pangea-sdk 6.1.1__py3-none-any.whl → 6.2.0__py3-none-any.whl

pangea/__init__.py

@@ -1,7 +1,15 @@
-__version__ = "6.1.1"
+__version__ = "6.2.0"
 
 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

@@ -1,23 +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, Optional, Sequence, Tuple, Type, Union, cast
 
 import aiohttp
 from aiohttp import FormData
 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)
 
 
@@ -34,7 +54,7 @@ class PangeaRequestAsync(PangeaRequestBase):
         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,
@@ -80,8 +100,11 @@ 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)
@@ -103,7 +126,9 @@ class PangeaRequestAsync(PangeaRequestBase):
 
                 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 = await self._handle_queued_result(pangea_response)
@@ -164,7 +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(
@@ -278,35 +303,44 @@ class PangeaRequestAsync(PangeaRequestBase):
     async def _http_post(
         self,
         url: str,
-        headers: Dict = {},
-        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)
@@ -318,8 +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")
@@ -343,7 +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:
@@ -394,7 +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"}))
 
@@ -426,6 +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

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
 

pangea/asyncio/services/audit.py

@@ -1,8 +1,13 @@
 # Copyright 2022 Pangea Cyber Corporation
 # Author: Pangea Cyber Corporation
+
+# TODO: Modernize.
+# ruff: noqa: UP006, UP035
+
 from __future__ import annotations
 
 import datetime
+from collections.abc import Mapping
 from typing import Any, Dict, Iterable, List, Optional, Sequence, Union
 
 import pangea.exceptions as pexc
@@ -63,7 +68,7 @@ class AuditAsync(ServiceBaseAsync, AuditBase):
         token: str,
         config: PangeaConfig | None = None,
         private_key_file: str = "",
-        public_key_info: dict[str, str] = {},
+        public_key_info: Mapping[str, str] = {},
         tenant_id: str | None = None,
         logger_name: str = "pangea",
         config_id: str | None = None,

pangea/asyncio/services/authn.py

@@ -1,7 +1,12 @@
 # Copyright 2022 Pangea Cyber Corporation
 # Author: Pangea Cyber Corporation
+
+# TODO: Modernize.
+# ruff: noqa: UP006, UP035
+
 from __future__ import annotations
 
+from collections.abc import Mapping
 from typing import Dict, List, Literal, Optional, Union
 
 import pangea.services.authn.models as m
@@ -475,7 +480,7 @@ class AuthNAsync(ServiceBaseAsync):
         async def create(
             self,
             email: str,
-            profile: m.Profile,
+            profile: Mapping[str, str],
             *,
             username: str | None = None,
         ) -> PangeaResponse[m.UserCreateResult]:
@@ -866,7 +871,7 @@ class AuthNAsync(ServiceBaseAsync):
 
             async def update(
                 self,
-                profile: m.Profile,
+                profile: Mapping[str, str],
                 id: str | None = None,
                 email: str | None = None,
                 *,
@@ -995,7 +1000,10 @@ class AuthNAsync(ServiceBaseAsync):
             )
 
         async def restart(
-            self, flow_id: str, choice: m.FlowChoice, data: m.FlowRestartData = {}
+            self,
+            flow_id: str,
+            choice: m.FlowChoice,
+            data: m.FlowRestartData = {},  # noqa: B006
         ) -> PangeaResponse[m.FlowRestartResult]:
             """
             Restart a sign-up/sign-in flow
@@ -1068,7 +1076,10 @@ class AuthNAsync(ServiceBaseAsync):
             return await self.request.post("v2/flow/start", m.FlowStartResult, data=input.model_dump(exclude_none=True))
 
         async def update(
-            self, flow_id: str, choice: m.FlowChoice, data: m.FlowUpdateData = {}
+            self,
+            flow_id: str,
+            choice: m.FlowChoice,
+            data: m.FlowUpdateData = {},  # noqa: B006
         ) -> PangeaResponse[m.FlowUpdateResult]:
             """
             Update a sign-up/sign-in flow

pangea/asyncio/services/base.py

@@ -1,5 +1,9 @@
 # Copyright 2022 Pangea Cyber Corporation
 # Author: Pangea Cyber Corporation
+
+# TODO: Modernize.
+# ruff: noqa: UP006, UP035
+
 from __future__ import annotations
 
 from typing import Dict, Optional, Type, Union

pangea/asyncio/services/file_scan.py

@@ -1,5 +1,11 @@
 # Copyright 2022 Pangea Cyber Corporation
 # Author: Pangea Cyber Corporation
+
+# TODO: Modernize.
+# ruff: noqa: UP006, UP035
+
+from __future__ import annotations
+
 import io
 import logging
 from typing import Dict, List, Optional, Tuple
@@ -98,7 +104,7 @@ class FileScanAsync(ServiceBaseAsync):
         files: Optional[List[Tuple]] = None
         if file or file_path:
             if file_path:
-                file = open(file_path, "rb")
+                file = open(file_path, "rb")  # noqa: SIM115
             if transfer_method == TransferMethod.POST_URL:
                 params = get_file_upload_params(file)  # type: ignore[arg-type]
                 crc = params.crc_hex

pangea/asyncio/services/intel.py

@@ -1,7 +1,12 @@
 # Copyright 2022 Pangea Cyber Corporation
 # Author: Pangea Cyber Corporation
+
+# TODO: Modernize.
+# ruff: noqa: UP006, UP035
+from __future__ import annotations
+
 import hashlib
-from typing import List, Optional
+from typing import List, Literal, Optional
 
 import pangea.services.intel as m
 from pangea.asyncio.services.base import ServiceBaseAsync
@@ -73,11 +78,11 @@ class FileIntelAsync(ServiceBaseAsync):
 
     async def hash_reputation_bulk(
         self,
-        hashes: List[str],
-        hash_type: str,
-        provider: Optional[str] = None,
-        verbose: Optional[bool] = None,
-        raw: Optional[bool] = None,
+        hashes: list[str],
+        hash_type: Literal["sha256", "sha", "md5"],
+        provider: Literal["reversinglabs", "crowdstrike"] | None = None,
+        verbose: bool | None = None,
+        raw: bool | None = None,
     ) -> PangeaResponse[m.FileReputationBulkResult]:
         """
         Reputation check
@@ -85,11 +90,11 @@ class FileIntelAsync(ServiceBaseAsync):
         Retrieve hash-based file reputation from a provider, including an optional detailed report.
 
         Args:
-            hashes (List[str]): The hash of each file to be looked up
-            hash_type (str): One of "sha256", "sha", "md5"
-            provider (str, optional): Use reputation data from these providers: "reversinglabs" or "crowdstrike"
-            verbose (bool, optional): Echo the API parameters in the response
-            raw (bool, optional): Include raw data from this provider
+            hashes: The hash of each file to be looked up
+            hash_type: One of "sha256", "sha", "md5"
+            provider: Use reputation data from these providers: "reversinglabs" or "crowdstrike"
+            verbose: Echo the API parameters in the response
+            raw: Include raw data from this provider
 
         Raises:
             PangeaAPIException: If an API Error happens
@@ -97,12 +102,8 @@ class FileIntelAsync(ServiceBaseAsync):
         Returns:
             A PangeaResponse where the sanctioned source(s) are in the
                 response.result field.  Available response fields can be found in our [API documentation](https://pangea.cloud/docs/api/file-intel).
-
-        Examples:
-            FIXME:
-
         """
-        input = m.FileReputationBulkRequest(  # type: ignore[call-arg]
+        input = m.FileReputationBulkRequest(
             hashes=hashes, hash_type=hash_type, verbose=verbose, raw=raw, provider=provider
         )
         return await self.request.post(
@@ -144,8 +145,8 @@ class FileIntelAsync(ServiceBaseAsync):
             )
         """
 
-        data = open(filepath, "rb")
-        hash = hashlib.sha256(data.read()).hexdigest()
+        with open(filepath, "rb") as data:
+            hash = hashlib.sha256(data.read()).hexdigest()
 
         input = m.FileReputationRequest(hash=hash, hash_type="sha256", verbose=verbose, raw=raw, provider=provider)
         return await self.request.post(
@@ -154,8 +155,8 @@ class FileIntelAsync(ServiceBaseAsync):
 
     async def filepath_reputation_bulk(
         self,
-        filepaths: List[str],
-        provider: Optional[str] = None,
+        filepaths: list[str],
+        provider: Literal["reversinglabs", "crowdstrike"] | None = None,
         verbose: Optional[bool] = None,
         raw: Optional[bool] = None,
     ) -> PangeaResponse[m.FileReputationBulkResult]:
@@ -168,10 +169,10 @@ class FileIntelAsync(ServiceBaseAsync):
         OperationId: file_intel_post_v1_reputation
 
         Args:
-            filepaths (List[str]): The path list to the files to be looked up
-            provider (str, optional): Use reputation data from these providers: "reversinglabs" or "crowdstrike"
-            verbose (bool, optional): Echo the API parameters in the response
-            raw (bool, optional): Include raw data from this provider
+            filepaths: The path list to the files to be looked up
+            provider: Use reputation data from these providers: "reversinglabs" or "crowdstrike"
+            verbose: Echo the API parameters in the response
+            raw: Include raw data from this provider
 
         Raises:
             PangeaAPIException: If an API Error happens
@@ -179,9 +180,6 @@ class FileIntelAsync(ServiceBaseAsync):
         Returns:
             A PangeaResponse where the sanctioned source(s) are in the
                 response.result field.  Available response fields can be found in our [API documentation](https://pangea.cloud/docs/api/file-intel).
-
-        Examples:
-            FIXME:
         """
         hashes = []
         for filepath in filepaths:
@@ -317,7 +315,7 @@ class DomainIntelAsync(ServiceBaseAsync):
                 provider="whoisxml",
             )
         """
-        input = m.DomainWhoIsRequest(domain=domain, verbose=verbose, provider=provider, raw=raw)  # type: ignore[call-arg]
+        input = m.DomainWhoIsRequest(domain=domain, verbose=verbose, provider=provider, raw=raw)
         return await self.request.post("v1/whois", m.DomainWhoIsResult, data=input.model_dump(exclude_none=True))
 
 

pangea/asyncio/services/redact.py

@@ -1,5 +1,9 @@
 # Copyright 2022 Pangea Cyber Corporation
 # Author: Pangea Cyber Corporation
+
+# TODO: Modernize.
+# ruff: noqa: UP006, UP035
+
 from __future__ import annotations
 
 from typing import Dict, List, Optional, Union

pangea/asyncio/services/sanitize.py

@@ -1,5 +1,9 @@
 # Copyright 2022 Pangea Cyber Corporation
 # Author: Pangea Cyber Corporation
+
+# TODO: Modernize.
+# ruff: noqa: UP006, UP035
+
 from __future__ import annotations
 
 import io
@@ -116,7 +120,7 @@ class SanitizeAsync(ServiceBaseAsync):
         files: Optional[List[Tuple]] = None
         if file or file_path:
             if file_path:
-                file = open(file_path, "rb")
+                file = open(file_path, "rb")  # noqa: SIM115
             if transfer_method == TransferMethod.POST_URL and (sha256 is None or crc32c is None or size is None):
                 params = get_file_upload_params(file)  # type: ignore[arg-type]
                 crc32c = params.crc_hex if crc32c is None else crc32c

pangea/asyncio/services/share.py

@@ -1,5 +1,9 @@
 # Copyright 2022 Pangea Cyber Corporation
 # Author: Pangea Cyber Corporation
+
+# TODO: Modernize.
+# ruff: noqa: UP006, UP035
+
 from __future__ import annotations
 
 import io
@@ -194,7 +198,7 @@ class ShareAsync(ServiceBaseAsync):
 
     async def get_archive(
         self,
-        ids: List[str] = [],
+        ids: list[str],
         format: Optional[m.ArchiveFormat] = None,
         transfer_method: Optional[TransferMethod] = None,
         bucket_id: Optional[str] = None,

pangea/asyncio/services/vault.py

@@ -1,5 +1,9 @@
 # Copyright 2022 Pangea Cyber Corporation
 # Author: Pangea Cyber Corporation
+
+# TODO: Modernize.
+# ruff: noqa: UP006, UP035
+
 from __future__ import annotations
 
 from collections.abc import Mapping

pangea/audit_logger.py

@@ -1,5 +1,7 @@
 # Copyright 2022 Pangea Cyber Corporation
 # Author: Pangea Cyber Corporation
+from __future__ import annotations
+
 import logging
 
 import pangea.exceptions as pe
@@ -20,7 +22,7 @@ class AuditLogger(logging.Logger):
     """
 
     def __init__(self, *args, **kwargs):
-        super(AuditLogger, self).__init__(*args, **kwargs)
+        super().__init__(*args, **kwargs)
 
     def set_auditor(self, auditor: Audit):
         """Sets the internal Pangea Audit Service client instance