pangea-sdk 6.2.0b1__py3-none-any.whl → 6.3.0__py3-none-any.whl

pangea/__init__.py

@@ -1,7 +1,15 @@
-__version__ = "6.2.0beta1"
+__version__ = "6.3.0"
 
 from pangea.asyncio.request import PangeaRequestAsync
 from pangea.config import PangeaConfig
 from pangea.file_uploader import FileUploader
 from pangea.request import PangeaRequest
 from pangea.response import PangeaResponse
+
+__all__ = (
+    "FileUploader",
+    "PangeaConfig",
+    "PangeaRequest",
+    "PangeaRequestAsync",
+    "PangeaResponse",
+)

pangea/asyncio/__init__.py

@@ -1 +1,2 @@
+# ruff: noqa: F401
 from .file_uploader import FileUploaderAsync

pangea/asyncio/file_uploader.py

@@ -1,8 +1,10 @@
 # Copyright 2022 Pangea Cyber Corporation
 # Author: Pangea Cyber Corporation
+
+from __future__ import annotations
+
 import io
 import logging
-from typing import Dict, Optional
 
 from pangea.asyncio.request import PangeaRequestAsync
 from pangea.request import PangeaConfig
@@ -24,7 +26,7 @@ class FileUploaderAsync:
         url: str,
         file: io.BufferedReader,
         transfer_method: TransferMethod = TransferMethod.PUT_URL,
-        file_details: Optional[Dict] = None,
+        file_details: dict | None = None,
     ) -> None:
         if transfer_method == TransferMethod.PUT_URL:
             files = [("file", ("filename", file, "application/octet-stream"))]

pangea/asyncio/request.py

@@ -1,24 +1,43 @@
 # Copyright 2022 Pangea Cyber Corporation
 # Author: Pangea Cyber Corporation
+
+# TODO: Modernize.
+# ruff: noqa: UP006, UP035
+
 from __future__ import annotations
 
 import asyncio
 import json
 import time
 from collections.abc import Iterable, Mapping
-from typing import Dict, List, Literal, Optional, Sequence, Tuple, Type, Union, cast, overload
+from typing import Dict, List, Optional, Sequence, Tuple, Type, Union, cast
 
 import aiohttp
 from aiohttp import FormData
-from pydantic import BaseModel, TypeAdapter
+from pydantic import BaseModel
 from pydantic_core import to_jsonable_python
-from typing_extensions import Any, TypeVar
+from typing_extensions import Any, TypeAlias, TypeVar, override
 
 import pangea.exceptions as pe
 from pangea.request import MultipartResponse, PangeaRequestBase
 from pangea.response import AttachedFile, PangeaResponse, PangeaResponseResult, ResponseStatus, TransferMethod
 from pangea.utils import default_encoder
 
+_FileName: TypeAlias = Union[str, None]
+_FileContent: TypeAlias = Union[str, bytes]
+_FileContentType: TypeAlias = str
+_FileCustomHeaders: TypeAlias = Mapping[str, str]
+_FileSpecTuple2: TypeAlias = tuple[_FileName, _FileContent]
+_FileSpecTuple3: TypeAlias = tuple[_FileName, _FileContent, _FileContentType]
+_FileSpecTuple4: TypeAlias = tuple[_FileName, _FileContent, _FileContentType, _FileCustomHeaders]
+_FileSpec: TypeAlias = Union[_FileContent, _FileSpecTuple2, _FileSpecTuple3, _FileSpecTuple4]
+_Files: TypeAlias = Union[Mapping[str, _FileSpec], Iterable[tuple[str, _FileSpec]]]
+_LooseHeaders = Union[
+    Mapping[str, str],
+    Iterable[tuple[str, str]],
+]
+
+
 TResult = TypeVar("TResult", bound=PangeaResponseResult)
 
 
@@ -31,86 +50,26 @@ class PangeaRequestAsync(PangeaRequestBase):
     be set in PangeaConfig.
     """
 
-    async def delete(self, endpoint: str) -> None:
-        """
-        Makes a DELETE call to a Pangea endpoint.
-
-        Args:
-            endpoint: The Pangea API endpoint.
-        """
-
-        url = self._url(endpoint)
-
-        self.logger.debug(
-            json.dumps({"service": self.service, "action": "delete", "url": url}, default=default_encoder)
-        )
-
-        requests_response = await self._http_delete(url, headers=self._headers())
-        await self._check_http_errors(requests_response)
-
-    @overload
     async def post(
         self,
         endpoint: str,
         result_class: Type[TResult],
-        data: str | BaseModel | dict[str, Any] | None = None,
+        data: str | BaseModel | Mapping[str, Any] | None = None,
         files: Optional[List[Tuple]] = None,
         poll_result: bool = True,
         url: Optional[str] = None,
-        *,
-        pangea_response: Literal[True] = True,
     ) -> PangeaResponse[TResult]:
-        """
-        Makes a POST call to a Pangea Service endpoint.
+        """Makes the POST call to a Pangea Service endpoint.
 
         Args:
-            endpoint: The Pangea Service API endpoint.
-            data: The POST body payload object
+            endpoint(str): The Pangea Service API endpoint.
+            data(dict): The POST body payload object
 
         Returns:
             PangeaResponse which contains the response in its entirety and
                various properties to retrieve individual fields
         """
 
-    @overload
-    async def post(
-        self,
-        endpoint: str,
-        result_class: Type[TResult],
-        data: str | BaseModel | dict[str, Any] | None = None,
-        files: Optional[List[Tuple]] = None,
-        poll_result: bool = True,
-        url: Optional[str] = None,
-        *,
-        pangea_response: Literal[False],
-    ) -> TResult:
-        """
-        Makes a POST call to a Pangea Service endpoint.
-
-        Args:
-            endpoint: The Pangea Service API endpoint.
-            data: The POST body payload object
-        """
-
-    async def post(
-        self,
-        endpoint: str,
-        result_class: Type[TResult],
-        data: str | BaseModel | dict[str, Any] | None = None,
-        files: Optional[List[Tuple]] = None,
-        poll_result: bool = True,
-        url: Optional[str] = None,
-        *,
-        pangea_response: bool = True,
-    ) -> PangeaResponse[TResult] | TResult:
-        """
-        Makes a POST call to a Pangea Service endpoint.
-
-        Args:
-            endpoint: The Pangea Service API endpoint.
-            data: The POST body payload object
-        """
-
         if isinstance(data, BaseModel):
             data = data.model_dump(exclude_none=True)
 
@@ -122,7 +81,7 @@ class PangeaRequestAsync(PangeaRequestBase):
             data = {}
 
         # Normalize.
-        data = cast(dict[str, Any], to_jsonable_python(data))
+        data = cast(dict[str, Any], to_jsonable_python(data, exclude_none=True))
 
         if url is None:
             url = self._url(endpoint)
@@ -141,19 +100,18 @@ class PangeaRequestAsync(PangeaRequestBase):
                 endpoint, result_class=result_class, data=data, files=files
             )
         else:
+            headers = self._headers()
+            if transfer_method == TransferMethod.MULTIPART.value:
+                del headers["Content-Type"]
             requests_response = await self._http_post(
-                url, headers=self._headers(), data=data, files=files, presigned_url_post=False
+                url, headers=headers, data=data, files=files, presigned_url_post=False
             )
 
         await self._check_http_errors(requests_response)
 
-        if not pangea_response:
-            type_adapter = TypeAdapter(result_class)
-            return type_adapter.validate_python(await requests_response.json())
-
         if "multipart/form-data" in requests_response.headers.get("content-type", ""):
             multipart_response = await self._process_multipart_response(requests_response)
-            pangea_response_obj: PangeaResponse = PangeaResponse(
+            pangea_response: PangeaResponse = PangeaResponse(
                 requests_response,
                 result_class=result_class,
                 json=multipart_response.pangea_json,
@@ -166,108 +124,49 @@ class PangeaRequestAsync(PangeaRequestBase):
                     json.dumps({"service": self.service, "action": "post", "url": url, "response": json_resp})
                 )
 
-                pangea_response_obj = PangeaResponse(requests_response, result_class=result_class, json=json_resp)
+                pangea_response = PangeaResponse(requests_response, result_class=result_class, json=json_resp)
             except aiohttp.ContentTypeError as e:
-                raise pe.PangeaException(f"Failed to decode json response. {e}. Body: {await requests_response.text()}")
+                raise pe.PangeaException(
+                    f"Failed to decode json response. {e}. Body: {await requests_response.text()}"
+                ) from e
 
         if poll_result:
-            pangea_response_obj = await self._handle_queued_result(pangea_response_obj)
+            pangea_response = await self._handle_queued_result(pangea_response)
 
-        return self._check_response(pangea_response_obj)
+        return self._check_response(pangea_response)
 
-    @overload
-    async def get(
-        self,
-        path: str,
-        result_class: Type[TResult],
-        check_response: bool = True,
-        *,
-        params: (
-            Mapping[str | bytes | int | float, str | bytes | int | float | Iterable[str | bytes | int | float] | None]
-            | None
-        ) = None,
-        pangea_response: Literal[True] = True,
-    ) -> PangeaResponse[TResult]:
-        """
-        Makes the GET call to a Pangea Service endpoint.
+    async def get(self, path: str, result_class: Type[TResult], check_response: bool = True) -> PangeaResponse[TResult]:
+        """Makes the GET call to a Pangea Service endpoint.
 
         Args:
-            path: Additional URL path
-            params: Dictionary of querystring data to attach to the request
+            endpoint(str): The Pangea Service API endpoint.
+            path(str): Additional URL path
 
         Returns:
             PangeaResponse which contains the response in its entirety and
                various properties to retrieve individual fields
         """
 
-    @overload
-    async def get(
-        self,
-        path: str,
-        result_class: Type[TResult],
-        check_response: bool = True,
-        *,
-        params: (
-            Mapping[str | bytes | int | float, str | bytes | int | float | Iterable[str | bytes | int | float] | None]
-            | None
-        ) = None,
-        pangea_response: Literal[False] = False,
-    ) -> TResult:
-        """
-        Makes the GET call to a Pangea Service endpoint.
-
-        Args:
-            path: Additional URL path
-            params: Dictionary of querystring data to attach to the request
-        """
-
-    async def get(
-        self,
-        path: str,
-        result_class: Type[TResult],
-        check_response: bool = True,
-        *,
-        params: (
-            Mapping[str | bytes | int | float, str | bytes | int | float | Iterable[str | bytes | int | float] | None]
-            | None
-        ) = None,
-        pangea_response: bool = True,
-    ) -> PangeaResponse[TResult] | TResult:
-        """
-        Makes the GET call to a Pangea Service endpoint.
-
-        Args:
-            path: Additional URL path
-            params: Dictionary of querystring data to attach to the request
-            pangea_response: Whether or not the response body follows Pangea's
-              standard response schema
-        """
-
         url = self._url(path)
         self.logger.debug(json.dumps({"service": self.service, "action": "get", "url": url}))
 
-        async with self.session.get(url, params=params, headers=self._headers()) as requests_response:
+        async with self.session.get(url, headers=self._headers()) as requests_response:
             await self._check_http_errors(requests_response)
-
-            if not pangea_response:
-                type_adapter = TypeAdapter(result_class)
-                return type_adapter.validate_python(await requests_response.json())
-
-            pangea_response_obj = PangeaResponse(
+            pangea_response = PangeaResponse(
                 requests_response, result_class=result_class, json=await requests_response.json()
             )
 
         self.logger.debug(
             json.dumps(
-                {"service": self.service, "action": "get", "url": url, "response": pangea_response_obj.json},
+                {"service": self.service, "action": "get", "url": url, "response": pangea_response.json},
                 default=default_encoder,
             )
         )
 
         if check_response is False:
-            return pangea_response_obj
+            return pangea_response
 
-        return self._check_response(pangea_response_obj)
+        return self._check_response(pangea_response)
 
     async def _check_http_errors(self, resp: aiohttp.ClientResponse):
         if resp.status == 503:
@@ -290,7 +189,7 @@ class PangeaRequestAsync(PangeaRequestBase):
 
         return await self.poll_result_by_id(request_id, response.result_class, check_response=check_response)
 
-    async def post_presigned_url(self, url: str, data: Dict, files: List[Tuple]):
+    async def post_presigned_url(self, url: str, data: dict[Any, Any], files: Sequence[Tuple]):
         # Send form request with file and upload_details as body
         resp = await self._http_post(url=url, data=data, files=files, presigned_url_post=True)
         self.logger.debug(
@@ -401,46 +300,47 @@ class PangeaRequestAsync(PangeaRequestBase):
         attached_files = await self._get_attached_files(multipart_reader)
         return MultipartResponse(pangea_json, attached_files)  # type: ignore[arg-type]
 
-    async def _http_delete(
-        self,
-        url: str,
-        *,
-        headers: Mapping[str, str | bytes | None] = {},
-    ) -> aiohttp.ClientResponse:
-        return await self.session.delete(url, headers=headers)
-
     async def _http_post(
         self,
         url: str,
-        headers: Mapping[str, str | bytes | None] = {},
-        data: Union[str, Dict] = {},
-        files: Optional[List[Tuple]] = [],
+        headers: _LooseHeaders = {},  # noqa: B006
+        data: str | dict[str, Any] | None = None,
+        files: _Files | None = None,
         presigned_url_post: bool = False,
     ) -> aiohttp.ClientResponse:
+        if data is None:
+            data = {}
+
         if files:
             form = FormData()
             if presigned_url_post:
-                for k, v in data.items():  # type: ignore[union-attr]
+                assert isinstance(data, dict)
+                assert isinstance(files, list)
+                for k, v in data.items():
                     form.add_field(k, v)
-                for name, value in files:
+                for _name, value in files:
                     form.add_field("file", value[1], filename=value[0], content_type=value[2])
             else:
-                data_send = json.dumps(data, default=default_encoder) if isinstance(data, dict) else data
+                assert isinstance(files, list)
+                data_send: str | FormData = (
+                    json.dumps(data, default=default_encoder) if isinstance(data, dict) else data
+                )
                 form.add_field("request", data_send, content_type="application/json")
                 for name, value in files:
                     form.add_field(name, value[1], filename=value[0], content_type=value[2])
 
-            data_send = form  # type: ignore[assignment]
+            data_send = form
         else:
             data_send = json.dumps(data, default=default_encoder) if isinstance(data, dict) else data
 
+        assert isinstance(self.session, aiohttp.ClientSession)
         return await self.session.post(url, headers=headers, data=data_send)
 
     async def _http_put(
         self,
         url: str,
         files: Sequence[Tuple],
-        headers: Dict = {},
+        headers: _LooseHeaders = {},  # noqa: B006
     ) -> aiohttp.ClientResponse:
         self.logger.debug(
             json.dumps({"service": self.service, "action": "http_put", "url": url}, default=default_encoder)
@@ -452,8 +352,8 @@ class PangeaRequestAsync(PangeaRequestBase):
         self,
         endpoint: str,
         result_class: Type[PangeaResponseResult],
-        data: Union[str, Dict] = {},
-        files: List[Tuple] = [],
+        data: Union[str, Mapping[str, Any]] = {},
+        files: Sequence[Tuple] = [],
     ):
         if len(files) == 0:
             raise AttributeError("files attribute should have at least 1 file")
@@ -477,7 +377,7 @@ class PangeaRequestAsync(PangeaRequestBase):
         self,
         endpoint: str,
         result_class: Type[PangeaResponseResult],
-        data: Union[str, Dict] = {},
+        data: Union[str, Mapping[str, Any]] = {},
     ) -> PangeaResponse:
         # Send request
         try:
@@ -528,7 +428,7 @@ class PangeaRequestAsync(PangeaRequestBase):
                         {"service": self.service, "action": "poll_presigned_url", "step": "exit", "cause": {str(e)}}
                     )
                 )
-                raise pe.PresignedURLException("Failed to pull Presigned URL", loop_exc.response, e)
+                raise pe.PresignedURLException("Failed to pull Presigned URL", loop_exc.response, e) from e
 
         self.logger.debug(json.dumps({"service": self.service, "action": "poll_presigned_url", "step": "exit"}))
 
@@ -560,6 +460,7 @@ class PangeaRequestAsync(PangeaRequestBase):
         self.logger.debug(json.dumps({"service": self.service, "action": "poll_result_retry", "step": "exit"}))
         return self._check_response(response)
 
+    @override
     def _init_session(self) -> aiohttp.ClientSession:
         # retry_config = Retry(
         #     total=self.config.request_retries,

pangea/asyncio/services/__init__.py

@@ -1,3 +1,5 @@
+# ruff: noqa: F401
+
 from .ai_guard import AIGuardAsync
 from .audit import AuditAsync
 from .authn import AuthNAsync
@@ -5,7 +7,6 @@ from .authz import AuthZAsync
 from .embargo import EmbargoAsync
 from .file_scan import FileScanAsync
 from .intel import DomainIntelAsync, FileIntelAsync, IpIntelAsync, UrlIntelAsync, UserIntelAsync
-from .management import ManagementAsync
 from .prompt_guard import PromptGuardAsync
 from .redact import RedactAsync
 from .sanitize import SanitizeAsync

pangea/asyncio/services/ai_guard.py

@@ -1,15 +1,12 @@
 from __future__ import annotations
 
+from collections.abc import Sequence
 from typing import overload
 
-from typing_extensions import TypeVar
-
 from pangea.asyncio.services.base import ServiceBaseAsync
 from pangea.config import PangeaConfig
 from pangea.response import PangeaResponse
-from pangea.services.ai_guard import LogFields, Overrides, TextGuardResult
-
-_T = TypeVar("_T")
+from pangea.services.ai_guard import LogFields, Message, Overrides, TextGuardResult
 
 
 class AIGuardAsync(ServiceBaseAsync):
@@ -60,7 +57,7 @@ class AIGuardAsync(ServiceBaseAsync):
         debug: bool | None = None,
         overrides: Overrides | None = None,
         log_fields: LogFields | None = None,
-    ) -> PangeaResponse[TextGuardResult[None]]:
+    ) -> PangeaResponse[TextGuardResult]:
         """
         Text Guard for scanning LLM inputs and outputs
 
@@ -88,12 +85,12 @@ class AIGuardAsync(ServiceBaseAsync):
     async def guard_text(
         self,
         *,
-        messages: _T,
+        messages: Sequence[Message],
         recipe: str | None = None,
         debug: bool | None = None,
         overrides: Overrides | None = None,
         log_fields: LogFields | None = None,
-    ) -> PangeaResponse[TextGuardResult[_T]]:
+    ) -> PangeaResponse[TextGuardResult]:
         """
         Text Guard for scanning LLM inputs and outputs
 
@@ -115,19 +112,19 @@ class AIGuardAsync(ServiceBaseAsync):
             log_field: Additional fields to include in activity log
 
         Examples:
-            response = await ai_guard.guard_text(messages=[{"role": "user", "content": "hello world"}])
+            response = await ai_guard.guard_text(messages=[Message(role="user", content="hello world")])
         """
 
-    async def guard_text(  # type: ignore[misc]
+    async def guard_text(
         self,
         text: str | None = None,
         *,
-        messages: _T | None = None,
+        messages: Sequence[Message] | None = None,
         recipe: str | None = None,
         debug: bool | None = None,
         overrides: Overrides | None = None,
         log_fields: LogFields | None = None,
-    ) -> PangeaResponse[TextGuardResult[None]]:
+    ) -> PangeaResponse[TextGuardResult]:
         """
         Text Guard for scanning LLM inputs and outputs