at
1575          * <something>.
1576          */
1577         if (1 < begin && (spec[0] == '+' || spec[0] == '-')) {
1578                 num = strtol(spec + 1, &term, 10);
1579                 if (term != spec + 1) {
1580                         if (spec[0] == '-')
1581                                 num = 0 - num;
1582                         if (0 < num)
1583                                 *ret = begin + num - 2;
1584                         else if (!num)
1585                                 *ret = begin;
1586                         else
1587                                 *ret = begin + num;
1588                         return term;
1589                 }
1590                 return spec;
1591         }
1592         num = strtol(spec, &term, 10);
1593         if (term != spec) {
1594                 *ret = num;
1595                 return term;
1596         }
1597         if (spec[0] != '/')
1598                 return spec;
1599
1600         /* it could be a regexp of form /.../ */
1601         for (term = (char*) spec + 1; *term && *term != '/'; term++) {
1602                 if (*term == '\\')
1603                         term++;
1604         }
1605         if (*term != '/')
1606                 return spec;
1607
1608         /* try [spec+1 .. term-1] as regexp */
1609         *term = 0;
1610         begin--; /* input is in human terms */
1611         line = nth_line(sb, begin);
1612
1613         if (!(reg_error = regcomp(&regexp, spec + 1, REG_NEWLINE)) &&
1614             !(reg_error = regexec(&regexp, line, 1, match, 0))) {
1615                 const char *cp = line + match[0].rm_so;
1616                 const char *nline;
1617
1618                 while (begin++ < lno) {
1619                         nline = nth_line(sb, begin);
1620                         if (line <= cp && cp < nline)
1621                                 break;
1622                         line = nline;
1623                 }
1624                 *ret = begin;
1625                 regfree(&regexp);
1626                 *term++ = '/';
1627                 return term;
1628         }
1629         else {
1630                 char errbuf[1024];
1631                 regerror(reg_error, &regexp, errbuf, 1024);
1632                 die("-L parameter '%s': %s", spec + 1, errbuf);
1633         }
1634 }
1635
1636 static void prepare_blame_range(struct scoreboard *sb,
1637                                 const char *bottomtop,
1638                                 long lno,
1639                                 long *bottom, long *top)
1640 {
1641         const char *term;
1642
1643         term = parse_loc(bottomtop, sb, lno, 1, bottom);
1644         if (*term == ',') {
1645                 term = parse_loc(term + 1, sb, lno, *bottom + 1, top);
1646                 if (*term)
1647                         usage(blame_usage);
1648         }
1649         if (*term)
1650                 usage(blame_usage);
1651 }
1652
1653 static int git_blame_config(const char *var, const char *value)
1654 {
1655         if (!strcmp(var, "blame.showroot")) {
1656                 show_root = git_config_bool(var, value);
1657                 return 0;
1658         }
1659         if (!strcmp(var, "blame.blankboundary")) {
1660                 blank_boundary = git_config_bool(var, value);
1661                 return 0;
1662         }
1663         return git_default_config(var, value);
1664 }
1665
1666 int cmd_blame(int argc, const char **argv, const char *prefix)
1667 {
1668         struct rev_info revs;
1669         const char *path;
1670         struct scoreboard sb;
1671         struct origin *o;
1672         struct blame_entry *ent;
1673         int i, seen_dashdash, unk, opt;
1674         long bottom, top, lno;
1675         int output_option = 0;
1676         const char *revs_file = NULL;
1677         const char *final_commit_name = NULL;
1678         char type[10];
1679         const char *bottomtop = NULL;
1680
1681         git_config(git_blame_config);
1682         save_commit_buffer = 0;
1683
1684         opt = 0;
1685         seen_dashdash = 0;
1686         for (unk = i = 1; i < argc; i++) {
1687                 const char *arg = argv[i];
1688                 if (*arg != '-')
1689                         break;
1690                 else if (!strcmp("-b", arg))
1691                         blank_boundary = 1;
1692                 else if (!strcmp("--root", arg))
1693                         show_root = 1;
1694                 else if (!strcmp("-c", arg))
1695                         output_option |= OUTPUT_ANNOTATE_COMPAT;
1696                 else if (!strcmp("-t", arg))
1697                         output_option |= OUTPUT_RAW_TIMESTAMP;
1698                 else if (!strcmp("-l", arg))
1699                         output_option |= OUTPUT_LONG_OBJECT_NAME;
1700                 else if (!strcmp("-S", arg) && ++i < argc)
1701                         revs_file = argv[i];
1702                 else if (!strncmp("-M", arg, 2)) {
1703                         opt |= PICKAXE_BLAME_MOVE;
1704                         blame_move_score = parse_score(arg+2);
1705                 }
1706                 else if (!strncmp("-C", arg, 2)) {
1707                         if (opt & PICKAXE_BLAME_COPY)
1708                                 opt |= PICKAXE_BLAME_COPY_HARDER;
1709                         opt |= PICKAXE_BLAME_COPY | PICKAXE_BLAME_MOVE;
1710                         blame_copy_score = parse_score(arg+2);
1711                 }
1712                 else if (!strncmp("-L", arg, 2)) {
1713                         if (!arg[2]) {
1714                                 if (++i >= argc)
1715                                         usage(blame_usage);
1716                                 arg = argv[i];
1717                         }
1718                         else
1719                                 arg += 2;
1720                         if (bottomtop)
1721                                 die("More than one '-L n,m' option given");
1722                         bottomtop = arg;
1723                 }
1724                 else if (!strcmp("--score-debug", arg))
1725                         output_option |= OUTPUT_SHOW_SCORE;
1726                 else if (!strcmp("-f", arg) ||
1727                          !strcmp("--show-name", arg))
1728                         output_option |= OUTPUT_SHOW_NAME;
1729                 else if (!strcmp("-n", arg) ||
1730                          !strcmp("--show-number", arg))
1731                         output_option |= OUTPUT_SHOW_NUMBER;
1732                 else if (!strcmp("-p", arg) ||
1733                          !strcmp("--porcelain", arg))
1734                         output_option |= OUTPUT_PORCELAIN;
1735                 else if (!strcmp("--", arg)) {
1736                         seen_dashdash = 1;
1737                         i++;
1738                         break;
1739                 }
1740                 else
1741                         argv[unk++] = arg;
1742         }
1743
1744         if (!blame_move_score)
1745                 blame_move_score = BLAME_DEFAULT_MOVE_SCORE;
1746         if (!blame_copy_score)
1747                 blame_copy_score = BLAME_DEFAULT_COPY_SCORE;
1748
1749         /* We have collected options unknown to us in argv[1..unk]
1750          * which are to be passed to revision machinery if we are
1751          * going to do the "bottom" procesing.
1752          *
1753          * The remaining are:
1754          *
1755          * (1) if seen_dashdash, its either
1756          *     "-options -- <path>" or
1757          *     "-options -- <path> <rev>".
1758          *     but the latter is allowed only if there is no
1759          *     options that we passed to revision machinery.
1760          *
1761          * (2) otherwise, we may have "--" somewhere later and
1762          *     might be looking at the first one of multiple 'rev'
1763          *     parameters (e.g. " master ^next ^maint -- path").
1764          *     See if there is a dashdash first, and give the
1765          *     arguments before that to revision machinery.
1766          *     After that there must be one 'path'.
1767          *
1768          * (3) otherwise, its one of the three:
1769          *     "-options <path> <rev>"
1770          *     "-options <rev> <path>"
1771          *     "-options <path>"
1772          *     but again the first one is allowed only if
1773          *     there is no options that we passed to revision
1774          *     machinery.
1775          */
1776
1777         if (seen_dashdash) {
1778                 /* (1) */
1779                 if (argc <= i)
1780                         usage(blame_usage);
1781                 path = add_prefix(prefix, argv[i]);
1782                 if (i + 1 == argc - 1) {
1783                         if (unk != 1)
1784                                 usage(blame_usage);
1785                         argv[unk++] = argv[i + 1];
1786                 }
1787                 else if (i + 1 != argc)
1788                         /* garbage at end */
1789                         usage(blame_usage);
1790         }
1791         else {
1792                 int j;
1793                 for (j = i; !seen_dashdash && j < argc; j++)
1794                         if (!strcmp(argv[j], "--"))
1795                                 seen_dashdash = j;
1796                 if (seen_dashdash) {
1797                         if (seen_dashdash + 1 != argc - 1)
1798                                 usage(blame_usage);
1799                         path = add_prefix(prefix, argv[seen_dashdash + 1]);
1800                         for (j = i; j < seen_dashdash; j++)
1801                                 argv[unk++] = argv[j];
1802                 }
1803                 else {
1804                         /* (3) */
1805                         path = add_prefix(prefix, argv[i]);
1806                         if (i + 1 == argc - 1) {
1807                                 final_commit_name = argv[i + 1];
1808
1809                                 /* if (unk == 1) we could be getting
1810                                  * old-style
1811                                  */
1812                                 if (unk == 1 && !has_path_in_work_tree(path)) {
1813                                         path = add_prefix(prefix, argv[i + 1]);
1814                                         final_commit_name = argv[i];
1815                                 }
1816                         }
1817                         else if (i != argc - 1)
1818                                 usage(blame_usage); /* garbage at end */
1819
1820                         if (!has_path_in_work_tree(path))
1821                                 die("cannot stat path %s: %s",
1822                                     path, strerror(errno));
1823                 }
1824         }
1825
1826         if (final_commit_name)
1827                 argv[unk++] = final_commit_name;
1828
1829         /* Now we got rev and path.  We do not want the path pruning
1830          * but we may want "bottom" processing.
1831          */
1832         argv[unk++] = "--"; /* terminate the rev name */
1833         argv[unk] = NULL;
1834
1835         init_revisions(&revs, NULL);
1836         setup_revisions(unk, argv, &revs, "HEAD");
1837         memset(&sb, 0, sizeof(sb));
1838
1839         /* There must be one and only one positive commit in the
1840          * revs->pending array.
1841          */
1842         for (i = 0; i < revs.pending.nr; i++) {
1843                 struct object *obj = revs.pending.objects[i].item;
1844                 if (obj->flags & UNINTERESTING)
1845                         continue;
1846                 while (obj->type == OBJ_TAG)
1847                         obj = deref_tag(obj, NULL, 0);
1848                 if (obj->type != OBJ_COMMIT)
1849                         die("Non commit %s?",
1850                             revs.pending.objects[i].name);
1851                 if (sb.final)
1852                         die("More than one commit to dig from %s and %s?",
1853                             revs.pending.objects[i].name,
1854                             final_commit_name);
1855                 sb.final = (struct commit *) obj;
1856                 final_commit_name = revs.pending.objects[i].name;
1857         }
1858
1859         if (!sb.final) {
1860                 /* "--not A B -- path" without anything positive */
1861                 unsigned char head_sha1[20];
1862
1863                 final_commit_name = "HEAD";
1864                 if (get_sha1(final_commit_name, head_sha1))
1865                         die("No such ref: HEAD");
1866                 sb.final = lookup_commit_reference(head_sha1);
1867                 add_pending_object(&revs, &(sb.final->object), "HEAD");
1868         }
1869
1870         /* If we have bottom, this will mark the ancestors of the
1871          * bottom commits we would reach while traversing as
1872          * uninteresting.
1873          */
1874         prepare_revision_walk(&revs);
1875
1876         o = get_origin(&sb, sb.final, path);
1877         if (fill_blob_sha1(o))
1878                 die("no such path %s in %s", path, final_commit_name);
1879
1880         sb.final_buf = read_sha1_file(o->blob_sha1, type, &sb.final_buf_size);
1881         num_read_blob++;
1882         lno = prepare_lines(&sb);
1883
1884         bottom = top = 0;
1885         if (bottomtop)
1886                 prepare_blame_range(&sb, bottomtop, lno, &bottom, &top);
1887         if (bottom && top && top < bottom) {
1888                 long tmp;
1889                 tmp = top; top = bottom; bottom = tmp;
1890         }
1891         if (bottom < 1)
1892                 bottom = 1;
1893         if (top < 1)
1894                 top = lno;
1895         bottom--;
1896         if (lno < top)
1897                 die("file %s has only %lu lines", path, lno);
1898
1899         ent = xcalloc(1, sizeof(*ent));
1900         ent->lno = bottom;
1901         ent->num_lines = top - bottom;
1902         ent->suspect = o;
1903         ent->s_lno = bottom;
1904
1905         sb.ent = ent;
1906         sb.path = path;
1907
1908         if (revs_file && read_ancestry(revs_file))
1909                 die("reading graft file %s failed: %s",
1910                     revs_file, strerror(errno));
1911
1912         assign_blame(&sb, &revs, opt);
1913
1914         coalesce(&sb);
1915
1916         if (!(output_option & OUTPUT_PORCELAIN))
1917                 find_alignment(&sb, &output_option);
1918
1919         output(&sb, output_option);
1920         free((void *)sb.final_buf);
1921         for (ent = sb.ent; ent; ) {
1922                 struct blame_entry *e = ent->next;
1923                 free(ent);
1924                 ent = e;
1925         }
1926
1927         if (DEBUG) {
1928                 printf("num read blob: %d\n", num_read_blob);
1929                 printf("num get patch: %d\n", num_get_patch);
1930                 printf("num commits: %d\n", num_commits);
1931         }
1932         return 0;
1933 }