src/blame.c

   1 /*
   2  * Copyright (C) the libgit2 contributors. All rights reserved.
   3  *
   4  * This file is part of libgit2, distributed under the GNU GPL v2 with
   5  * a Linking Exception. For full terms see the included COPYING file.
   6  */
   7
   8 #include "blame.h"
   9
  10 #include "git2/commit.h"
  11 #include "git2/revparse.h"
  12 #include "git2/revwalk.h"
  13 #include "git2/tree.h"
  14 #include "git2/diff.h"
  15 #include "git2/blob.h"
  16 #include "git2/signature.h"
  17 #include "git2/mailmap.h"
  18 #include "util.h"
  19 #include "repository.h"
  20 #include "blame_git.h"
  21
  22
  23 static int hunk_byfinalline_search_cmp(const void *key, const void *entry)
  24 {
  25         git_blame_hunk *hunk = (git_blame_hunk*)entry;
  26
  27         size_t lineno = *(size_t*)key;
  28         size_t lines_in_hunk = hunk->lines_in_hunk;
  29         size_t final_start_line_number = hunk->final_start_line_number;
  30
  31         if (lineno < final_start_line_number)
  32                 return -1;
  33         if (lineno >= final_start_line_number + lines_in_hunk)
  34                 return 1;
  35         return 0;
  36 }
  37
  38 static int paths_cmp(const void *a, const void *b) { return git__strcmp((char*)a, (char*)b); }
  39 static int hunk_cmp(const void *_a, const void *_b)
  40 {
  41         git_blame_hunk *a = (git_blame_hunk*)_a,
  42                                                 *b = (git_blame_hunk*)_b;
  43
  44         if (a->final_start_line_number > b->final_start_line_number)
  45                 return 1;
  46         else if (a->final_start_line_number < b->final_start_line_number)
  47                 return -1;
  48         else
  49                 return 0;
  50 }
  51
  52 static bool hunk_ends_at_or_before_line(git_blame_hunk *hunk, size_t line)
  53 {
  54         return line >= (hunk->final_start_line_number + hunk->lines_in_hunk - 1);
  55 }
  56
  57 static bool hunk_starts_at_or_after_line(git_blame_hunk *hunk, size_t line)
  58 {
  59         return line <= hunk->final_start_line_number;
  60 }
  61
  62 static git_blame_hunk* new_hunk(
  63                 size_t start,
  64                 size_t lines,
  65                 size_t orig_start,
  66                 const char *path)
  67 {
  68         git_blame_hunk *hunk = git__calloc(1, sizeof(git_blame_hunk));
  69         if (!hunk) return NULL;
  70
  71         hunk->lines_in_hunk = lines;
  72         hunk->final_start_line_number = start;
  73         hunk->orig_start_line_number = orig_start;
  74         hunk->orig_path = path ? git__strdup(path) : NULL;
  75
  76         return hunk;
  77 }
  78
  79 static git_blame_hunk* dup_hunk(git_blame_hunk *hunk)
  80 {
  81         git_blame_hunk *newhunk = new_hunk(
  82                         hunk->final_start_line_number,
  83                         hunk->lines_in_hunk,
  84                         hunk->orig_start_line_number,
  85                         hunk->orig_path);
  86
  87         if (!newhunk)
  88                 return NULL;
  89
  90         git_oid_cpy(&newhunk->orig_commit_id, &hunk->orig_commit_id);
  91         git_oid_cpy(&newhunk->final_commit_id, &hunk->final_commit_id);
  92         newhunk->boundary = hunk->boundary;
  93         git_signature_dup(&newhunk->final_signature, hunk->final_signature);
  94         git_signature_dup(&newhunk->orig_signature, hunk->orig_signature);
  95         return newhunk;
  96 }
  97
  98 static void free_hunk(git_blame_hunk *hunk)
  99 {
 100         git__free((void*)hunk->orig_path);
 101         git_signature_free(hunk->final_signature);
 102         git_signature_free(hunk->orig_signature);
 103         git__free(hunk);
 104 }
 105
 106 /* Starting with the hunk that includes start_line, shift all following hunks'
 107  * final_start_line by shift_by lines */
 108 static void shift_hunks_by(git_vector *v, size_t start_line, int shift_by)
 109 {
 110         size_t i;
 111
 112         if (!git_vector_bsearch2(&i, v, hunk_byfinalline_search_cmp, &start_line)) {
 113                 for (; i < v->length; i++) {
 114                         git_blame_hunk *hunk = (git_blame_hunk*)v->contents[i];
 115                         hunk->final_start_line_number += shift_by;
 116                 }
 117         }
 118 }
 119
 120 git_blame* git_blame__alloc(
 121         git_repository *repo,
 122         git_blame_options opts,
 123         const char *path)
 124 {
 125         git_blame *gbr = git__calloc(1, sizeof(git_blame));
 126         if (!gbr)
 127                 return NULL;
 128
 129         gbr->repository = repo;
 130         gbr->options = opts;
 131
 132         if (git_vector_init(&gbr->hunks, 8, hunk_cmp) < 0 ||
 133                 git_vector_init(&gbr->paths, 8, paths_cmp) < 0 ||
 134                 (gbr->path = git__strdup(path)) == NULL ||
 135                 git_vector_insert(&gbr->paths, git__strdup(path)) < 0)
 136         {
 137                 git_blame_free(gbr);
 138                 return NULL;
 139         }
 140
 141         if (opts.flags & GIT_BLAME_USE_MAILMAP &&
 142             git_mailmap_from_repository(&gbr->mailmap, repo) < 0) {
 143                 git_blame_free(gbr);
 144                 return NULL;
 145         }
 146
 147         return gbr;
 148 }
 149
 150 void git_blame_free(git_blame *blame)
 151 {
 152         size_t i;
 153         git_blame_hunk *hunk;
 154
 155         if (!blame) return;
 156
 157         git_vector_foreach(&blame->hunks, i, hunk)
 158                 free_hunk(hunk);
 159         git_vector_free(&blame->hunks);
 160
 161         git_vector_free_deep(&blame->paths);
 162
 163         git_array_clear(blame->line_index);
 164
 165         git_mailmap_free(blame->mailmap);
 166
 167         git__free(blame->path);
 168         git_blob_free(blame->final_blob);
 169         git__free(blame);
 170 }
 171
 172 uint32_t git_blame_get_hunk_count(git_blame *blame)
 173 {
 174         assert(blame);
 175         return (uint32_t)blame->hunks.length;
 176 }
 177
 178 const git_blame_hunk *git_blame_get_hunk_byindex(git_blame *blame, uint32_t index)
 179 {
 180         assert(blame);
 181         return (git_blame_hunk*)git_vector_get(&blame->hunks, index);
 182 }
 183
 184 const git_blame_hunk *git_blame_get_hunk_byline(git_blame *blame, size_t lineno)
 185 {
 186         size_t i, new_lineno = lineno;
 187         assert(blame);
 188
 189         if (!git_vector_bsearch2(&i, &blame->hunks, hunk_byfinalline_search_cmp, &new_lineno)) {
 190                 return git_blame_get_hunk_byindex(blame, (uint32_t)i);
 191         }
 192
 193         return NULL;
 194 }
 195
 196 static int normalize_options(
 197                 git_blame_options *out,
 198                 const git_blame_options *in,
 199                 git_repository *repo)
 200 {
 201         git_blame_options dummy = GIT_BLAME_OPTIONS_INIT;
 202         if (!in) in = &dummy;
 203
 204         memcpy(out, in, sizeof(git_blame_options));
 205
 206         /* No newest_commit => HEAD */
 207         if (git_oid_iszero(&out->newest_commit)) {
 208                 if (git_reference_name_to_id(&out->newest_commit, repo, "HEAD") < 0) {
 209                         return -1;
 210                 }
 211         }
 212
 213         /* min_line 0 really means 1 */
 214         if (!out->min_line) out->min_line = 1;
 215         /* max_line 0 really means N, but we don't know N yet */
 216
 217         /* Fix up option implications */
 218         if (out->flags & GIT_BLAME_TRACK_COPIES_ANY_COMMIT_COPIES)
 219                 out->flags |= GIT_BLAME_TRACK_COPIES_SAME_COMMIT_COPIES;
 220         if (out->flags & GIT_BLAME_TRACK_COPIES_SAME_COMMIT_COPIES)
 221                 out->flags |= GIT_BLAME_TRACK_COPIES_SAME_COMMIT_MOVES;
 222         if (out->flags & GIT_BLAME_TRACK_COPIES_SAME_COMMIT_MOVES)
 223                 out->flags |= GIT_BLAME_TRACK_COPIES_SAME_FILE;
 224
 225         return 0;
 226 }
 227
 228 static git_blame_hunk *split_hunk_in_vector(
 229                 git_vector *vec,
 230                 git_blame_hunk *hunk,
 231                 size_t rel_line,
 232                 bool return_new)
 233 {
 234         size_t new_line_count;
 235         git_blame_hunk *nh;
 236
 237         /* Don't split if already at a boundary */
 238         if (rel_line <= 0 ||
 239             rel_line >= hunk->lines_in_hunk)
 240         {
 241                 return hunk;
 242         }
 243
 244         new_line_count = hunk->lines_in_hunk - rel_line;
 245         nh = new_hunk(hunk->final_start_line_number + rel_line, new_line_count,
 246                         hunk->orig_start_line_number + rel_line, hunk->orig_path);
 247
 248         if (!nh)
 249                 return NULL;
 250
 251         git_oid_cpy(&nh->final_commit_id, &hunk->final_commit_id);
 252         git_oid_cpy(&nh->orig_commit_id, &hunk->orig_commit_id);
 253
 254         /* Adjust hunk that was split */
 255         hunk->lines_in_hunk -= new_line_count;
 256         git_vector_insert_sorted(vec, nh, NULL);
 257         {
 258                 git_blame_hunk *ret = return_new ? nh : hunk;
 259                 return ret;
 260         }
 261 }
 262
 263 /*
 264  * Construct a list of char indices for where lines begin
 265  * Adapted from core git:
 266  * https://github.com/gitster/git/blob/be5c9fb9049ed470e7005f159bb923a5f4de1309/builtin/blame.c#L1760-L1789
 267  */
 268 static int index_blob_lines(git_blame *blame)
 269 {
 270     const char *buf = blame->final_buf;
 271     git_off_t len = blame->final_buf_size;
 272     int num = 0, incomplete = 0, bol = 1;
 273     size_t *i;
 274
 275     if (len && buf[len-1] != '\n')
 276         incomplete++; /* incomplete line at the end */
 277     while (len--) {
 278         if (bol) {
 279             i = git_array_alloc(blame->line_index);
 280             GIT_ERROR_CHECK_ALLOC(i);
 281             *i = buf - blame->final_buf;
 282             bol = 0;
 283         }
 284         if (*buf++ == '\n') {
 285             num++;
 286             bol = 1;
 287         }
 288     }
 289     i = git_array_alloc(blame->line_index);
 290     GIT_ERROR_CHECK_ALLOC(i);
 291     *i = buf - blame->final_buf;
 292     blame->num_lines = num + incomplete;
 293     return blame->num_lines;
 294 }
 295
 296 static git_blame_hunk* hunk_from_entry(git_blame__entry *e, git_blame *blame)
 297 {
 298         git_blame_hunk *h = new_hunk(
 299                         e->lno+1, e->num_lines, e->s_lno+1, e->suspect->path);
 300
 301         if (!h)
 302                 return NULL;
 303
 304         git_oid_cpy(&h->final_commit_id, git_commit_id(e->suspect->commit));
 305         git_oid_cpy(&h->orig_commit_id, git_commit_id(e->suspect->commit));
 306         git_commit_author_with_mailmap(
 307                 &h->final_signature, e->suspect->commit, blame->mailmap);
 308         git_signature_dup(&h->orig_signature, h->final_signature);
 309         h->boundary = e->is_boundary ? 1 : 0;
 310         return h;
 311 }
 312
 313 static int load_blob(git_blame *blame)
 314 {
 315         int error;
 316
 317         if (blame->final_blob) return 0;
 318
 319         error = git_commit_lookup(&blame->final, blame->repository, &blame->options.newest_commit);
 320         if (error < 0)
 321                 goto cleanup;
 322         error = git_object_lookup_bypath((git_object**)&blame->final_blob,
 323                         (git_object*)blame->final, blame->path, GIT_OBJECT_BLOB);
 324
 325 cleanup:
 326         return error;
 327 }
 328
 329 static int blame_internal(git_blame *blame)
 330 {
 331         int error;
 332         git_blame__entry *ent = NULL;
 333         git_blame__origin *o;
 334
 335         if ((error = load_blob(blame)) < 0 ||
 336             (error = git_blame__get_origin(&o, blame, blame->final, blame->path)) < 0)
 337                 goto cleanup;
 338         blame->final_buf = git_blob_rawcontent(blame->final_blob);
 339         blame->final_buf_size = git_blob_rawsize(blame->final_blob);
 340
 341         ent = git__calloc(1, sizeof(git_blame__entry));
 342         GIT_ERROR_CHECK_ALLOC(ent);
 343
 344         ent->num_lines = index_blob_lines(blame);
 345         ent->lno = blame->options.min_line - 1;
 346         ent->num_lines = ent->num_lines - blame->options.min_line + 1;
 347         if (blame->options.max_line > 0)
 348                 ent->num_lines = blame->options.max_line - blame->options.min_line + 1;
 349         ent->s_lno = ent->lno;
 350         ent->suspect = o;
 351
 352         blame->ent = ent;
 353
 354         error = git_blame__like_git(blame, blame->options.flags);
 355
 356 cleanup:
 357         for (ent = blame->ent; ent; ) {
 358                 git_blame__entry *e = ent->next;
 359                 git_blame_hunk *h = hunk_from_entry(ent, blame);
 360
 361                 git_vector_insert(&blame->hunks, h);
 362
 363                 git_blame__free_entry(ent);
 364                 ent = e;
 365         }
 366
 367         return error;
 368 }
 369
 370 /*******************************************************************************
 371  * File blaming
 372  ******************************************************************************/
 373
 374 int git_blame_file(
 375                 git_blame **out,
 376                 git_repository *repo,
 377                 const char *path,
 378                 git_blame_options *options)
 379 {
 380         int error = -1;
 381         git_blame_options normOptions = GIT_BLAME_OPTIONS_INIT;
 382         git_blame *blame = NULL;
 383
 384         assert(out && repo && path);
 385         if ((error = normalize_options(&normOptions, options, repo)) < 0)
 386                 goto on_error;
 387
 388         blame = git_blame__alloc(repo, normOptions, path);
 389         GIT_ERROR_CHECK_ALLOC(blame);
 390
 391         if ((error = load_blob(blame)) < 0)
 392                 goto on_error;
 393
 394         if ((error = blame_internal(blame)) < 0)
 395                 goto on_error;
 396
 397         *out = blame;
 398         return 0;
 399
 400 on_error:
 401         git_blame_free(blame);
 402         return error;
 403 }
 404
 405 /*******************************************************************************
 406  * Buffer blaming
 407  *******************************************************************************/
 408
 409 static bool hunk_is_bufferblame(git_blame_hunk *hunk)
 410 {
 411         return git_oid_iszero(&hunk->final_commit_id);
 412 }
 413
 414 static int buffer_hunk_cb(
 415         const git_diff_delta *delta,
 416         const git_diff_hunk *hunk,
 417         void *payload)
 418 {
 419         git_blame *blame = (git_blame*)payload;
 420         uint32_t wedge_line;
 421
 422         GIT_UNUSED(delta);
 423
 424         wedge_line = (hunk->old_lines == 0) ? hunk->new_start : hunk->old_start;
 425         blame->current_diff_line = wedge_line;
 426
 427         blame->current_hunk = (git_blame_hunk*)git_blame_get_hunk_byline(blame, wedge_line);
 428         if (!blame->current_hunk) {
 429                 /* Line added at the end of the file */
 430                 blame->current_hunk = new_hunk(wedge_line, 0, wedge_line, blame->path);
 431                 GIT_ERROR_CHECK_ALLOC(blame->current_hunk);
 432
 433                 git_vector_insert(&blame->hunks, blame->current_hunk);
 434         } else if (!hunk_starts_at_or_after_line(blame->current_hunk, wedge_line)){
 435                 /* If this hunk doesn't start between existing hunks, split a hunk up so it does */
 436                 blame->current_hunk = split_hunk_in_vector(&blame->hunks, blame->current_hunk,
 437                                 wedge_line - blame->current_hunk->orig_start_line_number, true);
 438                 GIT_ERROR_CHECK_ALLOC(blame->current_hunk);
 439         }
 440
 441         return 0;
 442 }
 443
 444 static int ptrs_equal_cmp(const void *a, const void *b) { return a<b ? -1 : a>b ? 1 : 0; }
 445 static int buffer_line_cb(
 446         const git_diff_delta *delta,
 447         const git_diff_hunk *hunk,
 448         const git_diff_line *line,
 449         void *payload)
 450 {
 451         git_blame *blame = (git_blame*)payload;
 452
 453         GIT_UNUSED(delta);
 454         GIT_UNUSED(hunk);
 455         GIT_UNUSED(line);
 456
 457         if (line->origin == GIT_DIFF_LINE_ADDITION) {
 458                 if (hunk_is_bufferblame(blame->current_hunk) &&
 459                     hunk_ends_at_or_before_line(blame->current_hunk, blame->current_diff_line)) {
 460                         /* Append to the current buffer-blame hunk */
 461                         blame->current_hunk->lines_in_hunk++;
 462                         shift_hunks_by(&blame->hunks, blame->current_diff_line+1, 1);
 463                 } else {
 464                         /* Create a new buffer-blame hunk with this line */
 465                         shift_hunks_by(&blame->hunks, blame->current_diff_line, 1);
 466                         blame->current_hunk = new_hunk(blame->current_diff_line, 1, 0, blame->path);
 467                         GIT_ERROR_CHECK_ALLOC(blame->current_hunk);
 468
 469                         git_vector_insert_sorted(&blame->hunks, blame->current_hunk, NULL);
 470                 }
 471                 blame->current_diff_line++;
 472         }
 473
 474         if (line->origin == GIT_DIFF_LINE_DELETION) {
 475                 /* Trim the line from the current hunk; remove it if it's now empty */
 476                 size_t shift_base = blame->current_diff_line + blame->current_hunk->lines_in_hunk+1;
 477
 478                 if (--(blame->current_hunk->lines_in_hunk) == 0) {
 479                         size_t i;
 480                         shift_base--;
 481                         if (!git_vector_search2(&i, &blame->hunks, ptrs_equal_cmp, blame->current_hunk)) {
 482                                 git_vector_remove(&blame->hunks, i);
 483                                 free_hunk(blame->current_hunk);
 484                                 blame->current_hunk = (git_blame_hunk*)git_blame_get_hunk_byindex(blame, (uint32_t)i);
 485                         }
 486                 }
 487                 shift_hunks_by(&blame->hunks, shift_base, -1);
 488         }
 489         return 0;
 490 }
 491
 492 int git_blame_buffer(
 493                 git_blame **out,
 494                 git_blame *reference,
 495                 const char *buffer,
 496                 size_t buffer_len)
 497 {
 498         git_blame *blame;
 499         git_diff_options diffopts = GIT_DIFF_OPTIONS_INIT;
 500         size_t i;
 501         git_blame_hunk *hunk;
 502
 503         diffopts.context_lines = 0;
 504
 505         assert(out && reference && buffer && buffer_len);
 506
 507         blame = git_blame__alloc(reference->repository, reference->options, reference->path);
 508         GIT_ERROR_CHECK_ALLOC(blame);
 509
 510         /* Duplicate all of the hunk structures in the reference blame */
 511         git_vector_foreach(&reference->hunks, i, hunk) {
 512                 git_blame_hunk *h = dup_hunk(hunk);
 513                 GIT_ERROR_CHECK_ALLOC(h);
 514
 515                 git_vector_insert(&blame->hunks, h);
 516         }
 517
 518         /* Diff to the reference blob */
 519         git_diff_blob_to_buffer(reference->final_blob, blame->path,
 520                 buffer, buffer_len, blame->path, &diffopts,
 521                 NULL, NULL, buffer_hunk_cb, buffer_line_cb, blame);
 522
 523         *out = blame;
 524         return 0;
 525 }
 526
 527 int git_blame_init_options(git_blame_options *opts, unsigned int version)
 528 {
 529         GIT_INIT_STRUCTURE_FROM_TEMPLATE(
 530                 opts, version, git_blame_options, GIT_BLAME_OPTIONS_INIT);
 531         return 0;
 532 }