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         if ((error = git_diff_file_content__load(
 210                         &patch->ofile, &patch->base.diff_opts)) < 0 ||
 211                 should_skip_binary(patch, patch->ofile.file))
 212                 goto cleanup;
 213         if ((error = git_diff_file_content__load(
 214                         &patch->nfile, &patch->base.diff_opts)) < 0 ||
 215                 should_skip_binary(patch, patch->nfile.file))
 216                 goto cleanup;
 217
 218         /* if previously missing an oid, and now that we have it the two sides
 219          * are the same (and not submodules), update MODIFIED -> UNMODIFIED
 220          */
 221         if (incomplete_data &&
 222                 patch->ofile.file->mode == patch->nfile.file->mode &&
 223                 patch->ofile.file->mode != GIT_FILEMODE_COMMIT &&
 224                 git_oid_equal(&patch->ofile.file->id, &patch->nfile.file->id) &&
 225                 patch->base.delta->status == GIT_DELTA_MODIFIED) /* not RENAMED/COPIED! */
 226                 patch->base.delta->status = GIT_DELTA_UNMODIFIED;
 227
 228 cleanup:
 229         patch_generated_update_binary(patch);
 230
 231         if (!error) {
 232                 if (patch_generated_diffable(patch))
 233                         patch->flags |= GIT_PATCH_GENERATED_DIFFABLE;
 234
 235                 patch->flags |= GIT_PATCH_GENERATED_LOADED;
 236         }
 237
 238         return error;
 239 }
 240
 241 static int patch_generated_invoke_file_callback(
 242         git_patch_generated *patch, git_patch_generated_output *output)
 243 {
 244         float progress = patch->diff ?
 245                 ((float)patch->delta_index / patch->diff->deltas.length) : 1.0f;
 246
 247         if (!output->file_cb)
 248                 return 0;
 249
 250         return giterr_set_after_callback_function(
 251                 output->file_cb(patch->base.delta, progress, output->payload),
 252                 "git_patch");
 253 }
 254
 255 static int create_binary(
 256         git_diff_binary_t *out_type,
 257         char **out_data,
 258         size_t *out_datalen,
 259         size_t *out_inflatedlen,
 260         const char *a_data,
 261         size_t a_datalen,
 262         const char *b_data,
 263         size_t b_datalen)
 264 {
 265         git_buf deflate = GIT_BUF_INIT, delta = GIT_BUF_INIT;
 266         size_t delta_data_len = 0;
 267         int error;
 268
 269         /* The git_delta function accepts unsigned long only */
 270         if (!git__is_ulong(a_datalen) || !git__is_ulong(b_datalen))
 271                 return GIT_EBUFS;
 272
 273         if ((error = git_zstream_deflatebuf(&deflate, b_data, b_datalen)) < 0)
 274                 goto done;
 275
 276         /* The git_delta function accepts unsigned long only */
 277         if (!git__is_ulong(deflate.size)) {
 278                 error = GIT_EBUFS;
 279                 goto done;
 280         }
 281
 282         if (a_datalen && b_datalen) {
 283                 void *delta_data;
 284
 285                 error = git_delta(&delta_data, &delta_data_len,
 286                         a_data, a_datalen,
 287                         b_data, b_datalen,
 288                         deflate.size);
 289
 290                 if (error == 0) {
 291                         error = git_zstream_deflatebuf(
 292                                 &delta, delta_data, delta_data_len);
 293
 294                         git__free(delta_data);
 295                 } else if (error == GIT_EBUFS) {
 296                         error = 0;
 297                 }
 298
 299                 if (error < 0)
 300                         goto done;
 301         }
 302
 303         if (delta.size && delta.size < deflate.size) {
 304                 *out_type = GIT_DIFF_BINARY_DELTA;
 305                 *out_datalen = delta.size;
 306                 *out_data = git_buf_detach(&delta);
 307                 *out_inflatedlen = delta_data_len;
 308         } else {
 309                 *out_type = GIT_DIFF_BINARY_LITERAL;
 310                 *out_datalen = deflate.size;
 311                 *out_data = git_buf_detach(&deflate);
 312                 *out_inflatedlen = b_datalen;
 313         }
 314
 315 done:
 316         git_buf_free(&deflate);
 317         git_buf_free(&delta);
 318
 319         return error;
 320 }
 321
 322 static int diff_binary(git_patch_generated_output *output, git_patch_generated *patch)
 323 {
 324         git_diff_binary binary = {0};
 325         const char *old_data = patch->ofile.map.data;
 326         const char *new_data = patch->nfile.map.data;
 327         size_t old_len = patch->ofile.map.len,
 328                 new_len = patch->nfile.map.len;
 329         int error;
 330
 331         /* Only load contents if the user actually wants to diff
 332          * binary files. */
 333         if (patch->base.diff_opts.flags & GIT_DIFF_SHOW_BINARY) {
 334                 binary.contains_data = 1;
 335
 336                 /* Create the old->new delta (as the "new" side of the patch),
 337                  * and the new->old delta (as the "old" side)
 338                  */
 339                 if ((error = create_binary(&binary.old_file.type,
 340                                 (char **)&binary.old_file.data,
 341                                 &binary.old_file.datalen,
 342                                 &binary.old_file.inflatedlen,
 343                                 new_data, new_len, old_data, old_len)) < 0 ||
 344                         (error = create_binary(&binary.new_file.type,
 345                                 (char **)&binary.new_file.data,
 346                                 &binary.new_file.datalen,
 347                                 &binary.new_file.inflatedlen,
 348                                 old_data, old_len, new_data, new_len)) < 0)
 349                         return error;
 350         }
 351
 352         error = giterr_set_after_callback_function(
 353                 output->binary_cb(patch->base.delta, &binary, output->payload),
 354                 "git_patch");
 355
 356         git__free((char *) binary.old_file.data);
 357         git__free((char *) binary.new_file.data);
 358
 359         return error;
 360 }
 361
 362 static int patch_generated_create(
 363         git_patch_generated *patch,
 364         git_patch_generated_output *output)
 365 {
 366         int error = 0;
 367
 368         if ((patch->flags & GIT_PATCH_GENERATED_DIFFED) != 0)
 369                 return 0;
 370
 371         /* if we are not looking at the binary or text data, don't do the diff */
 372         if (!output->binary_cb && !output->hunk_cb && !output->data_cb)
 373                 return 0;
 374
 375         if ((patch->flags & GIT_PATCH_GENERATED_LOADED) == 0 &&
 376                 (error = patch_generated_load(patch, output)) < 0)
 377                 return error;
 378
 379         if ((patch->flags & GIT_PATCH_GENERATED_DIFFABLE) == 0)
 380                 return 0;
 381
 382         if ((patch->base.delta->flags & GIT_DIFF_FLAG_BINARY) != 0) {
 383                 if (output->binary_cb)
 384                         error = diff_binary(output, patch);
 385         }
 386         else {
 387                 if (output->diff_cb)
 388                         error = output->diff_cb(output, patch);
 389         }
 390
 391         patch->flags |= GIT_PATCH_GENERATED_DIFFED;
 392         return error;
 393 }
 394
 395 static int diff_required(git_diff *diff, const char *action)
 396 {
 397         if (diff)
 398                 return 0;
 399         giterr_set(GITERR_INVALID, "must provide valid diff to %s", action);
 400         return -1;
 401 }
 402
 403 typedef struct {
 404         git_patch_generated patch;
 405         git_diff_delta delta;
 406         char paths[GIT_FLEX_ARRAY];
 407 } patch_generated_with_delta;
 408
 409 static int diff_single_generate(patch_generated_with_delta *pd, git_xdiff_output *xo)
 410 {
 411         int error = 0;
 412         git_patch_generated *patch = &pd->patch;
 413         bool has_old = ((patch->ofile.flags & GIT_DIFF_FLAG__NO_DATA) == 0);
 414         bool has_new = ((patch->nfile.flags & GIT_DIFF_FLAG__NO_DATA) == 0);
 415
 416         pd->delta.status = has_new ?
 417                 (has_old ? GIT_DELTA_MODIFIED : GIT_DELTA_ADDED) :
 418                 (has_old ? GIT_DELTA_DELETED : GIT_DELTA_UNTRACKED);
 419
 420         if (git_oid_equal(&patch->nfile.file->id, &patch->ofile.file->id))
 421                 pd->delta.status = GIT_DELTA_UNMODIFIED;
 422
 423         patch->base.delta = &pd->delta;
 424
 425         patch_generated_init_common(patch);
 426
 427         if (pd->delta.status == GIT_DELTA_UNMODIFIED &&
 428                 !(patch->ofile.opts_flags & GIT_DIFF_INCLUDE_UNMODIFIED)) {
 429
 430                 /* Even empty patches are flagged as binary, and even though
 431                  * there's no difference, we flag this as "containing data"
 432                  * (the data is known to be empty, as opposed to wholly unknown).
 433                  */
 434                 if (patch->base.diff_opts.flags & GIT_DIFF_SHOW_BINARY)
 435                         patch->base.binary.contains_data = 1;
 436
 437                 return error;
 438         }
 439
 440         error = patch_generated_invoke_file_callback(patch, (git_patch_generated_output *)xo);
 441
 442         if (!error)
 443                 error = patch_generated_create(patch, (git_patch_generated_output *)xo);
 444
 445         return error;
 446 }
 447
 448 static int patch_generated_from_sources(
 449         patch_generated_with_delta *pd,
 450         git_xdiff_output *xo,
 451         git_diff_file_content_src *oldsrc,
 452         git_diff_file_content_src *newsrc,
 453         const git_diff_options *opts)
 454 {
 455         int error = 0;
 456         git_repository *repo =
 457                 oldsrc->blob ? git_blob_owner(oldsrc->blob) :
 458                 newsrc->blob ? git_blob_owner(newsrc->blob) : NULL;
 459         git_diff_file *lfile = &pd->delta.old_file, *rfile = &pd->delta.new_file;
 460         git_diff_file_content *ldata = &pd->patch.ofile, *rdata = &pd->patch.nfile;
 461
 462         if ((error = patch_generated_normalize_options(&pd->patch.base.diff_opts, opts)) < 0)
 463                 return error;
 464
 465         if (opts && (opts->flags & GIT_DIFF_REVERSE) != 0) {
 466                 void *tmp = lfile; lfile = rfile; rfile = tmp;
 467                 tmp = ldata; ldata = rdata; rdata = tmp;
 468         }
 469
 470         pd->patch.base.delta = &pd->delta;
 471
 472         if (!oldsrc->as_path) {
 473                 if (newsrc->as_path)
 474                         oldsrc->as_path = newsrc->as_path;
 475                 else
 476                         oldsrc->as_path = newsrc->as_path = "file";
 477         }
 478         else if (!newsrc->as_path)
 479                 newsrc->as_path = oldsrc->as_path;
 480
 481         lfile->path = oldsrc->as_path;
 482         rfile->path = newsrc->as_path;
 483
 484         if ((error = git_diff_file_content__init_from_src(
 485                         ldata, repo, opts, oldsrc, lfile)) < 0 ||
 486                 (error = git_diff_file_content__init_from_src(
 487                         rdata, repo, opts, newsrc, rfile)) < 0)
 488                 return error;
 489
 490         return diff_single_generate(pd, xo);
 491 }
 492
 493 static int patch_generated_with_delta_alloc(
 494         patch_generated_with_delta **out,
 495         const char **old_path,
 496         const char **new_path)
 497 {
 498         patch_generated_with_delta *pd;
 499         size_t old_len = *old_path ? strlen(*old_path) : 0;
 500         size_t new_len = *new_path ? strlen(*new_path) : 0;
 501         size_t alloc_len;
 502
 503         GITERR_CHECK_ALLOC_ADD(&alloc_len, sizeof(*pd), old_len);
 504         GITERR_CHECK_ALLOC_ADD(&alloc_len, alloc_len, new_len);
 505         GITERR_CHECK_ALLOC_ADD(&alloc_len, alloc_len, 2);
 506
 507         *out = pd = git__calloc(1, alloc_len);
 508         GITERR_CHECK_ALLOC(pd);
 509
 510         pd->patch.flags = GIT_PATCH_GENERATED_ALLOCATED;
 511
 512         if (*old_path) {
 513                 memcpy(&pd->paths[0], *old_path, old_len);
 514                 *old_path = &pd->paths[0];
 515         } else if (*new_path)
 516                 *old_path = &pd->paths[old_len + 1];
 517
 518         if (*new_path) {
 519                 memcpy(&pd->paths[old_len + 1], *new_path, new_len);
 520                 *new_path = &pd->paths[old_len + 1];
 521         } else if (*old_path)
 522                 *new_path = &pd->paths[0];
 523
 524         return 0;
 525 }
 526
 527 static int diff_from_sources(
 528         git_diff_file_content_src *oldsrc,
 529         git_diff_file_content_src *newsrc,
 530         const git_diff_options *opts,
 531         git_diff_file_cb file_cb,
 532         git_diff_binary_cb binary_cb,
 533         git_diff_hunk_cb hunk_cb,
 534         git_diff_line_cb data_cb,
 535         void *payload)
 536 {
 537         int error = 0;
 538         patch_generated_with_delta pd;
 539         git_xdiff_output xo;
 540
 541         memset(&xo, 0, sizeof(xo));
 542         diff_output_init(
 543                 &xo.output, opts, file_cb, binary_cb, hunk_cb, data_cb, payload);
 544         git_xdiff_init(&xo, opts);
 545
 546         memset(&pd, 0, sizeof(pd));
 547
 548         error = patch_generated_from_sources(&pd, &xo, oldsrc, newsrc, opts);
 549
 550         git_patch_free(&pd.patch.base);
 551
 552         return error;
 553 }
 554
 555 static int patch_from_sources(
 556         git_patch **out,
 557         git_diff_file_content_src *oldsrc,
 558         git_diff_file_content_src *newsrc,
 559         const git_diff_options *opts)
 560 {
 561         int error = 0;
 562         patch_generated_with_delta *pd;
 563         git_xdiff_output xo;
 564
 565         assert(out);
 566         *out = NULL;
 567
 568         if ((error = patch_generated_with_delta_alloc(
 569                         &pd, &oldsrc->as_path, &newsrc->as_path)) < 0)
 570                 return error;
 571
 572         memset(&xo, 0, sizeof(xo));
 573         diff_output_to_patch(&xo.output, &pd->patch);
 574         git_xdiff_init(&xo, opts);
 575
 576         if (!(error = patch_generated_from_sources(pd, &xo, oldsrc, newsrc, opts)))
 577                 *out = (git_patch *)pd;
 578         else
 579                 git_patch_free((git_patch *)pd);
 580
 581         return error;
 582 }
 583
 584 int git_diff_blobs(
 585         const git_blob *old_blob,
 586         const char *old_path,
 587         const git_blob *new_blob,
 588         const char *new_path,
 589         const git_diff_options *opts,
 590         git_diff_file_cb file_cb,
 591         git_diff_binary_cb binary_cb,
 592         git_diff_hunk_cb hunk_cb,
 593         git_diff_line_cb data_cb,
 594         void *payload)
 595 {
 596         git_diff_file_content_src osrc =
 597                 GIT_DIFF_FILE_CONTENT_SRC__BLOB(old_blob, old_path);
 598         git_diff_file_content_src nsrc =
 599                 GIT_DIFF_FILE_CONTENT_SRC__BLOB(new_blob, new_path);
 600         return diff_from_sources(
 601                 &osrc, &nsrc, opts, file_cb, binary_cb, hunk_cb, data_cb, payload);
 602 }
 603
 604 int git_patch_from_blobs(
 605         git_patch **out,
 606         const git_blob *old_blob,
 607         const char *old_path,
 608         const git_blob *new_blob,
 609         const char *new_path,
 610         const git_diff_options *opts)
 611 {
 612         git_diff_file_content_src osrc =
 613                 GIT_DIFF_FILE_CONTENT_SRC__BLOB(old_blob, old_path);
 614         git_diff_file_content_src nsrc =
 615                 GIT_DIFF_FILE_CONTENT_SRC__BLOB(new_blob, new_path);
 616         return patch_from_sources(out, &osrc, &nsrc, opts);
 617 }
 618
 619 int git_diff_blob_to_buffer(
 620         const git_blob *old_blob,
 621         const char *old_path,
 622         const char *buf,
 623         size_t buflen,
 624         const char *buf_path,
 625         const git_diff_options *opts,
 626         git_diff_file_cb file_cb,
 627         git_diff_binary_cb binary_cb,
 628         git_diff_hunk_cb hunk_cb,
 629         git_diff_line_cb data_cb,
 630         void *payload)
 631 {
 632         git_diff_file_content_src osrc =
 633                 GIT_DIFF_FILE_CONTENT_SRC__BLOB(old_blob, old_path);
 634         git_diff_file_content_src nsrc =
 635                 GIT_DIFF_FILE_CONTENT_SRC__BUF(buf, buflen, buf_path);
 636         return diff_from_sources(
 637                 &osrc, &nsrc, opts, file_cb, binary_cb, hunk_cb, data_cb, payload);
 638 }
 639
 640 int git_patch_from_blob_and_buffer(
 641         git_patch **out,
 642         const git_blob *old_blob,
 643         const char *old_path,
 644         const char *buf,
 645         size_t buflen,
 646         const char *buf_path,
 647         const git_diff_options *opts)
 648 {
 649         git_diff_file_content_src osrc =
 650                 GIT_DIFF_FILE_CONTENT_SRC__BLOB(old_blob, old_path);
 651         git_diff_file_content_src nsrc =
 652                 GIT_DIFF_FILE_CONTENT_SRC__BUF(buf, buflen, buf_path);
 653         return patch_from_sources(out, &osrc, &nsrc, opts);
 654 }
 655
 656 int git_diff_buffers(
 657         const void *old_buf,
 658         size_t old_len,
 659         const char *old_path,
 660         const void *new_buf,
 661         size_t new_len,
 662         const char *new_path,
 663         const git_diff_options *opts,
 664         git_diff_file_cb file_cb,
 665         git_diff_binary_cb binary_cb,
 666         git_diff_hunk_cb hunk_cb,
 667         git_diff_line_cb data_cb,
 668         void *payload)
 669 {
 670         git_diff_file_content_src osrc =
 671                 GIT_DIFF_FILE_CONTENT_SRC__BUF(old_buf, old_len, old_path);
 672         git_diff_file_content_src nsrc =
 673                 GIT_DIFF_FILE_CONTENT_SRC__BUF(new_buf, new_len, new_path);
 674         return diff_from_sources(
 675                 &osrc, &nsrc, opts, file_cb, binary_cb, hunk_cb, data_cb, payload);
 676 }
 677
 678 int git_patch_from_buffers(
 679         git_patch **out,
 680         const void *old_buf,
 681         size_t old_len,
 682         const char *old_path,
 683         const char *new_buf,
 684         size_t new_len,
 685         const char *new_path,
 686         const git_diff_options *opts)
 687 {
 688         git_diff_file_content_src osrc =
 689                 GIT_DIFF_FILE_CONTENT_SRC__BUF(old_buf, old_len, old_path);
 690         git_diff_file_content_src nsrc =
 691                 GIT_DIFF_FILE_CONTENT_SRC__BUF(new_buf, new_len, new_path);
 692         return patch_from_sources(out, &osrc, &nsrc, opts);
 693 }
 694
 695 int git_patch_generated_from_diff(
 696         git_patch **patch_ptr, git_diff *diff, size_t idx)
 697 {
 698         int error = 0;
 699         git_xdiff_output xo;
 700         git_diff_delta *delta = NULL;
 701         git_patch_generated *patch = NULL;
 702
 703         if (patch_ptr) *patch_ptr = NULL;
 704
 705         if (diff_required(diff, "git_patch_from_diff") < 0)
 706                 return -1;
 707
 708         delta = git_vector_get(&diff->deltas, idx);
 709         if (!delta) {
 710                 giterr_set(GITERR_INVALID, "index out of range for delta in diff");
 711                 return GIT_ENOTFOUND;
 712         }
 713
 714         if (git_diff_delta__should_skip(&diff->opts, delta))
 715                 return 0;
 716
 717         /* don't load the patch data unless we need it for binary check */
 718         if (!patch_ptr &&
 719                 ((delta->flags & DIFF_FLAGS_KNOWN_BINARY) != 0 ||
 720                  (diff->opts.flags & GIT_DIFF_SKIP_BINARY_CHECK) != 0))
 721                 return 0;
 722
 723         if ((error = patch_generated_alloc_from_diff(&patch, diff, idx)) < 0)
 724                 return error;
 725
 726         memset(&xo, 0, sizeof(xo));
 727         diff_output_to_patch(&xo.output, patch);
 728         git_xdiff_init(&xo, &diff->opts);
 729
 730         error = patch_generated_invoke_file_callback(patch, &xo.output);
 731
 732         if (!error)
 733                 error = patch_generated_create(patch, &xo.output);
 734
 735         if (!error) {
 736                 /* TODO: if cumulative diff size is < 0.5 total size, flatten patch */
 737                 /* TODO: and unload the file content */
 738         }
 739
 740         if (error || !patch_ptr)
 741                 git_patch_free(&patch->base);
 742         else
 743                 *patch_ptr = &patch->base;
 744
 745         return error;
 746 }
 747
 748 git_diff_driver *git_patch_generated_driver(git_patch_generated *patch)
 749 {
 750         /* ofile driver is representative for whole patch */
 751         return patch->ofile.driver;
 752 }
 753
 754 void git_patch_generated_old_data(
 755         char **ptr, size_t *len, git_patch_generated *patch)
 756 {
 757         *ptr = patch->ofile.map.data;
 758         *len = patch->ofile.map.len;
 759 }
 760
 761 void git_patch_generated_new_data(
 762         char **ptr, size_t *len, git_patch_generated *patch)
 763 {
 764         *ptr = patch->nfile.map.data;
 765         *len = patch->nfile.map.len;
 766 }
 767
 768 static int patch_generated_file_cb(
 769         const git_diff_delta *delta,
 770         float progress,
 771         void *payload)
 772 {
 773         GIT_UNUSED(delta); GIT_UNUSED(progress); GIT_UNUSED(payload);
 774         return 0;
 775 }
 776
 777 static int patch_generated_binary_cb(
 778         const git_diff_delta *delta,
 779         const git_diff_binary *binary,
 780         void *payload)
 781 {
 782         git_patch *patch = payload;
 783
 784         GIT_UNUSED(delta);
 785
 786         memcpy(&patch->binary, binary, sizeof(git_diff_binary));
 787
 788         if (binary->old_file.data) {
 789                 patch->binary.old_file.data = git__malloc(binary->old_file.datalen);
 790                 GITERR_CHECK_ALLOC(patch->binary.old_file.data);
 791
 792                 memcpy((char *)patch->binary.old_file.data,
 793                         binary->old_file.data, binary->old_file.datalen);
 794         }
 795
 796         if (binary->new_file.data) {
 797                 patch->binary.new_file.data = git__malloc(binary->new_file.datalen);
 798                 GITERR_CHECK_ALLOC(patch->binary.new_file.data);
 799
 800                 memcpy((char *)patch->binary.new_file.data,
 801                         binary->new_file.data, binary->new_file.datalen);
 802         }
 803
 804         return 0;
 805 }
 806
 807 static int git_patch_hunk_cb(
 808         const git_diff_delta *delta,
 809         const git_diff_hunk *hunk_,
 810         void *payload)
 811 {
 812         git_patch_generated  *patch = payload;
 813         git_patch_hunk *hunk;
 814
 815         GIT_UNUSED(delta);
 816
 817         hunk = git_array_alloc(patch->base.hunks);
 818         GITERR_CHECK_ALLOC(hunk);
 819
 820         memcpy(&hunk->hunk, hunk_, sizeof(hunk->hunk));
 821
 822         patch->base.header_size += hunk_->header_len;
 823
 824         hunk->line_start = git_array_size(patch->base.lines);
 825         hunk->line_count = 0;
 826
 827         return 0;
 828 }
 829
 830 static int patch_generated_line_cb(
 831         const git_diff_delta *delta,
 832         const git_diff_hunk *hunk_,
 833         const git_diff_line *line_,
 834         void *payload)
 835 {
 836         git_patch_generated  *patch = payload;
 837         git_patch_hunk *hunk;
 838         git_diff_line   *line;
 839
 840         GIT_UNUSED(delta);
 841         GIT_UNUSED(hunk_);
 842
 843         hunk = git_array_last(patch->base.hunks);
 844         assert(hunk); /* programmer error if no hunk is available */
 845
 846         line = git_array_alloc(patch->base.lines);
 847         GITERR_CHECK_ALLOC(line);
 848
 849         memcpy(line, line_, sizeof(*line));
 850
 851         /* do some bookkeeping so we can provide old/new line numbers */
 852
 853         patch->base.content_size += line->content_len;
 854
 855         if (line->origin == GIT_DIFF_LINE_ADDITION ||
 856                 line->origin == GIT_DIFF_LINE_DELETION)
 857                 patch->base.content_size += 1;
 858         else if (line->origin == GIT_DIFF_LINE_CONTEXT) {
 859                 patch->base.content_size += 1;
 860                 patch->base.context_size += line->content_len + 1;
 861         } else if (line->origin == GIT_DIFF_LINE_CONTEXT_EOFNL)
 862                 patch->base.context_size += line->content_len;
 863
 864         hunk->line_count++;
 865
 866         return 0;
 867 }
 868
 869 static void diff_output_init(
 870         git_patch_generated_output *out,
 871         const git_diff_options *opts,
 872         git_diff_file_cb file_cb,
 873         git_diff_binary_cb binary_cb,
 874         git_diff_hunk_cb hunk_cb,
 875         git_diff_line_cb data_cb,
 876         void *payload)
 877 {
 878         GIT_UNUSED(opts);
 879
 880         memset(out, 0, sizeof(*out));
 881
 882         out->file_cb = file_cb;
 883         out->binary_cb = binary_cb;
 884         out->hunk_cb = hunk_cb;
 885         out->data_cb = data_cb;
 886         out->payload = payload;
 887 }
 888
 889 static void diff_output_to_patch(
 890         git_patch_generated_output *out, git_patch_generated *patch)
 891 {
 892         diff_output_init(
 893                 out,
 894                 NULL,
 895                 patch_generated_file_cb,
 896                 patch_generated_binary_cb,
 897                 git_patch_hunk_cb,
 898                 patch_generated_line_cb,
 899                 patch);
 900 }