rapidata 2.42.1__py3-none-any.whl → 2.42.3__py3-none-any.whl

rapidata/rapidata_client/order/rapidata_order.py

@@ -53,14 +53,14 @@ class RapidataOrder:
         self.id = order_id
         self.name = name
         self.__created_at: datetime | None = None
-        self.__openapi_service = openapi_service
+        self._openapi_service = openapi_service
         self.__workflow_id: str = ""
         self.__campaign_id: str = ""
         self.__pipeline_id: str = ""
         self._max_retries = 10
         self._retry_delay = 2
         self.order_details_page = (
-            f"https://app.{self.__openapi_service.environment}/order/detail/{self.id}"
+            f"https://app.{self._openapi_service.environment}/order/detail/{self.id}"
         )
         logger.debug("RapidataOrder initialized")
 
@@ -68,7 +68,7 @@ class RapidataOrder:
     def created_at(self) -> datetime:
         """Returns the creation date of the order."""
         if not self.__created_at:
-            self.__created_at = self.__openapi_service.order_api.order_order_id_get(
+            self.__created_at = self._openapi_service.order_api.order_order_id_get(
                 self.id
             ).order_date
         return self.__created_at
@@ -77,7 +77,7 @@ class RapidataOrder:
         """Runs the order to start collecting responses."""
         with tracer.start_as_current_span("RapidataOrder.run"):
             logger.info("Starting order '%s'", self)
-            self.__openapi_service.order_api.order_order_id_submit_post(
+            self._openapi_service.order_api.order_order_id_submit_post(
                 self.id, SubmitOrderModel(ignoreFailedDatapoints=True)
             )
             logger.debug("Order '%s' has been started.", self)
@@ -90,7 +90,7 @@ class RapidataOrder:
         """Pauses the order."""
         with tracer.start_as_current_span("RapidataOrder.pause"):
             logger.info("Pausing order '%s'", self)
-            self.__openapi_service.order_api.order_order_id_pause_post(self.id)
+            self._openapi_service.order_api.order_order_id_pause_post(self.id)
             logger.debug("Order '%s' has been paused.", self)
             managed_print(f"Order '{self}' has been paused.")
 
@@ -98,7 +98,7 @@ class RapidataOrder:
         """Unpauses/resumes the order."""
         with tracer.start_as_current_span("RapidataOrder.unpause"):
             logger.info("Unpausing order '%s'", self)
-            self.__openapi_service.order_api.order_order_id_resume_post(self.id)
+            self._openapi_service.order_api.order_order_id_resume_post(self.id)
             logger.debug("Order '%s' has been unpaused.", self)
             managed_print(f"Order '{self}' has been unpaused.")
 
@@ -106,7 +106,7 @@ class RapidataOrder:
         """Deletes the order."""
         with tracer.start_as_current_span("RapidataOrder.delete"):
             logger.info("Deleting order '%s'", self)
-            self.__openapi_service.order_api.order_order_id_delete(self.id)
+            self._openapi_service.order_api.order_order_id_delete(self.id)
             logger.debug("Order '%s' has been deleted.", self)
             managed_print(f"Order '{self}' has been deleted.")
 
@@ -125,7 +125,7 @@ class RapidataOrder:
             Failed: The order has failed.
         """
         with tracer.start_as_current_span("RapidataOrder.get_status"):
-            return self.__openapi_service.order_api.order_order_id_get(self.id).state
+            return self._openapi_service.order_api.order_order_id_get(self.id).state
 
     def display_progress_bar(self, refresh_rate: int = 5) -> None:
         """
@@ -180,7 +180,7 @@ class RapidataOrder:
             try:
                 with suppress_rapidata_error_logging():
                     workflow_id = self.__get_workflow_id()
-                    progress = self.__openapi_service.workflow_api.workflow_workflow_id_progress_get(
+                    progress = self._openapi_service.workflow_api.workflow_workflow_id_progress_get(
                         workflow_id
                     )
                 break
@@ -223,7 +223,7 @@ class RapidataOrder:
             try:
                 return RapidataResults(
                     json.loads(
-                        self.__openapi_service.order_api.order_order_id_download_results_get(
+                        self._openapi_service.order_api.order_order_id_download_results_get(
                             order_id=self.id
                         )
                     )
@@ -260,13 +260,13 @@ class RapidataOrder:
         logger.info("Opening order preview in browser...")
         if self.get_status() == OrderState.CREATED:
             logger.info("Order is still in state created. Setting it to preview.")
-            self.__openapi_service.order_api.order_order_id_preview_post(
+            self._openapi_service.order_api.order_order_id_preview_post(
                 self.id, PreviewOrderModel(ignoreFailedDatapoints=True)
             )
             logger.info("Order is now in preview state.")
 
         campaign_id = self.__get_campaign_id()
-        auth_url = f"https://app.{self.__openapi_service.environment}/order/detail/{self.id}/preview?campaignId={campaign_id}"
+        auth_url = f"https://app.{self._openapi_service.environment}/order/detail/{self.id}/preview?campaignId={campaign_id}"
         could_open_browser = webbrowser.open(auth_url)
         if not could_open_browser:
             encoded_url = urllib.parse.quote(auth_url, safe="%/:=&?~#+!$,;'@()*[]")
@@ -282,7 +282,7 @@ class RapidataOrder:
             for _ in range(self._max_retries):
                 try:
                     self.__pipeline_id = (
-                        self.__openapi_service.order_api.order_order_id_get(
+                        self._openapi_service.order_api.order_order_id_get(
                             self.id
                         ).pipeline_id
                     )
@@ -312,7 +312,7 @@ class RapidataOrder:
         pipeline_id = self.__get_pipeline_id()
         for _ in range(self._max_retries):
             try:
-                pipeline = self.__openapi_service.pipeline_api.pipeline_pipeline_id_get(
+                pipeline = self._openapi_service.pipeline_api.pipeline_pipeline_id_get(
                     pipeline_id
                 )
                 self.__workflow_id = cast(
@@ -332,14 +332,14 @@ class RapidataOrder:
         """Internal method to fetch preliminary results."""
         try:
             pipeline_id = self.__get_pipeline_id()
-            download_id = self.__openapi_service.pipeline_api.pipeline_pipeline_id_preliminary_download_post(
+            download_id = self._openapi_service.pipeline_api.pipeline_pipeline_id_preliminary_download_post(
                 pipeline_id, PreliminaryDownloadModel(sendEmail=False)
             ).download_id
 
             elapsed = 0
             timeout = 60
             while elapsed < timeout:
-                preliminary_results = self.__openapi_service.pipeline_api.pipeline_preliminary_download_preliminary_download_id_get(
+                preliminary_results = self._openapi_service.pipeline_api.pipeline_preliminary_download_preliminary_download_id_get(
                     preliminary_download_id=download_id
                 )
                 if preliminary_results:

rapidata/rapidata_client/order/rapidata_order_manager.py

@@ -1,4 +1,4 @@
-from typing import Sequence, Optional, Literal
+from typing import Sequence, Optional, Literal, get_args
 from itertools import zip_longest
 
 from rapidata.rapidata_client.config.tracer import tracer
@@ -41,7 +41,7 @@ from rapidata.api_client.models.filter import Filter
 from rapidata.api_client.models.filter_operator import FilterOperator
 from rapidata.api_client.models.sort_criterion import SortCriterion
 from rapidata.api_client.models.sort_direction import SortDirection
-
+from rapidata.rapidata_client.order._rapidata_order_builder import StickyStateLiteral
 
 from tqdm import tqdm
 
@@ -61,7 +61,7 @@ class RapidataOrderManager:
         self.settings = RapidataSettings
         self.selections = RapidataSelections
         self.__priority: int | None = None
-        self.__sticky_state: Literal["None", "Temporary", "Permanent"] | None = None
+        self.__sticky_state: StickyStateLiteral | None = None
         self.__asset_uploader = AssetUploader(openapi_service)
         logger.debug("RapidataOrderManager initialized")
 
@@ -172,21 +172,21 @@ class RapidataOrderManager:
         logger.debug("Order created: %s", order)
         return order
 
-    def _set_priority(self, priority: int):
-        if not isinstance(priority, int):
-            raise TypeError("Priority must be an integer")
+    def _set_priority(self, priority: int | None):
+        if priority is not None and not isinstance(priority, int):
+            raise TypeError("Priority must be an integer or None")
 
-        if priority < 0:
-            raise ValueError("Priority must be greater than 0")
+        if priority is not None and priority < 0:
+            raise ValueError("Priority must be greater than 0 or None")
 
         self.__priority = priority
 
-    def _set_sticky_state(
-        self, sticky_state: Literal["None", "Temporary", "Permanent"]
-    ):
-        if sticky_state not in ["None", "Temporary", "Permanent"]:
+    def _set_sticky_state(self, sticky_state: StickyStateLiteral | None):
+        sticky_state_valid_values = get_args(StickyStateLiteral)
+
+        if sticky_state is not None and sticky_state not in sticky_state_valid_values:
             raise ValueError(
-                "Sticky state must be one of 'None', 'Temporary', 'Permanent'"
+                f"Sticky state must be one of {sticky_state_valid_values} or None"
             )
 
         self.__sticky_state = sticky_state
@@ -392,27 +392,15 @@ class RapidataOrderManager:
             if len(datapoints) < 2:
                 raise ValueError("At least two datapoints are required")
 
-            metadatas: list[Metadata] = []
-            if context:
-                if not isinstance(context, str) or context == "":
-                    raise ValueError("Context must be a non-empty string")
-                metadatas.append(PromptMetadata(context))
-            if media_context:
-                if not isinstance(media_context, str) or media_context == "":
-                    raise ValueError("Media context must be a non-empty string")
-                metadatas.append(
-                    MediaAssetMetadata(
-                        self.__asset_uploader.upload_asset(media_context)
-                    )
-                )
-
             return self._create_general_order(
                 name=name,
                 workflow=RankingWorkflow(
                     criteria=instruction,
                     total_comparison_budget=total_comparison_budget,
                     random_comparisons_ratio=random_comparisons_ratio,
-                    metadatas=metadatas,
+                    context=context,
+                    media_context=media_context,
+                    file_uploader=self.__asset_uploader,
                 ),
                 assets=datapoints,
                 data_type=data_type,

rapidata/rapidata_client/validation/rapids/rapids.py

@@ -1,5 +1,5 @@
 from rapidata.rapidata_client.settings._rapidata_setting import RapidataSetting
-from typing import Literal, Self, Any, Sequence
+from typing import Literal, Any, Sequence
 from pydantic import BaseModel, model_validator, ConfigDict
 
 
@@ -20,7 +20,7 @@ class Rapid(BaseModel):
     )
 
     @model_validator(mode="after")
-    def check_sentence_and_context(self) -> Self:
+    def check_sentence_and_context(self) -> "Rapid":
         if isinstance(self.sentence, str) and isinstance(self.context, str):
             raise ValueError(
                 "Both 'sentence' and 'context' cannot be strings at the same time."

rapidata/rapidata_client/validation/validation_set_manager.py

@@ -61,7 +61,7 @@ class ValidationSetManager:
             "ValidationSetManager._create_order_validation_set"
         ):
             rapids: list[Rapid] = []
-            for datapoint in datapoints:
+            for datapoint in workflow._format_datapoints(datapoints):
                 rapids.append(
                     Rapid(
                         asset=datapoint.asset,
@@ -672,13 +672,13 @@ class ValidationSetManager:
             )
 
     def find_validation_sets(
-        self, name: str = "", amount: int = 1
+        self, name: str = "", amount: int = 10
     ) -> list[RapidataValidationSet]:
         """Find validation sets by name.
 
         Args:
             name (str, optional): The name to search for. Defaults to "" to match with any set.
-            amount (int, optional): The amount of validation sets to return. Defaults to 1.
+            amount (int, optional): The amount of validation sets to return. Defaults to 10.
 
         Returns:
             list[RapidataValidationSet]: The list of validation sets.

rapidata/rapidata_client/workflow/_base_workflow.py

@@ -44,12 +44,19 @@ class Workflow(ABC):
     ):
         pass
 
+    @abstractmethod
+    def _get_instruction(self) -> str:
+        pass
+
     @abstractmethod
     def _to_model(
         self,
     ) -> SimpleWorkflowModel | CompareWorkflowModel | EvaluationWorkflowModel:
         pass
 
+    def _format_datapoints(self, datapoints: list[Datapoint]) -> list[Datapoint]:
+        return datapoints
+
     def __str__(self) -> str:
         return self._type
 

rapidata/rapidata_client/workflow/_classify_workflow.py

@@ -35,6 +35,9 @@ class ClassifyWorkflow(Workflow):
         self._instruction = instruction
         self._answer_options = answer_options
 
+    def _get_instruction(self) -> str:
+        return self._instruction
+
     def _to_dict(self) -> dict[str, Any]:
         return {
             **super()._to_dict(),

rapidata/rapidata_client/workflow/_compare_workflow.py

@@ -31,6 +31,9 @@ class CompareWorkflow(Workflow):
         self._instruction = instruction
         self._a_b_names = a_b_names
 
+    def _get_instruction(self) -> str:
+        return self._instruction
+
     def _to_dict(self) -> dict[str, Any]:
         return {
             **super()._to_dict(),

rapidata/rapidata_client/workflow/_draw_workflow.py

@@ -16,6 +16,9 @@ class DrawWorkflow(Workflow):
         super().__init__(type="SimpleWorkflowConfig")
         self._target = target
 
+    def _get_instruction(self) -> str:
+        return self._target
+
     def _to_model(self) -> SimpleWorkflowModel:
         blueprint = LineRapidBlueprint(_t="LineBlueprint", target=self._target)
 

rapidata/rapidata_client/workflow/_evaluation_workflow.py

@@ -22,6 +22,9 @@ class EvaluationWorkflow(Workflow):
         self.validation_set_id = validation_set_id
         self.should_accept_incorrect = should_accept_incorrect
 
+    def _get_instruction(self) -> str:
+        return ""
+
     def _to_model(self):
         return EvaluationWorkflowModel(
             _t="EvaluationWorkflow",

rapidata/rapidata_client/workflow/_free_text_workflow.py

@@ -33,6 +33,9 @@ class FreeTextWorkflow(Workflow):
         self._instruction = instruction
         self._validation_system_prompt = validation_system_prompt
 
+    def _get_instruction(self) -> str:
+        return self._instruction
+
     def _to_dict(self) -> dict[str, Any]:
         return {
             **super()._to_dict(),

rapidata/rapidata_client/workflow/_locate_workflow.py

@@ -16,6 +16,9 @@ class LocateWorkflow(Workflow):
         super().__init__(type="SimpleWorkflowConfig")
         self._target = target
 
+    def _get_instruction(self) -> str:
+        return self._target
+
     def _to_model(self) -> SimpleWorkflowModel:
         blueprint = LocateRapidBlueprint(_t="LocateBlueprint", target=self._target)
 

rapidata/rapidata_client/workflow/_ranking_workflow.py

@@ -8,10 +8,17 @@ from rapidata.rapidata_client.workflow._base_workflow import Workflow
 from rapidata.api_client import ComparePayload
 from rapidata.rapidata_client.datapoints._datapoint import Datapoint
 from rapidata.api_client.models.rapid_modality import RapidModality
-from rapidata.rapidata_client.datapoints.metadata import Metadata
+from rapidata.rapidata_client.datapoints.metadata import (
+    MediaAssetMetadata,
+    PromptMetadata,
+)
 from rapidata.api_client.models.create_datapoint_from_files_model_metadata_inner import (
     CreateDatapointFromFilesModelMetadataInner,
 )
+from rapidata.rapidata_client.datapoints._asset_uploader import AssetUploader
+import itertools
+import random
+from typing import cast
 
 
 class RankingWorkflow(Workflow):
@@ -25,11 +32,25 @@ class RankingWorkflow(Workflow):
         elo_start: int = 1200,
         elo_k_factor: int = 40,
         elo_scaling_factor: int = 400,
-        metadatas: list[Metadata] = [],
+        media_context: str | None = None,
+        context: str | None = None,
+        file_uploader: AssetUploader | None = None,
     ):
         super().__init__(type="CompareWorkflowConfig")
 
-        self.metadatas = metadatas
+        self.media_context = media_context
+        self.context = context
+
+        self.metadatas = []
+        if media_context:
+            assert (
+                file_uploader is not None
+            ), "File uploader is required if media_context is provided"
+            self.metadatas.append(
+                MediaAssetMetadata(file_uploader.upload_asset(media_context))
+            )
+        if context:
+            self.metadatas.append(PromptMetadata(context))
 
         self.criteria = criteria
         self.total_comparison_budget = total_comparison_budget
@@ -52,6 +73,9 @@ class RankingWorkflow(Workflow):
             scalingFactor=elo_scaling_factor,
         )
 
+    def _get_instruction(self) -> str:
+        return self.criteria
+
     def _to_model(self) -> CompareWorkflowModel:
 
         return CompareWorkflowModel(
@@ -71,6 +95,24 @@ class RankingWorkflow(Workflow):
             criteria=self.criteria,
         )
 
+    def _format_datapoints(self, datapoints: list[Datapoint]) -> list[Datapoint]:
+        if len(datapoints) < 3:
+            raise ValueError("RankingWorkflow requires at least three datapoints")
+        desired_length = len(datapoints)
+        assets = [datapoint.asset for datapoint in datapoints]
+        pairs = list(map(list, itertools.combinations(assets, 2)))
+        sampled_pairs = random.sample(pairs, desired_length)
+        formatted_datapoints = [
+            Datapoint(
+                asset=cast(list[str], pair),
+                data_type=datapoints[0].data_type,
+                context=self.context,
+                media_context=self.media_context,
+            )
+            for pair in sampled_pairs
+        ]
+        return formatted_datapoints
+
     def __str__(self) -> str:
         return (
             f"RankingWorkflow(criteria='{self.criteria}', metadatas={self.metadatas})"

rapidata/rapidata_client/workflow/_select_words_workflow.py

@@ -31,6 +31,9 @@ class SelectWordsWorkflow(Workflow):
         super().__init__(type="SimpleWorkflowConfig")
         self._instruction = instruction
 
+    def _get_instruction(self) -> str:
+        return self._instruction
+
     def _to_model(self) -> SimpleWorkflowModel:
         blueprint = TranscriptionRapidBlueprint(
             _t="TranscriptionBlueprint", title=self._instruction

rapidata/rapidata_client/workflow/_timestamp_workflow.py

@@ -29,6 +29,9 @@ class TimestampWorkflow(Workflow):
         super().__init__(type="SimpleWorkflowConfig")
         self._instruction = instruction
 
+    def _get_instruction(self) -> str:
+        return self._instruction
+
     def _to_model(self) -> SimpleWorkflowModel:
         blueprint = ScrubRapidBlueprint(_t="ScrubBlueprint", target=self._instruction)
 

{rapidata-2.42.1.dist-info → rapidata-2.42.3.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rapidata
-Version: 2.42.1
+Version: 2.42.3
 Summary: Rapidata package containing the Rapidata Python Client to interact with the Rapidata Web API in an easy way.
 License: Apache-2.0
 License-File: LICENSE