clerk-sdk 0.5.8__py3-none-any.whl → 0.5.10__py3-none-any.whl

clerk/__init__.py

@@ -1,4 +1,4 @@
 from .client import Clerk
 
 
-__version__ = "0.5.8"
+__version__ = "0.5.10"

clerk/client.py

@@ -2,6 +2,7 @@ from typing import Any, Dict, List, Literal
 
 from clerk.base import BaseClerk
 from clerk.models.document import Document, GetDocumentsRequest, UploadDocumentRequest
+from clerk.models.prevalidation import FileClassificationResponse
 from .models.file import ParsedFile, UploadFile
 
 
@@ -70,74 +71,23 @@ class Clerk(BaseClerk):
         files_data = [f.to_multipart_format() for f in files]
         self.post_request(endpoint, params=params, files=files_data)
 
-
-class ClerkDocument(BaseClerk):
-    endpoint: str = "/document"
-
-    def upload_document(self, request: UploadDocumentRequest) -> Document:
-        endpoint = "/document"
+    def prevalidate_file(
+        self, project_id: str, workflow_id: str, file: UploadFile
+    ) -> Dict[str, Any]:
+        """Submit a prevalidation for a specific file. Returns the id for fetching the results later."""
+        endpoint = f"/prevalidation/file"
+        files_data = [file.to_multipart_format()]
         res = self.post_request(
-            endpoint=endpoint, data=request.data, files=request.files_
+            endpoint,
+            files=files_data,
+            data={"project_id": project_id, "workflow_id": workflow_id},
         )
-        return Document(**res.data[0])
-
-    def reprocess_document(self, document_id: str, workflow_id: str) -> Document:
-        endpoint = f"/document/{document_id}/reprocess"
-        res = self.put_request(endpoint=endpoint, data={"workflow_id": workflow_id})
-        return Document(**res.data[0])
-
-    def cancel_document_run(self, document_id: str) -> Document:
-        endpoint = f"/document/{document_id}/cancel"
-        res = self.post_request(endpoint=endpoint)
-        return Document(**res.data[0])
-
-    def update_document_structured_data(
-        self, document_id: str, updated_structured_data: Dict[str, Any]
-    ) -> Document:
-        endpoint = f"/document/{document_id}"
-        payload = dict(structured_data=updated_structured_data)
-        res = self.put_request(endpoint, json=payload)
-
-        return Document(**res.data[0])
-
-    def get_document(self, document_id: str) -> Document:
-        endpoint = f"/document/{document_id}"
-        res = self.get_request(endpoint=endpoint)
-        return Document(**res.data[0])
-
-    def get_documents(self, request: GetDocumentsRequest) -> List[Document]:
-        if not any(
-            [
-                request.organization_id,
-                request.project_id,
-                request.start_date,
-                request.end_date,
-                request.query,
-                request.include_statuses,
-            ]
-        ):
-            raise ValueError(
-                "At least one query parameter (organization_id, project_id, start_date, end_date, query, or include_statuses) must be provided."
-            )
-
-        endpoint = f"/documents"
-        params = request.model_dump(mode="json")
-        res = self.get_request(endpoint, params=params)
-
-        return [Document(**d) for d in res.data]
-
-    def get_files_document(self, document_id: str) -> List[ParsedFile]:
-        endpoint = f"/document/{document_id}/files"
-        res = self.get_request(endpoint=endpoint)
-        return [ParsedFile(**d) for d in res.data]
-
-    def add_files_to_document(
-        self,
-        document_id: str,
-        type: Literal["input", "output"],
-        files: List[UploadFile],
-    ):
-        endpoint = f"/document/{document_id}/files/upload"
-        params = {"type": type}
-        files_data = [f.to_multipart_format() for f in files]
-        self.post_request(endpoint, params=params, files=files_data)
+        return res.data[0]
+
+    def get_prevalidation_results(
+        self, workflow_run_id: str
+    ) -> FileClassificationResponse:
+        """Fetch the results of a previously submitted prevalidation."""
+        endpoint = f"/prevalidation/file"
+        res = self.get_request(endpoint, params={"workflow_run_id": workflow_run_id})
+        return FileClassificationResponse(**res.data[0])

clerk/models/prevalidation.py

@@ -0,0 +1,33 @@
+from typing import Dict, List, Optional
+from pydantic import BaseModel, Field
+
+
+class LabelWithDescriptions(BaseModel):
+    label: str = Field(..., description="The label/category name")
+    description: Optional[str] = Field(
+        None, description="Optional description of the label/category"
+    )
+    attributes: Optional[List[str]] = Field(
+        None, description="Optional attributes for the label/category"
+    )
+    start_page: Optional[int] = Field(
+        None, description="Start page for classification (0-based, inclusive)"
+    )
+    end_page: Optional[int] = Field(
+        None, description="End page for classification (0-based, exclusive)"
+    )
+
+
+class FileClassificationResponse(BaseModel):
+    valid: bool = Field(False, description="Whether the file classification is valid")
+    filename: str = Field(..., description="The name of the file that was classified")
+    classification: Optional[str] = Field(
+        None, description="The selected classification category"
+    )
+    reason: Optional[str] = Field(
+        None,
+        description="The reasoning behind the classification. To be filled if no classification is made.",
+    )
+    possible_categories: Optional[Dict[str, LabelWithDescriptions]] = Field(
+        None, description="Possible categories for classification"
+    )

{clerk_sdk-0.5.8.dist-info → clerk_sdk-0.5.10.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: clerk-sdk
-Version: 0.5.8
+Version: 0.5.10
 Summary: Library for interacting with Clerk
 Project-URL: Homepage, https://github.com/F-ONE-Group/clerk_pypi
 Author-email: F-One <contact@f-one.group>

{clerk_sdk-0.5.8.dist-info → clerk_sdk-0.5.10.dist-info}/RECORD

@@ -1,6 +1,6 @@
-clerk/__init__.py,sha256=VzJGCDzhcxnxxL_3jMTrJkSpzpBvf-eENxpTCtkRKO8,50
+clerk/__init__.py,sha256=guyotbECs9C88el84Xes-MMJ_X80QdLi8y-tOQFXK9s,51
 clerk/base.py,sha256=lbFTdpdDfsmYIQUFH93S1aw0-L6GNJwAcubW1tdMFX4,3967
-clerk/client.py,sha256=JGhDUb4m8R7ZTHLxFsAwAw0vNA1kxewg5-GC6qOFYDk,5244
+clerk/client.py,sha256=Es6FyRlxXhiHBag2Hhy3gw5aQyP3Nlfn3Nt01C-VJt0,3519
 clerk/decorator/__init__.py,sha256=yGGcS17VsZ7cZ-hVGCm3I3vGDJMiJIAqmDGzriIi0DI,65
 clerk/decorator/models.py,sha256=nFMdVSG3nJ4hrEXs9YbI_GgjHbVjhSWZokOCzUh-lqQ,521
 clerk/decorator/task_decorator.py,sha256=H8caRvNvvl-IRwyREP66gBGVM-SpQJ1W7oAFImO-6Jw,3769
@@ -57,14 +57,15 @@ clerk/models/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 clerk/models/document.py,sha256=KCO2f255Nodi2JY4BeK3oGSaQ_aMHvrAAaKbOARu8Ww,3039
 clerk/models/document_statuses.py,sha256=67EMwh7FUNEOkb563mO5WtckK1EICY-sk9pRhD-BYK4,347
 clerk/models/file.py,sha256=wfMGoNTsIBFxpPUzOTOkUYZ-FGxXg6n2mZhDCBt6Jbc,733
+clerk/models/prevalidation.py,sha256=UKdpaQG1FuOVJtLLTgvlLMVA5U6kcUYAJ80F27YhDVo,1317
 clerk/models/remote_device.py,sha256=2jijS-9WWq7y6xIbAaDaPnzZo3-tjp2-dCQvNKP8YSU,109
 clerk/models/response_model.py,sha256=R62daUN1YVOwgnrh_epvFRsQcOwT7R4u97l73egvm-c,232
 clerk/models/ui_operator.py,sha256=mKTJUFZgv7PeEt5oys28HVZxHOJsofmRQOcRpqj0dbU,293
 clerk/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 clerk/utils/logger.py,sha256=NrMIlJfVmRjjRw_N_Jngkl0qqv7btXUbg5wxcRmFEH4,3800
 clerk/utils/save_artifact.py,sha256=94aYkYNVGcSUaSWZmdjiY6Oc-3yCKb2XWCZ56IAXQqk,1158
-clerk_sdk-0.5.8.dist-info/METADATA,sha256=TytUb6ss1EOrO1WG1_6qxaLxu4Fj8z2N8nUHQTuN8ro,9635
-clerk_sdk-0.5.8.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
-clerk_sdk-0.5.8.dist-info/entry_points.txt,sha256=VoUmW07sRRSioms5pqQ4A6CYxNEyhGA93GtyBlB_wGw,53
-clerk_sdk-0.5.8.dist-info/licenses/LICENSE,sha256=GTVQl3vH6ht70wJXKC0yMT8CmXKHxv_YyO_utAgm7EA,1065
-clerk_sdk-0.5.8.dist-info/RECORD,,
+clerk_sdk-0.5.10.dist-info/METADATA,sha256=QRgxgexHqT-Vp5kTDuMk6u66m-RVtxEdbEb9YSAvjsE,9636
+clerk_sdk-0.5.10.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
+clerk_sdk-0.5.10.dist-info/entry_points.txt,sha256=VoUmW07sRRSioms5pqQ4A6CYxNEyhGA93GtyBlB_wGw,53
+clerk_sdk-0.5.10.dist-info/licenses/LICENSE,sha256=GTVQl3vH6ht70wJXKC0yMT8CmXKHxv_YyO_utAgm7EA,1065
+clerk_sdk-0.5.10.dist-info/RECORD,,