retab 0.0.36__py3-none-any.whl → 0.0.37__py3-none-any.whl

retab/resources/evaluations/iterations.py

@@ -0,0 +1,452 @@
+from typing import Any, Dict, List, Optional
+
+from openai.types.chat.chat_completion_reasoning_effort import ChatCompletionReasoningEffort
+
+from ..._resource import AsyncAPIResource, SyncAPIResource
+from ...types.browser_canvas import BrowserCanvas
+from ...types.evaluations import CreateIterationRequest, Iteration, ProcessIterationRequest, IterationDocumentStatusResponse, PatchIterationRequest
+from ...types.inference_settings import InferenceSettings
+from ...types.metrics import DistancesResult
+from ...types.modalities import Modality
+from ...types.standards import DeleteResponse, PreparedRequest, FieldUnset
+from ...types.documents.extractions import UiParsedChatCompletion
+
+
+class IterationsMixin:
+    def prepare_get(self, evaluation_id: str, iteration_id: str) -> PreparedRequest:
+        return PreparedRequest(method="GET", url=f"/v1/evaluations/{evaluation_id}/iterations/{iteration_id}")
+
+    def prepare_list(self, evaluation_id: str, model: Optional[str] = None) -> PreparedRequest:
+        params = {}
+        if model:
+            params["model"] = model
+        return PreparedRequest(method="GET", url=f"/v1/evaluations/{evaluation_id}/iterations", params=params)
+
+    def prepare_create(
+        self,
+        evaluation_id: str,
+        model: str = FieldUnset,
+        json_schema: Optional[Dict[str, Any]] = None,
+        temperature: float = FieldUnset,
+        modality: Modality = FieldUnset,
+        reasoning_effort: ChatCompletionReasoningEffort = FieldUnset,
+        image_resolution_dpi: int = FieldUnset,
+        browser_canvas: BrowserCanvas = FieldUnset,
+        n_consensus: int = FieldUnset,
+    ) -> PreparedRequest:
+        inference_settings = InferenceSettings(
+            model=model,
+            temperature=temperature,
+            modality=modality,
+            reasoning_effort=reasoning_effort,
+            image_resolution_dpi=image_resolution_dpi,
+            browser_canvas=browser_canvas,
+            n_consensus=n_consensus,
+        )
+
+        request = CreateIterationRequest(inference_settings=inference_settings, json_schema=json_schema)
+
+        return PreparedRequest(method="POST", url=f"/v1/evaluations/{evaluation_id}/iterations", data=request.model_dump(exclude_unset=True, exclude_defaults=True, mode="json"))
+
+    def prepare_update(
+        self,
+        evaluation_id: str,
+        iteration_id: str,
+        json_schema: Dict[str, Any] = FieldUnset,
+        model: str = FieldUnset,
+        temperature: float = FieldUnset,
+        modality: Modality = FieldUnset,
+        reasoning_effort: ChatCompletionReasoningEffort = FieldUnset,
+        image_resolution_dpi: int = FieldUnset,
+        browser_canvas: BrowserCanvas = FieldUnset,
+        n_consensus: int = FieldUnset,
+    ) -> PreparedRequest:
+        inference_settings = InferenceSettings(
+            model=model,
+            temperature=temperature,
+            modality=modality,
+            reasoning_effort=reasoning_effort,
+            image_resolution_dpi=image_resolution_dpi,
+            browser_canvas=browser_canvas,
+            n_consensus=n_consensus,
+        )
+        if not inference_settings.model_dump(exclude_unset=True, mode="json"):
+            inference_settings = FieldUnset
+
+        iteration_data = PatchIterationRequest(json_schema=json_schema, inference_settings=inference_settings)
+
+        return PreparedRequest(
+            method="PATCH", url=f"/v1/evaluations/{evaluation_id}/iterations/{iteration_id}", data=iteration_data.model_dump(exclude_unset=True, exclude_defaults=True, mode="json")
+        )
+
+    def prepare_delete(self, evaluation_id: str, iteration_id: str) -> PreparedRequest:
+        return PreparedRequest(method="DELETE", url=f"/v1/evaluations/{evaluation_id}/iterations/{iteration_id}")
+
+    def prepare_compute_distances(self, evaluation_id: str, iteration_id: str, document_id: str) -> PreparedRequest:
+        return PreparedRequest(method="GET", url=f"/v1/evaluations/{evaluation_id}/iterations/{iteration_id}/documents/{document_id}/distances")
+
+    def prepare_process(
+        self,
+        evaluation_id: str,
+        iteration_id: str,
+        document_ids: Optional[List[str]] = None,
+        only_outdated: bool = True,
+    ) -> PreparedRequest:
+        request = ProcessIterationRequest(
+            document_ids=document_ids,
+            only_outdated=only_outdated,
+        )
+        return PreparedRequest(method="POST", url=f"/v1/evaluations/{evaluation_id}/iterations/{iteration_id}/process", data=request.model_dump(exclude_none=True, mode="json"))
+
+    def prepare_process_document(self, evaluation_id: str, iteration_id: str, document_id: str) -> PreparedRequest:
+        return PreparedRequest(method="POST", url=f"/v1/evaluations/{evaluation_id}/iterations/{iteration_id}/documents/{document_id}/process", data={"stream": False})
+
+    def prepare_status(self, evaluation_id: str, iteration_id: str) -> PreparedRequest:
+        return PreparedRequest(method="GET", url=f"/v1/evaluations/{evaluation_id}/iterations/{iteration_id}/status")
+
+
+class Iterations(SyncAPIResource, IterationsMixin):
+    """Iterations API wrapper for evaluations"""
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+
+    def get(self, evaluation_id: str, iteration_id: str) -> Iteration:
+        request = self.prepare_get(evaluation_id, iteration_id)
+        response = self._client._prepared_request(request)
+        return Iteration(**response)
+
+    def list(self, evaluation_id: str, model: Optional[str] = None) -> List[Iteration]:
+        """
+        List iterations for an evaluation.
+
+        Args:
+            evaluation_id: The ID of the evaluation
+            model: Optional model to filter by
+
+        Returns:
+            List[Iteration]: List of iterations
+        Raises:
+            HTTPException if the request fails
+        """
+        request = self.prepare_list(evaluation_id, model)
+        response = self._client._prepared_request(request)
+        return [Iteration(**item) for item in response.get("data", [])]
+
+    def create(
+        self,
+        evaluation_id: str,
+        model: str,
+        temperature: float = 0.0,
+        modality: Modality = "native",
+        json_schema: Optional[Dict[str, Any]] = None,
+        reasoning_effort: ChatCompletionReasoningEffort = "medium",
+        image_resolution_dpi: int = 96,
+        browser_canvas: BrowserCanvas = "A4",
+        n_consensus: int = 1,
+    ) -> Iteration:
+        """
+        Create a new iteration for an evaluation.
+
+        Args:
+            evaluation_id: The ID of the evaluation
+            json_schema: The JSON schema for the iteration (if not set, we use the one of the eval)
+            model: The model to use for the iteration
+            temperature: The temperature to use for the model
+            modality: The modality to use (text, image, etc.)
+            reasoning_effort: The reasoning effort setting for the model (auto, low, medium, high)
+            image_resolution_dpi: The DPI of the image. Defaults to 96.
+            browser_canvas: The canvas size of the browser. Must be one of:
+                - "A3" (11.7in x 16.54in)
+                - "A4" (8.27in x 11.7in)
+                - "A5" (5.83in x 8.27in)
+                Defaults to "A4".
+            n_consensus: Number of consensus iterations to perform
+
+        Returns:
+            Iteration: The created iteration
+        Raises:
+            HTTPException if the request fails
+        """
+        request = self.prepare_create(
+            evaluation_id=evaluation_id,
+            json_schema=json_schema,
+            model=model,
+            temperature=temperature,
+            modality=modality,
+            reasoning_effort=reasoning_effort,
+            image_resolution_dpi=image_resolution_dpi,
+            browser_canvas=browser_canvas,
+            n_consensus=n_consensus,
+        )
+        response = self._client._prepared_request(request)
+        return Iteration(**response)
+
+    def delete(self, evaluation_id: str, iteration_id: str) -> DeleteResponse:
+        """
+        Delete an iteration.
+
+        Args:
+            iteration_id: The ID of the iteration
+
+        Returns:
+            DeleteResponse: The response containing success status and ID
+        Raises:
+            HTTPException if the request fails
+        """
+        request = self.prepare_delete(evaluation_id, iteration_id)
+        return self._client._prepared_request(request)
+
+    def compute_distances(self, evaluation_id: str, iteration_id: str, document_id: str) -> DistancesResult:
+        """
+        Get distances for a document in an iteration.
+
+        Args:
+            iteration_id: The ID of the iteration
+            document_id: The ID of the document
+
+        Returns:
+            DistancesResult: The distances
+        Raises:
+            HTTPException if the request fails
+        """
+        request = self.prepare_compute_distances(evaluation_id, iteration_id, document_id)
+        response = self._client._prepared_request(request)
+        return DistancesResult(**response)
+
+    def process(
+        self,
+        evaluation_id: str,
+        iteration_id: str,
+        document_ids: Optional[List[str]] = None,
+        only_outdated: bool = True,
+    ) -> Iteration:
+        """
+        Process an iteration by running extractions on documents.
+
+        Args:
+            iteration_id: The ID of the iteration
+            document_ids: Optional list of specific document IDs to process
+            only_outdated: Whether to only process documents that need updates
+
+        Returns:
+            Iteration: The updated iteration
+        Raises:
+            HTTPException if the request fails
+        """
+        request = self.prepare_process(evaluation_id, iteration_id, document_ids, only_outdated)
+        response = self._client._prepared_request(request)
+        return Iteration(**response)
+
+    def process_document(self, evaluation_id: str, iteration_id: str, document_id: str) -> UiParsedChatCompletion:
+        """
+        Process a single document within an iteration.
+        This method updates the iteration document with the latest extraction.
+
+        Args:
+            iteration_id: The ID of the iteration
+            document_id: The ID of the document
+
+        Returns:
+            UiParsedChatCompletion: The parsed chat completion
+        Raises:
+            HTTPException if the request fails
+        """
+        request = self.prepare_process_document(evaluation_id, iteration_id, document_id)
+        response = self._client._prepared_request(request)
+        return UiParsedChatCompletion(**response)
+
+    def status(self, evaluation_id: str, iteration_id: str) -> IterationDocumentStatusResponse:
+        """
+        Get the status of documents in an iteration.
+
+        Args:
+            iteration_id: The ID of the iteration
+
+        Returns:
+            IterationDocumentStatusResponse: The status of documents
+        Raises:
+            HTTPException if the request fails
+        """
+        request = self.prepare_status(evaluation_id, iteration_id)
+        response = self._client._prepared_request(request)
+        return IterationDocumentStatusResponse(**response)
+
+
+class AsyncIterations(AsyncAPIResource, IterationsMixin):
+    """Async Iterations API wrapper for evaluations"""
+
+    def __init__(self, *args, **kwargs):
+        super().__init__(*args, **kwargs)
+
+    async def get(self, evaluation_id: str, iteration_id: str) -> Iteration:
+        """
+        Get an iteration by ID.
+
+        Args:
+            iteration_id: The ID of the iteration
+
+        Returns:
+            Iteration: The iteration
+        Raises:
+            HTTPException if the request fails
+        """
+        request = self.prepare_get(evaluation_id, iteration_id)
+        response = await self._client._prepared_request(request)
+        return Iteration(**response)
+
+    async def list(self, evaluation_id: str, model: Optional[str] = None) -> List[Iteration]:
+        """
+        List iterations for an evaluation.
+
+        Args:
+            evaluation_id: The ID of the evaluation
+            model: Optional model to filter by
+
+        Returns:
+            List[Iteration]: List of iterations
+        Raises:
+            HTTPException if the request fails
+        """
+        request = self.prepare_list(evaluation_id, model)
+        response = await self._client._prepared_request(request)
+        return [Iteration(**item) for item in response.get("data", [])]
+
+    async def create(
+        self,
+        evaluation_id: str,
+        model: str,
+        temperature: float = 0.0,
+        modality: Modality = "native",
+        json_schema: Optional[Dict[str, Any]] = None,
+        reasoning_effort: ChatCompletionReasoningEffort = "medium",
+        image_resolution_dpi: int = 96,
+        browser_canvas: BrowserCanvas = "A4",
+        n_consensus: int = 1,
+    ) -> Iteration:
+        """
+        Create a new iteration for an evaluation.
+
+        Args:
+            evaluation_id: The ID of the evaluation
+            json_schema: The JSON schema for the iteration
+            model: The model to use for the iteration
+            temperature: The temperature to use for the model
+            modality: The modality to use (text, image, etc.)
+            reasoning_effort: The reasoning effort setting for the model (auto, low, medium, high)
+            image_resolution_dpi: The DPI of the image. Defaults to 96.
+            browser_canvas: The canvas size of the browser. Must be one of:
+                - "A3" (11.7in x 16.54in)
+                - "A4" (8.27in x 11.7in)
+                - "A5" (5.83in x 8.27in)
+                Defaults to "A4".
+            n_consensus: Number of consensus iterations to perform
+
+        Returns:
+            Iteration: The created iteration
+        Raises:
+            HTTPException if the request fails
+        """
+        request = self.prepare_create(
+            evaluation_id=evaluation_id,
+            json_schema=json_schema,
+            model=model,
+            temperature=temperature,
+            modality=modality,
+            reasoning_effort=reasoning_effort,
+            image_resolution_dpi=image_resolution_dpi,
+            browser_canvas=browser_canvas,
+            n_consensus=n_consensus,
+        )
+        response = await self._client._prepared_request(request)
+        return Iteration(**response)
+
+    async def delete(self, evaluation_id: str, iteration_id: str) -> DeleteResponse:
+        """
+        Delete an iteration.
+
+        Args:
+            iteration_id: The ID of the iteration
+
+        Returns:
+            DeleteResponse: The response containing success status and ID
+        Raises:
+            HTTPException if the request fails
+        """
+        request = self.prepare_delete(evaluation_id, iteration_id)
+        return await self._client._prepared_request(request)
+
+    async def compute_distances(self, evaluation_id: str, iteration_id: str, document_id: str) -> DistancesResult:
+        """
+        Get distances for a document in an iteration.
+
+        Args:
+            iteration_id: The ID of the iteration
+            document_id: The ID of the document
+
+        Returns:
+            DistancesResult: The distances
+        Raises:
+            HTTPException if the request fails
+        """
+        request = self.prepare_compute_distances(evaluation_id, iteration_id, document_id)
+        response = await self._client._prepared_request(request)
+        return DistancesResult(**response)
+
+    async def process(
+        self,
+        evaluation_id: str,
+        iteration_id: str,
+        document_ids: Optional[List[str]] = None,
+        only_outdated: bool = True,
+    ) -> Iteration:
+        """
+        Process an iteration by running extractions on documents.
+
+        Args:
+            iteration_id: The ID of the iteration
+            document_ids: Optional list of specific document IDs to process
+            only_outdated: Whether to only process documents that need updates
+
+        Returns:
+            Iteration: The updated iteration
+        Raises:
+            HTTPException if the request fails
+        """
+        request = self.prepare_process(evaluation_id, iteration_id, document_ids, only_outdated)
+        response = await self._client._prepared_request(request)
+        return Iteration(**response)
+
+    async def process_document(self, evaluation_id: str, iteration_id: str, document_id: str) -> UiParsedChatCompletion:
+        """
+        Process a single document within an iteration.
+        This method updates the iteration document with the latest extraction.
+
+        Args:
+            iteration_id: The ID of the iteration
+            document_id: The ID of the document
+
+        Returns:
+            UiParsedChatCompletion: The parsed chat completion
+        Raises:
+            HTTPException if the request fails
+        """
+        request = self.prepare_process_document(evaluation_id, iteration_id, document_id)
+        response = await self._client._prepared_request(request)
+        return UiParsedChatCompletion(**response)
+
+    async def status(self, evaluation_id: str, iteration_id: str) -> IterationDocumentStatusResponse:
+        """
+        Get the status of documents in an iteration.
+
+        Args:
+            iteration_id: The ID of the iteration
+
+        Returns:
+            IterationDocumentStatusResponse: The status of documents
+        Raises:
+            HTTPException if the request fails
+        """
+        request = self.prepare_status(evaluation_id, iteration_id)
+        response = await self._client._prepared_request(request)
+        return IterationDocumentStatusResponse(**response)

{uiform → retab}/resources/files.py

@@ -11,7 +11,7 @@ class Files(SyncAPIResource):
     # Tell Pydantic that _client is not a "field" for serialization/validation,
     # but rather a private attribute to store arbitrary data.
 
-    def create(self, file: Any, purpose: Literal['assistants', 'batch', 'fine-tune', 'vision']) -> Any:
+    def create(self, file: Any, purpose: Literal["assistants", "batch", "fine-tune", "vision"]) -> Any:
         openai_client = OpenAI(api_key=self._client.headers["OpenAI-Api-Key"])
         return openai_client.files.create(file=file, purpose=purpose)
 
@@ -19,6 +19,6 @@ class Files(SyncAPIResource):
 class AsyncFiles(AsyncAPIResource):
     """Files Asyncronous API wrapper"""
 
-    async def create(self, file: Any, purpose: Literal['assistants', 'batch', 'fine-tune', 'vision']) -> Any:
+    async def create(self, file: Any, purpose: Literal["assistants", "batch", "fine-tune", "vision"]) -> Any:
         async with AsyncOpenAI(api_key=self._client.headers["OpenAI-Api-Key"]) as openai_client:
             return await openai_client.files.create(file=file, purpose=purpose)