rapidata 1.10.1__py3-none-any.whl → 2.1.0__py3-none-any.whl

rapidata/rapidata_client/metadata/{private_text_metadata.py → _private_text_metadata.py}

@@ -1,7 +1,7 @@
 from rapidata.api_client.models.private_text_metadata_input import (
     PrivateTextMetadataInput,
 )
-from rapidata.rapidata_client.metadata.base_metadata import Metadata
+from rapidata.rapidata_client.metadata._base_metadata import Metadata
 
 
 class PrivateTextMetadata(Metadata):
@@ -10,7 +10,7 @@ class PrivateTextMetadata(Metadata):
         super().__init__(identifier=identifier)
         self._text = text
 
-    def to_model(self):
+    def _to_model(self):
         return PrivateTextMetadataInput(
             _t="PrivateTextMetadataInput", identifier=self._identifier, text=self._text
         )

rapidata/rapidata_client/metadata/{prompt_metadata.py → _prompt_metadata.py}

@@ -1,8 +1,9 @@
-from rapidata.rapidata_client.metadata.base_metadata import Metadata
+from rapidata.rapidata_client.metadata._base_metadata import Metadata
 from rapidata.api_client.models.prompt_metadata_input import PromptMetadataInput
 
 
 class PromptMetadata(Metadata):
+    """The PromptMetadata class is used to display a prompt to the user."""
 
     def __init__(self, prompt: str, identifier: str = "prompt"):
         super().__init__(identifier=identifier)
@@ -13,5 +14,5 @@ class PromptMetadata(Metadata):
         self._prompt = prompt
     
 
-    def to_model(self):
+    def _to_model(self):
         return PromptMetadataInput(_t="PromptMetadataInput", identifier=self._identifier, prompt=self._prompt)

rapidata/rapidata_client/metadata/{public_text_metadata.py → _public_text_metadata.py}

@@ -1,7 +1,7 @@
 from rapidata.api_client.models.public_text_metadata_input import (
     PublicTextMetadataInput,
 )
-from rapidata.rapidata_client.metadata.base_metadata import Metadata
+from rapidata.rapidata_client.metadata._base_metadata import Metadata
 
 
 class PublicTextMetadata(Metadata):
@@ -10,7 +10,7 @@ class PublicTextMetadata(Metadata):
         super().__init__(identifier=identifier)
         self._text = text
 
-    def to_model(self):
+    def _to_model(self):
         return PublicTextMetadataInput(
             _t="PublicTextMetadataInput", identifier=self._identifier, text=self._text
         )

rapidata/rapidata_client/metadata/{select_words_metadata.py → _select_words_metadata.py}

@@ -1,17 +1,18 @@
 from rapidata.api_client.models.transcription_metadata_input import (
     TranscriptionMetadataInput,
 )
-from rapidata.rapidata_client.metadata.base_metadata import Metadata
+from rapidata.rapidata_client.metadata._base_metadata import Metadata
 
 
 class SelectWordsMetadata(Metadata):
+    """SelectWordsMetadata Class is used to define the Sentence that will be display to the user."""
 
     def __init__(self, select_words: str, identifier: str = "transcription"):
         super().__init__(identifier=identifier)
         self.identifier = identifier
         self.select_words = select_words
 
-    def to_model(self):
+    def _to_model(self):
         return TranscriptionMetadataInput(
             _t="TranscriptionMetadataInput",
             identifier=self.identifier,

rapidata/rapidata_client/{dataset/rapidata_dataset.py → order/_rapidata_dataset.py}

@@ -1,5 +1,4 @@
 from itertools import zip_longest
-import os
 
 from rapidata.api_client.models.datapoint_metadata_model import DatapointMetadataModel
 from rapidata.api_client.models.datapoint_metadata_model_metadata_inner import (
@@ -8,14 +7,14 @@ from rapidata.api_client.models.datapoint_metadata_model_metadata_inner import (
 from rapidata.api_client.models.upload_text_sources_to_dataset_model import (
     UploadTextSourcesToDatasetModel,
 )
-from rapidata.rapidata_client.metadata.base_metadata import Metadata
+from rapidata.rapidata_client.metadata._base_metadata import Metadata
 from rapidata.rapidata_client.assets import TextAsset, MediaAsset, MultiAsset
 from rapidata.service import LocalFileService
 from rapidata.service.openapi_service import OpenAPIService
 from concurrent.futures import ThreadPoolExecutor, as_completed
 from tqdm import tqdm
 
-from typing import cast
+from typing import cast, Sequence
 
 
 class RapidataDataset:
@@ -25,7 +24,7 @@ class RapidataDataset:
         self.openapi_service = openapi_service
         self.local_file_service = LocalFileService()
 
-    def add_texts(
+    def _add_texts(
         self,
         text_assets: list[TextAsset] | list[MultiAsset],
         max_workers: int = 10,
@@ -63,17 +62,17 @@ class RapidataDataset:
                     future.result()  # This will raise any exceptions that occurred during execution
                     pbar.update(1)
 
-    def add_media_from_paths(
+    def _add_media_from_paths(
         self,
         media_paths: list[MediaAsset] |  list[MultiAsset], # where multiasset is a list of media assets
-        metadata: list[Metadata] | None = None,
+        metadata: Sequence[Metadata] | None = None,
         max_workers: int = 10,
     ):
         if metadata is not None and len(metadata) != len(media_paths):
             raise ValueError(
                 "metadata must be None or have the same length as media_paths"
             )
-
+        
         for media_path in media_paths:
             if isinstance(media_path, MultiAsset):
                 assert all(
@@ -88,7 +87,7 @@ class RapidataDataset:
             else:
                 raise ValueError(f"Unsupported asset type: {type(media_asset)}")
 
-            meta_model = meta.to_model() if meta else None
+            meta_model = meta._to_model() if meta else None
             model = DatapointMetadataModel(
                 datasetId=self.dataset_id,
                 metadata=(

rapidata/rapidata_client/order/_rapidata_order_builder.py

@@ -0,0 +1,365 @@
+from rapidata.api_client.models.capped_selection_selections_inner import (
+    CappedSelectionSelectionsInner,
+)
+from rapidata.api_client.models.create_order_model import CreateOrderModel
+from rapidata.api_client.models.create_order_model_referee import (
+    CreateOrderModelReferee,
+)
+from rapidata.api_client.models.create_order_model_user_filters_inner import (
+    CreateOrderModelUserFiltersInner,
+)
+from rapidata.api_client.models.create_order_model_workflow import (
+    CreateOrderModelWorkflow,
+)
+
+from rapidata.rapidata_client.settings import RapidataSetting
+from rapidata.rapidata_client.metadata._base_metadata import Metadata
+from rapidata.rapidata_client.order._rapidata_dataset import RapidataDataset
+from rapidata.rapidata_client.referee._naive_referee import NaiveReferee
+from rapidata.rapidata_client.selection._base_selection import RapidataSelection
+from rapidata.rapidata_client.filter import RapidataFilter
+from rapidata.rapidata_client.workflow import Workflow
+from rapidata.rapidata_client.order.rapidata_order import RapidataOrder
+from rapidata.rapidata_client.referee import Referee
+from rapidata.service.openapi_service import OpenAPIService
+
+from rapidata.rapidata_client.workflow._compare_workflow import CompareWorkflow
+
+from rapidata.rapidata_client.assets import MediaAsset, TextAsset, MultiAsset, BaseAsset
+
+from typing import Optional, cast, Sequence
+
+
+class RapidataOrderBuilder:
+    """Builder object for creating Rapidata orders.
+
+    Use the fluent interface to set the desired configuration. Add a workflow to the order using `.workflow()` and finally call `.create()` to create the order.
+
+    Args:
+        openapi_service (OpenAPIService): The OpenAPIService instance.
+        name (str): The name of the order.
+    """
+
+    def __init__(
+        self,
+        name: str,
+        openapi_service: OpenAPIService,
+    ):
+        self._name = name
+        self.order_id: str | None = None
+        self.__openapi_service = openapi_service
+        self.__dataset: Optional[RapidataDataset]
+        self.__workflow: Workflow | None = None
+        self.__referee: Referee | None = None
+        self.__metadata: Sequence[Metadata] | None = None
+        self.__validation_set_id: str | None = None
+        self.__settings: Sequence[RapidataSetting] | None = None
+        self.__user_filters: list[RapidataFilter] = []
+        self.__selections: list[RapidataSelection] = []
+        self.__priority: int = 50
+        self.__assets: Sequence[BaseAsset] = []
+
+    def _to_model(self) -> CreateOrderModel:
+        """
+        Convert the builder configuration to a CreateOrderModel.
+
+        Raises:
+            ValueError: If no workflow is provided.
+
+        Returns:
+            CreateOrderModel: The model representing the order configuration.
+        """
+        if self.__workflow is None:
+            raise ValueError("You must provide a workflow to create an order.")
+
+        if self.__referee is None:
+            print("No referee provided, using default NaiveReferee.")
+            self.__referee = NaiveReferee()
+
+        return CreateOrderModel(
+            _t="CreateOrderModel",
+            orderName=self._name,
+            workflow=CreateOrderModelWorkflow(self.__workflow._to_model()),
+            userFilters=[
+                CreateOrderModelUserFiltersInner(user_filter._to_model())
+                for user_filter in self.__user_filters
+            ],
+            referee=CreateOrderModelReferee(self.__referee._to_model()),
+            validationSetId=self.__validation_set_id,
+            featureFlags=(
+                [setting._to_feature_flag() for setting in self.__settings]
+                if self.__settings is not None
+                else None
+            ),
+            selections=[
+                CappedSelectionSelectionsInner(selection._to_model())
+                for selection in self.__selections
+            ],
+            priority=self.__priority,
+        )
+
+    def _create(self, max_upload_workers: int = 10) -> RapidataOrder:
+        """
+        Create the Rapidata order by making the necessary API calls based on the builder's configuration.
+
+        Args:
+            max_upload_workers (int, optional): The maximum number of worker threads for processing media paths. Defaults to 10.
+
+        Raises:
+            ValueError: If both media paths and texts are provided, or if neither is provided.
+            AssertionError: If the workflow is a CompareWorkflow and media paths are not in pairs.
+
+        Returns:
+            RapidataOrder: The created RapidataOrder instance.
+        """
+        order_model = self._to_model()
+        if isinstance(
+            self.__workflow, CompareWorkflow
+        ):  # Temporary fix; will be handled by backend in the future
+            assert all(
+                isinstance(item, MultiAsset) for item in self.__assets
+            ), "The media paths must be of type MultiAsset for comparison tasks."
+
+        result = self.__openapi_service.order_api.order_create_post(
+            create_order_model=order_model
+        )
+
+        self.order_id = result.order_id
+
+        self.__dataset = (
+            RapidataDataset(result.dataset_id, self.__openapi_service)
+            if result.dataset_id
+            else None
+        )
+
+        order = RapidataOrder(
+            order_id=self.order_id,
+            dataset=self.__dataset,
+            openapi_service=self.__openapi_service,
+            name=self._name,
+        )
+
+        if all(isinstance(item, MediaAsset) for item in self.__assets) and order.dataset:
+            assets = cast(list[MediaAsset], self.__assets)
+            order.dataset._add_media_from_paths(assets, self.__metadata, max_upload_workers)
+
+        elif (
+            all(isinstance(item, TextAsset) for item in self.__assets) and order.dataset
+        ):
+            assets = cast(list[TextAsset], self.__assets)
+            order.dataset._add_texts(assets)
+
+        elif (
+            all(isinstance(item, MultiAsset) for item in self.__assets) and order.dataset
+        ):
+            multi_assets = cast(list[MultiAsset], self.__assets)
+
+            # Check if all MultiAssets contain the same type of assets
+            first_asset_type = type(multi_assets[0].assets[0])
+            if not all(
+                isinstance(asset, first_asset_type)
+                for multi_asset in multi_assets
+                for asset in multi_asset.assets
+            ):
+                raise ValueError(
+                    "All MultiAssets must contain the same type of assets (either all MediaAssets or all TextAssets)."
+                )
+
+            # Process based on the asset type
+            if issubclass(first_asset_type, MediaAsset):
+                order.dataset._add_media_from_paths(
+                    multi_assets, self.__metadata, max_upload_workers
+                )
+
+            elif issubclass(first_asset_type, TextAsset):
+                order.dataset._add_texts(multi_assets)
+
+            else:
+                raise ValueError(
+                    "MultiAsset must contain MediaAssets or TextAssets objects."
+                )
+
+        elif order.dataset:
+            raise ValueError(
+                "Media paths must all be of the same type: MediaAsset, TextAsset, or MultiAsset."
+            )
+
+        return order
+
+    def _workflow(self, workflow: Workflow) -> "RapidataOrderBuilder":
+        """
+        Set the workflow for the order.
+
+        Args:
+            workflow (Workflow): The workflow to be set.
+
+        Returns:
+            RapidataOrderBuilder: The updated RapidataOrderBuilder instance.
+        """
+        if not isinstance(workflow, Workflow):
+            raise TypeError("Workflow must be of type Workflow.")
+
+        self.__workflow = workflow
+        return self
+
+    def _referee(self, referee: Referee) -> "RapidataOrderBuilder":
+        """
+        Set the referee for the order.
+
+        Args:
+            referee (Referee): The referee to be set.
+
+        Returns:
+            RapidataOrderBuilder: The updated RapidataOrderBuilder instance.
+        """
+        if not isinstance(referee, Referee):
+            raise TypeError("Referee must be of type Referee.")
+
+        self.__referee = referee
+        return self
+
+    def _media(
+        self,
+        asset: Sequence[BaseAsset],
+        metadata: Sequence[Metadata] | None = None,
+    ) -> "RapidataOrderBuilder":
+        """
+        Set the media assets for the order.
+
+        Args:
+            asset: (list[MediaAsset] | list[TextAsset] | list[MultiAsset]): The paths of the media assets to be set.
+            metadata: (list[Metadata] | None, optional): Metadata for the media assets. Defaults to None.
+
+        Returns:
+            RapidataOrderBuilder: The updated RapidataOrderBuilder instance.
+        """
+        if not isinstance(asset, list):
+            raise TypeError("Media paths must be provided as a list of paths.")
+
+        # for a in asset:
+        #     if not isinstance(a, (MediaAsset, TextAsset, MultiAsset)):
+        #         raise TypeError(
+        #             "Media paths must be of type MediaAsset, TextAsset, or MultiAsset."
+        #         )
+
+        if metadata:
+            for data in metadata:
+                if not isinstance(data, Metadata):
+                    raise TypeError("Metadata must be of type Metadata.")
+
+        self.__assets = asset
+        self.__metadata = metadata
+        return self
+
+    def _settings(self, settings: Sequence[RapidataSetting]) -> "RapidataOrderBuilder":
+        """
+        Set the settings for the order.
+
+        Args:
+            settings (Settings): The settings to be set.
+
+        Returns:
+            RapidataOrderBuilder: The updated RapidataOrderBuilder instance.
+        """
+
+        if not isinstance(settings, list):
+            raise TypeError("Settings must be provided as a list of Setting objects.")
+        
+        for s in settings:
+            if not isinstance(s, RapidataSetting):
+                raise TypeError("The settings list must only contain Setting objects.")
+
+        self.__settings = settings
+        return self
+
+    def _filters(self, filters: Sequence[RapidataFilter]) -> "RapidataOrderBuilder":
+        """
+        Set the filters for the order, e.g., country, language, userscore, etc.
+
+        Args:
+            filters (Sequence[Filters]): The user filters to be set.
+
+        Returns:
+            RapidataOrderBuilder: The updated RapidataOrderBuilder instance.
+        """
+        if not isinstance(filters, list):
+            raise TypeError("Filters must be provided as a list of Filter objects.")
+
+        for f in filters:
+            if not isinstance(f, RapidataFilter):
+                raise TypeError("Filters must be of type Filter.")
+
+        if len(self.__user_filters) > 0:
+            print("Overwriting existing user filters.")
+
+        self.__user_filters = filters
+        return self
+
+    def _validation_set_id(self, validation_set_id: str) -> "RapidataOrderBuilder":
+        """
+        Set the validation set ID for the order.
+
+        Args:
+            validation_set_id (str): The validation set ID to be set.
+
+        Returns:
+            RapidataOrderBuilder: The updated RapidataOrderBuilder instance.
+        """
+        if not isinstance(validation_set_id, str):
+            raise TypeError("Validation set ID must be of type str.")
+
+        self.__validation_set_id = validation_set_id
+        return self
+
+    def _rapids_per_bag(self, amount: int) -> "RapidataOrderBuilder":
+        """
+        Define the number of tasks a user sees in a single session.
+
+        Args:
+            amount (int): The number of tasks a user sees in a single session.
+
+        Returns:
+            RapidataOrderBuilder: The updated RapidataOrderBuilder instance.
+
+        Raises:
+            NotImplementedError: This method is not implemented yet.
+        """
+        raise NotImplementedError("Not implemented yet.")
+
+    def _selections(self, selections: Sequence[RapidataSelection]) -> "RapidataOrderBuilder":
+        """
+        Set the selections for the order.
+
+        Args:
+            selections (Sequence[Selection]): The selections to be set.
+
+        Returns:
+            RapidataOrderBuilder: The updated RapidataOrderBuilder instance.
+        """
+        if not isinstance(selections, list):
+            raise TypeError(
+                "Selections must be provided as a list of Selection objects."
+            )
+
+        for selection in selections:
+            if not isinstance(selection, RapidataSelection):
+                raise TypeError("Selections must be of type Selection.")
+
+        self.__selections = selections  # type: ignore
+        return self
+
+    def _priority(self, priority: int) -> "RapidataOrderBuilder":
+        """
+        Set the priority for the order.
+
+        Args:
+            priority (int): The priority to be set.
+
+        Returns:
+            RapidataOrderBuilder: The updated RapidataOrderBuilder instance.
+        """
+        if not isinstance(priority, int):
+            raise TypeError("Priority must be of type int.")
+
+        self.__priority = priority
+        return self

rapidata/rapidata_client/order/rapidata_order.py

@@ -1,5 +1,5 @@
 from time import sleep
-from rapidata.rapidata_client.dataset.rapidata_dataset import RapidataDataset
+from rapidata.rapidata_client.order._rapidata_dataset import RapidataDataset
 from rapidata.service.openapi_service import OpenAPIService
 import json
 from rapidata.api_client.exceptions import ApiException
@@ -9,33 +9,48 @@ from tqdm import tqdm
 
 class RapidataOrder:
     """
-    Represents a Rapidata order.
+    An instance of a Rapidata order.
+
+    Used to interact with a specific order in the Rapidata system. 
+    Such as starting, pausing, and getting the results of the order.
 
     Args:
-        The ID of the order.
-        The optional Dataset associated with the order.
-        The OpenAPIService instance used to interact with the Rapidata API.
-        The name of the order.
+        name: The name of the order.
+        order_id: The ID of the order.
+        openapi_service: The OpenAPIService instance used to interact with the Rapidata API.
+        dataset: The optional Dataset associated with the order.
     """
 
     def __init__(
         self,
+        name: str,
         order_id: str,
-        dataset: Optional[RapidataDataset],
         openapi_service: OpenAPIService,
-        name: str,
+        dataset: Optional[RapidataDataset]=None,
     ):
-        self.openapi_service = openapi_service
         self.order_id = order_id
-        self._dataset = dataset
         self.name = name
-        self._workflow_id = None
+        self.__openapi_service = openapi_service
+        self.__dataset = dataset
+        self.__workflow_id = None
 
-    def submit(self):
+    def run(self, print_link: bool=True):
         """
-        Submits the order for processing.
+        Runs the order for to start collecting votes.
         """
-        self.openapi_service.order_api.order_submit_post(self.order_id)
+        self.__openapi_service.order_api.order_submit_post(self.order_id)
+
+        if print_link:
+            print(f"Order '{self.name}' is now viewable under: https://app.{self.__openapi_service.enviroment}/order/detail/{self.order_id}")
+        
+        return self
+    
+    def pause(self):
+        """
+        Pauses the order.
+        """
+        self.__openapi_service.order_api.order_pause_post(self.order_id)
+        print(f"Order '{self}' has been paused.")
 
     def get_status(self) -> str:
         """
@@ -44,19 +59,22 @@ class RapidataOrder:
         Returns: 
             The status of the order.
         """
-        return self.openapi_service.order_api.order_get_by_id_get(self.order_id).state
+        return self.__openapi_service.order_api.order_get_by_id_get(self.order_id).state
 
-    def display_progress_bar(self, refresh_rate=5):
+    def display_progress_bar(self, refresh_rate: int=5):
         """
         Displays a progress bar for the order processing using tqdm.
         
-        Prameter: 
-            How often to refresh the progress bar, in seconds.
+        Args: 
+            refresh_rate: How often to refresh the progress bar, in seconds.
         """
+        if refresh_rate < 1:
+            raise ValueError("refresh_rate must be at least 1")
+        
         with tqdm(total=100, desc="Processing order", unit="%", bar_format="{desc}: {percentage:3.0f}%|{bar}| completed [{elapsed}<{remaining}, {rate_fmt}]") as pbar:
             last_percentage = 0
             while True:
-                current_percentage = self._get_workflow_progress().completion_percentage
+                current_percentage = self.__get_workflow_progress().completion_percentage
                 if current_percentage > last_percentage:
                     pbar.update(current_percentage - last_percentage)
                     last_percentage = current_percentage
@@ -66,28 +84,28 @@ class RapidataOrder:
 
                 sleep(refresh_rate)
 
-    def _get_workflow_id(self):
-        if self._workflow_id:
-            return self._workflow_id
+    def __get_workflow_id(self):
+        if self.__workflow_id:
+            return self.__workflow_id
 
         for _ in range(10):
             try:
-                order_result = self.openapi_service.order_api.order_get_by_id_get(self.order_id)
-                pipeline = self.openapi_service.pipeline_api.pipeline_id_get(order_result.pipeline_id)
-                self._workflow_id = cast(WorkflowArtifactModel, pipeline.artifacts["workflow-artifact"].actual_instance).workflow_id
+                order_result = self.__openapi_service.order_api.order_get_by_id_get(self.order_id)
+                pipeline = self.__openapi_service.pipeline_api.pipeline_id_get(order_result.pipeline_id)
+                self.__workflow_id = cast(WorkflowArtifactModel, pipeline.artifacts["workflow-artifact"].actual_instance).workflow_id
                 break
             except Exception:
                 sleep(2)
-        if not self._workflow_id:
+        if not self.__workflow_id:
             raise Exception("Order has not started yet. Please start it or wait for a few seconds and try again.")
-        return self._workflow_id
+        return self.__workflow_id
     
-    def _get_workflow_progress(self):
-        workflow_id = self._get_workflow_id()
+    def __get_workflow_progress(self):
+        workflow_id = self.__get_workflow_id()
         progress = None
         for _ in range(2):
             try:
-                progress = self.openapi_service.workflow_api.workflow_get_progress_get(workflow_id)
+                progress = self.__openapi_service.workflow_api.workflow_get_progress_get(workflow_id)
                 break
             except Exception:
                 sleep(5)
@@ -111,7 +129,7 @@ class RapidataOrder:
 
         try:
             # Get the raw result string
-            result_str = self.openapi_service.order_api.order_result_get(id=self.order_id)
+            result_str = self.__openapi_service.order_api.order_result_get(id=self.order_id)
             # Parse the result string as JSON
             return json.loads(result_str)
         except ApiException as e:
@@ -128,7 +146,7 @@ class RapidataOrder:
         Returns: 
             The RapidataDataset instance.
         """
-        return self._dataset
+        return self.__dataset
     
     def __str__(self) -> str:
         return f"name: '{self.name}' order id: {self.order_id}"