airbyte-cdk 7.3.0__py3-none-any.whl → 7.3.1__py3-none-any.whl

airbyte_cdk/sources/file_based/file_based_stream_reader.py

@@ -3,13 +3,15 @@
 #
 
 import logging
+import time
 from abc import ABC, abstractmethod
 from datetime import datetime
 from enum import Enum
 from io import IOBase
 from os import makedirs, path
-from typing import Any, Callable, Iterable, List, MutableMapping, Optional, Set, Tuple
+from typing import Any, Iterable, List, MutableMapping, Optional, Set, Tuple
 
+from airbyte_protocol_dataclasses.models import FailureType
 from wcmatch.glob import GLOBSTAR, globmatch
 
 from airbyte_cdk.models import AirbyteRecordMessageFileReference
@@ -19,8 +21,9 @@ from airbyte_cdk.sources.file_based.config.validate_config_transfer_modes import
     preserve_directory_structure,
     use_file_transfer,
 )
+from airbyte_cdk.sources.file_based.exceptions import FileSizeLimitError
 from airbyte_cdk.sources.file_based.file_record_data import FileRecordData
-from airbyte_cdk.sources.file_based.remote_file import RemoteFile
+from airbyte_cdk.sources.file_based.remote_file import RemoteFile, UploadableRemoteFile
 
 
 class FileReadMode(Enum):
@@ -34,6 +37,7 @@ class AbstractFileBasedStreamReader(ABC):
     FILE_NAME = "file_name"
     LOCAL_FILE_PATH = "local_file_path"
     FILE_FOLDER = "file_folder"
+    FILE_SIZE_LIMIT = 1_500_000_000
 
     def __init__(self) -> None:
         self._config = None
@@ -113,16 +117,6 @@ class AbstractFileBasedStreamReader(ABC):
                     seen.add(file.uri)
                     yield file
 
-    @abstractmethod
-    def file_size(self, file: RemoteFile) -> int:
-        """Utility method to get size of the remote file.
-
-        This is required for connectors that will support writing to
-        files. If the connector does not support writing files, then the
-        subclass can simply `return 0`.
-        """
-        ...
-
     @staticmethod
     def file_matches_globs(file: RemoteFile, globs: List[str]) -> bool:
         # Use the GLOBSTAR flag to enable recursive ** matching
@@ -153,9 +147,8 @@ class AbstractFileBasedStreamReader(ABC):
             return include_identities_stream(self.config)
         return False
 
-    @abstractmethod
     def upload(
-        self, file: RemoteFile, local_directory: str, logger: logging.Logger
+        self, file: UploadableRemoteFile, local_directory: str, logger: logging.Logger
     ) -> Tuple[FileRecordData, AirbyteRecordMessageFileReference]:
         """
         This is required for connectors that will support writing to
@@ -173,7 +166,53 @@ class AbstractFileBasedStreamReader(ABC):
                    - file_size_bytes (int): The size of the referenced file in bytes.
                    - source_file_relative_path (str): The relative path to the referenced file in source.
         """
-        ...
+        if not isinstance(file, UploadableRemoteFile):
+            raise TypeError(f"Expected UploadableRemoteFile, got {type(file)}")
+
+        file_size = file.size
+
+        if file_size > self.FILE_SIZE_LIMIT:
+            message = f"File size exceeds the {self.FILE_SIZE_LIMIT / 1e9} GB limit."
+            raise FileSizeLimitError(
+                message=message, internal_message=message, failure_type=FailureType.config_error
+            )
+
+        file_paths = self._get_file_transfer_paths(
+            source_file_relative_path=file.source_file_relative_path,
+            staging_directory=local_directory,
+        )
+        local_file_path = file_paths[self.LOCAL_FILE_PATH]
+        file_relative_path = file_paths[self.FILE_RELATIVE_PATH]
+        file_name = file_paths[self.FILE_NAME]
+
+        logger.info(
+            f"Starting to download the file {file.file_uri_for_logging} with size: {file_size / (1024 * 1024):,.2f} MB ({file_size / (1024 * 1024 * 1024):.2f} GB)"
+        )
+        start_download_time = time.time()
+
+        file.download_to_local_directory(local_file_path)
+
+        write_duration = time.time() - start_download_time
+        logger.info(
+            f"Finished downloading the file {file.file_uri_for_logging} and saved to {local_file_path} in {write_duration:,.2f} seconds."
+        )
+
+        file_record_data = FileRecordData(
+            folder=file_paths[self.FILE_FOLDER],
+            file_name=file_name,
+            bytes=file_size,
+            id=file.id,
+            mime_type=file.mime_type,
+            created_at=file.created_at,
+            updated_at=file.updated_at,
+            source_uri=file.uri,
+        )
+        file_reference = AirbyteRecordMessageFileReference(
+            staging_file_url=local_file_path,
+            source_file_relative_path=file_relative_path,
+            file_size_bytes=file_size,
+        )
+        return file_record_data, file_reference
 
     def _get_file_transfer_paths(
         self, source_file_relative_path: str, staging_directory: str

airbyte_cdk/sources/file_based/file_types/file_transfer.py

@@ -7,7 +7,7 @@ from typing import Iterable, Tuple
 from airbyte_cdk.models import AirbyteRecordMessageFileReference
 from airbyte_cdk.sources.file_based.file_based_stream_reader import AbstractFileBasedStreamReader
 from airbyte_cdk.sources.file_based.file_record_data import FileRecordData
-from airbyte_cdk.sources.file_based.remote_file import RemoteFile
+from airbyte_cdk.sources.file_based.remote_file import UploadableRemoteFile
 from airbyte_cdk.sources.utils.files_directory import get_files_directory
 
 
@@ -17,7 +17,7 @@ class FileTransfer:
 
     def upload(
         self,
-        file: RemoteFile,
+        file: UploadableRemoteFile,
         stream_reader: AbstractFileBasedStreamReader,
         logger: logging.Logger,
     ) -> Iterable[Tuple[FileRecordData, AirbyteRecordMessageFileReference]]:

airbyte_cdk/sources/file_based/remote_file.py

@@ -1,7 +1,7 @@
 #
 # Copyright (c) 2023 Airbyte, Inc., all rights reserved.
 #
-
+from abc import ABC, abstractmethod
 from datetime import datetime
 from typing import Optional
 
@@ -16,3 +16,42 @@ class RemoteFile(BaseModel):
     uri: str
     last_modified: datetime
     mime_type: Optional[str] = None
+
+
+class UploadableRemoteFile(RemoteFile, ABC):
+    """
+    A file in a file-based stream that supports uploading(file transferring).
+    """
+
+    id: Optional[str] = None
+    created_at: Optional[str] = None
+    updated_at: Optional[str] = None
+
+    @property
+    @abstractmethod
+    def size(self) -> int:
+        """
+        Returns the file size in bytes.
+        """
+        ...
+
+    @abstractmethod
+    def download_to_local_directory(self, local_file_path: str) -> None:
+        """
+        Download the file from remote source to local storage.
+        """
+        ...
+
+    @property
+    def source_file_relative_path(self) -> str:
+        """
+        Returns the relative path of the source file.
+        """
+        return self.uri
+
+    @property
+    def file_uri_for_logging(self) -> str:
+        """
+        Returns the URI for the file being logged.
+        """
+        return self.uri

{airbyte_cdk-7.3.0.dist-info → airbyte_cdk-7.3.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: airbyte-cdk
-Version: 7.3.0
+Version: 7.3.1
 Summary: A framework for writing Airbyte Connectors.
 Home-page: https://airbyte.com
 License: MIT

{airbyte_cdk-7.3.0.dist-info → airbyte_cdk-7.3.1.dist-info}/RECORD

@@ -298,18 +298,18 @@ airbyte_cdk/sources/file_based/discovery_policy/default_discovery_policy.py,sha2
 airbyte_cdk/sources/file_based/exceptions.py,sha256=WP0qkG6fpWoBpOyyicgp5YNE393VWyegq5qSy0v4QtM,7362
 airbyte_cdk/sources/file_based/file_based_source.py,sha256=Xg8OYWnGc-OcVBglvS08uwAWGWHBhEqsBnyODIkOK-4,20051
 airbyte_cdk/sources/file_based/file_based_stream_permissions_reader.py,sha256=4e7FXqQ9hueacexC0SyrZyjF8oREYHza8pKF9CgKbD8,5050
-airbyte_cdk/sources/file_based/file_based_stream_reader.py,sha256=rwz8AhEIqYB9gBF7uW9eR--eUiHOntzuwLH8jFHNacE,7854
+airbyte_cdk/sources/file_based/file_based_stream_reader.py,sha256=Yg9KRXpyAtElBrUOO8oX4WHQH6k6Lk7keklrZmB5Klg,9614
 airbyte_cdk/sources/file_based/file_record_data.py,sha256=Vkr5AyZzlsOezjVCLhFrm_WpymlQdolWCnFAwqLJ9Iw,453
 airbyte_cdk/sources/file_based/file_types/__init__.py,sha256=blCLn0-2LC-ZdgcNyDEhqM2RiUvEjEBh-G4-t32ZtuM,1268
 airbyte_cdk/sources/file_based/file_types/avro_parser.py,sha256=USEYqiICXBWpDV443VtNOCmUA-GINzY_Zah74_5w3qQ,10860
 airbyte_cdk/sources/file_based/file_types/csv_parser.py,sha256=QlCXB-ry3np67Q_VerQEPoWDOTcPTB6Go4ydZxY9ae4,20445
 airbyte_cdk/sources/file_based/file_types/excel_parser.py,sha256=BeplCq0hmojELU6bZCvvpRLpQ9us81TqbGYwrhd3INo,7188
-airbyte_cdk/sources/file_based/file_types/file_transfer.py,sha256=5l2Jo6bp6neDmgM427PrZMZeqU0hCIZVWnzUZ_7BT10,1100
+airbyte_cdk/sources/file_based/file_types/file_transfer.py,sha256=rFxWaqItBux9tPf4xU03LT6b-wDZf1QolM92mP8Diuk,1120
 airbyte_cdk/sources/file_based/file_types/file_type_parser.py,sha256=JgpH21PrbRqwK92BJklZWvh2TndA6xZ-eP1LPMo44oQ,2832
 airbyte_cdk/sources/file_based/file_types/jsonl_parser.py,sha256=GwyNyxmST4RX-XpXy7xVH0D-znYWWBmGv_pVAu95oHQ,5886
 airbyte_cdk/sources/file_based/file_types/parquet_parser.py,sha256=XenFg5sJ-UBnIkSmsiNJRou11NO0zZXx-RXgPHMT2NA,10487
 airbyte_cdk/sources/file_based/file_types/unstructured_parser.py,sha256=2TYOQl62FQPCa8otLbkDIk_j01EP3oWaKSfXGhCjCHg,19492
-airbyte_cdk/sources/file_based/remote_file.py,sha256=yqRz93vPe8PBXLIMJ5W5u2JRlZRhg6sBrAjn3pPjJ8A,315
+airbyte_cdk/sources/file_based/remote_file.py,sha256=1Afzr2WFWwjiUz8R2vNFepeI192UNeHOZAXIGTWOzOM,1248
 airbyte_cdk/sources/file_based/schema_helpers.py,sha256=dKXAOTmMI3YmC5u7PeHC9AaZmlL6ft7CYSFQKCg0sXw,9911
 airbyte_cdk/sources/file_based/schema_validation_policies/__init__.py,sha256=FkByIyEy56x2_awYnxGPqGaOp7zAzpAoRkPZHKySI9M,536
 airbyte_cdk/sources/file_based/schema_validation_policies/abstract_schema_validation_policy.py,sha256=kjvX7nOmUALYd7HuZHilUzgJPZ-MnZ08mtvuBnt2tQ0,618
@@ -457,9 +457,9 @@ airbyte_cdk/utils/slice_hasher.py,sha256=EDxgROHDbfG-QKQb59m7h_7crN1tRiawdf5uU7G
 airbyte_cdk/utils/spec_schema_transformations.py,sha256=9YDJmnIGFsT51CVQf2tSSvTapGimITjEFGbUTSZAGTI,963
 airbyte_cdk/utils/stream_status_utils.py,sha256=ZmBoiy5HVbUEHAMrUONxZvxnvfV9CesmQJLDTAIWnWw,1171
 airbyte_cdk/utils/traced_exception.py,sha256=C8uIBuCL_E4WnBAOPSxBicD06JAldoN9fGsQDp463OY,6292
-airbyte_cdk-7.3.0.dist-info/LICENSE.txt,sha256=Wfe61S4BaGPj404v8lrAbvhjYR68SHlkzeYrg3_bbuM,1051
-airbyte_cdk-7.3.0.dist-info/LICENSE_SHORT,sha256=aqF6D1NcESmpn-cqsxBtszTEnHKnlsp8L4x9wAh3Nxg,55
-airbyte_cdk-7.3.0.dist-info/METADATA,sha256=KA6giadtbEGz9PBTe3tCmAha99is-QSuz0YULFMoxMU,6798
-airbyte_cdk-7.3.0.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
-airbyte_cdk-7.3.0.dist-info/entry_points.txt,sha256=eLZ2UYvJZGm1s07Pplcs--1Gim60YhZWTb53j_dghwU,195
-airbyte_cdk-7.3.0.dist-info/RECORD,,
+airbyte_cdk-7.3.1.dist-info/LICENSE.txt,sha256=Wfe61S4BaGPj404v8lrAbvhjYR68SHlkzeYrg3_bbuM,1051
+airbyte_cdk-7.3.1.dist-info/LICENSE_SHORT,sha256=aqF6D1NcESmpn-cqsxBtszTEnHKnlsp8L4x9wAh3Nxg,55
+airbyte_cdk-7.3.1.dist-info/METADATA,sha256=_n29oKSyO6A6mUMN1c6YqHvrJRFQrXOIhhAh1E0PuXo,6798
+airbyte_cdk-7.3.1.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
+airbyte_cdk-7.3.1.dist-info/entry_points.txt,sha256=eLZ2UYvJZGm1s07Pplcs--1Gim60YhZWTb53j_dghwU,195
+airbyte_cdk-7.3.1.dist-info/RECORD,,