builtin-pickaxe.con commit git-pickaxe: rename detection optimization (2f3f8b2)
   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
  21static char pickaxe_usage[] =
  22"git-pickaxe [-c] [-l] [-t] [-f] [-n] [-p] [-L n,m] [-S <revs-file>] [-M] [-C] [-C] [commit] [--] file\n"
  23"  -c, --compatibility Use the same output mode as git-annotate (Default: off)\n"
  24"  -l, --long          Show long commit SHA1 (Default: off)\n"
  25"  -t, --time          Show raw timestamp (Default: off)\n"
  26"  -f, --show-name     Show original filename (Default: auto)\n"
  27"  -n, --show-number   Show original linenumber (Default: off)\n"
  28"  -p, --porcelain     Show in a format designed for machine consumption\n"
  29"  -L n,m              Process only line range n,m, counting from 1\n"
  30"  -M, -C              Find line movements within and across files\n"
  31"  -S revs-file        Use revisions from revs-file instead of calling git-rev-list\n";
  32
  33static int longest_file;
  34static int longest_author;
  35static int max_orig_digits;
  36static int max_digits;
  37static int max_score_digits;
  38
  39#ifndef DEBUG
  40#define DEBUG 0
  41#endif
  42
  43#define PICKAXE_BLAME_MOVE              01
  44#define PICKAXE_BLAME_COPY              02
  45#define PICKAXE_BLAME_COPY_HARDER       04
  46
  47/*
  48 * blame for a blame_entry with score lower than these thresholds
  49 * is not passed to the parent using move/copy logic.
  50 */
  51static unsigned blame_move_score;
  52static unsigned blame_copy_score;
  53#define BLAME_DEFAULT_MOVE_SCORE        20
  54#define BLAME_DEFAULT_COPY_SCORE        40
  55
  56/* bits #0..7 in revision.h, #8..11 used for merge_bases() in commit.c */
  57#define METAINFO_SHOWN          (1u<<12)
  58#define MORE_THAN_ONE_PATH      (1u<<13)
  59
  60/*
  61 * One blob in a commit that is being suspected
  62 */
  63struct origin {
  64        int refcnt;
  65        struct commit *commit;
  66        unsigned char blob_sha1[20];
  67        char path[FLEX_ARRAY];
  68};
  69
  70static inline struct origin *origin_incref(struct origin *o)
  71{
  72        if (o)
  73                o->refcnt++;
  74        return o;
  75}
  76
  77static void origin_decref(struct origin *o)
  78{
  79        if (o && --o->refcnt <= 0) {
  80                memset(o, 0, sizeof(*o));
  81                free(o);
  82        }
  83}
  84
  85struct blame_entry {
  86        struct blame_entry *prev;
  87        struct blame_entry *next;
  88
  89        /* the first line of this group in the final image;
  90         * internally all line numbers are 0 based.
  91         */
  92        int lno;
  93
  94        /* how many lines this group has */
  95        int num_lines;
  96
  97        /* the commit that introduced this group into the final image */
  98        struct origin *suspect;
  99
 100        /* true if the suspect is truly guilty; false while we have not
 101         * checked if the group came from one of its parents.
 102         */
 103        char guilty;
 104
 105        /* the line number of the first line of this group in the
 106         * suspect's file; internally all line numbers are 0 based.
 107         */
 108        int s_lno;
 109
 110        /* how significant this entry is -- cached to avoid
 111         * scanning the lines over and over
 112         */
 113        unsigned score;
 114};
 115
 116struct scoreboard {
 117        /* the final commit (i.e. where we started digging from) */
 118        struct commit *final;
 119
 120        const char *path;
 121
 122        /* the contents in the final; pointed into by buf pointers of
 123         * blame_entries
 124         */
 125        const char *final_buf;
 126        unsigned long final_buf_size;
 127
 128        /* linked list of blames */
 129        struct blame_entry *ent;
 130
 131        /* look-up a line in the final buffer */
 132        int num_lines;
 133        int *lineno;
 134};
 135
 136static int cmp_suspect(struct origin *a, struct origin *b)
 137{
 138        int cmp = hashcmp(a->commit->object.sha1, b->commit->object.sha1);
 139        if (cmp)
 140                return cmp;
 141        return strcmp(a->path, b->path);
 142}
 143
 144#define cmp_suspect(a, b) ( ((a)==(b)) ? 0 : cmp_suspect(a,b) )
 145
 146static void sanity_check_refcnt(struct scoreboard *);
 147
 148static void coalesce(struct scoreboard *sb)
 149{
 150        struct blame_entry *ent, *next;
 151
 152        for (ent = sb->ent; ent && (next = ent->next); ent = next) {
 153                if (!cmp_suspect(ent->suspect, next->suspect) &&
 154                    ent->guilty == next->guilty &&
 155                    ent->s_lno + ent->num_lines == next->s_lno) {
 156                        ent->num_lines += next->num_lines;
 157                        ent->next = next->next;
 158                        if (ent->next)
 159                                ent->next->prev = ent;
 160                        origin_decref(next->suspect);
 161                        free(next);
 162                        ent->score = 0;
 163                        next = ent; /* again */
 164                }
 165        }
 166
 167        if (DEBUG) /* sanity */
 168                sanity_check_refcnt(sb);
 169}
 170
 171static struct origin *get_origin(struct scoreboard *sb,
 172                                 struct commit *commit,
 173                                 const char *path)
 174{
 175        struct blame_entry *e;
 176        struct origin *o;
 177
 178        for (e = sb->ent; e; e = e->next) {
 179                if (e->suspect->commit == commit &&
 180                    !strcmp(e->suspect->path, path))
 181                        return origin_incref(e->suspect);
 182        }
 183        o = xcalloc(1, sizeof(*o) + strlen(path) + 1);
 184        o->commit = commit;
 185        o->refcnt = 1;
 186        strcpy(o->path, path);
 187        return o;
 188}
 189
 190static int fill_blob_sha1(struct origin *origin)
 191{
 192        unsigned mode;
 193        char type[10];
 194
 195        if (!is_null_sha1(origin->blob_sha1))
 196                return 0;
 197        if (get_tree_entry(origin->commit->object.sha1,
 198                           origin->path,
 199                           origin->blob_sha1, &mode))
 200                goto error_out;
 201        if (sha1_object_info(origin->blob_sha1, type, NULL) ||
 202            strcmp(type, blob_type))
 203                goto error_out;
 204        return 0;
 205 error_out:
 206        hashclr(origin->blob_sha1);
 207        return -1;
 208}
 209
 210static struct origin *find_origin(struct scoreboard *sb,
 211                                  struct commit *parent,
 212                                  struct origin *origin)
 213{
 214        struct origin *porigin = NULL;
 215        struct diff_options diff_opts;
 216        const char *paths[2];
 217
 218        if (parent->util) {
 219                struct origin *cached = parent->util;
 220                if (!strcmp(cached->path, origin->path))
 221                        return origin_incref(cached);
 222        }
 223
 224        /* See if the origin->path is different between parent
 225         * and origin first.  Most of the time they are the
 226         * same and diff-tree is fairly efficient about this.
 227         */
 228        diff_setup(&diff_opts);
 229        diff_opts.recursive = 1;
 230        diff_opts.detect_rename = 0;
 231        diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT;
 232        paths[0] = origin->path;
 233        paths[1] = NULL;
 234
 235        diff_tree_setup_paths(paths, &diff_opts);
 236        if (diff_setup_done(&diff_opts) < 0)
 237                die("diff-setup");
 238        diff_tree_sha1(parent->tree->object.sha1,
 239                       origin->commit->tree->object.sha1,
 240                       "", &diff_opts);
 241        diffcore_std(&diff_opts);
 242
 243        /* It is either one entry that says "modified", or "created",
 244         * or nothing.
 245         */
 246        if (!diff_queued_diff.nr) {
 247                /* The path is the same as parent */
 248                porigin = get_origin(sb, parent, origin->path);
 249                hashcpy(porigin->blob_sha1, origin->blob_sha1);
 250        }
 251        else if (diff_queued_diff.nr != 1)
 252                die("internal error in pickaxe::find_origin");
 253        else {
 254                struct diff_filepair *p = diff_queued_diff.queue[0];
 255                switch (p->status) {
 256                default:
 257                        die("internal error in pickaxe::find_origin (%c)",
 258                            p->status);
 259                case 'M':
 260                        porigin = get_origin(sb, parent, origin->path);
 261                        hashcpy(porigin->blob_sha1, p->one->sha1);
 262                        break;
 263                case 'A':
 264                case 'T':
 265                        /* Did not exist in parent, or type changed */
 266                        break;
 267                }
 268        }
 269        diff_flush(&diff_opts);
 270        if (porigin) {
 271                origin_incref(porigin);
 272                if (parent->util)
 273                        origin_decref(parent->util);
 274                parent->util = porigin;
 275        }
 276        return porigin;
 277}
 278
 279static struct origin *find_rename(struct scoreboard *sb,
 280                                  struct commit *parent,
 281                                  struct origin *origin)
 282{
 283        struct origin *porigin = NULL;
 284        struct diff_options diff_opts;
 285        int i;
 286        const char *paths[2];
 287
 288        diff_setup(&diff_opts);
 289        diff_opts.recursive = 1;
 290        diff_opts.detect_rename = DIFF_DETECT_RENAME;
 291        diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT;
 292        diff_opts.single_follow = origin->path;
 293        paths[0] = NULL;
 294        diff_tree_setup_paths(paths, &diff_opts);
 295        if (diff_setup_done(&diff_opts) < 0)
 296                die("diff-setup");
 297        diff_tree_sha1(parent->tree->object.sha1,
 298                       origin->commit->tree->object.sha1,
 299                       "", &diff_opts);
 300        diffcore_std(&diff_opts);
 301
 302        for (i = 0; i < diff_queued_diff.nr; i++) {
 303                struct diff_filepair *p = diff_queued_diff.queue[i];
 304                if ((p->status == 'R' || p->status == 'C') &&
 305                    !strcmp(p->two->path, origin->path)) {
 306                        porigin = get_origin(sb, parent, p->one->path);
 307                        hashcpy(porigin->blob_sha1, p->one->sha1);
 308                        break;
 309                }
 310        }
 311        diff_flush(&diff_opts);
 312        return porigin;
 313}
 314
 315struct chunk {
 316        /* line number in postimage; up to but not including this
 317         * line is the same as preimage
 318         */
 319        int same;
 320
 321        /* preimage line number after this chunk */
 322        int p_next;
 323
 324        /* postimage line number after this chunk */
 325        int t_next;
 326};
 327
 328struct patch {
 329        struct chunk *chunks;
 330        int num;
 331};
 332
 333struct blame_diff_state {
 334        struct xdiff_emit_state xm;
 335        struct patch *ret;
 336        unsigned hunk_post_context;
 337        unsigned hunk_in_pre_context : 1;
 338};
 339
 340static void process_u_diff(void *state_, char *line, unsigned long len)
 341{
 342        struct blame_diff_state *state = state_;
 343        struct chunk *chunk;
 344        int off1, off2, len1, len2, num;
 345
 346        num = state->ret->num;
 347        if (len < 4 || line[0] != '@' || line[1] != '@') {
 348                if (state->hunk_in_pre_context && line[0] == ' ')
 349                        state->ret->chunks[num - 1].same++;
 350                else {
 351                        state->hunk_in_pre_context = 0;
 352                        if (line[0] == ' ')
 353                                state->hunk_post_context++;
 354                        else
 355                                state->hunk_post_context = 0;
 356                }
 357                return;
 358        }
 359
 360        if (num && state->hunk_post_context) {
 361                chunk = &state->ret->chunks[num - 1];
 362                chunk->p_next -= state->hunk_post_context;
 363                chunk->t_next -= state->hunk_post_context;
 364        }
 365        state->ret->num = ++num;
 366        state->ret->chunks = xrealloc(state->ret->chunks,
 367                                      sizeof(struct chunk) * num);
 368        chunk = &state->ret->chunks[num - 1];
 369        if (parse_hunk_header(line, len, &off1, &len1, &off2, &len2)) {
 370                state->ret->num--;
 371                return;
 372        }
 373
 374        /* Line numbers in patch output are one based. */
 375        off1--;
 376        off2--;
 377
 378        chunk->same = len2 ? off2 : (off2 + 1);
 379
 380        chunk->p_next = off1 + (len1 ? len1 : 1);
 381        chunk->t_next = chunk->same + len2;
 382        state->hunk_in_pre_context = 1;
 383        state->hunk_post_context = 0;
 384}
 385
 386static struct patch *compare_buffer(mmfile_t *file_p, mmfile_t *file_o,
 387                                    int context)
 388{
 389        struct blame_diff_state state;
 390        xpparam_t xpp;
 391        xdemitconf_t xecfg;
 392        xdemitcb_t ecb;
 393
 394        xpp.flags = XDF_NEED_MINIMAL;
 395        xecfg.ctxlen = context;
 396        xecfg.flags = 0;
 397        ecb.outf = xdiff_outf;
 398        ecb.priv = &state;
 399        memset(&state, 0, sizeof(state));
 400        state.xm.consume = process_u_diff;
 401        state.ret = xmalloc(sizeof(struct patch));
 402        state.ret->chunks = NULL;
 403        state.ret->num = 0;
 404
 405        xdl_diff(file_p, file_o, &xpp, &xecfg, &ecb);
 406
 407        if (state.ret->num) {
 408                struct chunk *chunk;
 409                chunk = &state.ret->chunks[state.ret->num - 1];
 410                chunk->p_next -= state.hunk_post_context;
 411                chunk->t_next -= state.hunk_post_context;
 412        }
 413        return state.ret;
 414}
 415
 416static struct patch *get_patch(struct origin *parent, struct origin *origin)
 417{
 418        mmfile_t file_p, file_o;
 419        char type[10];
 420        char *blob_p, *blob_o;
 421        struct patch *patch;
 422
 423        blob_p = read_sha1_file(parent->blob_sha1, type,
 424                                (unsigned long *) &file_p.size);
 425        blob_o = read_sha1_file(origin->blob_sha1, type,
 426                                (unsigned long *) &file_o.size);
 427        file_p.ptr = blob_p;
 428        file_o.ptr = blob_o;
 429        if (!file_p.ptr || !file_o.ptr) {
 430                free(blob_p);
 431                free(blob_o);
 432                return NULL;
 433        }
 434
 435        patch = compare_buffer(&file_p, &file_o, 0);
 436        free(blob_p);
 437        free(blob_o);
 438        return patch;
 439}
 440
 441static void free_patch(struct patch *p)
 442{
 443        free(p->chunks);
 444        free(p);
 445}
 446
 447static void add_blame_entry(struct scoreboard *sb, struct blame_entry *e)
 448{
 449        struct blame_entry *ent, *prev = NULL;
 450
 451        origin_incref(e->suspect);
 452
 453        for (ent = sb->ent; ent && ent->lno < e->lno; ent = ent->next)
 454                prev = ent;
 455
 456        /* prev, if not NULL, is the last one that is below e */
 457        e->prev = prev;
 458        if (prev) {
 459                e->next = prev->next;
 460                prev->next = e;
 461        }
 462        else {
 463                e->next = sb->ent;
 464                sb->ent = e;
 465        }
 466        if (e->next)
 467                e->next->prev = e;
 468}
 469
 470static void dup_entry(struct blame_entry *dst, struct blame_entry *src)
 471{
 472        struct blame_entry *p, *n;
 473
 474        p = dst->prev;
 475        n = dst->next;
 476        origin_incref(src->suspect);
 477        origin_decref(dst->suspect);
 478        memcpy(dst, src, sizeof(*src));
 479        dst->prev = p;
 480        dst->next = n;
 481        dst->score = 0;
 482}
 483
 484static const char *nth_line(struct scoreboard *sb, int lno)
 485{
 486        return sb->final_buf + sb->lineno[lno];
 487}
 488
 489static void split_overlap(struct blame_entry *split,
 490                          struct blame_entry *e,
 491                          int tlno, int plno, int same,
 492                          struct origin *parent)
 493{
 494        /* it is known that lines between tlno to same came from
 495         * parent, and e has an overlap with that range.  it also is
 496         * known that parent's line plno corresponds to e's line tlno.
 497         *
 498         *                <---- e ----->
 499         *                   <------>
 500         *                   <------------>
 501         *             <------------>
 502         *             <------------------>
 503         *
 504         * Potentially we need to split e into three parts; before
 505         * this chunk, the chunk to be blamed for parent, and after
 506         * that portion.
 507         */
 508        int chunk_end_lno;
 509        memset(split, 0, sizeof(struct blame_entry [3]));
 510
 511        if (e->s_lno < tlno) {
 512                /* there is a pre-chunk part not blamed on parent */
 513                split[0].suspect = origin_incref(e->suspect);
 514                split[0].lno = e->lno;
 515                split[0].s_lno = e->s_lno;
 516                split[0].num_lines = tlno - e->s_lno;
 517                split[1].lno = e->lno + tlno - e->s_lno;
 518                split[1].s_lno = plno;
 519        }
 520        else {
 521                split[1].lno = e->lno;
 522                split[1].s_lno = plno + (e->s_lno - tlno);
 523        }
 524
 525        if (same < e->s_lno + e->num_lines) {
 526                /* there is a post-chunk part not blamed on parent */
 527                split[2].suspect = origin_incref(e->suspect);
 528                split[2].lno = e->lno + (same - e->s_lno);
 529                split[2].s_lno = e->s_lno + (same - e->s_lno);
 530                split[2].num_lines = e->s_lno + e->num_lines - same;
 531                chunk_end_lno = split[2].lno;
 532        }
 533        else
 534                chunk_end_lno = e->lno + e->num_lines;
 535        split[1].num_lines = chunk_end_lno - split[1].lno;
 536
 537        if (split[1].num_lines < 1)
 538                return;
 539        split[1].suspect = origin_incref(parent);
 540}
 541
 542static void split_blame(struct scoreboard *sb,
 543                        struct blame_entry *split,
 544                        struct blame_entry *e)
 545{
 546        struct blame_entry *new_entry;
 547
 548        if (split[0].suspect && split[2].suspect) {
 549                /* we need to split e into two and add another for parent */
 550                dup_entry(e, &split[0]);
 551
 552                new_entry = xmalloc(sizeof(*new_entry));
 553                memcpy(new_entry, &(split[2]), sizeof(struct blame_entry));
 554                add_blame_entry(sb, new_entry);
 555
 556                new_entry = xmalloc(sizeof(*new_entry));
 557                memcpy(new_entry, &(split[1]), sizeof(struct blame_entry));
 558                add_blame_entry(sb, new_entry);
 559        }
 560        else if (!split[0].suspect && !split[2].suspect)
 561                /* parent covers the entire area */
 562                dup_entry(e, &split[1]);
 563        else if (split[0].suspect) {
 564                dup_entry(e, &split[0]);
 565
 566                new_entry = xmalloc(sizeof(*new_entry));
 567                memcpy(new_entry, &(split[1]), sizeof(struct blame_entry));
 568                add_blame_entry(sb, new_entry);
 569        }
 570        else {
 571                dup_entry(e, &split[1]);
 572
 573                new_entry = xmalloc(sizeof(*new_entry));
 574                memcpy(new_entry, &(split[2]), sizeof(struct blame_entry));
 575                add_blame_entry(sb, new_entry);
 576        }
 577
 578        if (DEBUG) { /* sanity */
 579                struct blame_entry *ent;
 580                int lno = sb->ent->lno, corrupt = 0;
 581
 582                for (ent = sb->ent; ent; ent = ent->next) {
 583                        if (lno != ent->lno)
 584                                corrupt = 1;
 585                        if (ent->s_lno < 0)
 586                                corrupt = 1;
 587                        lno += ent->num_lines;
 588                }
 589                if (corrupt) {
 590                        lno = sb->ent->lno;
 591                        for (ent = sb->ent; ent; ent = ent->next) {
 592                                printf("L %8d l %8d n %8d\n",
 593                                       lno, ent->lno, ent->num_lines);
 594                                lno = ent->lno + ent->num_lines;
 595                        }
 596                        die("oops");
 597                }
 598        }
 599}
 600
 601static void decref_split(struct blame_entry *split)
 602{
 603        int i;
 604
 605        for (i = 0; i < 3; i++)
 606                origin_decref(split[i].suspect);
 607}
 608
 609static void blame_overlap(struct scoreboard *sb, struct blame_entry *e,
 610                          int tlno, int plno, int same,
 611                          struct origin *parent)
 612{
 613        struct blame_entry split[3];
 614
 615        split_overlap(split, e, tlno, plno, same, parent);
 616        if (split[1].suspect)
 617                split_blame(sb, split, e);
 618        decref_split(split);
 619}
 620
 621static int find_last_in_target(struct scoreboard *sb, struct origin *target)
 622{
 623        struct blame_entry *e;
 624        int last_in_target = -1;
 625
 626        for (e = sb->ent; e; e = e->next) {
 627                if (e->guilty || cmp_suspect(e->suspect, target))
 628                        continue;
 629                if (last_in_target < e->s_lno + e->num_lines)
 630                        last_in_target = e->s_lno + e->num_lines;
 631        }
 632        return last_in_target;
 633}
 634
 635static void blame_chunk(struct scoreboard *sb,
 636                        int tlno, int plno, int same,
 637                        struct origin *target, struct origin *parent)
 638{
 639        struct blame_entry *e;
 640
 641        for (e = sb->ent; e; e = e->next) {
 642                if (e->guilty || cmp_suspect(e->suspect, target))
 643                        continue;
 644                if (same <= e->s_lno)
 645                        continue;
 646                if (tlno < e->s_lno + e->num_lines)
 647                        blame_overlap(sb, e, tlno, plno, same, parent);
 648        }
 649}
 650
 651static int pass_blame_to_parent(struct scoreboard *sb,
 652                                struct origin *target,
 653                                struct origin *parent)
 654{
 655        int i, last_in_target, plno, tlno;
 656        struct patch *patch;
 657
 658        last_in_target = find_last_in_target(sb, target);
 659        if (last_in_target < 0)
 660                return 1; /* nothing remains for this target */
 661
 662        patch = get_patch(parent, target);
 663        plno = tlno = 0;
 664        for (i = 0; i < patch->num; i++) {
 665                struct chunk *chunk = &patch->chunks[i];
 666
 667                blame_chunk(sb, tlno, plno, chunk->same, target, parent);
 668                plno = chunk->p_next;
 669                tlno = chunk->t_next;
 670        }
 671        /* rest (i.e. anything above tlno) are the same as parent */
 672        blame_chunk(sb, tlno, plno, last_in_target, target, parent);
 673
 674        free_patch(patch);
 675        return 0;
 676}
 677
 678static unsigned ent_score(struct scoreboard *sb, struct blame_entry *e)
 679{
 680        unsigned score;
 681        const char *cp, *ep;
 682
 683        if (e->score)
 684                return e->score;
 685
 686        score = 1;
 687        cp = nth_line(sb, e->lno);
 688        ep = nth_line(sb, e->lno + e->num_lines);
 689        while (cp < ep) {
 690                unsigned ch = *((unsigned char *)cp);
 691                if (isalnum(ch))
 692                        score++;
 693                cp++;
 694        }
 695        e->score = score;
 696        return score;
 697}
 698
 699static void copy_split_if_better(struct scoreboard *sb,
 700                                 struct blame_entry *best_so_far,
 701                                 struct blame_entry *this)
 702{
 703        int i;
 704
 705        if (!this[1].suspect)
 706                return;
 707        if (best_so_far[1].suspect) {
 708                if (ent_score(sb, &this[1]) < ent_score(sb, &best_so_far[1]))
 709                        return;
 710        }
 711
 712        for (i = 0; i < 3; i++)
 713                origin_incref(this[i].suspect);
 714        decref_split(best_so_far);
 715        memcpy(best_so_far, this, sizeof(struct blame_entry [3]));
 716}
 717
 718static void find_copy_in_blob(struct scoreboard *sb,
 719                              struct blame_entry *ent,
 720                              struct origin *parent,
 721                              struct blame_entry *split,
 722                              mmfile_t *file_p)
 723{
 724        const char *cp;
 725        int cnt;
 726        mmfile_t file_o;
 727        struct patch *patch;
 728        int i, plno, tlno;
 729
 730        cp = nth_line(sb, ent->lno);
 731        file_o.ptr = (char*) cp;
 732        cnt = ent->num_lines;
 733
 734        while (cnt && cp < sb->final_buf + sb->final_buf_size) {
 735                if (*cp++ == '\n')
 736                        cnt--;
 737        }
 738        file_o.size = cp - file_o.ptr;
 739
 740        patch = compare_buffer(file_p, &file_o, 1);
 741
 742        memset(split, 0, sizeof(struct blame_entry [3]));
 743        plno = tlno = 0;
 744        for (i = 0; i < patch->num; i++) {
 745                struct chunk *chunk = &patch->chunks[i];
 746
 747                /* tlno to chunk->same are the same as ent */
 748                if (ent->num_lines <= tlno)
 749                        break;
 750                if (tlno < chunk->same) {
 751                        struct blame_entry this[3];
 752                        split_overlap(this, ent,
 753                                      tlno + ent->s_lno, plno,
 754                                      chunk->same + ent->s_lno,
 755                                      parent);
 756                        copy_split_if_better(sb, split, this);
 757                        decref_split(this);
 758                }
 759                plno = chunk->p_next;
 760                tlno = chunk->t_next;
 761        }
 762        free_patch(patch);
 763}
 764
 765static int find_move_in_parent(struct scoreboard *sb,
 766                               struct origin *target,
 767                               struct origin *parent)
 768{
 769        int last_in_target;
 770        struct blame_entry *e, split[3];
 771        mmfile_t file_p;
 772        char type[10];
 773        char *blob_p;
 774
 775        last_in_target = find_last_in_target(sb, target);
 776        if (last_in_target < 0)
 777                return 1; /* nothing remains for this target */
 778
 779        blob_p = read_sha1_file(parent->blob_sha1, type,
 780                                (unsigned long *) &file_p.size);
 781        file_p.ptr = blob_p;
 782        if (!file_p.ptr) {
 783                free(blob_p);
 784                return 0;
 785        }
 786
 787        for (e = sb->ent; e; e = e->next) {
 788                if (e->guilty || cmp_suspect(e->suspect, target))
 789                        continue;
 790                find_copy_in_blob(sb, e, parent, split, &file_p);
 791                if (split[1].suspect &&
 792                    blame_move_score < ent_score(sb, &split[1]))
 793                        split_blame(sb, split, e);
 794                decref_split(split);
 795        }
 796        free(blob_p);
 797        return 0;
 798}
 799
 800static int find_copy_in_parent(struct scoreboard *sb,
 801                               struct origin *target,
 802                               struct commit *parent,
 803                               struct origin *porigin,
 804                               int opt)
 805{
 806        struct diff_options diff_opts;
 807        const char *paths[1];
 808        struct blame_entry *e;
 809        int i, j;
 810        struct blame_list {
 811                struct blame_entry *ent;
 812                struct blame_entry split[3];
 813        } *blame_list;
 814        int num_ents;
 815
 816        /* Count the number of entries the target is suspected for,
 817         * and prepare a list of entry and the best split.
 818         */
 819        for (e = sb->ent, num_ents = 0; e; e = e->next)
 820                if (!e->guilty && !cmp_suspect(e->suspect, target))
 821                        num_ents++;
 822        if (!num_ents)
 823                return 1; /* nothing remains for this target */
 824
 825        blame_list = xcalloc(num_ents, sizeof(struct blame_list));
 826        for (e = sb->ent, i = 0; e; e = e->next)
 827                if (!e->guilty && !cmp_suspect(e->suspect, target))
 828                        blame_list[i++].ent = e;
 829
 830        diff_setup(&diff_opts);
 831        diff_opts.recursive = 1;
 832        diff_opts.output_format = DIFF_FORMAT_NO_OUTPUT;
 833
 834        /* Try "find copies harder" on new path */
 835        if ((opt & PICKAXE_BLAME_COPY_HARDER) &&
 836            (!porigin || strcmp(target->path, porigin->path))) {
 837                diff_opts.detect_rename = DIFF_DETECT_COPY;
 838                diff_opts.find_copies_harder = 1;
 839        }
 840        paths[0] = NULL;
 841        diff_tree_setup_paths(paths, &diff_opts);
 842        if (diff_setup_done(&diff_opts) < 0)
 843                die("diff-setup");
 844        diff_tree_sha1(parent->tree->object.sha1,
 845                       target->commit->tree->object.sha1,
 846                       "", &diff_opts);
 847        diffcore_std(&diff_opts);
 848
 849        for (i = 0; i < diff_queued_diff.nr; i++) {
 850                struct diff_filepair *p = diff_queued_diff.queue[i];
 851                struct origin *norigin;
 852                mmfile_t file_p;
 853                char type[10];
 854                char *blob;
 855                struct blame_entry this[3];
 856
 857                if (!DIFF_FILE_VALID(p->one))
 858                        continue; /* does not exist in parent */
 859                if (porigin && !strcmp(p->one->path, porigin->path))
 860                        /* find_move already dealt with this path */
 861                        continue;
 862
 863                norigin = get_origin(sb, parent, p->one->path);
 864                hashcpy(norigin->blob_sha1, p->one->sha1);
 865                blob = read_sha1_file(norigin->blob_sha1, type,
 866                                      (unsigned long *) &file_p.size);
 867                file_p.ptr = blob;
 868                if (!file_p.ptr)
 869                        continue;
 870
 871                for (j = 0; j < num_ents; j++) {
 872                        find_copy_in_blob(sb, blame_list[j].ent, norigin,
 873                                          this, &file_p);
 874                        copy_split_if_better(sb, blame_list[j].split,
 875                                             this);
 876                        decref_split(this);
 877                }
 878                free(blob);
 879                origin_decref(norigin);
 880        }
 881        diff_flush(&diff_opts);
 882
 883        for (j = 0; j < num_ents; j++) {
 884                struct blame_entry *split = blame_list[j].split;
 885                if (split[1].suspect &&
 886                    blame_copy_score < ent_score(sb, &split[1]))
 887                        split_blame(sb, split, blame_list[j].ent);
 888                decref_split(split);
 889        }
 890        free(blame_list);
 891
 892        return 0;
 893}
 894
 895#define MAXPARENT 16
 896
 897static void pass_blame(struct scoreboard *sb, struct origin *origin, int opt)
 898{
 899        int i, pass;
 900        struct commit *commit = origin->commit;
 901        struct commit_list *parent;
 902        struct origin *parent_origin[MAXPARENT], *porigin;
 903
 904        memset(parent_origin, 0, sizeof(parent_origin));
 905
 906        /* The first pass looks for unrenamed path to optimize for
 907         * common cases, then we look for renames in the second pass.
 908         */
 909        for (pass = 0; pass < 2; pass++) {
 910                struct origin *(*find)(struct scoreboard *,
 911                                       struct commit *, struct origin *);
 912                find = pass ? find_rename : find_origin;
 913
 914                for (i = 0, parent = commit->parents;
 915                     i < MAXPARENT && parent;
 916                     parent = parent->next, i++) {
 917                        struct commit *p = parent->item;
 918
 919                        if (parent_origin[i])
 920                                continue;
 921                        if (parse_commit(p))
 922                                continue;
 923                        porigin = find(sb, p, origin);
 924                        if (!porigin)
 925                                continue;
 926                        if (!hashcmp(porigin->blob_sha1, origin->blob_sha1)) {
 927                                struct blame_entry *e;
 928                                for (e = sb->ent; e; e = e->next)
 929                                        if (e->suspect == origin) {
 930                                                origin_incref(porigin);
 931                                                origin_decref(e->suspect);
 932                                                e->suspect = porigin;
 933                                        }
 934                                origin_decref(porigin);
 935                                goto finish;
 936                        }
 937                        parent_origin[i] = porigin;
 938                }
 939        }
 940
 941        for (i = 0, parent = commit->parents;
 942             i < MAXPARENT && parent;
 943             parent = parent->next, i++) {
 944                struct origin *porigin = parent_origin[i];
 945                if (!porigin)
 946                        continue;
 947                if (pass_blame_to_parent(sb, origin, porigin))
 948                        goto finish;
 949        }
 950
 951        /*
 952         * Optionally run "miff" to find moves in parents' files here.
 953         */
 954        if (opt & PICKAXE_BLAME_MOVE)
 955                for (i = 0, parent = commit->parents;
 956                     i < MAXPARENT && parent;
 957                     parent = parent->next, i++) {
 958                        struct origin *porigin = parent_origin[i];
 959                        if (!porigin)
 960                                continue;
 961                        if (find_move_in_parent(sb, origin, porigin))
 962                                goto finish;
 963                }
 964
 965        /*
 966         * Optionally run "ciff" to find copies from parents' files here.
 967         */
 968        if (opt & PICKAXE_BLAME_COPY)
 969                for (i = 0, parent = commit->parents;
 970                     i < MAXPARENT && parent;
 971                     parent = parent->next, i++) {
 972                        struct origin *porigin = parent_origin[i];
 973                        if (find_copy_in_parent(sb, origin, parent->item,
 974                                                porigin, opt))
 975                                goto finish;
 976                }
 977
 978 finish:
 979        for (i = 0; i < MAXPARENT; i++)
 980                origin_decref(parent_origin[i]);
 981}
 982
 983static void assign_blame(struct scoreboard *sb, struct rev_info *revs, int opt)
 984{
 985        while (1) {
 986                struct blame_entry *ent;
 987                struct commit *commit;
 988                struct origin *suspect = NULL;
 989
 990                /* find one suspect to break down */
 991                for (ent = sb->ent; !suspect && ent; ent = ent->next)
 992                        if (!ent->guilty)
 993                                suspect = ent->suspect;
 994                if (!suspect)
 995                        return; /* all done */
 996
 997                origin_incref(suspect);
 998                commit = suspect->commit;
 999                parse_commit(commit);
1000                if (!(commit->object.flags & UNINTERESTING) &&
1001                    !(revs->max_age != -1 && commit->date  < revs->max_age))
1002                        pass_blame(sb, suspect, opt);
1003
1004                /* Take responsibility for the remaining entries */
1005                for (ent = sb->ent; ent; ent = ent->next)
1006                        if (!cmp_suspect(ent->suspect, suspect))
1007                                ent->guilty = 1;
1008                origin_decref(suspect);
1009
1010                if (DEBUG) /* sanity */
1011                        sanity_check_refcnt(sb);
1012        }
1013}
1014
1015static const char *format_time(unsigned long time, const char *tz_str,
1016                               int show_raw_time)
1017{
1018        static char time_buf[128];
1019        time_t t = time;
1020        int minutes, tz;
1021        struct tm *tm;
1022
1023        if (show_raw_time) {
1024                sprintf(time_buf, "%lu %s", time, tz_str);
1025                return time_buf;
1026        }
1027
1028        tz = atoi(tz_str);
1029        minutes = tz < 0 ? -tz : tz;
1030        minutes = (minutes / 100)*60 + (minutes % 100);
1031        minutes = tz < 0 ? -minutes : minutes;
1032        t = time + minutes * 60;
1033        tm = gmtime(&t);
1034
1035        strftime(time_buf, sizeof(time_buf), "%Y-%m-%d %H:%M:%S ", tm);
1036        strcat(time_buf, tz_str);
1037        return time_buf;
1038}
1039
1040struct commit_info
1041{
1042        char *author;
1043        char *author_mail;
1044        unsigned long author_time;
1045        char *author_tz;
1046
1047        /* filled only when asked for details */
1048        char *committer;
1049        char *committer_mail;
1050        unsigned long committer_time;
1051        char *committer_tz;
1052
1053        char *summary;
1054};
1055
1056static void get_ac_line(const char *inbuf, const char *what,
1057                        int bufsz, char *person, char **mail,
1058                        unsigned long *time, char **tz)
1059{
1060        int len;
1061        char *tmp, *endp;
1062
1063        tmp = strstr(inbuf, what);
1064        if (!tmp)
1065                goto error_out;
1066        tmp += strlen(what);
1067        endp = strchr(tmp, '\n');
1068        if (!endp)
1069                len = strlen(tmp);
1070        else
1071                len = endp - tmp;
1072        if (bufsz <= len) {
1073        error_out:
1074                /* Ugh */
1075                person = *mail = *tz = "(unknown)";
1076                *time = 0;
1077                return;
1078        }
1079        memcpy(person, tmp, len);
1080
1081        tmp = person;
1082        tmp += len;
1083        *tmp = 0;
1084        while (*tmp != ' ')
1085                tmp--;
1086        *tz = tmp+1;
1087
1088        *tmp = 0;
1089        while (*tmp != ' ')
1090                tmp--;
1091        *time = strtoul(tmp, NULL, 10);
1092
1093        *tmp = 0;
1094        while (*tmp != ' ')
1095                tmp--;
1096        *mail = tmp + 1;
1097        *tmp = 0;
1098}
1099
1100static void get_commit_info(struct commit *commit,
1101                            struct commit_info *ret,
1102                            int detailed)
1103{
1104        int len;
1105        char *tmp, *endp;
1106        static char author_buf[1024];
1107        static char committer_buf[1024];
1108        static char summary_buf[1024];
1109
1110        /* We've operated without save_commit_buffer, so
1111         * we now need to populate them for output.
1112         */
1113        if (!commit->buffer) {
1114                char type[20];
1115                unsigned long size;
1116                commit->buffer =
1117                        read_sha1_file(commit->object.sha1, type, &size);
1118        }
1119        ret->author = author_buf;
1120        get_ac_line(commit->buffer, "\nauthor ",
1121                    sizeof(author_buf), author_buf, &ret->author_mail,
1122                    &ret->author_time, &ret->author_tz);
1123
1124        if (!detailed)
1125                return;
1126
1127        ret->committer = committer_buf;
1128        get_ac_line(commit->buffer, "\ncommitter ",
1129                    sizeof(committer_buf), committer_buf, &ret->committer_mail,
1130                    &ret->committer_time, &ret->committer_tz);
1131
1132        ret->summary = summary_buf;
1133        tmp = strstr(commit->buffer, "\n\n");
1134        if (!tmp) {
1135        error_out:
1136                sprintf(summary_buf, "(%s)", sha1_to_hex(commit->object.sha1));
1137                return;
1138        }
1139        tmp += 2;
1140        endp = strchr(tmp, '\n');
1141        if (!endp)
1142                goto error_out;
1143        len = endp - tmp;
1144        if (len >= sizeof(summary_buf))
1145                goto error_out;
1146        memcpy(summary_buf, tmp, len);
1147        summary_buf[len] = 0;
1148}
1149
1150#define OUTPUT_ANNOTATE_COMPAT  001
1151#define OUTPUT_LONG_OBJECT_NAME 002
1152#define OUTPUT_RAW_TIMESTAMP    004
1153#define OUTPUT_PORCELAIN        010
1154#define OUTPUT_SHOW_NAME        020
1155#define OUTPUT_SHOW_NUMBER      040
1156#define OUTPUT_SHOW_SCORE      0100
1157
1158static void emit_porcelain(struct scoreboard *sb, struct blame_entry *ent)
1159{
1160        int cnt;
1161        const char *cp;
1162        struct origin *suspect = ent->suspect;
1163        char hex[41];
1164
1165        strcpy(hex, sha1_to_hex(suspect->commit->object.sha1));
1166        printf("%s%c%d %d %d\n",
1167               hex,
1168               ent->guilty ? ' ' : '*', // purely for debugging
1169               ent->s_lno + 1,
1170               ent->lno + 1,
1171               ent->num_lines);
1172        if (!(suspect->commit->object.flags & METAINFO_SHOWN)) {
1173                struct commit_info ci;
1174                suspect->commit->object.flags |= METAINFO_SHOWN;
1175                get_commit_info(suspect->commit, &ci, 1);
1176                printf("author %s\n", ci.author);
1177                printf("author-mail %s\n", ci.author_mail);
1178                printf("author-time %lu\n", ci.author_time);
1179                printf("author-tz %s\n", ci.author_tz);
1180                printf("committer %s\n", ci.committer);
1181                printf("committer-mail %s\n", ci.committer_mail);
1182                printf("committer-time %lu\n", ci.committer_time);
1183                printf("committer-tz %s\n", ci.committer_tz);
1184                printf("filename %s\n", suspect->path);
1185                printf("summary %s\n", ci.summary);
1186        }
1187        else if (suspect->commit->object.flags & MORE_THAN_ONE_PATH)
1188                printf("filename %s\n", suspect->path);
1189
1190        cp = nth_line(sb, ent->lno);
1191        for (cnt = 0; cnt < ent->num_lines; cnt++) {
1192                char ch;
1193                if (cnt)
1194                        printf("%s %d %d\n", hex,
1195                               ent->s_lno + 1 + cnt,
1196                               ent->lno + 1 + cnt);
1197                putchar('\t');
1198                do {
1199                        ch = *cp++;
1200                        putchar(ch);
1201                } while (ch != '\n' &&
1202                         cp < sb->final_buf + sb->final_buf_size);
1203        }
1204}
1205
1206static void emit_other(struct scoreboard *sb, struct blame_entry *ent, int opt)
1207{
1208        int cnt;
1209        const char *cp;
1210        struct origin *suspect = ent->suspect;
1211        struct commit_info ci;
1212        char hex[41];
1213        int show_raw_time = !!(opt & OUTPUT_RAW_TIMESTAMP);
1214
1215        get_commit_info(suspect->commit, &ci, 1);
1216        strcpy(hex, sha1_to_hex(suspect->commit->object.sha1));
1217
1218        cp = nth_line(sb, ent->lno);
1219        for (cnt = 0; cnt < ent->num_lines; cnt++) {
1220                char ch;
1221
1222                printf("%.*s", (opt & OUTPUT_LONG_OBJECT_NAME) ? 40 : 8, hex);
1223                if (opt & OUTPUT_ANNOTATE_COMPAT)
1224                        printf("\t(%10s\t%10s\t%d)", ci.author,
1225                               format_time(ci.author_time, ci.author_tz,
1226                                           show_raw_time),
1227                               ent->lno + 1 + cnt);
1228                else {
1229                        if (opt & OUTPUT_SHOW_SCORE)
1230                                printf(" %*d %02d",
1231                                       max_score_digits, ent->score,
1232                                       ent->suspect->refcnt);
1233                        if (opt & OUTPUT_SHOW_NAME)
1234                                printf(" %-*.*s", longest_file, longest_file,
1235                                       suspect->path);
1236                        if (opt & OUTPUT_SHOW_NUMBER)
1237                                printf(" %*d", max_orig_digits,
1238                                       ent->s_lno + 1 + cnt);
1239                        printf(" (%-*.*s %10s %*d) ",
1240                               longest_author, longest_author, ci.author,
1241                               format_time(ci.author_time, ci.author_tz,
1242                                           show_raw_time),
1243                               max_digits, ent->lno + 1 + cnt);
1244                }
1245                do {
1246                        ch = *cp++;
1247                        putchar(ch);
1248                } while (ch != '\n' &&
1249                         cp < sb->final_buf + sb->final_buf_size);
1250        }
1251}
1252
1253static void output(struct scoreboard *sb, int option)
1254{
1255        struct blame_entry *ent;
1256
1257        if (option & OUTPUT_PORCELAIN) {
1258                for (ent = sb->ent; ent; ent = ent->next) {
1259                        struct blame_entry *oth;
1260                        struct origin *suspect = ent->suspect;
1261                        struct commit *commit = suspect->commit;
1262                        if (commit->object.flags & MORE_THAN_ONE_PATH)
1263                                continue;
1264                        for (oth = ent->next; oth; oth = oth->next) {
1265                                if ((oth->suspect->commit != commit) ||
1266                                    !strcmp(oth->suspect->path, suspect->path))
1267                                        continue;
1268                                commit->object.flags |= MORE_THAN_ONE_PATH;
1269                                break;
1270                        }
1271                }
1272        }
1273
1274        for (ent = sb->ent; ent; ent = ent->next) {
1275                if (option & OUTPUT_PORCELAIN)
1276                        emit_porcelain(sb, ent);
1277                else {
1278                        emit_other(sb, ent, option);
1279                }
1280        }
1281}
1282
1283static int prepare_lines(struct scoreboard *sb)
1284{
1285        const char *buf = sb->final_buf;
1286        unsigned long len = sb->final_buf_size;
1287        int num = 0, incomplete = 0, bol = 1;
1288
1289        if (len && buf[len-1] != '\n')
1290                incomplete++; /* incomplete line at the end */
1291        while (len--) {
1292                if (bol) {
1293                        sb->lineno = xrealloc(sb->lineno,
1294                                              sizeof(int* ) * (num + 1));
1295                        sb->lineno[num] = buf - sb->final_buf;
1296                        bol = 0;
1297                }
1298                if (*buf++ == '\n') {
1299                        num++;
1300                        bol = 1;
1301                }
1302        }
1303        sb->lineno = xrealloc(sb->lineno,
1304                              sizeof(int* ) * (num + incomplete + 1));
1305        sb->lineno[num + incomplete] = buf - sb->final_buf;
1306        sb->num_lines = num + incomplete;
1307        return sb->num_lines;
1308}
1309
1310static int read_ancestry(const char *graft_file)
1311{
1312        FILE *fp = fopen(graft_file, "r");
1313        char buf[1024];
1314        if (!fp)
1315                return -1;
1316        while (fgets(buf, sizeof(buf), fp)) {
1317                /* The format is just "Commit Parent1 Parent2 ...\n" */
1318                int len = strlen(buf);
1319                struct commit_graft *graft = read_graft_line(buf, len);
1320                register_commit_graft(graft, 0);
1321        }
1322        fclose(fp);
1323        return 0;
1324}
1325
1326static int lineno_width(int lines)
1327{
1328        int i, width;
1329
1330        for (width = 1, i = 10; i <= lines + 1; width++)
1331                i *= 10;
1332        return width;
1333}
1334
1335static void find_alignment(struct scoreboard *sb, int *option)
1336{
1337        int longest_src_lines = 0;
1338        int longest_dst_lines = 0;
1339        unsigned largest_score = 0;
1340        struct blame_entry *e;
1341
1342        for (e = sb->ent; e; e = e->next) {
1343                struct origin *suspect = e->suspect;
1344                struct commit_info ci;
1345                int num;
1346
1347                if (!(suspect->commit->object.flags & METAINFO_SHOWN)) {
1348                        suspect->commit->object.flags |= METAINFO_SHOWN;
1349                        get_commit_info(suspect->commit, &ci, 1);
1350                        if (strcmp(suspect->path, sb->path))
1351                                *option |= OUTPUT_SHOW_NAME;
1352                        num = strlen(suspect->path);
1353                        if (longest_file < num)
1354                                longest_file = num;
1355                        num = strlen(ci.author);
1356                        if (longest_author < num)
1357                                longest_author = num;
1358                }
1359                num = e->s_lno + e->num_lines;
1360                if (longest_src_lines < num)
1361                        longest_src_lines = num;
1362                num = e->lno + e->num_lines;
1363                if (longest_dst_lines < num)
1364                        longest_dst_lines = num;
1365                if (largest_score < ent_score(sb, e))
1366                        largest_score = ent_score(sb, e);
1367        }
1368        max_orig_digits = lineno_width(longest_src_lines);
1369        max_digits = lineno_width(longest_dst_lines);
1370        max_score_digits = lineno_width(largest_score);
1371}
1372
1373static void sanity_check_refcnt(struct scoreboard *sb)
1374{
1375        int baa = 0;
1376        struct blame_entry *ent;
1377
1378        for (ent = sb->ent; ent; ent = ent->next) {
1379                /* Nobody should have zero or negative refcnt */
1380                if (ent->suspect->refcnt <= 0)
1381                        baa = 1;
1382        }
1383        for (ent = sb->ent; ent; ent = ent->next) {
1384                /* Mark the ones that haven't been checked */
1385                if (0 < ent->suspect->refcnt)
1386                        ent->suspect->refcnt = -ent->suspect->refcnt;
1387        }
1388        for (ent = sb->ent; ent; ent = ent->next) {
1389                /* then pick each and see if they have the the correct
1390                 * refcnt; note that ->util caching means origin's refcnt
1391                 * may well be greater than the number of blame entries
1392                 * that use it.
1393                 */
1394                int found;
1395                struct blame_entry *e;
1396                struct origin *suspect = ent->suspect;
1397
1398                if (0 < suspect->refcnt)
1399                        continue;
1400                suspect->refcnt = -suspect->refcnt; /* Unmark */
1401                for (found = 0, e = sb->ent; e; e = e->next) {
1402                        if (e->suspect != suspect)
1403                                continue;
1404                        found++;
1405                }
1406                if (suspect->refcnt < found)
1407                        baa = 1;
1408        }
1409        if (baa) {
1410                int opt = 0160;
1411                find_alignment(sb, &opt);
1412                output(sb, opt);
1413                die("Baa!");
1414        }
1415}
1416
1417static int has_path_in_work_tree(const char *path)
1418{
1419        struct stat st;
1420        return !lstat(path, &st);
1421}
1422
1423static unsigned parse_score(const char *arg)
1424{
1425        char *end;
1426        unsigned long score = strtoul(arg, &end, 10);
1427        if (*end)
1428                return 0;
1429        return score;
1430}
1431
1432static const char *add_prefix(const char *prefix, const char *path)
1433{
1434        if (!prefix || !prefix[0])
1435                return path;
1436        return prefix_path(prefix, strlen(prefix), path);
1437}
1438
1439int cmd_pickaxe(int argc, const char **argv, const char *prefix)
1440{
1441        struct rev_info revs;
1442        const char *path;
1443        struct scoreboard sb;
1444        struct origin *o;
1445        struct blame_entry *ent;
1446        int i, seen_dashdash, unk, opt;
1447        long bottom, top, lno;
1448        int output_option = 0;
1449        const char *revs_file = NULL;
1450        const char *final_commit_name = NULL;
1451        char type[10];
1452
1453        save_commit_buffer = 0;
1454
1455        opt = 0;
1456        bottom = top = 0;
1457        seen_dashdash = 0;
1458        for (unk = i = 1; i < argc; i++) {
1459                const char *arg = argv[i];
1460                if (*arg != '-')
1461                        break;
1462                else if (!strcmp("-c", arg))
1463                        output_option |= OUTPUT_ANNOTATE_COMPAT;
1464                else if (!strcmp("-t", arg))
1465                        output_option |= OUTPUT_RAW_TIMESTAMP;
1466                else if (!strcmp("-l", arg))
1467                        output_option |= OUTPUT_LONG_OBJECT_NAME;
1468                else if (!strcmp("-S", arg) && ++i < argc)
1469                        revs_file = argv[i];
1470                else if (!strncmp("-M", arg, 2)) {
1471                        opt |= PICKAXE_BLAME_MOVE;
1472                        blame_move_score = parse_score(arg+2);
1473                }
1474                else if (!strncmp("-C", arg, 2)) {
1475                        if (opt & PICKAXE_BLAME_COPY)
1476                                opt |= PICKAXE_BLAME_COPY_HARDER;
1477                        opt |= PICKAXE_BLAME_COPY | PICKAXE_BLAME_MOVE;
1478                        blame_copy_score = parse_score(arg+2);
1479                }
1480                else if (!strncmp("-L", arg, 2)) {
1481                        char *term;
1482                        if (!arg[2]) {
1483                                if (++i >= argc)
1484                                        usage(pickaxe_usage);
1485                                arg = argv[i];
1486                        }
1487                        else
1488                                arg += 2;
1489                        if (bottom || top)
1490                                die("More than one '-L n,m' option given");
1491                        bottom = strtol(arg, &term, 10);
1492                        if (*term == ',') {
1493                                top = strtol(term + 1, &term, 10);
1494                                if (*term)
1495                                        usage(pickaxe_usage);
1496                        }
1497                        if (bottom && top && top < bottom) {
1498                                unsigned long tmp;
1499                                tmp = top; top = bottom; bottom = tmp;
1500                        }
1501                }
1502                else if (!strcmp("--score-debug", arg))
1503                        output_option |= OUTPUT_SHOW_SCORE;
1504                else if (!strcmp("-f", arg) ||
1505                         !strcmp("--show-name", arg))
1506                        output_option |= OUTPUT_SHOW_NAME;
1507                else if (!strcmp("-n", arg) ||
1508                         !strcmp("--show-number", arg))
1509                        output_option |= OUTPUT_SHOW_NUMBER;
1510                else if (!strcmp("-p", arg) ||
1511                         !strcmp("--porcelain", arg))
1512                        output_option |= OUTPUT_PORCELAIN;
1513                else if (!strcmp("--", arg)) {
1514                        seen_dashdash = 1;
1515                        i++;
1516                        break;
1517                }
1518                else
1519                        argv[unk++] = arg;
1520        }
1521
1522        if (!blame_move_score)
1523                blame_move_score = BLAME_DEFAULT_MOVE_SCORE;
1524        if (!blame_copy_score)
1525                blame_copy_score = BLAME_DEFAULT_COPY_SCORE;
1526
1527        /* We have collected options unknown to us in argv[1..unk]
1528         * which are to be passed to revision machinery if we are
1529         * going to do the "bottom" procesing.
1530         *
1531         * The remaining are:
1532         *
1533         * (1) if seen_dashdash, its either
1534         *     "-options -- <path>" or
1535         *     "-options -- <path> <rev>".
1536         *     but the latter is allowed only if there is no
1537         *     options that we passed to revision machinery.
1538         *
1539         * (2) otherwise, we may have "--" somewhere later and
1540         *     might be looking at the first one of multiple 'rev'
1541         *     parameters (e.g. " master ^next ^maint -- path").
1542         *     See if there is a dashdash first, and give the
1543         *     arguments before that to revision machinery.
1544         *     After that there must be one 'path'.
1545         *
1546         * (3) otherwise, its one of the three:
1547         *     "-options <path> <rev>"
1548         *     "-options <rev> <path>"
1549         *     "-options <path>"
1550         *     but again the first one is allowed only if
1551         *     there is no options that we passed to revision
1552         *     machinery.
1553         */
1554
1555        if (seen_dashdash) {
1556                /* (1) */
1557                if (argc <= i)
1558                        usage(pickaxe_usage);
1559                path = add_prefix(prefix, argv[i]);
1560                if (i + 1 == argc - 1) {
1561                        if (unk != 1)
1562                                usage(pickaxe_usage);
1563                        argv[unk++] = argv[i + 1];
1564                }
1565                else if (i + 1 != argc)
1566                        /* garbage at end */
1567                        usage(pickaxe_usage);
1568        }
1569        else {
1570                int j;
1571                for (j = i; !seen_dashdash && j < argc; j++)
1572                        if (!strcmp(argv[j], "--"))
1573                                seen_dashdash = j;
1574                if (seen_dashdash) {
1575                        if (seen_dashdash + 1 != argc - 1)
1576                                usage(pickaxe_usage);
1577                        path = add_prefix(prefix, argv[seen_dashdash + 1]);
1578                        for (j = i; j < seen_dashdash; j++)
1579                                argv[unk++] = argv[j];
1580                }
1581                else {
1582                        /* (3) */
1583                        path = add_prefix(prefix, argv[i]);
1584                        if (i + 1 == argc - 1) {
1585                                final_commit_name = argv[i + 1];
1586
1587                                /* if (unk == 1) we could be getting
1588                                 * old-style
1589                                 */
1590                                if (unk == 1 && !has_path_in_work_tree(path)) {
1591                                        path = add_prefix(prefix, argv[i + 1]);
1592                                        final_commit_name = argv[i];
1593                                }
1594                        }
1595                        else if (i != argc - 1)
1596                                usage(pickaxe_usage); /* garbage at end */
1597
1598                        if (!has_path_in_work_tree(path))
1599                                die("cannot stat path %s: %s",
1600                                    path, strerror(errno));
1601                }
1602        }
1603
1604        if (final_commit_name)
1605                argv[unk++] = final_commit_name;
1606
1607        /* Now we got rev and path.  We do not want the path pruning
1608         * but we may want "bottom" processing.
1609         */
1610        argv[unk] = NULL;
1611
1612        init_revisions(&revs, NULL);
1613        setup_revisions(unk, argv, &revs, "HEAD");
1614        memset(&sb, 0, sizeof(sb));
1615
1616        /* There must be one and only one positive commit in the
1617         * revs->pending array.
1618         */
1619        for (i = 0; i < revs.pending.nr; i++) {
1620                struct object *obj = revs.pending.objects[i].item;
1621                if (obj->flags & UNINTERESTING)
1622                        continue;
1623                while (obj->type == OBJ_TAG)
1624                        obj = deref_tag(obj, NULL, 0);
1625                if (obj->type != OBJ_COMMIT)
1626                        die("Non commit %s?",
1627                            revs.pending.objects[i].name);
1628                if (sb.final)
1629                        die("More than one commit to dig from %s and %s?",
1630                            revs.pending.objects[i].name,
1631                            final_commit_name);
1632                sb.final = (struct commit *) obj;
1633                final_commit_name = revs.pending.objects[i].name;
1634        }
1635
1636        if (!sb.final) {
1637                /* "--not A B -- path" without anything positive */
1638                unsigned char head_sha1[20];
1639
1640                final_commit_name = "HEAD";
1641                if (get_sha1(final_commit_name, head_sha1))
1642                        die("No such ref: HEAD");
1643                sb.final = lookup_commit_reference(head_sha1);
1644                add_pending_object(&revs, &(sb.final->object), "HEAD");
1645        }
1646
1647        /* If we have bottom, this will mark the ancestors of the
1648         * bottom commits we would reach while traversing as
1649         * uninteresting.
1650         */
1651        prepare_revision_walk(&revs);
1652
1653        o = get_origin(&sb, sb.final, path);
1654        if (fill_blob_sha1(o))
1655                die("no such path %s in %s", path, final_commit_name);
1656
1657        sb.final_buf = read_sha1_file(o->blob_sha1, type, &sb.final_buf_size);
1658        lno = prepare_lines(&sb);
1659
1660        if (bottom < 1)
1661                bottom = 1;
1662        if (top < 1)
1663                top = lno;
1664        bottom--;
1665        if (lno < top)
1666                die("file %s has only %lu lines", path, lno);
1667
1668        ent = xcalloc(1, sizeof(*ent));
1669        ent->lno = bottom;
1670        ent->num_lines = top - bottom;
1671        ent->suspect = o;
1672        ent->s_lno = bottom;
1673
1674        sb.ent = ent;
1675        sb.path = path;
1676
1677        if (revs_file && read_ancestry(revs_file))
1678                die("reading graft file %s failed: %s",
1679                    revs_file, strerror(errno));
1680
1681        assign_blame(&sb, &revs, opt);
1682
1683        coalesce(&sb);
1684
1685        if (!(output_option & OUTPUT_PORCELAIN))
1686                find_alignment(&sb, &output_option);
1687
1688        output(&sb, output_option);
1689        free((void *)sb.final_buf);
1690        for (ent = sb.ent; ent; ) {
1691                struct blame_entry *e = ent->next;
1692                free(ent);
1693                ent = e;
1694        }
1695        return 0;
1696}