synapse-sdk 1.0.0a50__tar.gz → 1.0.0a53__tar.gz

{synapse_sdk-1.0.0a50 → synapse_sdk-1.0.0a53}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: synapse-sdk
-Version: 1.0.0a50
+Version: 1.0.0a53
 Summary: synapse sdk
 Author-email: datamaker <developer@datamaker.io>
 License: MIT

{synapse_sdk-1.0.0a50 → synapse_sdk-1.0.0a53}/synapse_sdk/clients/backend/__init__.py

@@ -1,6 +1,6 @@
 from synapse_sdk.clients.backend.annotation import AnnotationClientMixin
 from synapse_sdk.clients.backend.core import CoreClientMixin
-from synapse_sdk.clients.backend.dataset import DatasetClientMixin
+from synapse_sdk.clients.backend.data_collection import DataCollectionClientMixin
 from synapse_sdk.clients.backend.hitl import HITLClientMixin
 from synapse_sdk.clients.backend.integration import IntegrationClientMixin
 from synapse_sdk.clients.backend.ml import MLClientMixin
@@ -9,7 +9,7 @@ from synapse_sdk.clients.backend.ml import MLClientMixin
 class BackendClient(
     AnnotationClientMixin,
     CoreClientMixin,
-    DatasetClientMixin,
+    DataCollectionClientMixin,
     IntegrationClientMixin,
     MLClientMixin,
     HITLClientMixin,

synapse_sdk-1.0.0a50/synapse_sdk/clients/backend/dataset.py → synapse_sdk-1.0.0a53/synapse_sdk/clients/backend/data_collection.py

@@ -8,18 +8,18 @@ from synapse_sdk.clients.base import BaseClient
 from synapse_sdk.clients.utils import get_batched_list
 
 
-class DatasetClientMixin(BaseClient):
-    def list_dataset(self):
-        path = 'datasets/'
+class DataCollectionClientMixin(BaseClient):
+    def list_data_collection(self):
+        path = 'data_collections/'
         return self._list(path)
 
-    def get_dataset(self, dataset_id):
-        """Get dataset from synapse-backend.
+    def get_data_collection(self, data_collection_id):
+        """Get data_collection from synapse-backend.
 
         Args:
-            dataset_id: The dataset id to get.
+            data_collection_id: The data_collection id to get.
         """
-        path = f'datasets/{dataset_id}/?expand=file_specifications'
+        path = f'data_collections/{data_collection_id}/?expand=file_specifications'
         return self._get(path)
 
     def create_data_file(self, file_path: Path):
@@ -40,19 +40,19 @@ class DatasetClientMixin(BaseClient):
         path = 'data_units/'
         return self._post(path, data=data)
 
-    def upload_dataset(
+    def upload_data_collection(
         self,
-        dataset_id: int,
-        dataset: Dict,
+        data_collection_id: int,
+        data_collection: Dict,
         project_id: Optional[int] = None,
         batch_size: int = 1000,
         process_pool: int = 10,
     ):
-        """Upload dataset to synapse-backend.
+        """Upload data_collection to synapse-backend.
 
         Args:
-            dataset_id: The dataset id to upload the data to.
-            dataset: The dataset to upload.
+            data_collection_id: The data_collection id to upload the data to.
+            data_collection: The data_collection to upload.
                 * structure:
                     - files: The files to upload. (key: file name, value: file pathlib object)
                     - meta: The meta data to upload.
@@ -60,14 +60,14 @@ class DatasetClientMixin(BaseClient):
             batch_size: The batch size to upload the data.
             process_pool: The process pool to upload the data.
         """
-        # TODO validate dataset with schema
+        # TODO validate data_collection with schema
 
-        params = [(data, dataset_id) for data in dataset]
+        params = [(data, data_collection_id) for data in data_collection]
 
         with Pool(processes=process_pool) as pool:
-            dataset = pool.starmap(self.upload_data_file, tqdm(params))
+            data_collection = pool.starmap(self.upload_data_file, tqdm(params))
 
-        batches = get_batched_list(dataset, batch_size)
+        batches = get_batched_list(data_collection, batch_size)
 
         for batch in tqdm(batches):
             data_units = self.create_data_units(batch)
@@ -82,7 +82,7 @@ class DatasetClientMixin(BaseClient):
 
                 self.create_tasks(tasks_data)
 
-    def upload_data_file(self, data: Dict, dataset_id: int) -> Dict:
+    def upload_data_file(self, data: Dict, data_collection_id: int) -> Dict:
         """Upload files to synapse-backend.
 
         Args:
@@ -90,13 +90,13 @@ class DatasetClientMixin(BaseClient):
                 * structure:
                     - files: The files to upload. (key: file name, value: file pathlib object)
                     - meta: The meta data to upload.
-            dataset_id: The dataset id to upload the data to.
+            data_collection_id: The data_collection id to upload the data to.
 
         Returns:
             Dict: The result of the upload.
         """
         for name, path in data['files'].items():
             data_file = self.create_data_file(path)
-            data['dataset'] = dataset_id
+            data['data_collection'] = data_collection_id
             data['files'][name] = {'checksum': data_file['checksum'], 'path': str(path)}
         return data

{synapse_sdk-1.0.0a50 → synapse_sdk-1.0.0a53}/synapse_sdk/plugins/categories/upload/actions/task_pre_annotation.py

@@ -5,6 +5,7 @@ from pydantic import AfterValidator, BaseModel, field_validator
 from pydantic_core import PydanticCustomError
 
 from synapse_sdk.clients.exceptions import ClientError
+from synapse_sdk.plugins.categories.base import Action
 from synapse_sdk.plugins.categories.decorators import register_action
 from synapse_sdk.plugins.enums import PluginCategory, RunMethod
 from synapse_sdk.plugins.models import Run
@@ -44,7 +45,7 @@ class TaskPreAnnotationParams(BaseModel):
         action = info.context['action']
         client = action.client
         try:
-            client.get_dataset(value)
+            client.get_data_collection(value)
         except ClientError:
             raise PydanticCustomError('client_error', 'Error occurred while checking data collection exists.')
         return value
@@ -66,7 +67,7 @@ class TaskPreAnnotationParams(BaseModel):
 
 
 @register_action
-class TaskPreAnnotationAction(TaskPreAnnotationRun):
+class TaskPreAnnotationAction(Action):
     """TaskPreAnnotation action class.
 
     * Annotate data to tasks.

{synapse_sdk-1.0.0a50 → synapse_sdk-1.0.0a53}/synapse_sdk/plugins/categories/upload/actions/upload.py

@@ -129,7 +129,7 @@ class UploadParams(BaseModel):
         action = info.context['action']
         client = action.client
         try:
-            client.get_dataset(value)
+            client.get_data_collection(value)
         except ClientError:
             raise PydanticCustomError('client_error', _('Error occurred while checking collection exists.'))
         return value
@@ -244,7 +244,7 @@ class UploadAction(Action):
 
         client = self.run.client
         collection_id = self.params['collection']
-        collection = client.get_dataset(collection_id)
+        collection = client.get_data_collection(collection_id)
 
         # Finish progress
         self.run.set_progress(1, 1, category='analyze_collection')

{synapse_sdk-1.0.0a50 → synapse_sdk-1.0.0a53}/synapse_sdk.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: synapse-sdk
-Version: 1.0.0a50
+Version: 1.0.0a53
 Summary: synapse sdk
 Author-email: datamaker <developer@datamaker.io>
 License: MIT

{synapse_sdk-1.0.0a50 → synapse_sdk-1.0.0a53}/synapse_sdk.egg-info/SOURCES.txt

@@ -48,7 +48,7 @@ synapse_sdk/clients/agent/service.py
 synapse_sdk/clients/backend/__init__.py
 synapse_sdk/clients/backend/annotation.py
 synapse_sdk/clients/backend/core.py
-synapse_sdk/clients/backend/dataset.py
+synapse_sdk/clients/backend/data_collection.py
 synapse_sdk/clients/backend/hitl.py
 synapse_sdk/clients/backend/integration.py
 synapse_sdk/clients/backend/ml.py