int git_blob_filtered_content(
git_buf *out,
git_blob *blob,
const char *path,
int check_for_binary_data)
{
int error = 0;
git_filter_list *fl = NULL;
assert(blob && path && out);
git_buf_sanitize(out);
if (check_for_binary_data && git_blob_is_binary(blob))
return 0;
if (!(error = git_filter_list_load(
&fl, git_blob_owner(blob), blob, path,
GIT_FILTER_TO_WORKTREE, GIT_FILTER_DEFAULT))) {
error = git_filter_list_apply_to_blob(out, fl, blob);
git_filter_list_free(fl);
}
return error;
}
static int diff_file_content_load_workdir_file(
git_diff_file_content *fc,
git_buf *path,
git_diff_options *diff_opts)
{
int error = 0;
git_filter_list *fl = NULL;
git_file fd = git_futils_open_ro(git_buf_cstr(path));
git_buf raw = GIT_BUF_INIT;
if (fd < 0)
return fd;
if (!fc->file->size &&
!(fc->file->size = git_futils_filesize(fd)))
goto cleanup;
if ((diff_opts->flags & GIT_DIFF_SHOW_BINARY) == 0 &&
diff_file_content_binary_by_size(fc))
goto cleanup;
if ((error = git_filter_list_load(
&fl, fc->repo, NULL, fc->file->path,
GIT_FILTER_TO_ODB, GIT_FILTER_ALLOW_UNSAFE)) < 0)
goto cleanup;
/* if there are no filters, try to mmap the file */
if (fl == NULL) {
if (!(error = git_futils_mmap_ro(
&fc->map, fd, 0, (size_t)fc->file->size))) {
fc->flags |= GIT_DIFF_FLAG__UNMAP_DATA;
goto cleanup;
}
/* if mmap failed, fall through to try readbuffer below */
giterr_clear();
}
if (!(error = git_futils_readbuffer_fd(&raw, fd, (size_t)fc->file->size))) {
git_buf out = GIT_BUF_INIT;
error = git_filter_list_apply_to_data(&out, fl, &raw);
if (out.ptr != raw.ptr)
git_buf_dispose(&raw);
if (!error) {
fc->map.len = out.size;
fc->map.data = out.ptr;
fc->flags |= GIT_DIFF_FLAG__FREE_DATA;
}
}
cleanup:
git_filter_list_free(fl);
p_close(fd);
return error;
}
static int filter_for(const char *filename, const char *filter)
{
git_filter_list *fl;
int filtered;
cl_git_pass(git_filter_list_load(
&fl, g_repo, NULL, filename, GIT_FILTER_TO_WORKTREE, 0));
filtered = git_filter_list_contains(fl, filter);
git_filter_list_free(fl);
return filtered;
}
void test_filter_custom__can_register_a_custom_filter_in_the_repository(void)
{
git_filter_list *fl;
cl_git_pass(git_filter_list_load(
&fl, g_repo, NULL, "herofile", GIT_FILTER_TO_WORKTREE));
/* expect: bitflip, reverse, crlf */
cl_assert_equal_sz(3, git_filter_list_length(fl));
git_filter_list_free(fl);
cl_git_pass(git_filter_list_load(
&fl, g_repo, NULL, "herocorp", GIT_FILTER_TO_WORKTREE));
/* expect: bitflip, reverse - possibly crlf depending on global config */
{
size_t flen = git_filter_list_length(fl);
cl_assert(flen == 2 || flen == 3);
}
git_filter_list_free(fl);
cl_git_pass(git_filter_list_load(
&fl, g_repo, NULL, "hero.bin", GIT_FILTER_TO_WORKTREE));
/* expect: bitflip, reverse */
cl_assert_equal_sz(2, git_filter_list_length(fl));
git_filter_list_free(fl);
cl_git_pass(git_filter_list_load(
&fl, g_repo, NULL, "heroflip", GIT_FILTER_TO_WORKTREE));
/* expect: bitflip (because of -reverse) */
cl_assert_equal_sz(1, git_filter_list_length(fl));
git_filter_list_free(fl);
cl_git_pass(git_filter_list_load(
&fl, g_repo, NULL, "doesntapplytome.bin", GIT_FILTER_TO_WORKTREE));
/* expect: none */
cl_assert_equal_sz(0, git_filter_list_length(fl));
git_filter_list_free(fl);
}
void test_filter_custom__to_odb(void)
{
git_filter_list *fl;
git_buf out = { 0 };
git_buf in = GIT_BUF_INIT_CONST(workdir_data, strlen(workdir_data));
cl_git_pass(git_filter_list_load(
&fl, g_repo, NULL, "herofile", GIT_FILTER_TO_ODB));
cl_git_pass(git_filter_list_apply_to_data(&out, fl, &in));
cl_assert_equal_i(BITFLIPPED_AND_REVERSED_DATA_LEN, out.size);
cl_assert_equal_i(
0, memcmp(bitflipped_and_reversed_data, out.ptr, out.size));
git_filter_list_free(fl);
git_buf_free(&out);
}
void test_object_blob_filter__to_odb(void)
{
git_filter_list *fl = NULL;
git_config *cfg;
int i;
git_blob *blob;
git_buf out = GIT_BUF_INIT, zeroed;
cl_git_pass(git_repository_config(&cfg, g_repo));
cl_assert(cfg);
git_attr_cache_flush(g_repo);
cl_git_append2file("empty_standard_repo/.gitattributes", "*.txt text\n");
cl_git_pass(git_filter_list_load(
&fl, g_repo, NULL, "filename.txt", GIT_FILTER_TO_ODB, 0));
cl_assert(fl != NULL);
for (i = 0; i < CRLF_NUM_TEST_OBJECTS; i++) {
cl_git_pass(git_blob_lookup(&blob, g_repo, &g_crlf_oids[i]));
/* try once with allocated blob */
cl_git_pass(git_filter_list_apply_to_blob(&out, fl, blob));
cl_assert_equal_sz(g_crlf_filtered[i].size, out.size);
cl_assert_equal_i(
0, memcmp(out.ptr, g_crlf_filtered[i].ptr, out.size));
/* try again with zeroed blob */
memset(&zeroed, 0, sizeof(zeroed));
cl_git_pass(git_filter_list_apply_to_blob(&zeroed, fl, blob));
cl_assert_equal_sz(g_crlf_filtered[i].size, zeroed.size);
cl_assert_equal_i(
0, memcmp(zeroed.ptr, g_crlf_filtered[i].ptr, zeroed.size));
git_buf_free(&zeroed);
git_blob_free(blob);
}
git_filter_list_free(fl);
git_buf_free(&out);
git_config_free(cfg);
}
void test_filter_wildcard__none(void)
{
git_filter_list *fl;
git_buf in = GIT_BUF_INIT, out = GIT_BUF_INIT;
cl_git_pass(git_filter_list_load(
&fl, g_repo, NULL, "none-foo", GIT_FILTER_TO_ODB, 0));
cl_git_pass(git_buf_put(&in, (char *)input, DATA_LEN));
cl_git_pass(git_filter_list_apply_to_data(&out, fl, &in));
cl_assert_equal_i(DATA_LEN, out.size);
cl_assert_equal_i(
0, memcmp(input, out.ptr, out.size));
git_filter_list_free(fl);
git_buf_dispose(&out);
git_buf_dispose(&in);
}
static int workdir_reader_read(
git_buf *out,
git_oid *out_id,
git_filemode_t *out_filemode,
git_reader *_reader,
const char *filename)
{
workdir_reader *reader = (workdir_reader *)_reader;
git_buf path = GIT_BUF_INIT;
struct stat st;
git_filemode_t filemode;
git_filter_list *filters = NULL;
const git_index_entry *idx_entry;
git_oid id;
int error;
if ((error = git_buf_joinpath(&path,
git_repository_workdir(reader->repo), filename)) < 0)
goto done;
if ((error = p_lstat(path.ptr, &st)) < 0) {
if (error == -1 && errno == ENOENT)
error = GIT_ENOTFOUND;
giterr_set(GITERR_OS, "could not stat '%s'", path.ptr);
goto done;
}
filemode = git_futils_canonical_mode(st.st_mode);
/*
* Patch application - for example - uses the filtered version of
* the working directory data to match git. So we will run the
* workdir -> ODB filter on the contents in this workdir reader.
*/
if ((error = git_filter_list_load(&filters, reader->repo, NULL, filename,
GIT_FILTER_TO_ODB, GIT_FILTER_DEFAULT)) < 0)
goto done;
if ((error = git_filter_list_apply_to_file(out,
filters, reader->repo, path.ptr)) < 0)
goto done;
if (out_id || reader->index) {
if ((error = git_odb_hash(&id, out->ptr, out->size, GIT_OBJ_BLOB)) < 0)
goto done;
}
if (reader->index) {
if (!(idx_entry = git_index_get_bypath(reader->index, filename, 0)) ||
filemode != idx_entry->mode ||
!git_oid_equal(&id, &idx_entry->id)) {
error = GIT_READER_MISMATCH;
goto done;
}
}
if (out_id)
git_oid_cpy(out_id, &id);
if (out_filemode)
*out_filemode = filemode;
done:
git_filter_list_free(filters);
git_buf_dispose(&path);
return error;
}
int git_blob__create_from_paths(
git_oid *id,
struct stat *out_st,
git_repository *repo,
const char *content_path,
const char *hint_path,
mode_t hint_mode,
bool try_load_filters)
{
int error;
struct stat st;
git_odb *odb = NULL;
git_off_t size;
mode_t mode;
git_buf path = GIT_BUF_INIT;
assert(hint_path || !try_load_filters);
if (!content_path) {
if (git_repository__ensure_not_bare(repo, "create blob from file") < 0)
return GIT_EBAREREPO;
if (git_buf_joinpath(
&path, git_repository_workdir(repo), hint_path) < 0)
return -1;
content_path = path.ptr;
}
if ((error = git_path_lstat(content_path, &st)) < 0 ||
(error = git_repository_odb(&odb, repo)) < 0)
goto done;
if (S_ISDIR(st.st_mode)) {
giterr_set(GITERR_ODB, "cannot create blob from '%s': it is a directory", content_path);
error = GIT_EDIRECTORY;
goto done;
}
if (out_st)
memcpy(out_st, &st, sizeof(st));
size = st.st_size;
mode = hint_mode ? hint_mode : st.st_mode;
if (S_ISLNK(mode)) {
error = write_symlink(id, odb, content_path, (size_t)size);
} else {
git_filter_list *fl = NULL;
if (try_load_filters)
/* Load the filters for writing this file to the ODB */
error = git_filter_list_load(
&fl, repo, NULL, hint_path,
GIT_FILTER_TO_ODB, GIT_FILTER_DEFAULT);
if (error < 0)
/* well, that didn't work */;
else if (fl == NULL)
/* No filters need to be applied to the document: we can stream
* directly from disk */
error = write_file_stream(id, odb, content_path, size);
else {
/* We need to apply one or more filters */
error = write_file_filtered(id, &size, odb, content_path, fl);
git_filter_list_free(fl);
}
/*
* TODO: eventually support streaming filtered files, for files
* which are bigger than a given threshold. This is not a priority
* because applying a filter in streaming mode changes the final
* size of the blob, and without knowing its final size, the blob
* cannot be written in stream mode to the ODB.
*
* The plan is to do streaming writes to a tempfile on disk and then
* opening streaming that file to the ODB, using
* `write_file_stream`.
*
* CAREFULLY DESIGNED APIS YO
*/
}
done:
git_odb_free(odb);
git_buf_dispose(&path);
return error;
}
