src/patch_generate.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 #include "common.h"
   8 #include "git2/blob.h"
   9 #include "diff.h"
  10 #include "diff_generate.h"
  11 #include "diff_file.h"
  12 #include "diff_driver.h"
  13 #include "patch_generate.h"
  14 #include "diff_xdiff.h"
  15 #include "delta.h"
  16 #include "zstream.h"
  17 #include "fileops.h"
  18
  19 static void diff_output_init(
  20         git_patch_generated_output *, const git_diff_options *, git_diff_file_cb,
  21         git_diff_binary_cb, git_diff_hunk_cb, git_diff_line_cb, void*);
  22
  23 static void diff_output_to_patch(
  24         git_patch_generated_output *, git_patch_generated *);
  25
  26 static void patch_generated_free(git_patch *p)
  27 {
  28         git_patch_generated *patch = (git_patch_generated *)p;
  29
  30         git_array_clear(patch->base.lines);
  31         git_array_clear(patch->base.hunks);
  32
  33         git__free((char *)patch->base.binary.old_file.data);
  34         git__free((char *)patch->base.binary.new_file.data);
  35
  36         git_diff_file_content__clear(&patch->ofile);
  37         git_diff_file_content__clear(&patch->nfile);
  38
  39         git_diff_free(patch->diff); /* decrements refcount */
  40         patch->diff = NULL;
  41
  42         git_pool_clear(&patch->flattened);
  43
  44         git__free((char *)patch->base.diff_opts.old_prefix);
  45         git__free((char *)patch->base.diff_opts.new_prefix);
  46
  47         if (patch->flags & GIT_PATCH_GENERATED_ALLOCATED)
  48                 git__free(patch);
  49 }
  50
  51 static void patch_generated_update_binary(git_patch_generated *patch)
  52 {
  53         if ((patch->base.delta->flags & DIFF_FLAGS_KNOWN_BINARY) != 0)
  54                 return;
  55
  56         if ((patch->ofile.file->flags & GIT_DIFF_FLAG_BINARY) != 0 ||
  57                 (patch->nfile.file->flags & GIT_DIFF_FLAG_BINARY) != 0)
  58                 patch->base.delta->flags |= GIT_DIFF_FLAG_BINARY;
  59
  60         else if (patch->ofile.file->size > GIT_XDIFF_MAX_SIZE ||
  61                 patch->nfile.file->size > GIT_XDIFF_MAX_SIZE)
  62                 patch->base.delta->flags |= GIT_DIFF_FLAG_BINARY;
  63
  64         else if ((patch->ofile.file->flags & DIFF_FLAGS_NOT_BINARY) != 0 &&
  65                 (patch->nfile.file->flags & DIFF_FLAGS_NOT_BINARY) != 0)
  66                 patch->base.delta->flags |= GIT_DIFF_FLAG_NOT_BINARY;
  67 }
  68
  69 static void patch_generated_init_common(git_patch_generated *patch)
  70 {
  71         patch->base.free_fn = patch_generated_free;
  72
  73         patch_generated_update_binary(patch);
  74
  75         patch->flags |= GIT_PATCH_GENERATED_INITIALIZED;
  76
  77         if (patch->diff)
  78                 git_diff_addref(patch->diff);
  79 }
  80
  81 static int patch_generated_normalize_options(
  82         git_diff_options *out,
  83         const git_diff_options *opts)
  84 {
  85         if (opts) {
  86                 GITERR_CHECK_VERSION(opts, GIT_DIFF_OPTIONS_VERSION, "git_diff_options");
  87                 memcpy(out, opts, sizeof(git_diff_options));
  88         } else {
  89                 git_diff_options default_opts = GIT_DIFF_OPTIONS_INIT;
  90                 memcpy(out, &default_opts, sizeof(git_diff_options));
  91         }
  92
  93         out->old_prefix = opts && opts->old_prefix ?
  94                 git__strdup(opts->old_prefix) :
  95                 git__strdup(DIFF_OLD_PREFIX_DEFAULT);
  96
  97         out->new_prefix = opts && opts->new_prefix ?
  98                 git__strdup(opts->new_prefix) :
  99                 git__strdup(DIFF_NEW_PREFIX_DEFAULT);
 100
 101         GITERR_CHECK_ALLOC(out->old_prefix);
 102         GITERR_CHECK_ALLOC(out->new_prefix);
 103
 104         return 0;
 105 }
 106
 107 static int patch_generated_init(
 108         git_patch_generated *patch, git_diff *diff, size_t delta_index)
 109 {
 110         int error = 0;
 111
 112         memset(patch, 0, sizeof(*patch));
 113
 114         patch->diff = diff;
 115         patch->base.repo = diff->repo;
 116         patch->base.delta = git_vector_get(&diff->deltas, delta_index);
 117         patch->delta_index = delta_index;
 118
 119         if ((error = patch_generated_normalize_options(
 120                         &patch->base.diff_opts, &diff->opts)) < 0 ||
 121                 (error = git_diff_file_content__init_from_diff(
 122                         &patch->ofile, diff, patch->base.delta, true)) < 0 ||
 123                 (error = git_diff_file_content__init_from_diff(
 124                         &patch->nfile, diff, patch->base.delta, false)) < 0)
 125                 return error;
 126
 127         patch_generated_init_common(patch);
 128
 129         return 0;
 130 }
 131
 132 static int patch_generated_alloc_from_diff(
 133         git_patch_generated **out, git_diff *diff, size_t delta_index)
 134 {
 135         int error;
 136         git_patch_generated *patch = git__calloc(1, sizeof(git_patch_generated));
 137         GITERR_CHECK_ALLOC(patch);
 138
 139         if (!(error = patch_generated_init(patch, diff, delta_index))) {
 140                 patch->flags |= GIT_PATCH_GENERATED_ALLOCATED;
 141                 GIT_REFCOUNT_INC(patch);
 142         } else {
 143                 git__free(patch);
 144                 patch = NULL;
 145         }
 146
 147         *out = patch;
 148         return error;
 149 }
 150
 151 GIT_INLINE(bool) should_skip_binary(git_patch_generated *patch, git_diff_file *file)
 152 {
 153         if ((patch->base.diff_opts.flags & GIT_DIFF_SHOW_BINARY) != 0)
 154                 return false;
 155
 156         return (file->flags & GIT_DIFF_FLAG_BINARY) != 0;
 157 }
 158
 159 static bool patch_generated_diffable(git_patch_generated *patch)
 160 {
 161         size_t olen, nlen;
 162
 163         if (patch->base.delta->status == GIT_DELTA_UNMODIFIED)
 164                 return false;
 165
 166         /* if we've determined this to be binary (and we are not showing binary
 167          * data) then we have skipped loading the map data.  instead, query the
 168          * file data itself.
 169          */
 170         if ((patch->base.delta->flags & GIT_DIFF_FLAG_BINARY) != 0 &&
 171                 (patch->base.diff_opts.flags & GIT_DIFF_SHOW_BINARY) == 0) {
 172                 olen = (size_t)patch->ofile.file->size;
 173                 nlen = (size_t)patch->nfile.file->size;
 174         } else {
 175                 olen = patch->ofile.map.len;
 176                 nlen = patch->nfile.map.len;
 177         }
 178
 179         /* if both sides are empty, files are identical */
 180         if (!olen && !nlen)
 181                 return false;
 182
 183         /* otherwise, check the file sizes and the oid */
 184         return (olen != nlen ||
 185                 !git_oid_equal(&patch->ofile.file->id, &patch->nfile.file->id));
 186 }
 187
 188 static int patch_generated_load(git_patch_generated *patch, git_patch_generated_output *output)
 189 {
 190         int error = 0;
 191         bool incomplete_data;
 192
 193         if ((patch->flags & GIT_PATCH_GENERATED_LOADED) != 0)
 194                 return 0;
 195
 196         /* if no hunk and data callbacks and user doesn't care if data looks
 197          * binary, then there is no need to actually load the data
 198          */
 199         if ((patch->ofile.opts_flags & GIT_DIFF_SKIP_BINARY_CHECK) != 0 &&
 200                 output && !output->binary_cb && !output->hunk_cb && !output->data_cb)
 201                 return 0;
 202
 203         incomplete_data =
 204                 (((patch->ofile.flags & GIT_DIFF_FLAG__NO_DATA) != 0 ||
 205                   (patch->ofile.file->flags & GIT_DIFF_FLAG_VALID_ID) != 0) &&
 206                  ((patch->nfile.flags & GIT_DIFF_FLAG__NO_DATA) != 0 ||
 207                   (patch->nfile.file->flags & GIT_DIFF_FLAG_VALID_ID) != 0));
 208
 209         /* always try to load workdir content first because filtering may
 210          * need 2x data size and this minimizes peak memory footprint
 211          */
 212         if (patch->ofile.src == GIT_ITERATOR_TYPE_WORKDIR) {
 213                 if ((error = git_diff_file_content__load(
 214                                 &patch->ofile, &patch->base.diff_opts)) < 0 ||
 215                         should_skip_binary(patch, patch->ofile.file))
 216                         goto cleanup;
 217         }
 218         if (patch->nfile.src == GIT_ITERATOR_TYPE_WORKDIR) {
 219                 if ((error = git_diff_file_content__load(
 220                                 &patch->nfile, &patch->base.diff_opts)) < 0 ||
 221                         should_skip_binary(patch, patch->nfile.file))
 222                         goto cleanup;
 223         }
 224
 225         /* once workdir has been tried, load other data as needed */
 226         if (patch->ofile.src != GIT_ITERATOR_TYPE_WORKDIR) {
 227                 if ((error = git_diff_file_content__load(
 228                                 &patch->ofile, &patch->base.diff_opts)) < 0 ||
 229                         should_skip_binary(patch, patch->ofile.file))
 230                         goto cleanup;
 231         }
 232         if (patch->nfile.src != GIT_ITERATOR_TYPE_WORKDIR) {
 233                 if ((error = git_diff_file_content__load(
 234                                 &patch->nfile, &patch->base.diff_opts)) < 0 ||
 235                         should_skip_binary(patch, patch->nfile.file))
 236                         goto cleanup;
 237         }
 238
 239         /* if previously missing an oid, and now that we have it the two sides
 240          * are the same (and not submodules), update MODIFIED -> UNMODIFIED
 241          */
 242         if (incomplete_data &&
 243                 patch->ofile.file->mode == patch->nfile.file->mode &&
 244                 patch->ofile.file->mode != GIT_FILEMODE_COMMIT &&
 245                 git_oid_equal(&patch->ofile.file->id, &patch->nfile.file->id) &&
 246                 patch->base.delta->status == GIT_DELTA_MODIFIED) /* not RENAMED/COPIED! */
 247                 patch->base.delta->status = GIT_DELTA_UNMODIFIED;
 248
 249 cleanup:
 250         patch_generated_update_binary(patch);
 251
 252         if (!error) {
 253                 if (patch_generated_diffable(patch))
 254                         patch->flags |= GIT_PATCH_GENERATED_DIFFABLE;
 255
 256                 patch->flags |= GIT_PATCH_GENERATED_LOADED;
 257         }
 258
 259         return error;
 260 }
 261
 262 static int patch_generated_invoke_file_callback(
 263         git_patch_generated *patch, git_patch_generated_output *output)
 264 {
 265         float progress = patch->diff ?
 266                 ((float)patch->delta_index / patch->diff->deltas.length) : 1.0f;
 267
 268         if (!output->file_cb)
 269                 return 0;
 270
 271         return giterr_set_after_callback_function(
 272                 output->file_cb(patch->base.delta, progress, output->payload),
 273                 "git_patch");
 274 }
 275
 276 static int create_binary(
 277         git_diff_binary_t *out_type,
 278         char **out_data,
 279         size_t *out_datalen,
 280         size_t *out_inflatedlen,
 281         const char *a_data,
 282         size_t a_datalen,
 283         const char *b_data,
 284         size_t b_datalen)
 285 {
 286         git_buf deflate = GIT_BUF_INIT, delta = GIT_BUF_INIT;
 287         size_t delta_data_len = 0;
 288         int error;
 289
 290         /* The git_delta function accepts unsigned long only */
 291         if (!git__is_ulong(a_datalen) || !git__is_ulong(b_datalen))
 292                 return GIT_EBUFS;
 293
 294         if ((error = git_zstream_deflatebuf(&deflate, b_data, b_datalen)) < 0)
 295                 goto done;
 296
 297         /* The git_delta function accepts unsigned long only */
 298         if (!git__is_ulong(deflate.size)) {
 299                 error = GIT_EBUFS;
 300                 goto done;
 301         }
 302
 303         if (a_datalen && b_datalen) {
 304                 void *delta_data;
 305
 306                 error = git_delta(&delta_data, &delta_data_len,
 307                         a_data, a_datalen,
 308                         b_data, b_datalen,
 309                         deflate.size);
 310
 311                 if (error == 0) {
 312                         error = git_zstream_deflatebuf(
 313                                 &delta, delta_data, delta_data_len);
 314
 315                         git__free(delta_data);
 316                 } else if (error == GIT_EBUFS) {
 317                         error = 0;
 318                 }
 319
 320                 if (error < 0)
 321                         goto done;
 322         }
 323
 324         if (delta.size && delta.size < deflate.size) {
 325                 *out_type = GIT_DIFF_BINARY_DELTA;
 326                 *out_datalen = delta.size;
 327                 *out_data = git_buf_detach(&delta);
 328                 *out_inflatedlen = delta_data_len;
 329         } else {
 330                 *out_type = GIT_DIFF_BINARY_LITERAL;
 331                 *out_datalen = deflate.size;
 332                 *out_data = git_buf_detach(&deflate);
 333                 *out_inflatedlen = b_datalen;
 334         }
 335
 336 done:
 337         git_buf_free(&deflate);
 338         git_buf_free(&delta);
 339
 340         return error;
 341 }
 342
 343 static int diff_binary(git_patch_generated_output *output, git_patch_generated *patch)
 344 {
 345         git_diff_binary binary = {0};
 346         const char *old_data = patch->ofile.map.data;
 347         const char *new_data = patch->nfile.map.data;
 348         size_t old_len = patch->ofile.map.len,
 349                 new_len = patch->nfile.map.len;
 350         int error;
 351
 352         /* Only load contents if the user actually wants to diff
 353          * binary files. */
 354         if (patch->base.diff_opts.flags & GIT_DIFF_SHOW_BINARY) {
 355                 binary.contains_data = 1;
 356
 357                 /* Create the old->new delta (as the "new" side of the patch),
 358                  * and the new->old delta (as the "old" side)
 359                  */
 360                 if ((error = create_binary(&binary.old_file.type,
 361                                 (char **)&binary.old_file.data,
 362                                 &binary.old_file.datalen,
 363                                 &binary.old_file.inflatedlen,
 364                                 new_data, new_len, old_data, old_len)) < 0 ||
 365                         (error = create_binary(&binary.new_file.type,
 366                                 (char **)&binary.new_file.data,
 367                                 &binary.new_file.datalen,
 368                                 &binary.new_file.inflatedlen,
 369                                 old_data, old_len, new_data, new_len)) < 0)
 370                         return error;
 371         }
 372
 373         error = giterr_set_after_callback_function(
 374                 output->binary_cb(patch->base.delta, &binary, output->payload),
 375                 "git_patch");
 376
 377         git__free((char *) binary.old_file.data);
 378         git__free((char *) binary.new_file.data);
 379
 380         return error;
 381 }
 382
 383 static int patch_generated_create(
 384         git_patch_generated *patch,
 385         git_patch_generated_output *output)
 386 {
 387         int error = 0;
 388
 389         if ((patch->flags & GIT_PATCH_GENERATED_DIFFED) != 0)
 390                 return 0;
 391
 392         /* if we are not looking at the binary or text data, don't do the diff */
 393         if (!output->binary_cb && !output->hunk_cb && !output->data_cb)
 394                 return 0;
 395
 396         if ((patch->flags & GIT_PATCH_GENERATED_LOADED) == 0 &&
 397                 (error = patch_generated_load(patch, output)) < 0)
 398                 return error;
 399
 400         if ((patch->flags & GIT_PATCH_GENERATED_DIFFABLE) == 0)
 401                 return 0;
 402
 403         if ((patch->base.delta->flags & GIT_DIFF_FLAG_BINARY) != 0) {
 404                 if (output->binary_cb)
 405                         error = diff_binary(output, patch);
 406         }
 407         else {
 408                 if (output->diff_cb)
 409                         error = output->diff_cb(output, patch);
 410         }
 411
 412         patch->flags |= GIT_PATCH_GENERATED_DIFFED;
 413         return error;
 414 }
 415
 416 static int diff_required(git_diff *diff, const char *action)
 417 {
 418         if (diff)
 419                 return 0;
 420         giterr_set(GITERR_INVALID, "Must provide valid diff to %s", action);
 421         return -1;
 422 }
 423
 424 int git_diff_foreach(
 425         git_diff *diff,
 426         git_diff_file_cb file_cb,
 427         git_diff_binary_cb binary_cb,
 428         git_diff_hunk_cb hunk_cb,
 429         git_diff_line_cb data_cb,
 430         void *payload)
 431 {
 432         int error = 0;
 433         git_xdiff_output xo;
 434         size_t idx;
 435         git_patch_generated patch;
 436
 437         if ((error = diff_required(diff, "git_diff_foreach")) < 0)
 438                 return error;
 439
 440         memset(&xo, 0, sizeof(xo));
 441         memset(&patch, 0, sizeof(patch));
 442         diff_output_init(
 443                 &xo.output, &diff->opts, file_cb, binary_cb, hunk_cb, data_cb, payload);
 444         git_xdiff_init(&xo, &diff->opts);
 445
 446         git_vector_foreach(&diff->deltas, idx, patch.base.delta) {
 447
 448                 /* check flags against patch status */
 449                 if (git_diff_delta__should_skip(&diff->opts, patch.base.delta))
 450                         continue;
 451
 452                 if (binary_cb || hunk_cb || data_cb) {
 453                         if ((error = patch_generated_init(&patch, diff, idx)) != 0 ||
 454                                 (error = patch_generated_load(&patch, &xo.output)) != 0)
 455                                 return error;
 456                 }
 457
 458                 if ((error = patch_generated_invoke_file_callback(&patch, &xo.output)) == 0) {
 459                         if (binary_cb || hunk_cb || data_cb)
 460                                         error = patch_generated_create(&patch, &xo.output);
 461                 }
 462
 463                 git_patch_free(&patch.base);
 464
 465                 if (error)
 466                         break;
 467         }
 468
 469         return error;
 470 }
 471
 472 typedef struct {
 473         git_patch_generated patch;
 474         git_diff_delta delta;
 475         char paths[GIT_FLEX_ARRAY];
 476 } patch_generated_with_delta;
 477
 478 static int diff_single_generate(patch_generated_with_delta *pd, git_xdiff_output *xo)
 479 {
 480         int error = 0;
 481         git_patch_generated *patch = &pd->patch;
 482         bool has_old = ((patch->ofile.flags & GIT_DIFF_FLAG__NO_DATA) == 0);
 483         bool has_new = ((patch->nfile.flags & GIT_DIFF_FLAG__NO_DATA) == 0);
 484
 485         pd->delta.status = has_new ?
 486                 (has_old ? GIT_DELTA_MODIFIED : GIT_DELTA_ADDED) :
 487                 (has_old ? GIT_DELTA_DELETED : GIT_DELTA_UNTRACKED);
 488
 489         if (git_oid_equal(&patch->nfile.file->id, &patch->ofile.file->id))
 490                 pd->delta.status = GIT_DELTA_UNMODIFIED;
 491
 492         patch->base.delta = &pd->delta;
 493
 494         patch_generated_init_common(patch);
 495
 496         if (pd->delta.status == GIT_DELTA_UNMODIFIED &&
 497                 !(patch->ofile.opts_flags & GIT_DIFF_INCLUDE_UNMODIFIED)) {
 498
 499                 /* Even empty patches are flagged as binary, and even though
 500                  * there's no difference, we flag this as "containing data"
 501                  * (the data is known to be empty, as opposed to wholly unknown).
 502                  */
 503                 if (patch->base.diff_opts.flags & GIT_DIFF_SHOW_BINARY)
 504                         patch->base.binary.contains_data = 1;
 505
 506                 return error;
 507         }
 508
 509         error = patch_generated_invoke_file_callback(patch, (git_patch_generated_output *)xo);
 510
 511         if (!error)
 512                 error = patch_generated_create(patch, (git_patch_generated_output *)xo);
 513
 514         return error;
 515 }
 516
 517 static int patch_generated_from_sources(
 518         patch_generated_with_delta *pd,
 519         git_xdiff_output *xo,
 520         git_diff_file_content_src *oldsrc,
 521         git_diff_file_content_src *newsrc,
 522         const git_diff_options *opts)
 523 {
 524         int error = 0;
 525         git_repository *repo =
 526                 oldsrc->blob ? git_blob_owner(oldsrc->blob) :
 527                 newsrc->blob ? git_blob_owner(newsrc->blob) : NULL;
 528         git_diff_file *lfile = &pd->delta.old_file, *rfile = &pd->delta.new_file;
 529         git_diff_file_content *ldata = &pd->patch.ofile, *rdata = &pd->patch.nfile;
 530
 531         if ((error = patch_generated_normalize_options(&pd->patch.base.diff_opts, opts)) < 0)
 532                 return error;
 533
 534         if (opts && (opts->flags & GIT_DIFF_REVERSE) != 0) {
 535                 void *tmp = lfile; lfile = rfile; rfile = tmp;
 536                 tmp = ldata; ldata = rdata; rdata = tmp;
 537         }
 538
 539         pd->patch.base.delta = &pd->delta;
 540
 541         if (!oldsrc->as_path) {
 542                 if (newsrc->as_path)
 543                         oldsrc->as_path = newsrc->as_path;
 544                 else
 545                         oldsrc->as_path = newsrc->as_path = "file";
 546         }
 547         else if (!newsrc->as_path)
 548                 newsrc->as_path = oldsrc->as_path;
 549
 550         lfile->path = oldsrc->as_path;
 551         rfile->path = newsrc->as_path;
 552
 553         if ((error = git_diff_file_content__init_from_src(
 554                         ldata, repo, opts, oldsrc, lfile)) < 0 ||
 555                 (error = git_diff_file_content__init_from_src(
 556                         rdata, repo, opts, newsrc, rfile)) < 0)
 557                 return error;
 558
 559         return diff_single_generate(pd, xo);
 560 }
 561
 562 static int patch_generated_with_delta_alloc(
 563         patch_generated_with_delta **out,
 564         const char **old_path,
 565         const char **new_path)
 566 {
 567         patch_generated_with_delta *pd;
 568         size_t old_len = *old_path ? strlen(*old_path) : 0;
 569         size_t new_len = *new_path ? strlen(*new_path) : 0;
 570         size_t alloc_len;
 571
 572         GITERR_CHECK_ALLOC_ADD(&alloc_len, sizeof(*pd), old_len);
 573         GITERR_CHECK_ALLOC_ADD(&alloc_len, alloc_len, new_len);
 574         GITERR_CHECK_ALLOC_ADD(&alloc_len, alloc_len, 2);
 575
 576         *out = pd = git__calloc(1, alloc_len);
 577         GITERR_CHECK_ALLOC(pd);
 578
 579         pd->patch.flags = GIT_PATCH_GENERATED_ALLOCATED;
 580
 581         if (*old_path) {
 582                 memcpy(&pd->paths[0], *old_path, old_len);
 583                 *old_path = &pd->paths[0];
 584         } else if (*new_path)
 585                 *old_path = &pd->paths[old_len + 1];
 586
 587         if (*new_path) {
 588                 memcpy(&pd->paths[old_len + 1], *new_path, new_len);
 589                 *new_path = &pd->paths[old_len + 1];
 590         } else if (*old_path)
 591                 *new_path = &pd->paths[0];
 592
 593         return 0;
 594 }
 595
 596 static int diff_from_sources(
 597         git_diff_file_content_src *oldsrc,
 598         git_diff_file_content_src *newsrc,
 599         const git_diff_options *opts,
 600         git_diff_file_cb file_cb,
 601         git_diff_binary_cb binary_cb,
 602         git_diff_hunk_cb hunk_cb,
 603         git_diff_line_cb data_cb,
 604         void *payload)
 605 {
 606         int error = 0;
 607         patch_generated_with_delta pd;
 608         git_xdiff_output xo;
 609
 610         memset(&xo, 0, sizeof(xo));
 611         diff_output_init(
 612                 &xo.output, opts, file_cb, binary_cb, hunk_cb, data_cb, payload);
 613         git_xdiff_init(&xo, opts);
 614
 615         memset(&pd, 0, sizeof(pd));
 616
 617         error = patch_generated_from_sources(&pd, &xo, oldsrc, newsrc, opts);
 618
 619         git_patch_free(&pd.patch.base);
 620
 621         return error;
 622 }
 623
 624 static int patch_from_sources(
 625         git_patch **out,
 626         git_diff_file_content_src *oldsrc,
 627         git_diff_file_content_src *newsrc,
 628         const git_diff_options *opts)
 629 {
 630         int error = 0;
 631         patch_generated_with_delta *pd;
 632         git_xdiff_output xo;
 633
 634         assert(out);
 635         *out = NULL;
 636
 637         if ((error = patch_generated_with_delta_alloc(
 638                         &pd, &oldsrc->as_path, &newsrc->as_path)) < 0)
 639                 return error;
 640
 641         memset(&xo, 0, sizeof(xo));
 642         diff_output_to_patch(&xo.output, &pd->patch);
 643         git_xdiff_init(&xo, opts);
 644
 645         if (!(error = patch_generated_from_sources(pd, &xo, oldsrc, newsrc, opts)))
 646                 *out = (git_patch *)pd;
 647         else
 648                 git_patch_free((git_patch *)pd);
 649
 650         return error;
 651 }
 652
 653 int git_diff_blobs(
 654         const git_blob *old_blob,
 655         const char *old_path,
 656         const git_blob *new_blob,
 657         const char *new_path,
 658         const git_diff_options *opts,
 659         git_diff_file_cb file_cb,
 660         git_diff_binary_cb binary_cb,
 661         git_diff_hunk_cb hunk_cb,
 662         git_diff_line_cb data_cb,
 663         void *payload)
 664 {
 665         git_diff_file_content_src osrc =
 666                 GIT_DIFF_FILE_CONTENT_SRC__BLOB(old_blob, old_path);
 667         git_diff_file_content_src nsrc =
 668                 GIT_DIFF_FILE_CONTENT_SRC__BLOB(new_blob, new_path);
 669         return diff_from_sources(
 670                 &osrc, &nsrc, opts, file_cb, binary_cb, hunk_cb, data_cb, payload);
 671 }
 672
 673 int git_patch_from_blobs(
 674         git_patch **out,
 675         const git_blob *old_blob,
 676         const char *old_path,
 677         const git_blob *new_blob,
 678         const char *new_path,
 679         const git_diff_options *opts)
 680 {
 681         git_diff_file_content_src osrc =
 682                 GIT_DIFF_FILE_CONTENT_SRC__BLOB(old_blob, old_path);
 683         git_diff_file_content_src nsrc =
 684                 GIT_DIFF_FILE_CONTENT_SRC__BLOB(new_blob, new_path);
 685         return patch_from_sources(out, &osrc, &nsrc, opts);
 686 }
 687
 688 int git_diff_blob_to_buffer(
 689         const git_blob *old_blob,
 690         const char *old_path,
 691         const char *buf,
 692         size_t buflen,
 693         const char *buf_path,
 694         const git_diff_options *opts,
 695         git_diff_file_cb file_cb,
 696         git_diff_binary_cb binary_cb,
 697         git_diff_hunk_cb hunk_cb,
 698         git_diff_line_cb data_cb,
 699         void *payload)
 700 {
 701         git_diff_file_content_src osrc =
 702                 GIT_DIFF_FILE_CONTENT_SRC__BLOB(old_blob, old_path);
 703         git_diff_file_content_src nsrc =
 704                 GIT_DIFF_FILE_CONTENT_SRC__BUF(buf, buflen, buf_path);
 705         return diff_from_sources(
 706                 &osrc, &nsrc, opts, file_cb, binary_cb, hunk_cb, data_cb, payload);
 707 }
 708
 709 int git_patch_from_blob_and_buffer(
 710         git_patch **out,
 711         const git_blob *old_blob,
 712         const char *old_path,
 713         const char *buf,
 714         size_t buflen,
 715         const char *buf_path,
 716         const git_diff_options *opts)
 717 {
 718         git_diff_file_content_src osrc =
 719                 GIT_DIFF_FILE_CONTENT_SRC__BLOB(old_blob, old_path);
 720         git_diff_file_content_src nsrc =
 721                 GIT_DIFF_FILE_CONTENT_SRC__BUF(buf, buflen, buf_path);
 722         return patch_from_sources(out, &osrc, &nsrc, opts);
 723 }
 724
 725 int git_diff_buffers(
 726         const void *old_buf,
 727         size_t old_len,
 728         const char *old_path,
 729         const void *new_buf,
 730         size_t new_len,
 731         const char *new_path,
 732         const git_diff_options *opts,
 733         git_diff_file_cb file_cb,
 734         git_diff_binary_cb binary_cb,
 735         git_diff_hunk_cb hunk_cb,
 736         git_diff_line_cb data_cb,
 737         void *payload)
 738 {
 739         git_diff_file_content_src osrc =
 740                 GIT_DIFF_FILE_CONTENT_SRC__BUF(old_buf, old_len, old_path);
 741         git_diff_file_content_src nsrc =
 742                 GIT_DIFF_FILE_CONTENT_SRC__BUF(new_buf, new_len, new_path);
 743         return diff_from_sources(
 744                 &osrc, &nsrc, opts, file_cb, binary_cb, hunk_cb, data_cb, payload);
 745 }
 746
 747 int git_patch_from_buffers(
 748         git_patch **out,
 749         const void *old_buf,
 750         size_t old_len,
 751         const char *old_path,
 752         const char *new_buf,
 753         size_t new_len,
 754         const char *new_path,
 755         const git_diff_options *opts)
 756 {
 757         git_diff_file_content_src osrc =
 758                 GIT_DIFF_FILE_CONTENT_SRC__BUF(old_buf, old_len, old_path);
 759         git_diff_file_content_src nsrc =
 760                 GIT_DIFF_FILE_CONTENT_SRC__BUF(new_buf, new_len, new_path);
 761         return patch_from_sources(out, &osrc, &nsrc, opts);
 762 }
 763
 764 int git_patch_generated_from_diff(
 765         git_patch **patch_ptr, git_diff *diff, size_t idx)
 766 {
 767         int error = 0;
 768         git_xdiff_output xo;
 769         git_diff_delta *delta = NULL;
 770         git_patch_generated *patch = NULL;
 771
 772         if (patch_ptr) *patch_ptr = NULL;
 773
 774         if (diff_required(diff, "git_patch_from_diff") < 0)
 775                 return -1;
 776
 777         delta = git_vector_get(&diff->deltas, idx);
 778         if (!delta) {
 779                 giterr_set(GITERR_INVALID, "Index out of range for delta in diff");
 780                 return GIT_ENOTFOUND;
 781         }
 782
 783         if (git_diff_delta__should_skip(&diff->opts, delta))
 784                 return 0;
 785
 786         /* don't load the patch data unless we need it for binary check */
 787         if (!patch_ptr &&
 788                 ((delta->flags & DIFF_FLAGS_KNOWN_BINARY) != 0 ||
 789                  (diff->opts.flags & GIT_DIFF_SKIP_BINARY_CHECK) != 0))
 790                 return 0;
 791
 792         if ((error = patch_generated_alloc_from_diff(&patch, diff, idx)) < 0)
 793                 return error;
 794
 795         memset(&xo, 0, sizeof(xo));
 796         diff_output_to_patch(&xo.output, patch);
 797         git_xdiff_init(&xo, &diff->opts);
 798
 799         error = patch_generated_invoke_file_callback(patch, &xo.output);
 800
 801         if (!error)
 802                 error = patch_generated_create(patch, &xo.output);
 803
 804         if (!error) {
 805                 /* TODO: if cumulative diff size is < 0.5 total size, flatten patch */
 806                 /* TODO: and unload the file content */
 807         }
 808
 809         if (error || !patch_ptr)
 810                 git_patch_free(&patch->base);
 811         else
 812                 *patch_ptr = &patch->base;
 813
 814         return error;
 815 }
 816
 817 git_diff_driver *git_patch_generated_driver(git_patch_generated *patch)
 818 {
 819         /* ofile driver is representative for whole patch */
 820         return patch->ofile.driver;
 821 }
 822
 823 void git_patch_generated_old_data(
 824         char **ptr, size_t *len, git_patch_generated *patch)
 825 {
 826         *ptr = patch->ofile.map.data;
 827         *len = patch->ofile.map.len;
 828 }
 829
 830 void git_patch_generated_new_data(
 831         char **ptr, size_t *len, git_patch_generated *patch)
 832 {
 833         *ptr = patch->nfile.map.data;
 834         *len = patch->nfile.map.len;
 835 }
 836
 837 static int patch_generated_file_cb(
 838         const git_diff_delta *delta,
 839         float progress,
 840         void *payload)
 841 {
 842         GIT_UNUSED(delta); GIT_UNUSED(progress); GIT_UNUSED(payload);
 843         return 0;
 844 }
 845
 846 static int patch_generated_binary_cb(
 847         const git_diff_delta *delta,
 848         const git_diff_binary *binary,
 849         void *payload)
 850 {
 851         git_patch *patch = payload;
 852
 853         GIT_UNUSED(delta);
 854
 855         memcpy(&patch->binary, binary, sizeof(git_diff_binary));
 856
 857         if (binary->old_file.data) {
 858                 patch->binary.old_file.data = git__malloc(binary->old_file.datalen);
 859                 GITERR_CHECK_ALLOC(patch->binary.old_file.data);
 860
 861                 memcpy((char *)patch->binary.old_file.data,
 862                         binary->old_file.data, binary->old_file.datalen);
 863         }
 864
 865         if (binary->new_file.data) {
 866                 patch->binary.new_file.data = git__malloc(binary->new_file.datalen);
 867                 GITERR_CHECK_ALLOC(patch->binary.new_file.data);
 868
 869                 memcpy((char *)patch->binary.new_file.data,
 870                         binary->new_file.data, binary->new_file.datalen);
 871         }
 872
 873         return 0;
 874 }
 875
 876 static int git_patch_hunk_cb(
 877         const git_diff_delta *delta,
 878         const git_diff_hunk *hunk_,
 879         void *payload)
 880 {
 881         git_patch_generated  *patch = payload;
 882         git_patch_hunk *hunk;
 883
 884         GIT_UNUSED(delta);
 885
 886         hunk = git_array_alloc(patch->base.hunks);
 887         GITERR_CHECK_ALLOC(hunk);
 888
 889         memcpy(&hunk->hunk, hunk_, sizeof(hunk->hunk));
 890
 891         patch->base.header_size += hunk_->header_len;
 892
 893         hunk->line_start = git_array_size(patch->base.lines);
 894         hunk->line_count = 0;
 895
 896         return 0;
 897 }
 898
 899 static int patch_generated_line_cb(
 900         const git_diff_delta *delta,
 901         const git_diff_hunk *hunk_,
 902         const git_diff_line *line_,
 903         void *payload)
 904 {
 905         git_patch_generated  *patch = payload;
 906         git_patch_hunk *hunk;
 907         git_diff_line   *line;
 908
 909         GIT_UNUSED(delta);
 910         GIT_UNUSED(hunk_);
 911
 912         hunk = git_array_last(patch->base.hunks);
 913         assert(hunk); /* programmer error if no hunk is available */
 914
 915         line = git_array_alloc(patch->base.lines);
 916         GITERR_CHECK_ALLOC(line);
 917
 918         memcpy(line, line_, sizeof(*line));
 919
 920         /* do some bookkeeping so we can provide old/new line numbers */
 921
 922         patch->base.content_size += line->content_len;
 923
 924         if (line->origin == GIT_DIFF_LINE_ADDITION ||
 925                 line->origin == GIT_DIFF_LINE_DELETION)
 926                 patch->base.content_size += 1;
 927         else if (line->origin == GIT_DIFF_LINE_CONTEXT) {
 928                 patch->base.content_size += 1;
 929                 patch->base.context_size += line->content_len + 1;
 930         } else if (line->origin == GIT_DIFF_LINE_CONTEXT_EOFNL)
 931                 patch->base.context_size += line->content_len;
 932
 933         hunk->line_count++;
 934
 935         return 0;
 936 }
 937
 938 static void diff_output_init(
 939         git_patch_generated_output *out,
 940         const git_diff_options *opts,
 941         git_diff_file_cb file_cb,
 942         git_diff_binary_cb binary_cb,
 943         git_diff_hunk_cb hunk_cb,
 944         git_diff_line_cb data_cb,
 945         void *payload)
 946 {
 947         GIT_UNUSED(opts);
 948
 949         memset(out, 0, sizeof(*out));
 950
 951         out->file_cb = file_cb;
 952         out->binary_cb = binary_cb;
 953         out->hunk_cb = hunk_cb;
 954         out->data_cb = data_cb;
 955         out->payload = payload;
 956 }
 957
 958 static void diff_output_to_patch(
 959         git_patch_generated_output *out, git_patch_generated *patch)
 960 {
 961         diff_output_init(
 962                 out,
 963                 NULL,
 964                 patch_generated_file_cb,
 965                 patch_generated_binary_cb,
 966                 git_patch_hunk_cb,
 967                 patch_generated_line_cb,
 968                 patch);
 969 }