Andrew's git / gitweb.git / blame.c
?
blame.con commit Merge part of 'jc/pack' into 'next' (6013f17)
   1/*
   2 * Copyright (C) 2006, Fredrik Kuivinen <freku045@student.liu.se>
   3 */
   4
   5#include <assert.h>
   6#include <time.h>
   7#include <sys/time.h>
   8
   9#include "cache.h"
  10#include "refs.h"
  11#include "tag.h"
  12#include "commit.h"
  13#include "tree.h"
  14#include "blob.h"
  15#include "diff.h"
  16#include "revision.h"
  17
  18#define DEBUG 0
  19
  20struct commit **blame_lines;
  21int num_blame_lines;
  22
  23struct util_info {
  24        int *line_map;
  25        unsigned char sha1[20]; /* blob sha, not commit! */
  26        char *buf;
  27        unsigned long size;
  28        int num_lines;
  29//    const char* path;
  30};
  31
  32struct chunk {
  33        int off1, len1; // ---
  34        int off2, len2; // +++
  35};
  36
  37struct patch {
  38        struct chunk *chunks;
  39        int num;
  40};
  41
  42static void get_blob(struct commit *commit);
  43
  44/* Only used for statistics */
  45static int num_get_patch = 0;
  46static int num_commits = 0;
  47static int patch_time = 0;
  48
  49#define TEMPFILE_PATH_LEN 60
  50static struct patch *get_patch(struct commit *commit, struct commit *other)
  51{
  52        struct patch *ret;
  53        struct util_info *info_c = (struct util_info *)commit->object.util;
  54        struct util_info *info_o = (struct util_info *)other->object.util;
  55        char tmp_path1[TEMPFILE_PATH_LEN], tmp_path2[TEMPFILE_PATH_LEN];
  56        char diff_cmd[TEMPFILE_PATH_LEN*2 + 20];
  57        struct timeval tv_start, tv_end;
  58        int fd;
  59        FILE *fin;
  60        char buf[1024];
  61
  62        ret = xmalloc(sizeof(struct patch));
  63        ret->chunks = NULL;
  64        ret->num = 0;
  65
  66        get_blob(commit);
  67        get_blob(other);
  68
  69        gettimeofday(&tv_start, NULL);
  70
  71        fd = git_mkstemp(tmp_path1, TEMPFILE_PATH_LEN, "git-blame-XXXXXX");
  72        if (fd < 0)
  73                die("unable to create temp-file: %s", strerror(errno));
  74
  75        if (xwrite(fd, info_c->buf, info_c->size) != info_c->size)
  76                die("write failed: %s", strerror(errno));
  77        close(fd);
  78
  79        fd = git_mkstemp(tmp_path2, TEMPFILE_PATH_LEN, "git-blame-XXXXXX");
  80        if (fd < 0)
  81                die("unable to create temp-file: %s", strerror(errno));
  82
  83        if (xwrite(fd, info_o->buf, info_o->size) != info_o->size)
  84                die("write failed: %s", strerror(errno));
  85        close(fd);
  86
  87        sprintf(diff_cmd, "diff -u0 %s %s", tmp_path1, tmp_path2);
  88        fin = popen(diff_cmd, "r");
  89        if (!fin)
  90                die("popen failed: %s", strerror(errno));
  91
  92        while (fgets(buf, sizeof(buf), fin)) {
  93                struct chunk *chunk;
  94                char *start, *sp;
  95
  96                if (buf[0] != '@' || buf[1] != '@')
  97                        continue;
  98
  99                if (DEBUG)
 100                        printf("chunk line: %s", buf);
 101                ret->num++;
 102                ret->chunks = xrealloc(ret->chunks,
 103                                       sizeof(struct chunk) * ret->num);
 104                chunk = &ret->chunks[ret->num - 1];
 105
 106                assert(!strncmp(buf, "@@ -", 4));
 107
 108                start = buf + 4;
 109                sp = index(start, ' ');
 110                *sp = '\0';
 111                if (index(start, ',')) {
 112                        int ret =
 113                            sscanf(start, "%d,%d", &chunk->off1, &chunk->len1);
 114                        assert(ret == 2);
 115                } else {
 116                        int ret = sscanf(start, "%d", &chunk->off1);
 117                        assert(ret == 1);
 118                        chunk->len1 = 1;
 119                }
 120                *sp = ' ';
 121
 122                start = sp + 1;
 123                sp = index(start, ' ');
 124                *sp = '\0';
 125                if (index(start, ',')) {
 126                        int ret =
 127                            sscanf(start, "%d,%d", &chunk->off2, &chunk->len2);
 128                        assert(ret == 2);
 129                } else {
 130                        int ret = sscanf(start, "%d", &chunk->off2);
 131                        assert(ret == 1);
 132                        chunk->len2 = 1;
 133                }
 134                *sp = ' ';
 135
 136                if (chunk->len1 == 0)
 137                        chunk->off1++;
 138                if (chunk->len2 == 0)
 139                        chunk->off2++;
 140
 141                if (chunk->off1 > 0)
 142                        chunk->off1--;
 143                if (chunk->off2 > 0)
 144                        chunk->off2--;
 145
 146                assert(chunk->off1 >= 0);
 147                assert(chunk->off2 >= 0);
 148        }
 149        pclose(fin);
 150        unlink(tmp_path1);
 151        unlink(tmp_path2);
 152
 153        gettimeofday(&tv_end, NULL);
 154        patch_time += 1000000 * (tv_end.tv_sec - tv_start.tv_sec) +
 155                tv_end.tv_usec - tv_start.tv_usec;
 156
 157        num_get_patch++;
 158        return ret;
 159}
 160
 161static void free_patch(struct patch *p)
 162{
 163        free(p->chunks);
 164        free(p);
 165}
 166
 167static int get_blob_sha1_internal(unsigned char *sha1, const char *base,
 168                                  int baselen, const char *pathname,
 169                                  unsigned mode, int stage);
 170
 171static unsigned char blob_sha1[20];
 172static int get_blob_sha1(struct tree *t, const char *pathname,
 173                         unsigned char *sha1)
 174{
 175        int i;
 176        const char *pathspec[2];
 177        pathspec[0] = pathname;
 178        pathspec[1] = NULL;
 179        memset(blob_sha1, 0, sizeof(blob_sha1));
 180        read_tree_recursive(t, "", 0, 0, pathspec, get_blob_sha1_internal);
 181
 182        for (i = 0; i < 20; i++) {
 183                if (blob_sha1[i] != 0)
 184                        break;
 185        }
 186
 187        if (i == 20)
 188                return -1;
 189
 190        memcpy(sha1, blob_sha1, 20);
 191        return 0;
 192}
 193
 194static int get_blob_sha1_internal(unsigned char *sha1, const char *base,
 195                                  int baselen, const char *pathname,
 196                                  unsigned mode, int stage)
 197{
 198        if (S_ISDIR(mode))
 199                return READ_TREE_RECURSIVE;
 200
 201        memcpy(blob_sha1, sha1, 20);
 202        return -1;
 203}
 204
 205static void get_blob(struct commit *commit)
 206{
 207        struct util_info *info = commit->object.util;
 208        char type[20];
 209
 210        if (info->buf)
 211                return;
 212
 213        info->buf = read_sha1_file(info->sha1, type, &info->size);
 214
 215        assert(!strcmp(type, "blob"));
 216}
 217
 218/* For debugging only */
 219static void print_patch(struct patch *p)
 220{
 221        int i;
 222        printf("Num chunks: %d\n", p->num);
 223        for (i = 0; i < p->num; i++) {
 224                printf("%d,%d %d,%d\n", p->chunks[i].off1, p->chunks[i].len1,
 225                       p->chunks[i].off2, p->chunks[i].len2);
 226        }
 227}
 228
 229#if 0
 230/* For debugging only */
 231static void print_map(struct commit *cmit, struct commit *other)
 232{
 233        struct util_info *util = cmit->object.util;
 234        struct util_info *util2 = other->object.util;
 235
 236        int i;
 237        int max =
 238            util->num_lines >
 239            util2->num_lines ? util->num_lines : util2->num_lines;
 240        int num;
 241
 242        for (i = 0; i < max; i++) {
 243                printf("i: %d ", i);
 244                num = -1;
 245
 246                if (i < util->num_lines) {
 247                        num = util->line_map[i];
 248                        printf("%d\t", num);
 249                } else
 250                        printf("\t");
 251
 252                if (i < util2->num_lines) {
 253                        int num2 = util2->line_map[i];
 254                        printf("%d\t", num2);
 255                        if (num != -1 && num2 != num)
 256                                printf("---");
 257                } else
 258                        printf("\t");
 259
 260                printf("\n");
 261        }
 262}
 263#endif
 264
 265// p is a patch from commit to other.
 266static void fill_line_map(struct commit *commit, struct commit *other,
 267                          struct patch *p)
 268{
 269        struct util_info *util = commit->object.util;
 270        struct util_info *util2 = other->object.util;
 271        int *map = util->line_map;
 272        int *map2 = util2->line_map;
 273        int cur_chunk = 0;
 274        int i1, i2;
 275
 276        if (p->num && DEBUG)
 277                print_patch(p);
 278
 279        if (DEBUG)
 280                printf("num lines 1: %d num lines 2: %d\n", util->num_lines,
 281                       util2->num_lines);
 282
 283        for (i1 = 0, i2 = 0; i1 < util->num_lines; i1++, i2++) {
 284                struct chunk *chunk = NULL;
 285                if (cur_chunk < p->num)
 286                        chunk = &p->chunks[cur_chunk];
 287
 288                if (chunk && chunk->off1 == i1) {
 289                        if (DEBUG && i2 != chunk->off2)
 290                                printf("i2: %d off2: %d\n", i2, chunk->off2);
 291
 292                        assert(i2 == chunk->off2);
 293
 294                        i1--;
 295                        i2--;
 296                        if (chunk->len1 > 0)
 297                                i1 += chunk->len1;
 298
 299                        if (chunk->len2 > 0)
 300                                i2 += chunk->len2;
 301
 302                        cur_chunk++;
 303                } else {
 304                        if (i2 >= util2->num_lines)
 305                                break;
 306
 307                        if (map[i1] != map2[i2] && map[i1] != -1) {
 308                                if (DEBUG)
 309                                        printf("map: i1: %d %d %p i2: %d %d %p\n",
 310                                               i1, map[i1],
 311                                               i1 != -1 ? blame_lines[map[i1]] : NULL,
 312                                               i2, map2[i2],
 313                                               i2 != -1 ? blame_lines[map2[i2]] : NULL);
 314                                if (map2[i2] != -1 &&
 315                                    blame_lines[map[i1]] &&
 316                                    !blame_lines[map2[i2]])
 317                                        map[i1] = map2[i2];
 318                        }
 319
 320                        if (map[i1] == -1 && map2[i2] != -1)
 321                                map[i1] = map2[i2];
 322                }
 323
 324                if (DEBUG > 1)
 325                        printf("l1: %d l2: %d i1: %d i2: %d\n",
 326                               map[i1], map2[i2], i1, i2);
 327        }
 328}
 329
 330static int map_line(struct commit *commit, int line)
 331{
 332        struct util_info *info = commit->object.util;
 333        assert(line >= 0 && line < info->num_lines);
 334        return info->line_map[line];
 335}
 336
 337static int fill_util_info(struct commit *commit, const char *path)
 338{
 339        struct util_info *util;
 340        if (commit->object.util)
 341                return 0;
 342
 343        util = xmalloc(sizeof(struct util_info));
 344
 345        if (get_blob_sha1(commit->tree, path, util->sha1)) {
 346                free(util);
 347                return 1;
 348        } else {
 349                util->buf = NULL;
 350                util->size = 0;
 351                util->line_map = NULL;
 352                util->num_lines = -1;
 353                commit->object.util = util;
 354                return 0;
 355        }
 356}
 357
 358static void alloc_line_map(struct commit *commit)
 359{
 360        struct util_info *util = commit->object.util;
 361        int i;
 362
 363        if (util->line_map)
 364                return;
 365
 366        get_blob(commit);
 367
 368        util->num_lines = 0;
 369        for (i = 0; i < util->size; i++) {
 370                if (util->buf[i] == '\n')
 371                        util->num_lines++;
 372        }
 373        if(util->buf[util->size - 1] != '\n')
 374                util->num_lines++;
 375
 376        util->line_map = xmalloc(sizeof(int) * util->num_lines);
 377
 378        for (i = 0; i < util->num_lines; i++)
 379                util->line_map[i] = -1;
 380}
 381
 382static void init_first_commit(struct commit* commit, const char* filename)
 383{
 384        struct util_info* util;
 385        int i;
 386
 387        if (fill_util_info(commit, filename))
 388                die("fill_util_info failed");
 389
 390        alloc_line_map(commit);
 391
 392        util = commit->object.util;
 393        num_blame_lines = util->num_lines;
 394
 395        for (i = 0; i < num_blame_lines; i++)
 396                util->line_map[i] = i;
 397}
 398
 399
 400static void process_commits(struct rev_info *rev, const char *path,
 401                            struct commit** initial)
 402{
 403        int i;
 404        struct util_info* util;
 405        int lines_left;
 406        int *blame_p;
 407        int *new_lines;
 408        int new_lines_len;
 409
 410        struct commit* commit = get_revision(rev);
 411        assert(commit);
 412        init_first_commit(commit, path);
 413
 414        util = commit->object.util;
 415        num_blame_lines = util->num_lines;
 416        blame_lines = xmalloc(sizeof(struct commit *) * num_blame_lines);
 417        for (i = 0; i < num_blame_lines; i++)
 418                blame_lines[i] = NULL;
 419
 420        lines_left = num_blame_lines;
 421        blame_p = xmalloc(sizeof(int) * num_blame_lines);
 422        new_lines = xmalloc(sizeof(int) * num_blame_lines);
 423        do {
 424                struct commit_list *parents;
 425                int num_parents;
 426                struct util_info *util;
 427
 428                if (DEBUG)
 429                        printf("\nProcessing commit: %d %s\n", num_commits,
 430                               sha1_to_hex(commit->object.sha1));
 431
 432                if (lines_left == 0)
 433                        return;
 434
 435                num_commits++;
 436                memset(blame_p, 0, sizeof(int) * num_blame_lines);
 437                new_lines_len = 0;
 438                num_parents = 0;
 439                for (parents = commit->parents;
 440                     parents != NULL; parents = parents->next)
 441                        num_parents++;
 442
 443                if(num_parents == 0)
 444                        *initial = commit;
 445
 446                if(fill_util_info(commit, path))
 447                        continue;
 448
 449                alloc_line_map(commit);
 450                util = commit->object.util;
 451
 452                for (parents = commit->parents;
 453                     parents != NULL; parents = parents->next) {
 454                        struct commit *parent = parents->item;
 455                        struct patch *patch;
 456
 457                        if (parse_commit(parent) < 0)
 458                                die("parse_commit error");
 459
 460                        if (DEBUG)
 461                                printf("parent: %s\n",
 462                                       sha1_to_hex(parent->object.sha1));
 463
 464                        if(fill_util_info(parent, path)) {
 465                                num_parents--;
 466                                continue;
 467                        }
 468
 469                        patch = get_patch(parent, commit);
 470                        alloc_line_map(parent);
 471                        fill_line_map(parent, commit, patch);
 472
 473                        for (i = 0; i < patch->num; i++) {
 474                            int l;
 475                            for (l = 0; l < patch->chunks[i].len2; l++) {
 476                                int mapped_line =
 477                                    map_line(commit, patch->chunks[i].off2 + l);
 478                                if (mapped_line != -1) {
 479                                    blame_p[mapped_line]++;
 480                                    if (blame_p[mapped_line] == num_parents)
 481                                        new_lines[new_lines_len++] = mapped_line;
 482                                }
 483                            }
 484                        }
 485                        free_patch(patch);
 486                }
 487
 488                if (DEBUG)
 489                        printf("parents: %d\n", num_parents);
 490
 491                for (i = 0; i < new_lines_len; i++) {
 492                        int mapped_line = new_lines[i];
 493                        if (blame_lines[mapped_line] == NULL) {
 494                                blame_lines[mapped_line] = commit;
 495                                lines_left--;
 496                                if (DEBUG)
 497                                        printf("blame: mapped: %d i: %d\n",
 498                                               mapped_line, i);
 499                        }
 500                }
 501        } while ((commit = get_revision(rev)) != NULL);
 502}
 503
 504int main(int argc, const char **argv)
 505{
 506        int i;
 507        struct commit *initial = NULL;
 508        unsigned char sha1[20];
 509        const char* filename;
 510        int num_args;
 511        const char* args[10];
 512        struct rev_info rev;
 513
 514        setup_git_directory();
 515
 516        if (argc != 3)
 517                die("Usage: blame commit-ish file");
 518
 519
 520        filename = argv[2];
 521
 522        {
 523                struct commit* commit;
 524                if (get_sha1(argv[1], sha1))
 525                        die("get_sha1 failed");
 526                commit = lookup_commit_reference(sha1);
 527
 528                if (fill_util_info(commit, filename)) {
 529                        printf("%s not found in %s\n", filename, argv[1]);
 530                        return 1;
 531                }
 532        }
 533
 534        num_args = 0;
 535        args[num_args++] = NULL;
 536        args[num_args++] = "--topo-order";
 537        args[num_args++] = "--remove-empty";
 538        args[num_args++] = argv[1];
 539        args[num_args++] = "--";
 540        args[num_args++] = filename;
 541        args[num_args] = NULL;
 542
 543        setup_revisions(num_args, args, &rev, "HEAD");
 544        prepare_revision_walk(&rev);
 545        process_commits(&rev, filename, &initial);
 546
 547        for (i = 0; i < num_blame_lines; i++) {
 548                struct commit *c = blame_lines[i];
 549                if (!c)
 550                        c = initial;
 551
 552                printf("%d %.8s\n", i, sha1_to_hex(c->object.sha1));
 553// printf("%d %s\n", i, find_unique_abbrev(blame_lines[i]->object.sha1, 6));
 554        }
 555
 556        if (DEBUG) {
 557                printf("num get patch: %d\n", num_get_patch);
 558                printf("num commits: %d\n", num_commits);
 559                printf("patch time: %f\n", patch_time / 1000000.0);
 560                printf("initial: %s\n", sha1_to_hex(initial->object.sha1));
 561        }
 562
 563        return 0;
 564}