|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
from pyarrow.includes.libarrow_dataset cimport * |
|
from pyarrow.includes.libparquet_encryption cimport * |
|
|
|
from pyarrow._parquet cimport * |
|
|
|
|
|
cdef extern from "arrow/dataset/parquet_encryption_config.h" namespace "arrow::dataset" nogil: |
|
cdef cppclass CParquetEncryptionConfig "arrow::dataset::ParquetEncryptionConfig": |
|
shared_ptr[CCryptoFactory] crypto_factory |
|
shared_ptr[CKmsConnectionConfig] kms_connection_config |
|
shared_ptr[CEncryptionConfiguration] encryption_config |
|
|
|
cdef cppclass CParquetDecryptionConfig "arrow::dataset::ParquetDecryptionConfig": |
|
shared_ptr[CCryptoFactory] crypto_factory |
|
shared_ptr[CKmsConnectionConfig] kms_connection_config |
|
shared_ptr[CDecryptionConfiguration] decryption_config |
|
|
|
|
|
cdef extern from "arrow/dataset/api.h" namespace "arrow::dataset" nogil: |
|
|
|
cdef cppclass CParquetFileWriter \ |
|
"arrow::dataset::ParquetFileWriter"(CFileWriter): |
|
const shared_ptr[FileWriter]& parquet_writer() const |
|
|
|
cdef cppclass CParquetFileWriteOptions \ |
|
"arrow::dataset::ParquetFileWriteOptions"(CFileWriteOptions): |
|
shared_ptr[WriterProperties] writer_properties |
|
shared_ptr[ArrowWriterProperties] arrow_writer_properties |
|
shared_ptr[CParquetEncryptionConfig] parquet_encryption_config |
|
|
|
cdef cppclass CParquetFileFragment "arrow::dataset::ParquetFileFragment"( |
|
CFileFragment): |
|
const vector[int]& row_groups() const |
|
shared_ptr[CFileMetaData] metadata() const |
|
CResult[vector[shared_ptr[CFragment]]] SplitByRowGroup( |
|
CExpression predicate) |
|
CResult[shared_ptr[CFragment]] SubsetWithFilter "Subset"( |
|
CExpression predicate) |
|
CResult[shared_ptr[CFragment]] SubsetWithIds "Subset"( |
|
vector[int] row_group_ids) |
|
CStatus EnsureCompleteMetadata() |
|
|
|
cdef cppclass CParquetFileFormatReaderOptions \ |
|
"arrow::dataset::ParquetFileFormat::ReaderOptions": |
|
unordered_set[c_string] dict_columns |
|
TimeUnit coerce_int96_timestamp_unit |
|
|
|
cdef cppclass CParquetFileFormat "arrow::dataset::ParquetFileFormat"( |
|
CFileFormat): |
|
CParquetFileFormatReaderOptions reader_options |
|
CResult[shared_ptr[CFileFragment]] MakeFragment( |
|
CFileSource source, |
|
CExpression partition_expression, |
|
shared_ptr[CSchema] physical_schema, |
|
vector[int] row_groups) |
|
|
|
cdef cppclass CParquetFragmentScanOptions \ |
|
"arrow::dataset::ParquetFragmentScanOptions"(CFragmentScanOptions): |
|
shared_ptr[CReaderProperties] reader_properties |
|
shared_ptr[ArrowReaderProperties] arrow_reader_properties |
|
shared_ptr[CParquetDecryptionConfig] parquet_decryption_config |
|
|
|
cdef cppclass CParquetFactoryOptions \ |
|
"arrow::dataset::ParquetFactoryOptions": |
|
CPartitioningOrFactory partitioning |
|
c_string partition_base_dir |
|
c_bool validate_column_chunk_paths |
|
|
|
cdef cppclass CParquetDatasetFactory \ |
|
"arrow::dataset::ParquetDatasetFactory"(CDatasetFactory): |
|
@staticmethod |
|
CResult[shared_ptr[CDatasetFactory]] MakeFromMetaDataPath "Make"( |
|
const c_string& metadata_path, |
|
shared_ptr[CFileSystem] filesystem, |
|
shared_ptr[CParquetFileFormat] format, |
|
CParquetFactoryOptions options |
|
) |
|
|
|
@staticmethod |
|
CResult[shared_ptr[CDatasetFactory]] MakeFromMetaDataSource "Make"( |
|
const CFileSource& metadata_path, |
|
const c_string& base_path, |
|
shared_ptr[CFileSystem] filesystem, |
|
shared_ptr[CParquetFileFormat] format, |
|
CParquetFactoryOptions options |
|
) |
|
|