rapidata 1.7.1__py3-none-any.whl → 1.8.0__py3-none-any.whl

rapidata/rapidata_client/dataset/rapidata_dataset.py

@@ -106,10 +106,12 @@ class RapidataDataset:
                     else:
                         files.append(cast(str, asset.path))
 
-            self.openapi_service.dataset_api.dataset_create_datapoint_post(
+            upload_response = self.openapi_service.dataset_api.dataset_create_datapoint_post(
                 model=model,
                 files=files # type: ignore
             )
+            if upload_response.errors:
+                raise ValueError(f"Error uploading datapoint: {upload_response.errors}")
 
         total_uploads = len(media_paths)
         with ThreadPoolExecutor(max_workers=max_workers) as executor:

rapidata/rapidata_client/dataset/rapidata_validation_set.py

@@ -52,6 +52,27 @@ class RapidataValidationSet:
         self.openapi_service = openapi_service
         self.name = name
 
+    def upload_files(self, model: AddValidationRapidModel, assets: list[MediaAsset]):
+        """Upload a file to the validation set.
+
+        Args:
+            asset list[(MediaAsset)]: The asset to upload.
+
+        Returns:
+            str: The path to the uploaded file.
+        """
+        files = []
+        for asset in assets:
+            if isinstance(asset.path, str):
+                files.append(asset.path)
+            elif isinstance(asset.path, bytes):
+                files.append((asset.name, asset.path))
+            else:
+                raise ValueError("upload file failed")
+        self.openapi_service.validation_api.validation_add_validation_rapid_post(
+            model=model, files=files
+        )        
+
     def add_general_validation_rapid(
         self,
         payload: (
@@ -107,9 +128,7 @@ class RapidataValidationSet:
             randomCorrectProbability=randomCorrectProbability,
         )
         if isinstance(asset, MediaAsset):
-            self.openapi_service.validation_api.validation_add_validation_rapid_post(
-                model=model, files=[asset.path]
-            )
+            self.upload_files(model=model, assets=[asset])
 
         elif isinstance(asset, TextAsset):
             model = AddValidationTextRapidModel(
@@ -128,12 +147,10 @@ class RapidataValidationSet:
             )
 
         elif isinstance(asset, MultiAsset):
-            files = [a.path for a in asset if isinstance(a, MediaAsset)]
+            files = [a for a in asset if isinstance(a, MediaAsset)]
             texts = [a.text for a in asset if isinstance(a, TextAsset)]
             if files:
-                self.openapi_service.validation_api.validation_add_validation_rapid_post(
-                    model=model, files=files # type: ignore
-                )
+                self.upload_files(model=model, assets=files)
             if texts:
                 model = AddValidationTextRapidModel(
                     validationSetId=self.id,

rapidata/rapidata_client/dataset/validation_set_builder.py

@@ -16,6 +16,14 @@ from rapidata.rapidata_client.dataset.validation_rapid_parts import ValidatioRap
 from rapidata.rapidata_client.metadata.base_metadata import Metadata
 from rapidata.service.openapi_service import OpenAPIService
 
+from rapidata.rapidata_client.dataset.rapid_builders.rapids import (
+    Rapid, 
+    ClassificationRapid,
+    CompareRapid,
+    TranscriptionRapid
+)
+from deprecated import deprecated
+
 
 class ValidationSetBuilder:
     """The ValidationSetBuilder is used to build a validation set.
@@ -71,7 +79,29 @@ class ValidationSetBuilder:
             )
 
         return validation_set
+    
+    def add_rapid(self, rapid: Rapid):
+        """Add a rapid to the validation set.
+        To create the Rapid, use the RapidataClient.rapid_builder instance.
+
+        Args:
+            rapid (Rapid): The rapid to add to the validation set.
+        """
+        if not isinstance(rapid, Rapid):
+            raise ValueError("This method only accepts Rapid instances")
+        
+        if isinstance(rapid, ClassificationRapid):
+            self._add_classify_rapid(rapid.asset, rapid.question, rapid.options, rapid.truths, rapid.metadata)
+
+        if isinstance(rapid, CompareRapid):
+            self._add_compare_rapid(rapid.asset, rapid.criteria, rapid.truth, rapid.metadata)
 
+        if isinstance(rapid, TranscriptionRapid):
+            self._add_transcription_rapid(rapid.asset, rapid.instruction, rapid.transcription, rapid.truths, rapid.strict_grading)
+
+        return self
+
+    @deprecated("Use add_rapid instead")
     def add_classify_rapid(
         self,
         asset: MediaAsset | TextAsset,
@@ -82,6 +112,33 @@ class ValidationSetBuilder:
     ):
         """Add a classify rapid to the validation set.
 
+        Args:
+            asset (MediaAsset | TextAsset): The asset for the rapid.
+            question (str): The question for the rapid.
+            categories (list[str]): The list of categories for the rapid.
+            truths (list[str]): The list of truths for the rapid.
+            metadata (list[Metadata], optional): The metadata for the rapid. Defaults to an empty list.
+
+        Returns:
+            ValidationSetBuilder: The ValidationSetBuilder instance.
+
+        Raises:
+            ValueError: If the lengths of categories and truths are inconsistent.
+        """
+        self._add_classify_rapid(asset, question, categories, truths, metadata)
+
+        return self
+    
+    def _add_classify_rapid(
+        self,
+        asset: MediaAsset | TextAsset,
+        question: str,
+        categories: list[str],
+        truths: list[str],
+        metadata: list[Metadata] = [],
+    ):
+        """Add a classify rapid to the validation set.
+
         Args:
             asset (MediaAsset | TextAsset): The asset for the rapid.
             question (str): The question for the rapid.
@@ -113,8 +170,7 @@ class ValidationSetBuilder:
             )
         )
 
-        return self
-
+    @deprecated("Use add_rapid instead")
     def add_compare_rapid(
         self,
         asset: MultiAsset,
@@ -136,7 +192,32 @@ class ValidationSetBuilder:
         Raises:
             ValueError: If the number of assets is not exactly two.
         """
-        payload = ComparePayload(_t="ComparePayload", criteria=question)
+        self._add_compare_rapid(asset, question, truth, metadata)
+
+        return self
+    
+    def _add_compare_rapid(
+        self,
+        asset: MultiAsset,
+        criteria: str,
+        truth: str,
+        metadata: list[Metadata] = [],
+    ):
+        """Add a compare rapid to the validation set.
+
+        Args:
+            asset (MultiAsset): The assets for the rapid.
+            criteria (str): The criteria for the comparison.
+            truth (str): The truth identifier for the rapid.
+            metadata (list[Metadata], optional): The metadata for the rapid. Defaults to an empty list.
+
+        Returns:
+            ValidationSetBuilder: The ValidationSetBuilder instance.
+
+        Raises:
+            ValueError: If the number of assets is not exactly two.
+        """
+        payload = ComparePayload(_t="ComparePayload", criteria=criteria)
         # take only last part of truth path
         truth = os.path.basename(truth)
         model_truth = CompareTruth(_t="CompareTruth", winnerId=truth)
@@ -146,7 +227,7 @@ class ValidationSetBuilder:
 
         self._rapid_parts.append(
             ValidatioRapidParts(
-                question=question,
+                question=criteria,
                 payload=payload,
                 truths=model_truth,
                 metadata=metadata,
@@ -155,8 +236,7 @@ class ValidationSetBuilder:
             )
         )
 
-        return self
-
+    @deprecated("Use add_rapid instead")
     def add_transcription_rapid(
         self,
         asset: MediaAsset | TextAsset,
@@ -168,6 +248,35 @@ class ValidationSetBuilder:
     ):
         """Add a transcription rapid to the validation set.
 
+        Args:
+            asset (MediaAsset | TextAsset): The asset for the rapid.
+            question (str): The question for the rapid.
+            transcription (list[str]): The transcription for the rapid.
+            truths (list[int]): The list of indices of the true word selections.
+            strict_grading (bool | None, optional): The strict grading for the rapid. Defaults to None.
+            metadata (list[Metadata], optional): The metadata for the rapid.
+
+        Returns:
+            ValidationSetBuilder: The ValidationSetBuilder instance.
+
+        Raises:
+            ValueError: If a correct word is not found in the transcription.
+        """
+        self._add_transcription_rapid(asset, question, transcription, truths, strict_grading, metadata)
+
+        return self
+    
+    def _add_transcription_rapid(
+        self,
+        asset: MediaAsset | TextAsset,
+        question: str,
+        transcription: str,
+        truths: list[int],
+        strict_grading: bool | None = None,
+        metadata: list[Metadata] = [],
+    ):
+        """Add a transcription rapid to the validation set.
+
         Args:
             asset (MediaAsset | TextAsset): The asset for the rapid.
             question (str): The question for the rapid.
@@ -213,5 +322,3 @@ class ValidationSetBuilder:
                 randomCorrectProbability = 1 / len(transcription_words),
             )
         )
-
-        return self

rapidata/rapidata_client/filter/country_filter.py

@@ -7,6 +7,9 @@ class CountryFilter(Filter):
 
     def __init__(self, country_codes: list[str]):
         # check that all characters in the country codes are uppercase
+        if not isinstance(country_codes, list):
+            raise ValueError("Country codes must be a list")
+        
         if not all([code.isupper() for code in country_codes]):
             raise ValueError("Country codes must be uppercase")
 

rapidata/rapidata_client/filter/language_filter.py

@@ -8,6 +8,9 @@ from rapidata.api_client.models.language_user_filter_model import (
 class LanguageFilter(Filter):
 
     def __init__(self, language_codes: list[str]):
+        if not isinstance(language_codes, list):
+            raise ValueError("Language codes must be a list")
+        
         # check that all characters in the language codes are lowercase
         if not all([code.islower() for code in language_codes]):
             raise ValueError("Language codes must be lowercase")

rapidata/rapidata_client/metadata/prompt_metadata.py

@@ -6,8 +6,12 @@ class PromptMetadata(Metadata):
 
     def __init__(self, prompt: str, identifier: str = "prompt"):
         super().__init__(identifier=identifier)
+
+        if not isinstance(prompt, str):
+            raise ValueError("Prompt must be a string")
+        
         self._prompt = prompt
     
 
     def to_model(self):
-        return PromptMetadataInput(_t="PromptMetadataInput", identifier=self._identifier, prompt=self._prompt)
+        return PromptMetadataInput(_t="PromptMetadataInput", identifier=self._identifier, prompt=self._prompt)

rapidata/rapidata_client/order/rapidata_order.py

@@ -106,7 +106,7 @@ class RapidataOrder:
         Returns: 
             The results of the order.
         """
-        while self.get_status() == "Processing":
+        while self.get_status() not in ["Completed", "Paused", "ManuelReview"]:
             sleep(5)
 
         try:

rapidata/rapidata_client/order/rapidata_order_builder.py

@@ -27,7 +27,7 @@ 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
+from rapidata.rapidata_client.assets import MediaAsset, TextAsset, MultiAsset, BaseAsset
 
 from typing import Optional, cast, Sequence
 
@@ -69,7 +69,7 @@ class RapidataOrderBuilder:
         self._selections: list[Selection] = []
         self._rapids_per_bag: int = 2
         self._priority: int = 50
-        self._assets: list[MediaAsset] | list[TextAsset] | list[MultiAsset] = []
+        self._assets: Sequence[BaseAsset] = []
 
     def _to_model(self) -> CreateOrderModel:
         """
@@ -202,7 +202,7 @@ class RapidataOrderBuilder:
             order.submit()
 
         if not disable_link:
-            print(f"Order '{self._name}' is now viewable under https://app.rapidata.ai/order/detail/{order.order_id}.")
+            print(f"Order '{self._name}' is now viewable under: https://app.{self._openapi_service.enviroment}/order/detail/{order.order_id}")
 
         return order
 
@@ -240,8 +240,8 @@ class RapidataOrderBuilder:
 
     def media(
         self,
-        asset: list[MediaAsset] | list[TextAsset] | list[MultiAsset],
-        metadata: Sequence[Metadata] | None = None,
+        asset: Sequence[BaseAsset],
+        metadata: Sequence[Metadata] | None = None, # make this a list of metadata on next major release
     ) -> "RapidataOrderBuilder":
         """
         Set the media assets for the order.

rapidata/rapidata_client/rapidata_client.py

@@ -1,6 +1,4 @@
-from rapidata.rapidata_client.dataset.rapidata_validation_set import (
-    RapidataValidationSet,
-)
+from rapidata.rapidata_client.dataset.rapidata_validation_set import RapidataValidationSet
 from rapidata.rapidata_client.dataset.validation_set_builder import ValidationSetBuilder
 from rapidata.rapidata_client.order.rapidata_order_builder import RapidataOrderBuilder
 from rapidata.service.openapi_service import OpenAPIService
@@ -9,6 +7,10 @@ from rapidata.rapidata_client.dataset.rapidata_dataset import RapidataDataset
 
 from rapidata.rapidata_client.simple_builders.simple_classification_builders import ClassificationQuestionBuilder
 from rapidata.rapidata_client.simple_builders.simple_compare_builders import CompareCriteriaBuilder
+from rapidata.rapidata_client.simple_builders.simple_free_text_builders import FreeTextQuestionBuilder
+from rapidata.rapidata_client.simple_builders.simple_transcription_builders import TranscriptionInstructionBuilder
+
+from rapidata.rapidata_client.dataset.rapid_builders import BaseRapidBuilder
 
 from rapidata.api_client.exceptions import BadRequestException
 from urllib3._collections import HTTPHeaderDict
@@ -21,16 +23,19 @@ from rapidata.api_client.models.sort_criterion import SortCriterion
 
 from rapidata.api_client.models.query_validation_set_model import QueryValidationSetModel
 
+from deprecated import deprecated
+
 
 class RapidataClient:
-    """The Rapidata client is the main entry point for interacting with the Rapidata API. It allows you to create orders and validation sets. For creating a new order, check out `new_order()`. For creating a new validation set, check out `new_validation_set()`."""
+    """The Rapidata client is the main entry point for interacting with the Rapidata API. It allows you to create orders and validation sets."""
 
+    rapid_builder = BaseRapidBuilder()
+    
     def __init__(
         self,
         client_id: str | None = None,
         client_secret: str | None = None,
-        endpoint: str = "https://api.rapidata.ai",
-        token_url: str = "https://auth.rapidata.ai",
+        enviroment: str = "rapidata.ai",
         oauth_scope: str = "openid",
         cert_path: str | None = None,
     ):
@@ -44,12 +49,12 @@ class RapidataClient:
         self.openapi_service = OpenAPIService(
             client_id=client_id,
             client_secret=client_secret,
-            endpoint=endpoint,
-            token_url=token_url,
+            enviroment=enviroment,
             oauth_scope=oauth_scope,
             cert_path=cert_path
         )
-
+    
+    @deprecated("Use the specific builder methods instead.")
     def new_order(self, name: str) -> RapidataOrderBuilder:
         """Create a new order using a RapidataOrderBuilder instance.
 
@@ -177,3 +182,26 @@ class RapidataClient:
             CompareQuestionBuilder: A CompareQuestionBuilder instance.
         """
         return CompareCriteriaBuilder(name=name, openapi_service=self.openapi_service)
+    
+    def create_free_text_order(self, name: str) -> FreeTextQuestionBuilder:
+        """Create a new free text order where people are asked to provide a free text answer.
+
+        Args:
+            name (str): The name of the order.
+
+        Returns:
+            FreeTextQuestionBuilder: A FreeTextQuestionBuilder instance.
+        """
+        return FreeTextQuestionBuilder(name=name, openapi_service=self.openapi_service)
+    
+    def create_transcription_order(self, name: str) -> TranscriptionInstructionBuilder:
+        """Create a new transcription order where people are asked to transcribe an audio file.
+
+        Args:
+            name (str): The name of the order.
+
+        Returns:
+            TranscriptionInstructionBuilder: A TranscriptionInstructionBuilder instance.
+        """
+        return TranscriptionInstructionBuilder(name=name, openapi_service=self.openapi_service)
+

rapidata/rapidata_client/settings/__init__.py

@@ -1,2 +1,2 @@
 from .feature_flags import FeatureFlags
-from .settings import Settings
+from .settings import Settings, TranslationBehaviour

rapidata/rapidata_client/settings/settings.py

@@ -96,17 +96,18 @@ class Settings:
         """
         self._settings["no_shuffle"] = str(value)
         return self
-
-    def compare_with_prompt_design(self, value: bool = True):
-        """A special design to compare two texts/images based on a criteria and a given prompt.
-
+    
+    def play_video_until_the_end(self, additional_time=0):
+        """Allows users to only answer once the video has finished playing.
+        The additional time gets added on top. Can be negative to allow answers before the video ends.
+        
         Args:
-            value (bool, optional): Whether to enable compare with prompt design. Defaults to True.
-
+            additional_time (int, optional): Additional time in milliseconds. Defaults to 0.
+            
         Returns:
-            Settings: The current Settings instance for method chaining.
-        """
-        self._settings["claire"] = str(value)
+            Settings: The current Settings instance for method chaining."""
+            
+        self._settings["alert_on_fast_response_add_media_duration"] = str(additional_time)
         return self
 
     def key_value(self, key: str, value: str):