release: 0.2.0-alpha.15

.release-please-manifest.json

@@ -1,3 +1,3 @@
 {
-  ".": "0.2.0-alpha.14"
+  ".": "0.2.0-alpha.15"
 }

CHANGELOG.md

@@ -20,6 +20,22 @@ and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.
 ### Removed
 * Deprecated and removed `publish_ground_truths` method. Use `update_data` instead.
 
+## 0.2.0-alpha.15 (2024-07-31)
+
+Full Changelog: [v0.2.0-alpha.14...v0.2.0-alpha.15](https://github.com/openlayer-ai/openlayer-python/compare/v0.2.0-alpha.14...v0.2.0-alpha.15)
+
+### Features
+
+* improvement: allow specifying dataset as path for uploads ([a4d126f](https://github.com/openlayer-ai/openlayer-python/commit/a4d126f2c0b3bdf67fefbb06fb3ffa9107ea1387))
+* improvement: include method to update batch of inferences ([a8f3d82](https://github.com/openlayer-ai/openlayer-python/commit/a8f3d8246c75ff8ebff8f5e92212044fd3433d47))
+
+
+### Chores
+
+* **internal:** add type construction helper ([#287](https://github.com/openlayer-ai/openlayer-python/issues/287)) ([39fbda1](https://github.com/openlayer-ai/openlayer-python/commit/39fbda1bcaacbd8546926e7d32b7fc2ae1ad058e))
+* **internal:** version bump ([#284](https://github.com/openlayer-ai/openlayer-python/issues/284)) ([73c3067](https://github.com/openlayer-ai/openlayer-python/commit/73c30676b1e49e2355cffd232305c5aab1a0b309))
+* **tests:** update prism version ([#285](https://github.com/openlayer-ai/openlayer-python/issues/285)) ([3c0fcbb](https://github.com/openlayer-ai/openlayer-python/commit/3c0fcbbe9199b68ef5bc92247df751bfd4ae3649))
+
 ## 0.2.0-alpha.14 (2024-07-29)
 
 Full Changelog: [v0.2.0-alpha.13...v0.2.0-alpha.14](https://github.com/openlayer-ai/openlayer-python/compare/v0.2.0-alpha.13...v0.2.0-alpha.14)

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "openlayer"
-version = "0.2.0-alpha.14"
+version = "0.2.0-alpha.15"
 description = "The official Python library for the openlayer API"
 dynamic = ["readme"]
 license = "Apache-2.0"

src/openlayer/_models.py

@@ -406,6 +406,15 @@ def build(
     return cast(_BaseModelT, construct_type(type_=base_model_cls, value=kwargs))
 
 
+def construct_type_unchecked(*, value: object, type_: type[_T]) -> _T:
+    """Loose coercion to the expected type with construction of nested values.
+
+    Note: the returned value from this function is not guaranteed to match the
+    given type.
+    """
+    return cast(_T, construct_type(value=value, type_=type_))
+
+
 def construct_type(*, value: object, type_: object) -> object:
     """Loose coercion to the expected type with construction of nested values.
 

src/openlayer/_version.py

@@ -1,4 +1,4 @@
 # File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
 
 __title__ = "openlayer"
-__version__ = "0.2.0-alpha.14"  # x-release-please-version
+__version__ = "0.2.0-alpha.15"  # x-release-please-version

src/openlayer/lib/data/__init__.py

@@ -1,7 +1,12 @@
 """Data upload functions."""
 
-__all__ = ["StorageType", "upload_reference_dataframe", "upload_batch_inferences"]
+__all__ = [
+    "StorageType",
+    "upload_reference_dataframe",
+    "upload_batch_inferences",
+    "update_batch_inferences",
+]
 
 from ._upload import StorageType
+from .batch_inferences import update_batch_inferences, upload_batch_inferences
 from .reference_dataset import upload_reference_dataframe
-from .batch_inferences import upload_batch_inferences

src/openlayer/lib/data/batch_inferences.py

@@ -5,8 +5,8 @@
 import tempfile
 import time
 from typing import Optional
-import httpx
 
+import httpx
 import pandas as pd
 
 from ... import Openlayer
@@ -19,11 +19,18 @@
 def upload_batch_inferences(
     client: Openlayer,
     inference_pipeline_id: str,
-    dataset_df: pd.DataFrame,
     config: data_stream_params.Config,
+    dataset_df: Optional[pd.DataFrame] = None,
+    dataset_path: Optional[str] = None,
     storage_type: Optional[StorageType] = None,
+    merge: bool = False,
 ) -> None:
     """Uploads a batch of inferences to the Openlayer platform."""
+    if dataset_df is None and dataset_path is None:
+        raise ValueError("Either dataset_df or dataset_path must be provided.")
+    if dataset_df is not None and dataset_path is not None:
+        raise ValueError("Only one of dataset_df or dataset_path should be provided.")
+
     uploader = _upload.Uploader(client, storage_type)
     object_name = f"batch_data_{time.time()}_{inference_pipeline_id}.tar.gz"
 
@@ -34,8 +41,11 @@ def upload_batch_inferences(
 
     # Write dataset and config to temp directory
     with tempfile.TemporaryDirectory() as tmp_dir:
-        temp_file_path = f"{tmp_dir}/dataset.csv"
-        dataset_df.to_csv(temp_file_path, index=False)
+        if dataset_df is not None:
+            temp_file_path = f"{tmp_dir}/dataset.csv"
+            dataset_df.to_csv(temp_file_path, index=False)
+        else:
+            temp_file_path = dataset_path
 
         # Copy relevant files to tmp dir
         config["label"] = "production"
@@ -46,7 +56,11 @@ def upload_batch_inferences(
 
         tar_file_path = os.path.join(tmp_dir, object_name)
         with tarfile.open(tar_file_path, mode="w:gz") as tar:
-            tar.add(tmp_dir, arcname=os.path.basename("monitoring_data"))
+            tar.add(temp_file_path, arcname=os.path.basename("dataset.csv"))
+            tar.add(
+                f"{tmp_dir}/dataset_config.yaml",
+                arcname=os.path.basename("dataset_config.yaml"),
+            )
 
         # Upload to storage
         uploader.upload(
@@ -61,6 +75,26 @@ def upload_batch_inferences(
         cast_to=httpx.Response,
         body={
             "storageUri": presigned_url_response.storage_uri,
-            "performDataMerge": False,
+            "performDataMerge": merge,
         },
     )
+
+
+def update_batch_inferences(
+    client: Openlayer,
+    inference_pipeline_id: str,
+    dataset_df: pd.DataFrame,
+    config: data_stream_params.Config,
+    storage_type: Optional[StorageType] = None,
+) -> None:
+    """Updates a batch of inferences on the Openlayer platform."""
+    if config["inference_id_column_name"] is None:
+        raise ValueError("inference_id_column_name must be set in config")
+    upload_batch_inferences(
+        client=client,
+        inference_pipeline_id=inference_pipeline_id,
+        dataset_df=dataset_df,
+        config=config,
+        storage_type=storage_type,
+        merge=True,
+    )