add jupyter notebook

biosim_client/dataset.py

@@ -1,4 +1,3 @@
-import json
 from typing import Optional, Any
 
 import numpy as np
@@ -38,16 +37,14 @@ def to_numpy(self) -> np.ndarray:
     def to_pandas(self) -> pd.DataFrame:
         dataframe = pd.DataFrame(self.to_numpy().transpose())
         if self.column_names is not None:
-            dataframe.columns = self.column_names
+            dataframe.columns = self.column_names  # type: ignore
         return dataframe
 
     @classmethod
     def from_api(cls, data: DatasetData, hdf5_dataset: HDF5Dataset) -> 'Dataset':
         values = data.to_dict()['values']
         shape = hdf5_dataset.shape
 
-        # print(f"in from_api(), len(values)={len(values)}, shape={shape}, hdf5_dataset.attributes={ {attr.to_dict()['key']: attr.to_dict()['value'] for attr in hdf5_dataset.attributes} }")
-
         attributes: dict[str, Any] = {}
         column_names: Optional[list[str]] = None
         for attr in hdf5_dataset.attributes:

biosim_client/example.py

@@ -21,4 +21,3 @@
     print()
     print(dataframe)
     print()
-

biosim_client/sim_data.py

@@ -1,8 +1,8 @@
 from typing import Optional, get_args
 
 import biosim_client.api_clients.simdata.openapi_client as simdata_client
-from biosim_client.api_clients.simdata.openapi_client import HDF5File, DatasetData, HDF5Dataset, HDF5Group, \
-    Configuration, HDF5Attribute
+from biosim_client.api_clients.simdata.openapi_client import HDF5File, DatasetData, HDF5Dataset, Configuration, \
+    HDF5Attribute
 from biosim_client.dataset import Dataset, AttributeValueTypes
 
 attribute_value_types = get_args(AttributeValueTypes)
@@ -24,16 +24,15 @@ def dataset_names(self) -> list[str]:
         return [dataset.name for group in self.hdf5_file.groups for dataset in group.datasets]
 
     def dataset_uris(self) -> list[str]:
-        return [attr.to_dict()['value'] for group in self.hdf5_file.groups for dataset in group.datasets for attr in dataset.attributes if attr.to_dict()['key'] == "uri"]
+        return [attr.to_dict()['value'] for group in self.hdf5_file.groups for dataset in group.datasets for attr in
+                dataset.attributes if attr.to_dict()['key'] == "uri"]
 
     def get_dataset(self, name: str) -> Dataset:
         if name in self.datasets:
             # print("cache hit, returning cached dataset")
             return self.datasets[name]
 
         dataset_uri = None
-        column_names = None
-        hdf5_group: Optional[HDF5Group] = None
         hdf5_dataset: Optional[HDF5Dataset] = None
         hdf5_attribute: HDF5Attribute
         for hdf5_group in self.hdf5_file.groups:
@@ -42,25 +41,19 @@ def get_dataset(self, name: str) -> Dataset:
                     for hdf5_attribute in hdf5_dataset.attributes:
                         if hdf5_attribute.to_dict()['key'] == "uri":
                             dataset_uri = hdf5_attribute.to_dict()["value"]
-                            # print(f"Found dataset '{name}' with URI {dataset_uri}")
-                        if hdf5_attribute.to_dict()['key'] == "sedmlDataSetLabels":
-                            column_names = hdf5_attribute.to_dict()["value"]
-                            # print(f"Found dataset '{name}' with column names {column_names}")
                     break
 
         if dataset_uri is None:
             raise ValueError(f"Dataset '{name}' not found")
         if hdf5_dataset is None:
             raise ValueError(f"Dataset '{name}' not found")
 
-        # print(f"Reading dataset '{name}' with URI '''{dataset_uri}''' and column names '''{column_names}'''")
         with (simdata_client.api_client.ApiClient(self.configuration) as api_client):
             api_instance = simdata_client.DefaultApi(api_client)
             dataset_data: DatasetData = api_instance.read_dataset(run_id=self.run_id, dataset_name=dataset_uri)
-            # print(f"Read dataset '{name}' with URI '''{dataset_uri}''' and column names {column_names} and data shape {dataset_data.shape} and shape from metadata {hdf5_dataset.shape}")
             dataset = Dataset.from_api(data=dataset_data, hdf5_dataset=hdf5_dataset)
             self.datasets[name] = dataset
             return dataset
 
     def __str__(self) -> str:
-        return f"SimResults(run_id='{self.run_id}', dataset_names={self.dataset_names()})"
+        return f"SimResults(run_id='{self.run_id}', dataset_names={self.dataset_names()})"

biosim_client/simdataclient.py

@@ -15,8 +15,8 @@ def get_health(self) -> str:
             api_response: StatusResponse = api_instance.get_health()
             return api_response.to_str()
 
-    def get_simdata(self, id: str) -> SimData:
+    def get_simdata(self, run_id: str) -> SimData:
         with simdata_client.api_client.ApiClient(self.configuration) as api_client:
             api_instance = simdata_client.DefaultApi(api_client)
-            hdf5_file: HDF5File = api_instance.get_metadata(id)
-            return SimData(configuration=self.configuration, run_id=id, hdf5_file=hdf5_file)
+            hdf5_file: HDF5File = api_instance.get_metadata(run_id)
+            return SimData(configuration=self.configuration, run_id=run_id, hdf5_file=hdf5_file)