airbyte-cdk 6.38.3.dev4100__py3-none-any.whl → 6.38.4__py3-none-any.whl

airbyte_cdk/sources/file_based/file_based_source.py

@@ -48,9 +48,6 @@ from airbyte_cdk.sources.file_based.exceptions import (
     FileBasedErrorsCollector,
     FileBasedSourceError,
 )
-from airbyte_cdk.sources.file_based.file_based_stream_permissions_reader import (
-    AbstractFileBasedStreamPermissionsReader,
-)
 from airbyte_cdk.sources.file_based.file_based_stream_reader import AbstractFileBasedStreamReader
 from airbyte_cdk.sources.file_based.file_types import default_parsers
 from airbyte_cdk.sources.file_based.file_types.file_type_parser import FileTypeParser
@@ -103,10 +100,8 @@ class FileBasedSource(ConcurrentSourceAdapter, ABC):
         cursor_cls: Type[
             Union[AbstractConcurrentFileBasedCursor, AbstractFileBasedCursor]
         ] = FileBasedConcurrentCursor,
-        stream_permissions_reader: Optional[AbstractFileBasedStreamPermissionsReader] = None,
     ):
         self.stream_reader = stream_reader
-        self.stream_permissions_reader = stream_permissions_reader
         self.spec_class = spec_class
         self.config = config
         self.catalog = catalog
@@ -342,23 +337,9 @@ class FileBasedSource(ConcurrentSourceAdapter, ABC):
             preserve_directory_structure=preserve_directory_structure(parsed_config),
         )
 
-    def _ensure_permissions_reader_available(self) -> None:
-        """
-        Validates that a stream permissions reader is available.
-        Raises a ValueError if the reader is not provided.
-        """
-        if not self.stream_permissions_reader:
-            raise ValueError(
-                "Stream permissions reader is required for streams that use permissions transfer mode."
-            )
-
     def _make_permissions_stream(
         self, stream_config: FileBasedStreamConfig, cursor: Optional[AbstractFileBasedCursor]
     ) -> AbstractFileBasedStream:
-        """
-        Creates a stream that reads permissions from files.
-        """
-        self._ensure_permissions_reader_available()
         return PermissionsFileBasedStream(
             config=stream_config,
             catalog_schema=self.stream_schemas.get(stream_config.name),
@@ -369,7 +350,6 @@ class FileBasedSource(ConcurrentSourceAdapter, ABC):
             validation_policy=self._validate_and_get_validation_policy(stream_config),
             errors_collector=self.errors_collector,
             cursor=cursor,
-            stream_permissions_reader=self.stream_permissions_reader,  # type: ignore
         )
 
     def _make_file_based_stream(
@@ -390,10 +370,9 @@ class FileBasedSource(ConcurrentSourceAdapter, ABC):
     def _make_identities_stream(
         self,
     ) -> Stream:
-        self._ensure_permissions_reader_available()
         return FileIdentitiesStream(
             catalog_schema=self.stream_schemas.get(FileIdentitiesStream.IDENTITIES_STREAM_NAME),
-            stream_permissions_reader=self.stream_permissions_reader,  # type: ignore
+            stream_reader=self.stream_reader,
             discovery_policy=self.discovery_policy,
             errors_collector=self.errors_collector,
         )

airbyte_cdk/sources/file_based/file_based_stream_reader.py

@@ -184,3 +184,97 @@ class AbstractFileBasedStreamReader(ABC):
         makedirs(path.dirname(local_file_path), exist_ok=True)
         absolute_file_path = path.abspath(local_file_path)
         return [file_relative_path, local_file_path, absolute_file_path]
+
+    @abstractmethod
+    def get_file_acl_permissions(self, file: RemoteFile, logger: logging.Logger) -> Dict[str, Any]:
+        """
+        This function should return the allow list for a given file, i.e. the list of all identities and their permission levels associated with it
+
+        e.g.
+        def get_file_acl_permissions(self, file: RemoteFile, logger: logging.Logger):
+            api_conn = some_api.conn(credentials=SOME_CREDENTIALS)
+            result = api_conn.get_file_permissions_info(file.id)
+            return MyPermissionsModel(
+                id=result["id"],
+                access_control_list = result["access_control_list"],
+                is_public = result["is_public"],
+                ).dict()
+        """
+        raise NotImplementedError(
+            f"{self.__class__.__name__} does not implement get_file_acl_permissions(). To support ACL permissions, implement this method and update file_permissions_schema."
+        )
+
+    @abstractmethod
+    def load_identity_groups(self, logger: logging.Logger) -> Iterable[Dict[str, Any]]:
+        """
+        This function should return the Identities in a determined "space" or "domain" where the file metadata (ACLs) are fetched and ACLs items (Identities) exists.
+
+        e.g.
+        def load_identity_groups(self, logger: logging.Logger) -> Dict[str, Any]:
+            api_conn = some_api.conn(credentials=SOME_CREDENTIALS)
+            users_api = api_conn.users()
+            groups_api = api_conn.groups()
+            members_api = self.google_directory_service.members()
+            for user in users_api.list():
+                yield my_identity_model(id=user.id, name=user.name, email_address=user.email, type="user").dict()
+            for group in groups_api.list():
+                group_obj = my_identity_model(id=group.id, name=groups.name, email_address=user.email, type="group").dict()
+                for member in members_api.list(group=group):
+                    group_obj.member_email_addresses = group_obj.member_email_addresses or []
+                    group_obj.member_email_addresses.append(member.email)
+                yield group_obj.dict()
+        """
+        raise NotImplementedError(
+            f"{self.__class__.__name__} does not implement load_identity_groups(). To support identities, implement this method and update identities_schema."
+        )
+
+    @property
+    @abstractmethod
+    def file_permissions_schema(self) -> Dict[str, Any]:
+        """
+        This function should return the permissions schema for file permissions stream.
+
+        e.g.
+        def file_permissions_schema(self) -> Dict[str, Any]:
+            # you can also follow the patter we have for python connectors and have a json file and read from there e.g. schemas/identities.json
+            return {
+                  "type": "object",
+                  "properties": {
+                    "id": { "type": "string" },
+                    "file_path": { "type": "string" },
+                    "access_control_list": {
+                      "type": "array",
+                      "items": { "type": "string" }
+                    },
+                    "publicly_accessible": { "type": "boolean" }
+                  }
+                }
+        """
+        raise NotImplementedError(
+            f"{self.__class__.__name__} does not implement file_permissions_schema, please return json schema for your permissions streams."
+        )
+
+    @property
+    @abstractmethod
+    def identities_schema(self) -> Dict[str, Any]:
+        """
+        This function should return the identities schema for file identity stream.
+
+        e.g.
+        def identities_schema(self) -> Dict[str, Any]:
+            # you can also follow the patter we have for python connectors and have a json file and read from there e.g. schemas/identities.json
+            return {
+              "type": "object",
+              "properties": {
+                "id": { "type": "string" },
+                "remote_id": { "type": "string" },
+                "name": { "type": ["null", "string"] },
+                "email_address": { "type": ["null", "string"] },
+                "member_email_addresses": { "type": ["null", "array"] },
+                "type": { "type": "string" },
+              }
+            }
+        """
+        raise NotImplementedError(
+            f"{self.__class__.__name__} does not implement identities_schema, please return json schema for your identities stream."
+        )

airbyte_cdk/sources/file_based/stream/identities_stream.py

@@ -8,9 +8,7 @@ from typing import Any, Dict, Iterable, Mapping, MutableMapping, Optional
 from airbyte_cdk.sources.file_based.config.file_based_stream_config import PrimaryKeyType
 from airbyte_cdk.sources.file_based.discovery_policy import AbstractDiscoveryPolicy
 from airbyte_cdk.sources.file_based.exceptions import FileBasedErrorsCollector
-from airbyte_cdk.sources.file_based.file_based_stream_permissions_reader import (
-    AbstractFileBasedStreamPermissionsReader,
-)
+from airbyte_cdk.sources.file_based.file_based_stream_reader import AbstractFileBasedStreamReader
 from airbyte_cdk.sources.streams.core import JsonSchema
 from airbyte_cdk.sources.streams.permissions.identities_stream import IdentitiesStream
 
@@ -26,13 +24,13 @@ class FileIdentitiesStream(IdentitiesStream):
     def __init__(
         self,
         catalog_schema: Optional[Mapping[str, Any]],
-        stream_permissions_reader: AbstractFileBasedStreamPermissionsReader,
+        stream_reader: AbstractFileBasedStreamReader,
         discovery_policy: AbstractDiscoveryPolicy,
         errors_collector: FileBasedErrorsCollector,
     ) -> None:
         super().__init__()
         self.catalog_schema = catalog_schema
-        self.stream_permissions_reader = stream_permissions_reader
+        self.stream_reader = stream_reader
         self._discovery_policy = discovery_policy
         self.errors_collector = errors_collector
         self._cursor: MutableMapping[str, Any] = {}
@@ -42,8 +40,8 @@ class FileIdentitiesStream(IdentitiesStream):
         return None
 
     def load_identity_groups(self) -> Iterable[Dict[str, Any]]:
-        return self.stream_permissions_reader.load_identity_groups(logger=self.logger)
+        return self.stream_reader.load_identity_groups(logger=self.logger)
 
     @cache
     def get_json_schema(self) -> JsonSchema:
-        return self.stream_permissions_reader.identities_schema
+        return self.stream_reader.identities_schema

airbyte_cdk/sources/file_based/stream/permissions_file_based_stream.py

@@ -7,9 +7,6 @@ from typing import Any, Dict, Iterable
 
 from airbyte_cdk.models import AirbyteLogMessage, AirbyteMessage, Level
 from airbyte_cdk.models import Type as MessageType
-from airbyte_cdk.sources.file_based.file_based_stream_permissions_reader import (
-    AbstractFileBasedStreamPermissionsReader,
-)
 from airbyte_cdk.sources.file_based.stream import DefaultFileBasedStream
 from airbyte_cdk.sources.file_based.types import StreamSlice
 from airbyte_cdk.sources.streams.core import JsonSchema
@@ -29,16 +26,10 @@ class PermissionsFileBasedStream(DefaultFileBasedStream):
     and schema definition, while this class handles the streaming interface.
     """
 
-    def __init__(
-        self, stream_permissions_reader: AbstractFileBasedStreamPermissionsReader, **kwargs: Any
-    ):
-        super().__init__(**kwargs)
-        self.stream_permissions_reader = stream_permissions_reader
-
     def _filter_schema_invalid_properties(
         self, configured_catalog_json_schema: Dict[str, Any]
     ) -> Dict[str, Any]:
-        return self.stream_permissions_reader.file_permissions_schema
+        return self.stream_reader.file_permissions_schema
 
     def read_records_from_slice(self, stream_slice: StreamSlice) -> Iterable[AirbyteMessage]:
         """
@@ -49,7 +40,7 @@ class PermissionsFileBasedStream(DefaultFileBasedStream):
             no_permissions = False
             file_datetime_string = file.last_modified.strftime(self.DATE_TIME_FORMAT)
             try:
-                permissions_record = self.stream_permissions_reader.get_file_acl_permissions(
+                permissions_record = self.stream_reader.get_file_acl_permissions(
                     file, logger=self.logger
                 )
                 if not permissions_record:
@@ -91,4 +82,4 @@ class PermissionsFileBasedStream(DefaultFileBasedStream):
         Returns:
            The file permissions schema that defines the structure of permission records
         """
-        return self.stream_permissions_reader.file_permissions_schema
+        return self.stream_reader.file_permissions_schema

{airbyte_cdk-6.38.3.dev4100.dist-info → airbyte_cdk-6.38.4.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: airbyte-cdk
-Version: 6.38.3.dev4100
+Version: 6.38.4
 Summary: A framework for writing Airbyte Connectors.
 Home-page: https://airbyte.com
 License: MIT

{airbyte_cdk-6.38.3.dev4100.dist-info → airbyte_cdk-6.38.4.dist-info}/RECORD

@@ -214,9 +214,8 @@ airbyte_cdk/sources/file_based/discovery_policy/__init__.py,sha256=gl3ey6mZbyfra
 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=WP0qkG6fpWoBpOyyicgp5YNE393VWyegq5qSy0v4QtM,7362
-airbyte_cdk/sources/file_based/file_based_source.py,sha256=HG4wok4kTEQYzoMxa3u_JwkgdUp_eHkqFMEx5Is7jOM,19934
-airbyte_cdk/sources/file_based/file_based_stream_permissions_reader.py,sha256=hwwVTi5Ue5cPCzVM8jWOVbsTdjVRW4eDyVv8JMWFnyc,4846
-airbyte_cdk/sources/file_based/file_based_stream_reader.py,sha256=0cmppYO3pZlFiJrs5oorF4JXv4ErhOeEMrdLG7P-Gdk,6742
+airbyte_cdk/sources/file_based/file_based_source.py,sha256=JXfwc9KaW7PvjAbm2GJ7Ra3DJnCZH4KaE3WytYvtM1Q,18925
+airbyte_cdk/sources/file_based/file_based_stream_reader.py,sha256=d2UZ3C8M-A591KvBvg8kDpVdpox0rKVlRhVy5bi-auc,11209
 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
@@ -243,8 +242,8 @@ airbyte_cdk/sources/file_based/stream/cursor/__init__.py,sha256=MhFB5hOo8sjwvCh8
 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=XLU5cNqQ-5mj243gNzMyXtm_oCtg1ORyoqbCsUo9Dn4,18044
-airbyte_cdk/sources/file_based/stream/identities_stream.py,sha256=FZH83Geoy3K3nwUk2VVNJERFcXUTnl-4XljjucUM23s,1893
-airbyte_cdk/sources/file_based/stream/permissions_file_based_stream.py,sha256=ke82qgm7snOlQTDx94Lqsc0cDkHWi3OJDTrPxffpFqc,3914
+airbyte_cdk/sources/file_based/stream/identities_stream.py,sha256=DwgNU-jDp5vZ_WloQSUzBciDnAFMo8bXPjXpQx5-eko,1790
+airbyte_cdk/sources/file_based/stream/permissions_file_based_stream.py,sha256=i0Jn0zuAPomLa4pHSu9TQ3gAN5xXhNzPTYVwUDiDEyE,3523
 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=H93kPAujHhPmXNX0JSFG3D-SL6yEFA5PtKot9Hu3TYA,1690
@@ -358,9 +357,9 @@ airbyte_cdk/utils/slice_hasher.py,sha256=EDxgROHDbfG-QKQb59m7h_7crN1tRiawdf5uU7G
 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-6.38.3.dev4100.dist-info/LICENSE.txt,sha256=Wfe61S4BaGPj404v8lrAbvhjYR68SHlkzeYrg3_bbuM,1051
-airbyte_cdk-6.38.3.dev4100.dist-info/LICENSE_SHORT,sha256=aqF6D1NcESmpn-cqsxBtszTEnHKnlsp8L4x9wAh3Nxg,55
-airbyte_cdk-6.38.3.dev4100.dist-info/METADATA,sha256=w60PKl3DfFjzvG3uKUtLCCEf0s83i-osbdqy2doggrc,6021
-airbyte_cdk-6.38.3.dev4100.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
-airbyte_cdk-6.38.3.dev4100.dist-info/entry_points.txt,sha256=fj-e3PAQvsxsQzyyq8UkG1k8spunWnD4BAH2AwlR6NM,95
-airbyte_cdk-6.38.3.dev4100.dist-info/RECORD,,
+airbyte_cdk-6.38.4.dist-info/LICENSE.txt,sha256=Wfe61S4BaGPj404v8lrAbvhjYR68SHlkzeYrg3_bbuM,1051
+airbyte_cdk-6.38.4.dist-info/LICENSE_SHORT,sha256=aqF6D1NcESmpn-cqsxBtszTEnHKnlsp8L4x9wAh3Nxg,55
+airbyte_cdk-6.38.4.dist-info/METADATA,sha256=r625YvczxwtgbhW8dS2s_xslYC-2sxWgmj_FkGvHO0Q,6013
+airbyte_cdk-6.38.4.dist-info/WHEEL,sha256=Nq82e9rUAnEjt98J6MlVmMCZb-t9cYE2Ir1kpBmnWfs,88
+airbyte_cdk-6.38.4.dist-info/entry_points.txt,sha256=fj-e3PAQvsxsQzyyq8UkG1k8spunWnD4BAH2AwlR6NM,95
+airbyte_cdk-6.38.4.dist-info/RECORD,,

airbyte_cdk/sources/file_based/file_based_stream_permissions_reader.py

@@ -1,109 +0,0 @@
-#
-# Copyright (c) 2025 Airbyte, Inc., all rights reserved.
-#
-
-import logging
-from abc import ABC, abstractmethod
-from typing import Any, Dict, Iterable
-
-from airbyte_cdk.sources.file_based.remote_file import RemoteFile
-
-
-class AbstractFileBasedStreamPermissionsReader(ABC):
-    """
-    This class is responsible for reading file permissions and Identities from a source.
-    """
-
-    @abstractmethod
-    def get_file_acl_permissions(self, file: RemoteFile, logger: logging.Logger) -> Dict[str, Any]:
-        """
-        This function should return the allow list for a given file, i.e. the list of all identities and their permission levels associated with it
-
-        e.g.
-        def get_file_acl_permissions(self, file: RemoteFile, logger: logging.Logger):
-            api_conn = some_api.conn(credentials=SOME_CREDENTIALS)
-            result = api_conn.get_file_permissions_info(file.id)
-            return MyPermissionsModel(
-                id=result["id"],
-                access_control_list = result["access_control_list"],
-                is_public = result["is_public"],
-                ).dict()
-        """
-        raise NotImplementedError(
-            f"{self.__class__.__name__} does not implement get_file_acl_permissions(). To support ACL permissions, implement this method and update file_permissions_schema."
-        )
-
-    @abstractmethod
-    def load_identity_groups(self, logger: logging.Logger) -> Iterable[Dict[str, Any]]:
-        """
-        This function should return the Identities in a determined "space" or "domain" where the file metadata (ACLs) are fetched and ACLs items (Identities) exists.
-
-        e.g.
-        def load_identity_groups(self, logger: logging.Logger) -> Dict[str, Any]:
-            api_conn = some_api.conn(credentials=SOME_CREDENTIALS)
-            users_api = api_conn.users()
-            groups_api = api_conn.groups()
-            members_api = self.google_directory_service.members()
-            for user in users_api.list():
-                yield my_identity_model(id=user.id, name=user.name, email_address=user.email, type="user").dict()
-            for group in groups_api.list():
-                group_obj = my_identity_model(id=group.id, name=groups.name, email_address=user.email, type="group").dict()
-                for member in members_api.list(group=group):
-                    group_obj.member_email_addresses = group_obj.member_email_addresses or []
-                    group_obj.member_email_addresses.append(member.email)
-                yield group_obj.dict()
-        """
-        raise NotImplementedError(
-            f"{self.__class__.__name__} does not implement load_identity_groups(). To support identities, implement this method and update identities_schema."
-        )
-
-    @property
-    @abstractmethod
-    def file_permissions_schema(self) -> Dict[str, Any]:
-        """
-        This function should return the permissions schema for file permissions stream.
-
-        e.g.
-        def file_permissions_schema(self) -> Dict[str, Any]:
-            # you can also follow the patter we have for python connectors and have a json file and read from there e.g. schemas/identities.json
-            return {
-                  "type": "object",
-                  "properties": {
-                    "id": { "type": "string" },
-                    "file_path": { "type": "string" },
-                    "access_control_list": {
-                      "type": "array",
-                      "items": { "type": "string" }
-                    },
-                    "publicly_accessible": { "type": "boolean" }
-                  }
-                }
-        """
-        raise NotImplementedError(
-            f"{self.__class__.__name__} does not implement file_permissions_schema, please return json schema for your permissions streams."
-        )
-
-    @property
-    @abstractmethod
-    def identities_schema(self) -> Dict[str, Any]:
-        """
-        This function should return the identities schema for file identity stream.
-
-        e.g.
-        def identities_schema(self) -> Dict[str, Any]:
-            # you can also follow the patter we have for python connectors and have a json file and read from there e.g. schemas/identities.json
-            return {
-              "type": "object",
-              "properties": {
-                "id": { "type": "string" },
-                "remote_id": { "type": "string" },
-                "name": { "type": ["null", "string"] },
-                "email_address": { "type": ["null", "string"] },
-                "member_email_addresses": { "type": ["null", "array"] },
-                "type": { "type": "string" },
-              }
-            }
-        """
-        raise NotImplementedError(
-            f"{self.__class__.__name__} does not implement identities_schema, please return json schema for your identities stream."
-        )