airbyte-cdk 6.18.1__py3-none-any.whl → 6.18.2__py3-none-any.whl

airbyte_cdk/sources/file_based/config/abstract_file_based_spec.py

@@ -31,6 +31,17 @@ class DeliverRawFiles(BaseModel):
 
     delivery_type: Literal["use_file_transfer"] = Field("use_file_transfer", const=True)
 
+    preserve_directory_structure: bool = Field(
+        title="Preserve Sub-Directories in File Paths",
+        description=(
+            "If enabled, sends subdirectory folder structure "
+            "along with source file names to the destination. "
+            "Otherwise, files will be synced by their names only. "
+            "This option is ignored when file-based replication is not enabled."
+        ),
+        default=True,
+    )
+
 
 class AbstractFileBasedSpec(BaseModel):
     """

airbyte_cdk/sources/file_based/exceptions.py

@@ -111,6 +111,40 @@ class ErrorListingFiles(BaseFileBasedSourceError):
     pass
 
 
+class DuplicatedFilesError(BaseFileBasedSourceError):
+    def __init__(self, duplicated_files_names: List[dict[str, List[str]]], **kwargs: Any):
+        self._duplicated_files_names = duplicated_files_names
+        self._stream_name: str = kwargs["stream"]
+        super().__init__(self._format_duplicate_files_error_message(), **kwargs)
+
+    def _format_duplicate_files_error_message(self) -> str:
+        duplicated_files_messages = []
+        for duplicated_file in self._duplicated_files_names:
+            for duplicated_file_name, file_paths in duplicated_file.items():
+                file_duplicated_message = (
+                    f"{len(file_paths)} duplicates found for file name {duplicated_file_name}:\n\n"
+                    + "".join(f"\n - {file_paths}")
+                )
+                duplicated_files_messages.append(file_duplicated_message)
+
+        error_message = (
+            f"ERROR: Duplicate filenames found for stream {self._stream_name}. "
+            "Duplicate file names are not allowed if the Preserve Sub-Directories in File Paths option is disabled. "
+            "Please remove or rename the duplicate files before attempting to re-run the sync.\n\n"
+            + "\n".join(duplicated_files_messages)
+        )
+
+        return error_message
+
+    def __repr__(self) -> str:
+        """Return a string representation of the exception."""
+        class_name = self.__class__.__name__
+        properties_str = ", ".join(
+            f"{k}={v!r}" for k, v in self.__dict__.items() if not k.startswith("_")
+        )
+        return f"{class_name}({properties_str})"
+
+
 class CustomFileBasedException(AirbyteTracedException):
     """
     A specialized exception for file-based connectors.

airbyte_cdk/sources/file_based/file_based_source.py

@@ -242,7 +242,7 @@ class FileBasedSource(ConcurrentSourceAdapter, ABC):
                         stream=self._make_default_stream(
                             stream_config=stream_config,
                             cursor=cursor,
-                            use_file_transfer=self._use_file_transfer(parsed_config),
+                            parsed_config=parsed_config,
                         ),
                         source=self,
                         logger=self.logger,
@@ -273,7 +273,7 @@ class FileBasedSource(ConcurrentSourceAdapter, ABC):
                         stream=self._make_default_stream(
                             stream_config=stream_config,
                             cursor=cursor,
-                            use_file_transfer=self._use_file_transfer(parsed_config),
+                            parsed_config=parsed_config,
                         ),
                         source=self,
                         logger=self.logger,
@@ -285,7 +285,7 @@ class FileBasedSource(ConcurrentSourceAdapter, ABC):
                     stream = self._make_default_stream(
                         stream_config=stream_config,
                         cursor=cursor,
-                        use_file_transfer=self._use_file_transfer(parsed_config),
+                        parsed_config=parsed_config,
                     )
 
                 streams.append(stream)
@@ -298,7 +298,7 @@ class FileBasedSource(ConcurrentSourceAdapter, ABC):
         self,
         stream_config: FileBasedStreamConfig,
         cursor: Optional[AbstractFileBasedCursor],
-        use_file_transfer: bool = False,
+        parsed_config: AbstractFileBasedSpec,
     ) -> AbstractFileBasedStream:
         return DefaultFileBasedStream(
             config=stream_config,
@@ -310,7 +310,8 @@ class FileBasedSource(ConcurrentSourceAdapter, ABC):
             validation_policy=self._validate_and_get_validation_policy(stream_config),
             errors_collector=self.errors_collector,
             cursor=cursor,
-            use_file_transfer=use_file_transfer,
+            use_file_transfer=self._use_file_transfer(parsed_config),
+            preserve_directory_structure=self._preserve_directory_structure(parsed_config),
         )
 
     def _get_stream_from_catalog(
@@ -385,3 +386,25 @@ class FileBasedSource(ConcurrentSourceAdapter, ABC):
             and parsed_config.delivery_method.delivery_type == "use_file_transfer"
         )
         return use_file_transfer
+
+    @staticmethod
+    def _preserve_directory_structure(parsed_config: AbstractFileBasedSpec) -> bool:
+        """
+        Determines whether to preserve directory structure during file transfer.
+
+        When enabled, files maintain their subdirectory paths in the destination.
+        When disabled, files are flattened to the root of the destination.
+
+        Args:
+            parsed_config: The parsed configuration containing delivery method settings
+
+        Returns:
+            True if directory structure should be preserved (default), False otherwise
+        """
+        if (
+            FileBasedSource._use_file_transfer(parsed_config)
+            and hasattr(parsed_config.delivery_method, "preserve_directory_structure")
+            and parsed_config.delivery_method.preserve_directory_structure is not None
+        ):
+            return parsed_config.delivery_method.preserve_directory_structure
+        return True

airbyte_cdk/sources/file_based/file_based_stream_reader.py

@@ -135,6 +135,17 @@ class AbstractFileBasedStreamReader(ABC):
             return use_file_transfer
         return False
 
+    def preserve_directory_structure(self) -> bool:
+        # fall back to preserve subdirectories if config is not present or incomplete
+        if (
+            self.use_file_transfer()
+            and self.config
+            and hasattr(self.config.delivery_method, "preserve_directory_structure")
+            and self.config.delivery_method.preserve_directory_structure is not None
+        ):
+            return self.config.delivery_method.preserve_directory_structure
+        return True
+
     @abstractmethod
     def get_file(
         self, file: RemoteFile, local_directory: str, logger: logging.Logger
@@ -159,10 +170,13 @@ class AbstractFileBasedStreamReader(ABC):
         """
         ...
 
-    @staticmethod
-    def _get_file_transfer_paths(file: RemoteFile, local_directory: str) -> List[str]:
-        # Remove left slashes from source path format to make relative path for writing locally
-        file_relative_path = file.uri.lstrip("/")
+    def _get_file_transfer_paths(self, file: RemoteFile, local_directory: str) -> List[str]:
+        preserve_directory_structure = self.preserve_directory_structure()
+        if preserve_directory_structure:
+            # Remove left slashes from source path format to make relative path for writing locally
+            file_relative_path = file.uri.lstrip("/")
+        else:
+            file_relative_path = path.basename(file.uri)
         local_file_path = path.join(local_directory, file_relative_path)
 
         # Ensure the local directory exists

airbyte_cdk/sources/file_based/file_types/unstructured_parser.py

@@ -2,6 +2,7 @@
 # Copyright (c) 2023 Airbyte, Inc., all rights reserved.
 #
 import logging
+import os
 import traceback
 from datetime import datetime
 from io import BytesIO, IOBase
@@ -42,12 +43,34 @@ unstructured_partition_pdf = None
 unstructured_partition_docx = None
 unstructured_partition_pptx = None
 
+AIRBYTE_NLTK_DATA_DIR = "/airbyte/nltk_data"
+TMP_NLTK_DATA_DIR = "/tmp/nltk_data"
+
+
+def get_nltk_temp_folder() -> str:
+    """
+    For non-root connectors /tmp is not currently writable, but we should allow it in the future.
+    It's safe to use /airbyte for now. Fallback to /tmp for local development.
+    """
+    try:
+        nltk_data_dir = AIRBYTE_NLTK_DATA_DIR
+        os.makedirs(nltk_data_dir, exist_ok=True)
+    except OSError:
+        nltk_data_dir = TMP_NLTK_DATA_DIR
+        os.makedirs(nltk_data_dir, exist_ok=True)
+    return nltk_data_dir
+
+
 try:
+    nltk_data_dir = get_nltk_temp_folder()
+    nltk.data.path.append(nltk_data_dir)
     nltk.data.find("tokenizers/punkt.zip")
     nltk.data.find("tokenizers/punkt_tab.zip")
+    nltk.data.find("tokenizers/averaged_perceptron_tagger_eng.zip")
 except LookupError:
-    nltk.download("punkt")
-    nltk.download("punkt_tab")
+    nltk.download("punkt", download_dir=nltk_data_dir, quiet=True)
+    nltk.download("punkt_tab", download_dir=nltk_data_dir, quiet=True)
+    nltk.download("averaged_perceptron_tagger_eng", download_dir=nltk_data_dir, quiet=True)
 
 
 def optional_decode(contents: Union[str, bytes]) -> str:

airbyte_cdk/sources/file_based/stream/default_file_based_stream.py

@@ -5,14 +5,17 @@
 import asyncio
 import itertools
 import traceback
+from collections import defaultdict
 from copy import deepcopy
 from functools import cache
-from typing import Any, Dict, Iterable, List, Mapping, MutableMapping, Optional, Set, Union
+from os import path
+from typing import Any, Dict, Iterable, List, Mapping, MutableMapping, Optional, Set, Tuple, Union
 
 from airbyte_cdk.models import AirbyteLogMessage, AirbyteMessage, FailureType, Level
 from airbyte_cdk.models import Type as MessageType
 from airbyte_cdk.sources.file_based.config.file_based_stream_config import PrimaryKeyType
 from airbyte_cdk.sources.file_based.exceptions import (
+    DuplicatedFilesError,
     FileBasedSourceError,
     InvalidSchemaError,
     MissingSchemaError,
@@ -43,6 +46,8 @@ class DefaultFileBasedStream(AbstractFileBasedStream, IncrementalMixin):
     """
 
     FILE_TRANSFER_KW = "use_file_transfer"
+    PRESERVE_DIRECTORY_STRUCTURE_KW = "preserve_directory_structure"
+    FILES_KEY = "files"
     DATE_TIME_FORMAT = "%Y-%m-%dT%H:%M:%S.%fZ"
     ab_last_mod_col = "_ab_source_file_last_modified"
     ab_file_name_col = "_ab_source_file_url"
@@ -50,10 +55,15 @@ class DefaultFileBasedStream(AbstractFileBasedStream, IncrementalMixin):
     source_file_url = "source_file_url"
     airbyte_columns = [ab_last_mod_col, ab_file_name_col]
     use_file_transfer = False
+    preserve_directory_structure = True
 
     def __init__(self, **kwargs: Any):
         if self.FILE_TRANSFER_KW in kwargs:
             self.use_file_transfer = kwargs.pop(self.FILE_TRANSFER_KW, False)
+        if self.PRESERVE_DIRECTORY_STRUCTURE_KW in kwargs:
+            self.preserve_directory_structure = kwargs.pop(
+                self.PRESERVE_DIRECTORY_STRUCTURE_KW, True
+            )
         super().__init__(**kwargs)
 
     @property
@@ -98,15 +108,33 @@ class DefaultFileBasedStream(AbstractFileBasedStream, IncrementalMixin):
         else:
             return super()._filter_schema_invalid_properties(configured_catalog_json_schema)
 
+    def _duplicated_files_names(
+        self, slices: List[dict[str, List[RemoteFile]]]
+    ) -> List[dict[str, List[str]]]:
+        seen_file_names: Dict[str, List[str]] = defaultdict(list)
+        for file_slice in slices:
+            for file_found in file_slice[self.FILES_KEY]:
+                file_name = path.basename(file_found.uri)
+                seen_file_names[file_name].append(file_found.uri)
+        return [
+            {file_name: paths} for file_name, paths in seen_file_names.items() if len(paths) > 1
+        ]
+
     def compute_slices(self) -> Iterable[Optional[Mapping[str, Any]]]:
         # Sort files by last_modified, uri and return them grouped by last_modified
         all_files = self.list_files()
         files_to_read = self._cursor.get_files_to_sync(all_files, self.logger)
         sorted_files_to_read = sorted(files_to_read, key=lambda f: (f.last_modified, f.uri))
         slices = [
-            {"files": list(group[1])}
+            {self.FILES_KEY: list(group[1])}
             for group in itertools.groupby(sorted_files_to_read, lambda f: f.last_modified)
         ]
+        if slices and not self.preserve_directory_structure:
+            duplicated_files_names = self._duplicated_files_names(slices)
+            if duplicated_files_names:
+                raise DuplicatedFilesError(
+                    stream=self.name, duplicated_files_names=duplicated_files_names
+                )
         return slices
 
     def transform_record(

{airbyte_cdk-6.18.1.dist-info → airbyte_cdk-6.18.2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: airbyte-cdk
-Version: 6.18.1
+Version: 6.18.2
 Summary: A framework for writing Airbyte Connectors.
 License: MIT
 Keywords: airbyte,connector-development-kit,cdk

{airbyte_cdk-6.18.1.dist-info → airbyte_cdk-6.18.2.dist-info}/RECORD

@@ -196,7 +196,7 @@ airbyte_cdk/sources/file_based/availability_strategy/__init__.py,sha256=ddKQfUmk
 airbyte_cdk/sources/file_based/availability_strategy/abstract_file_based_availability_strategy.py,sha256=01Nd4b7ERAbp-OZo_8rrAzFXWPTMwr02SnWiN17nx8Q,2363
 airbyte_cdk/sources/file_based/availability_strategy/default_file_based_availability_strategy.py,sha256=j9T5TimfWFUz7nqsaj-83G3xWmDpsmeSbDnaUNmz0UM,5849
 airbyte_cdk/sources/file_based/config/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-airbyte_cdk/sources/file_based/config/abstract_file_based_spec.py,sha256=tj-M1L5BTa5yIQ3jHo09CtCTSq_eR-68zgyOPqwsurw,6455
+airbyte_cdk/sources/file_based/config/abstract_file_based_spec.py,sha256=gXlZwnEKLWknnK_n7j14lANgR6vkqhlLJ-G3rRu-ox4,6897
 airbyte_cdk/sources/file_based/config/avro_format.py,sha256=NxTF96ewzn6HuhgodsY7Rpb-ybr1ZEWW5d4Vid64g5A,716
 airbyte_cdk/sources/file_based/config/csv_format.py,sha256=NWekkyT8dTwiVK0mwa_krQD4FJPHSDfILo8kPAg3-Vs,8006
 airbyte_cdk/sources/file_based/config/excel_format.py,sha256=9qAmTsT6SoVzNfNv0oBVkVCmiyqQuVAbfRKajjoa7Js,378
@@ -207,9 +207,9 @@ airbyte_cdk/sources/file_based/config/unstructured_format.py,sha256=tIbB9Pn1HqU6
 airbyte_cdk/sources/file_based/discovery_policy/__init__.py,sha256=gl3ey6mZbyfraB9P3pFhf9UJp2JeTZ1SUFAopy2iBvY,301
 airbyte_cdk/sources/file_based/discovery_policy/abstract_discovery_policy.py,sha256=dCfXX529Rd5rtopg4VeEgTPJjFtqjtjzPq6LCw18Wt0,605
 airbyte_cdk/sources/file_based/discovery_policy/default_discovery_policy.py,sha256=-xujTidtrq6HC00WKbjQh1CZdT5LMuzkp5BLjqDmfTY,1007
-airbyte_cdk/sources/file_based/exceptions.py,sha256=AEELNIRzKPX6eopKd_2jhE7WiNeR0Aw7nQWVOL8fvkc,5760
-airbyte_cdk/sources/file_based/file_based_source.py,sha256=RfpctRNLJ_EHKKEc2E1EZGYRfhG0Z9o6TgsKS4XrSNY,16652
-airbyte_cdk/sources/file_based/file_based_stream_reader.py,sha256=ohxKlqPuV7TGwjyRy_gaWUol8QN5lBSoCYoaqBtRh1c,6179
+airbyte_cdk/sources/file_based/exceptions.py,sha256=WP0qkG6fpWoBpOyyicgp5YNE393VWyegq5qSy0v4QtM,7362
+airbyte_cdk/sources/file_based/file_based_source.py,sha256=Biv2QufYQtHZQCBZs4iCUpqTd82rk7xo8SDYkEeau3k,17616
+airbyte_cdk/sources/file_based/file_based_stream_reader.py,sha256=e1KhgTh7mzvkBOz9DjLwzOsDwevrTmbxSYIcvhgWgGM,6856
 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=XNx-JC-sgzH9u3nOJ2M59FxBXvtig8LN6BIkeDOavZA,10858
 airbyte_cdk/sources/file_based/file_types/csv_parser.py,sha256=QlCXB-ry3np67Q_VerQEPoWDOTcPTB6Go4ydZxY9ae4,20445
@@ -218,7 +218,7 @@ airbyte_cdk/sources/file_based/file_types/file_transfer.py,sha256=HyGRihJxcb_lEs
 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=r5FNcJadiI5PTyl1-doIodPCwW7xZWOTHl4Epd-w0-8,18602
+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/schema_helpers.py,sha256=Cf8FH1bDFP0qCDDfEYir_WjP4exXUnikz8hZ40y1Ek0,9601
 airbyte_cdk/sources/file_based/schema_validation_policies/__init__.py,sha256=FkByIyEy56x2_awYnxGPqGaOp7zAzpAoRkPZHKySI9M,536
@@ -235,7 +235,7 @@ airbyte_cdk/sources/file_based/stream/concurrent/cursor/file_based_final_state_c
 airbyte_cdk/sources/file_based/stream/cursor/__init__.py,sha256=MhFB5hOo8sjwvCh8gangaymdg3EJWYt_72brFOZt068,191
 airbyte_cdk/sources/file_based/stream/cursor/abstract_file_based_cursor.py,sha256=om-x3gZFPgWDpi15S9RxZmR36VHnk8sytgN6LlBQhAw,1934
 airbyte_cdk/sources/file_based/stream/cursor/default_file_based_cursor.py,sha256=VGV7xLyBribuBMVrXtO1xqkWJD86bl7yhXtjnwLMohM,7051
-airbyte_cdk/sources/file_based/stream/default_file_based_stream.py,sha256=rpwU6AOyhFLuXtcFKkcOHFWbRQ4kLCOKzAjcID_M87k,16770
+airbyte_cdk/sources/file_based/stream/default_file_based_stream.py,sha256=XLU5cNqQ-5mj243gNzMyXtm_oCtg1ORyoqbCsUo9Dn4,18044
 airbyte_cdk/sources/file_based/types.py,sha256=INxG7OPnkdUP69oYNKMAbwhvV1AGvLRHs1J6pIia2FI,218
 airbyte_cdk/sources/http_config.py,sha256=OBZeuyFilm6NlDlBhFQvHhTWabEvZww6OHDIlZujIS0,730
 airbyte_cdk/sources/http_logger.py,sha256=TyBmtRA6D9g0XDkKGvdM415b36RXDjgfkwRewDsH8-0,1576
@@ -343,8 +343,8 @@ airbyte_cdk/utils/slice_hasher.py,sha256=-pHexlNYoWYPnXNH-M7HEbjmeJe9Zk7SJijdQ7d
 airbyte_cdk/utils/spec_schema_transformations.py,sha256=-5HTuNsnDBAhj-oLeQXwpTGA0HdcjFOf2zTEMUTTg_Y,816
 airbyte_cdk/utils/stream_status_utils.py,sha256=ZmBoiy5HVbUEHAMrUONxZvxnvfV9CesmQJLDTAIWnWw,1171
 airbyte_cdk/utils/traced_exception.py,sha256=C8uIBuCL_E4WnBAOPSxBicD06JAldoN9fGsQDp463OY,6292
-airbyte_cdk-6.18.1.dist-info/LICENSE.txt,sha256=Wfe61S4BaGPj404v8lrAbvhjYR68SHlkzeYrg3_bbuM,1051
-airbyte_cdk-6.18.1.dist-info/METADATA,sha256=OMpca59Gc1MJOlwEgvDJX0uwp7skSel83qkbtcan6hE,6000
-airbyte_cdk-6.18.1.dist-info/WHEEL,sha256=IYZQI976HJqqOpQU6PHkJ8fb3tMNBFjg-Cn-pwAbaFM,88
-airbyte_cdk-6.18.1.dist-info/entry_points.txt,sha256=fj-e3PAQvsxsQzyyq8UkG1k8spunWnD4BAH2AwlR6NM,95
-airbyte_cdk-6.18.1.dist-info/RECORD,,
+airbyte_cdk-6.18.2.dist-info/LICENSE.txt,sha256=Wfe61S4BaGPj404v8lrAbvhjYR68SHlkzeYrg3_bbuM,1051
+airbyte_cdk-6.18.2.dist-info/METADATA,sha256=Uwnd37XQTCsP3n7GzWDM5mQJsphngbGzpyvqnRyMt7I,6000
+airbyte_cdk-6.18.2.dist-info/WHEEL,sha256=IYZQI976HJqqOpQU6PHkJ8fb3tMNBFjg-Cn-pwAbaFM,88
+airbyte_cdk-6.18.2.dist-info/entry_points.txt,sha256=fj-e3PAQvsxsQzyyq8UkG1k8spunWnD4BAH2AwlR6NM,95
+airbyte_cdk-6.18.2.dist-info/RECORD,,