UncountablePythonSDK 0.0.158__tar.gz → 0.0.159__tar.gz

{uncountablepythonsdk-0.0.158 → uncountablepythonsdk-0.0.159}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: UncountablePythonSDK
-Version: 0.0.158
+Version: 0.0.159
 Summary: Uncountable SDK
 Project-URL: Homepage, https://github.com/uncountableinc/uncountable-python-sdk
 Project-URL: Repository, https://github.com/uncountableinc/uncountable-python-sdk.git

{uncountablepythonsdk-0.0.158 → uncountablepythonsdk-0.0.159}/UncountablePythonSDK.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: UncountablePythonSDK
-Version: 0.0.158
+Version: 0.0.159
 Summary: Uncountable SDK
 Project-URL: Homepage, https://github.com/uncountableinc/uncountable-python-sdk
 Project-URL: Repository, https://github.com/uncountableinc/uncountable-python-sdk.git

{uncountablepythonsdk-0.0.158 → uncountablepythonsdk-0.0.159}/UncountablePythonSDK.egg-info/SOURCES.txt

@@ -133,6 +133,7 @@ uncountable/integration/construct_client.py
 uncountable/integration/cron.py
 uncountable/integration/entrypoint.py
 uncountable/integration/job.py
+uncountable/integration/request_context.py
 uncountable/integration/scan_profiles.py
 uncountable/integration/scheduler.py
 uncountable/integration/server.py

{uncountablepythonsdk-0.0.158 → uncountablepythonsdk-0.0.159}/examples/async_batch_large_parse.py

@@ -2,6 +2,7 @@ import os
 from decimal import Decimal
 
 from uncountable.core import AsyncBatchProcessor, AuthDetailsApiKey, Client
+from uncountable.core.file_upload import UploaderFileUpload
 from uncountable.types.generic_upload_t import UploadDestinationRecipe
 from uncountable.types.identifier import IdentifierKeyId
 from uncountable.types.uploader_t import (
@@ -37,8 +38,13 @@ parsed_file_data = [
 ]
 
 batch_loader = AsyncBatchProcessor(client=client)
+
+uploaded = client.upload_files(
+    file_uploads=[UploaderFileUpload(parsed_file_data=parsed_file_data)]
+)
+
 batch_loader.complete_async_parse(
-    parsed_file_data=parsed_file_data,
+    file_id=uploaded[0].file_id,
     async_job_key=IdentifierKeyId(id=1),
     upload_destination=UploadDestinationRecipe(
         recipe_key=IdentifierKeyId(id=1),

{uncountablepythonsdk-0.0.158 → uncountablepythonsdk-0.0.159}/uncountable/core/file_upload.py

@@ -8,8 +8,11 @@ from typing import Generator, Literal, Self, assert_never
 
 import aiohttp
 import aiotus
+import simplejson as json
 
+from pkgs.serialization_util import serialize_for_storage
 from uncountable.integration.telemetry import Logger, push_scope_optional
+from uncountable.types import uploader_t
 
 from .types import AuthDetailsAll, AuthDetailsApiKey
 
@@ -19,6 +22,7 @@ _CHUNK_SIZE = 5 * 1024 * 1024  # s3 requires 5MiB minimum
 class FileUploadType(StrEnum):
     MEDIA_FILE_UPLOAD = "MEDIA_FILE_UPLOAD"
     DATA_FILE_UPLOAD = "DATA_FILE_UPLOAD"
+    UPLOADER_FILE_UPLOAD = "UPLOADER_FILE_UPLOAD"
 
 
 @dataclass(kw_only=True)
@@ -36,7 +40,20 @@ class DataFileUpload:
     type: Literal[FileUploadType.DATA_FILE_UPLOAD] = FileUploadType.DATA_FILE_UPLOAD
 
 
-FileUpload = MediaFileUpload | DataFileUpload
+@dataclass(kw_only=True)
+class UploaderFileUpload:
+    parsed_file_data: list[uploader_t.ParsedFileData]
+    type: Literal[FileUploadType.UPLOADER_FILE_UPLOAD] = (
+        FileUploadType.UPLOADER_FILE_UPLOAD
+    )
+
+
+FileUpload = MediaFileUpload | DataFileUpload | UploaderFileUpload
+
+
+@dataclass(kw_only=True)
+class _ParsedFileDataWrapper:
+    parsed_file_data: list[uploader_t.ParsedFileData]
 
 
 @dataclass(kw_only=True)
@@ -55,6 +72,16 @@ def file_upload_data(file_upload: FileUpload) -> Generator[FileBytes, None, None
                 )
         case DataFileUpload():
             yield FileBytes(name=file_upload.name, bytes_data=file_upload.data)
+        case UploaderFileUpload():
+            wrapper = _ParsedFileDataWrapper(
+                parsed_file_data=file_upload.parsed_file_data
+            )
+            serialized = serialize_for_storage(wrapper)
+            json_bytes = json.dumps(serialized).encode("utf-8")
+            yield FileBytes(
+                name="parsed_file_data.json",
+                bytes_data=BytesIO(json_bytes),
+            )
 
 
 @dataclass(kw_only=True)
@@ -107,6 +134,8 @@ class FileUploader:
                     attributes["file_path"] = file_upload.path
                 case DataFileUpload():
                     attributes["file_name"] = file_upload.name
+                case UploaderFileUpload():
+                    attributes["file_name"] = "parsed_file_data.json"
                 case _:
                     assert_never(file_upload)
             with push_scope_optional(

{uncountablepythonsdk-0.0.158 → uncountablepythonsdk-0.0.159}/uncountable/integration/job.py

@@ -24,6 +24,7 @@ from uncountable.integration.queue_runner.command_server.command_client import (
 from uncountable.integration.queue_runner.command_server.types import (
     CommandServerException,
 )
+from uncountable.integration.request_context import extract_request_context
 from uncountable.integration.secret_retrieval.retrieve_secret import retrieve_secret
 from uncountable.integration.telemetry import JobLogger
 from uncountable.types import (
@@ -210,12 +211,15 @@ class WebhookJob[WPT](Job[webhook_job_t.WebhookEventPayload]):
         except (simplejson.JSONDecodeError, ValueError) as e:
             raise HttpException.body_parse_error() from e
 
+        request_context = extract_request_context(headers=request.headers)
+
         try:
             send_job_queue_message(
                 job_ref_name=job_definition.id,
                 payload=queued_job_t.QueuedJobPayload(
                     invocation_context=queued_job_t.InvocationContextWebhook(
-                        webhook_payload=webhook_payload
+                        webhook_payload=webhook_payload,
+                        request_context=request_context,
                     )
                 ),
                 port=get_local_admin_server_port(),

uncountablepythonsdk-0.0.159/uncountable/integration/request_context.py

@@ -0,0 +1,34 @@
+from uncountable.types import client_config_t
+
+MAX_HEADER_LENGTH = 100
+
+
+def validated_header(*, headers: dict[str, str], key: str) -> str | None:
+    value = headers.get(key)
+    if value is not None and len(value) > MAX_HEADER_LENGTH:
+        return None
+    return value
+
+
+def _parse_locale(*, raw_locale: str | None) -> client_config_t.SdkLocale | None:
+    if raw_locale is None:
+        return None
+    try:
+        return client_config_t.SdkLocale(raw_locale)
+    except ValueError:
+        return None
+
+
+def extract_request_context(
+    *, headers: dict[str, str]
+) -> client_config_t.RequestContext | None:
+    user_timezone = validated_header(headers=headers, key="X-User-Timezone")
+    user_locale = _parse_locale(
+        raw_locale=validated_header(headers=headers, key="X-User-Locale")
+    )
+    if user_timezone is None and user_locale is None:
+        return None
+    return client_config_t.RequestContext(
+        user_timezone=user_timezone,
+        user_locale=user_locale,
+    )

{uncountablepythonsdk-0.0.158 → uncountablepythonsdk-0.0.159}/uncountable/types/api/uploader/complete_async_parse.py

@@ -31,9 +31,10 @@ ENDPOINT_PATH = "api/external/uploader/complete_async_parse"
 )
 @dataclasses.dataclass(slots=base_t.ENABLE_SLOTS, kw_only=True)  # type: ignore[literal-required]
 class Arguments:
-    parsed_file_data: list[uploader_t.ParsedFileData]
     async_job_key: identifier_t.IdentifierKey
     upload_destination: generic_upload_t.UploadDestinationRecipe
+    parsed_file_data: list[uploader_t.ParsedFileData] | None = None
+    file_id: base_t.ObjectId | None = None
 
 
 # DO NOT MODIFY -- This file is generated by type_spec

{uncountablepythonsdk-0.0.158 → uncountablepythonsdk-0.0.159}/uncountable/types/async_batch_processor.py

@@ -205,18 +205,22 @@ class AsyncBatchProcessorBase(ABC):
     def complete_async_parse(
         self,
         *,
-        parsed_file_data: list[uploader_t.ParsedFileData],
         async_job_key: identifier_t.IdentifierKey,
         upload_destination: generic_upload_t.UploadDestinationRecipe,
+        parsed_file_data: list[uploader_t.ParsedFileData] | None = None,
+        file_id: base_t.ObjectId | None = None,
         depends_on: list[str] | None = None,
         _request_options: client_config_t.RequestOptions | None = None,
     ) -> async_batch_t.QueuedAsyncBatchRequest:
         """Parses uploaded files asynchronously
 
+        :param parsed_file_data: Inline parsed data. Provide either parsed_file_data or file_id, not both.
+        :param file_id: File ID containing parsed data. Provide either file_id or parsed_file_data, not both.
         :param depends_on: A list of batch reference keys to process before processing this request
         """
         args = complete_async_parse_t.Arguments(
             parsed_file_data=parsed_file_data,
+            file_id=file_id,
             async_job_key=async_job_key,
             upload_destination=upload_destination,
         )

{uncountablepythonsdk-0.0.158 → uncountablepythonsdk-0.0.159}/uncountable/types/client_base.py

@@ -340,16 +340,20 @@ class ClientMethods(ABC):
     def complete_async_parse(
         self,
         *,
-        parsed_file_data: list[uploader_t.ParsedFileData],
         async_job_key: identifier_t.IdentifierKey,
         upload_destination: generic_upload_t.UploadDestinationRecipe,
+        parsed_file_data: list[uploader_t.ParsedFileData] | None = None,
+        file_id: base_t.ObjectId | None = None,
         _request_options: client_config_t.RequestOptions | None = None,
     ) -> complete_async_parse_t.Data:
         """Parses uploaded files asynchronously
 
+        :param parsed_file_data: Inline parsed data. Provide either parsed_file_data or file_id, not both.
+        :param file_id: File ID containing parsed data. Provide either file_id or parsed_file_data, not both.
         """
         args = complete_async_parse_t.Arguments(
             parsed_file_data=parsed_file_data,
+            file_id=file_id,
             async_job_key=async_job_key,
             upload_destination=upload_destination,
         )

{uncountablepythonsdk-0.0.158 → uncountablepythonsdk-0.0.159}/uncountable/types/queued_job_t.py

@@ -11,6 +11,7 @@ import dataclasses
 from pkgs.serialization import serial_class
 from pkgs.serialization import serial_union_annotation
 from . import base_t
+from . import client_config_t
 from . import job_definition_t
 
 __all__: list[str] = [
@@ -74,6 +75,7 @@ class InvocationContextManual(InvocationContextBase):
 class InvocationContextWebhook(InvocationContextBase):
     type: typing.Literal[InvocationContextType.WEBHOOK] = InvocationContextType.WEBHOOK
     webhook_payload: base_t.JsonValue
+    request_context: client_config_t.RequestContext | None = None
 
 
 # DO NOT MODIFY -- This file is generated by type_spec