trismik 1.0.1__tar.gz → 1.0.2__tar.gz

{trismik-1.0.1 → trismik-1.0.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: trismik
-Version: 1.0.1
+Version: 1.0.2
 Summary: 
 License-File: LICENSE
 Author: Bartosz Kielczewski

{trismik-1.0.1 → trismik-1.0.2}/pyproject.toml

@@ -14,7 +14,7 @@ dependencies = [
     "httpx (>=0.27.2,<1.0.0)",
     "nest-asyncio (>=1.6.0,<2.0.0)",
 ]
-version = "1.0.1"
+version = "1.0.2"
 
 [project.optional-dependencies]
 examples = [

{trismik-1.0.1 → trismik-1.0.2}/src/trismik/_async/client.py

@@ -20,6 +20,7 @@ from trismik.types import (
     TrismikClassicEvalRequest,
     TrismikClassicEvalResponse,
     TrismikDataset,
+    TrismikDatasetInfo,
     TrismikItem,
     TrismikMeResponse,
     TrismikProject,
@@ -173,9 +174,35 @@ class TrismikAsyncClient:
         except httpx.HTTPError as e:
             raise TrismikApiError(str(e)) from e
 
+    async def get_dataset_info(self, dataset_id: str) -> TrismikDatasetInfo:
+        """
+        Get detailed information about a specific dataset.
+
+        Args:
+            dataset_id (str): ID of the dataset to retrieve.
+
+        Returns:
+            TrismikDatasetInfo: Detailed dataset information including datacard.
+
+        Raises:
+            TrismikApiError: If API request fails.
+        """
+        try:
+            url = "/datasets/get-dataset"
+            body = {"datasetId": dataset_id}
+            response = await self._http_client.post(url, json=body)
+            response.raise_for_status()
+            json = response.json()
+            return TrismikResponseMapper.to_dataset_info(json)
+        except httpx.HTTPStatusError as e:
+            raise TrismikApiError(TrismikUtils.get_error_message(e.response)) from e
+        except httpx.HTTPError as e:
+            raise TrismikApiError(str(e)) from e
+
     async def start_run(
         self,
         dataset_id: str,
+        split: str,
         project_id: str,
         experiment: str,
         metadata: Optional[TrismikRunMetadata] = None,
@@ -185,6 +212,7 @@ class TrismikAsyncClient:
 
         Args:
             dataset_id (str): ID of the dataset.
+            split (str): Split of the dataset to use (e.g., "train", "test", "val").
             project_id (str): ID of the project.
             experiment (str): Name of the experiment.
             metadata (Optional[TrismikRunMetadata]): Run metadata.
@@ -201,6 +229,7 @@ class TrismikAsyncClient:
             url = "/runs/start"
             body = {
                 "datasetId": dataset_id,
+                "split": split,
                 "projectId": project_id,
                 "experiment": experiment,
                 "metadata": metadata.toDict() if metadata else {},
@@ -444,6 +473,7 @@ class TrismikAsyncClient:
     async def run(  # noqa: E704
         self,
         test_id: str,
+        split: str,
         project_id: str,
         experiment: str,
         run_metadata: TrismikRunMetadata,
@@ -457,6 +487,7 @@ class TrismikAsyncClient:
     async def run(  # noqa: E704
         self,
         test_id: str,
+        split: str,
         project_id: str,
         experiment: str,
         run_metadata: TrismikRunMetadata,
@@ -469,6 +500,7 @@ class TrismikAsyncClient:
     async def run(
         self,
         test_id: str,
+        split: str,
         project_id: str,
         experiment: str,
         run_metadata: TrismikRunMetadata,
@@ -482,6 +514,7 @@ class TrismikAsyncClient:
 
         Args:
             test_id: ID of the test to run.
+            split: Split of the dataset to use (e.g., "train", "test", "val").
             project_id: ID of the project.
             experiment: Name of the experiment.
             run_metadata: Metadata for the run.
@@ -502,7 +535,7 @@ class TrismikAsyncClient:
             raise NotImplementedError("with_responses is not yet implemented for the new API flow")
 
         # Start run and get first item
-        start_response = await self.start_run(test_id, project_id, experiment, run_metadata)
+        start_response = await self.start_run(test_id, split, project_id, experiment, run_metadata)
 
         # Initialize state tracking
         states: List[TrismikAdaptiveTestState] = []

{trismik-1.0.1 → trismik-1.0.2}/src/trismik/_mapper.py

@@ -4,7 +4,10 @@ from typing import Any, Dict, List
 from trismik.exceptions import TrismikApiError
 from trismik.types import (
     TrismikClassicEvalResponse,
+    TrismikDatacard,
     TrismikDataset,
+    TrismikDatasetInfo,
+    TrismikDatasetVersion,
     TrismikItem,
     TrismikMeResponse,
     TrismikMultipleChoiceTextItem,
@@ -40,17 +43,64 @@ class TrismikResponseMapper:
             json (Dict[str, Any]): JSON response containing dataset data.
 
         Returns:
-            List[TrismikDataset]: List of
-            dataset objects with IDs and names.
+            List[TrismikDataset]: List of dataset objects with IDs, names,
+                adaptive status, and splits.
         """
         return [
             TrismikDataset(
                 id=item["id"],
                 name=item["name"],
+                isAdaptive=item["isAdaptive"],
+                splits=item["splits"],
             )
             for item in json["data"]
         ]
 
+    @staticmethod
+    def to_dataset_info(json: Dict[str, Any]) -> TrismikDatasetInfo:
+        """
+        Convert JSON response to a TrismikDatasetInfo object.
+
+        Args:
+            json (Dict[str, Any]): JSON response containing detailed dataset
+                information.
+
+        Returns:
+            TrismikDatasetInfo: Dataset info object with complete metadata
+                including datacard.
+        """
+        version = TrismikDatasetVersion(
+            year=json["datacard"]["version"]["year"],
+            month=json["datacard"]["version"]["month"],
+            revision=json["datacard"]["version"]["revision"],
+        )
+
+        datacard = TrismikDatacard(
+            task=json["datacard"]["task"],
+            license=json["datacard"]["license"],
+            version=version,
+            datasetHf=json["datacard"]["datasetHf"],
+            languages=json["datacard"]["languages"],
+            datasetUrl=json["datacard"]["datasetUrl"],
+            itemsCount=json["datacard"]["itemsCount"],
+            splitsUsed=json["datacard"]["splitsUsed"],
+            datasetName=json["datacard"]["datasetName"],
+            description=json["datacard"]["description"],
+            exampleItem=json["datacard"]["exampleItem"],
+            originalReadmeUrl=json["datacard"]["originalReadmeUrl"],
+            originalReadmeDownloadedAt=json["datacard"]["originalReadmeDownloadedAt"],
+            originalItemCount=json["datacard"]["originalItemCount"],
+            difficultyEstimators=json["datacard"]["difficultyEstimators"],
+        )
+
+        return TrismikDatasetInfo(
+            id=json["id"],
+            name=json["name"],
+            isAdaptive=json["isAdaptive"],
+            splits=json["splits"],
+            datacard=datacard,
+        )
+
     @staticmethod
     def to_run(json: Dict[str, Any]) -> TrismikRun:
         """

{trismik-1.0.1 → trismik-1.0.2}/src/trismik/_sync/client.py

@@ -20,6 +20,7 @@ from trismik.types import (
     TrismikClassicEvalRequest,
     TrismikClassicEvalResponse,
     TrismikDataset,
+    TrismikDatasetInfo,
     TrismikItem,
     TrismikMeResponse,
     TrismikProject,
@@ -173,9 +174,35 @@ class TrismikClient:
         except httpx.HTTPError as e:
             raise TrismikApiError(str(e)) from e
 
+    def get_dataset_info(self, dataset_id: str) -> TrismikDatasetInfo:
+        """
+        Get detailed information about a specific dataset.
+
+        Args:
+            dataset_id (str): ID of the dataset to retrieve.
+
+        Returns:
+            TrismikDatasetInfo: Detailed dataset information including datacard.
+
+        Raises:
+            TrismikApiError: If API request fails.
+        """
+        try:
+            url = "/datasets/get-dataset"
+            body = {"datasetId": dataset_id}
+            response = self._http_client.post(url, json=body)
+            response.raise_for_status()
+            json = response.json()
+            return TrismikResponseMapper.to_dataset_info(json)
+        except httpx.HTTPStatusError as e:
+            raise TrismikApiError(TrismikUtils.get_error_message(e.response)) from e
+        except httpx.HTTPError as e:
+            raise TrismikApiError(str(e)) from e
+
     def start_run(
         self,
         dataset_id: str,
+        split: str,
         project_id: str,
         experiment: str,
         metadata: Optional[TrismikRunMetadata] = None,
@@ -185,6 +212,7 @@ class TrismikClient:
 
         Args:
             dataset_id (str): ID of the dataset.
+            split (str): Split of the dataset to use (e.g., "train", "test", "val").
             project_id (str): ID of the project.
             experiment (str): Name of the experiment.
             metadata (Optional[TrismikRunMetadata]): Run metadata.
@@ -201,6 +229,7 @@ class TrismikClient:
             url = "/runs/start"
             body = {
                 "datasetId": dataset_id,
+                "split": split,
                 "projectId": project_id,
                 "experiment": experiment,
                 "metadata": metadata.toDict() if metadata else {},
@@ -444,6 +473,7 @@ class TrismikClient:
     def run(  # noqa: E704
         self,
         test_id: str,
+        split: str,
         project_id: str,
         experiment: str,
         run_metadata: TrismikRunMetadata,
@@ -457,6 +487,7 @@ class TrismikClient:
     def run(  # noqa: E704
         self,
         test_id: str,
+        split: str,
         project_id: str,
         experiment: str,
         run_metadata: TrismikRunMetadata,
@@ -469,6 +500,7 @@ class TrismikClient:
     def run(
         self,
         test_id: str,
+        split: str,
         project_id: str,
         experiment: str,
         run_metadata: TrismikRunMetadata,
@@ -482,6 +514,7 @@ class TrismikClient:
 
         Args:
             test_id: ID of the test to run.
+            split: Split of the dataset to use (e.g., "train", "test", "val").
             project_id: ID of the project.
             experiment: Name of the experiment.
             run_metadata: Metadata for the run.
@@ -502,7 +535,7 @@ class TrismikClient:
             raise NotImplementedError("with_responses is not yet implemented for the new API flow")
 
         # Start run and get first item
-        start_response = self.start_run(test_id, project_id, experiment, run_metadata)
+        start_response = self.start_run(test_id, split, project_id, experiment, run_metadata)
 
         # Initialize state tracking
         states: List[TrismikAdaptiveTestState] = []

{trismik-1.0.1 → trismik-1.0.2}/src/trismik/types.py

@@ -16,6 +16,49 @@ class TrismikDataset:
 
     id: str
     name: str
+    isAdaptive: bool
+    splits: List[str]
+
+
+@dataclass
+class TrismikDatasetVersion:
+    """Version information for a dataset."""
+
+    year: int
+    month: int
+    revision: int
+
+
+@dataclass
+class TrismikDatacard:
+    """Detailed metadata card for a dataset."""
+
+    task: str
+    license: str
+    version: "TrismikDatasetVersion"
+    datasetHf: str
+    languages: List[str]
+    datasetUrl: str
+    itemsCount: int
+    splitsUsed: List[str]
+    datasetName: str
+    description: str
+    exampleItem: str
+    originalReadmeUrl: str
+    originalReadmeDownloadedAt: str
+    originalItemCount: int
+    difficultyEstimators: List[str]
+
+
+@dataclass
+class TrismikDatasetInfo:
+    """Complete dataset information including datacard."""
+
+    id: str
+    name: str
+    isAdaptive: bool
+    splits: List[str]
+    datacard: TrismikDatacard
 
 
 @dataclass