synapse-sdk 2025.9.3__tar.gz → 2025.9.5__tar.gz

{synapse_sdk-2025.9.3 → synapse_sdk-2025.9.5}/.github/workflows/pypi-publish.yml

@@ -44,7 +44,7 @@ jobs:
     needs:
     - build
     environment: release
-    runs-on: ubuntu-latest
+    runs-on: runner-set
     permissions:
       # IMPORTANT: this permission is mandatory for trusted publishing
       id-token: write
@@ -54,5 +54,16 @@ jobs:
       with:
         name: python-package-distributions
         path: dist/
+    - name: Set up Python
+      uses: actions/setup-python@v5
+      with:
+        python-version: "3.x"
+    - name: Install publish dependencies
+      run: |
+        python -m pip install --upgrade pip
+        pip install twine
     - name: Publish distribution 📦 to PyPI
-      uses: pypa/gh-action-pypi-publish@release/v1
+      env:
+        TWINE_USERNAME: __token__
+        TWINE_PASSWORD: ${{ secrets.PYPI_API_TOKEN }}
+      run: python -m twine upload --non-interactive dist/*

{synapse_sdk-2025.9.3/synapse_sdk.egg-info → synapse_sdk-2025.9.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: synapse-sdk
-Version: 2025.9.3
+Version: 2025.9.5
 Summary: synapse sdk
 Author-email: datamaker <developer@datamaker.io>
 License: MIT

{synapse_sdk-2025.9.3 → synapse_sdk-2025.9.5}/synapse_sdk/plugins/categories/base.py

@@ -12,6 +12,7 @@ from synapse_sdk.plugins.enums import RunMethod
 from synapse_sdk.plugins.exceptions import ActionError
 from synapse_sdk.plugins.models import PluginRelease, Run
 from synapse_sdk.plugins.upload import archive_and_upload, build_and_upload, download_and_upload
+from synapse_sdk.shared import init_sentry, needs_sentry_init
 from synapse_sdk.utils.module_loading import import_string
 from synapse_sdk.utils.pydantic.errors import pydantic_to_drf_error
 
@@ -182,8 +183,8 @@ class Action:
             runtime_env[self.plugin_package_manager][key] = value
 
         # Sentry init if SENTRY_DSN is set
-        if self.envs.get('SENTRY_INIT_FUNCTION') is not None:
-            runtime_env['worker_process_setup_hook'] = self.envs.pop('SENTRY_INIT_FUNCTION')
+        if needs_sentry_init():
+            runtime_env['worker_process_setup_hook'] = 'synapse_sdk.shared'
 
         # 맨 마지막에 진행되어야 함
         runtime_env['env_vars'] = self.envs
@@ -312,27 +313,10 @@ class Action:
 
         return JobSubmissionClient(address=self.envs.get('RAY_DASHBOARD_URL'))
 
-    def _sentry_init(self, sentry_dsn):
-        import sentry_sdk
-        from sentry_sdk.integrations.ray import RayIntegration
-
-        sentry_sdk.init(
-            dsn=sentry_dsn,
-            environment=os.getenv('DEPLOYMENT_TARGET', 'development'),
-            integrations=[RayIntegration()],
-            send_default_pii=True,
-        )
-
-    def sentry_init(self):
-        sentry_dsn = os.getenv('SENTRY_DSN', self.envs.get('SENTRY_DSN') if self.envs else None)
-        if sentry_dsn:
-            self._sentry_init(sentry_dsn)
-            self.envs['SENTRY_INIT_FUNCTION'] = self._sentry_init
-
     def ray_init(self):
         import ray
 
-        self.sentry_init()
+        init_sentry()
 
         if not ray.is_initialized():
             ray.init(address=self.envs['RAY_ADDRESS'], ignore_reinit_error=True)

{synapse_sdk-2025.9.3 → synapse_sdk-2025.9.5}/synapse_sdk/plugins/categories/upload/actions/upload/strategies/file_discovery/recursive.py

@@ -52,26 +52,40 @@ class RecursiveFileDiscoveryStrategy(FileDiscoveryStrategy):
             for spec_name, dir_path in type_dirs.items():
                 dir_path_str = path_cache[dir_path]
                 if file_path_str.startswith(dir_path_str):
-                    file_name = file_path.stem
+                    # Create unique dataset key using relative path from spec directory
+                    relative_path = file_path.relative_to(dir_path)
+                    # Use parent directory + stem as unique key to group related files
+                    if relative_path.parent != Path('.'):
+                        dataset_key = f'{relative_path.parent}_{file_path.stem}'
+                    else:
+                        dataset_key = file_path.stem
 
-                    if file_name not in dataset_files:
-                        dataset_files[file_name] = {}
+                    if dataset_key not in dataset_files:
+                        dataset_files[dataset_key] = {}
 
-                    if spec_name not in dataset_files[file_name]:
-                        dataset_files[file_name][spec_name] = file_path
+                    if spec_name not in dataset_files[dataset_key]:
+                        dataset_files[dataset_key][spec_name] = file_path
                     else:
                         # Keep the most recent file - only stat when needed
-                        existing_file = dataset_files[file_name][spec_name]
+                        existing_file = dataset_files[dataset_key][spec_name]
                         try:
                             if file_path.stat().st_mtime > existing_file.stat().st_mtime:
-                                dataset_files[file_name][spec_name] = file_path
+                                dataset_files[dataset_key][spec_name] = file_path
                         except (OSError, IOError):
                             # If stat fails, keep existing file
                             pass
 
         # Create organized files for datasets with all required files
-        for file_name, files_dict in sorted(dataset_files.items()):
+        for dataset_key, files_dict in sorted(dataset_files.items()):
             if all(req in files_dict for req in required_specs):
+                # Extract original file stem from dataset_key
+                # If dataset_key contains path info (parent_stem), extract just the stem part
+                if '_' in dataset_key and len(dataset_key.split('_')) >= 2:
+                    # Extract the last part after the last underscore as the original stem
+                    original_stem = dataset_key.split('_')[-1]
+                else:
+                    original_stem = dataset_key
+
                 # Calculate most common file extension (optimized)
                 file_extensions = {}
                 for file_path in files_dict.values():
@@ -82,14 +96,15 @@ class RecursiveFileDiscoveryStrategy(FileDiscoveryStrategy):
                 origin_file_extension = max(file_extensions, key=file_extensions.get) if file_extensions else ''
 
                 meta_data = {
-                    'origin_file_stem': file_name,
+                    'origin_file_stem': original_stem,
                     'origin_file_extension': origin_file_extension,
                     'created_at': datetime.now().isoformat(),
+                    'dataset_key': dataset_key,  # Add dataset key for debugging
                 }
 
                 # Add metadata if available - using optimized index lookup
                 if metadata_index:
-                    matched_metadata = self._find_matching_metadata_optimized(file_name, files_dict, metadata_index)
+                    matched_metadata = self._find_matching_metadata_optimized(original_stem, files_dict, metadata_index)
                     if matched_metadata:
                         meta_data.update(matched_metadata)
 

{synapse_sdk-2025.9.3 → synapse_sdk-2025.9.5}/synapse_sdk/plugins/models.py

@@ -6,6 +6,7 @@ from synapse_sdk.clients.backend import BackendClient
 from synapse_sdk.devtools.config import get_backend_config
 from synapse_sdk.loggers import BackendLogger, ConsoleLogger
 from synapse_sdk.plugins.utils import read_plugin_config
+from synapse_sdk.shared import needs_sentry_init
 from synapse_sdk.shared.enums import Context
 from synapse_sdk.utils.storage import get_storage
 from synapse_sdk.utils.string import hash_text
@@ -84,8 +85,8 @@ class PluginRelease:
 
         extra_runtime_env = {}
 
-        if self.envs.get('SENTRY_INIT_FUNCTION') is not None:
-            extra_runtime_env['worker_process_setup_hook'] = self.envs['SENTRY_INIT_FUNCTION']
+        if needs_sentry_init():
+            extra_runtime_env['worker_process_setup_hook'] = 'synapse_sdk.shared'
 
         nodes = list_nodes(address=self.envs['RAY_DASHBOARD_URL'])
         node_ids = [n['node_id'] for n in nodes]
@@ -99,6 +100,7 @@ class PluginRelease:
                         ** self.package_manager_options
                     },
                     'working_dir': self.get_url(self.envs['SYNAPSE_PLUGIN_STORAGE']),
+                    **extra_runtime_env,
                 },
                 scheduling_strategy=strategy,
             ).remote()

synapse_sdk-2025.9.5/synapse_sdk/shared/__init__.py

@@ -0,0 +1,25 @@
+import os
+
+
+def needs_sentry_init():
+    return os.getenv('SENTRY_DSN') is not None
+
+
+def init_sentry():
+    import sentry_sdk
+    from sentry_sdk.integrations.ray import RayIntegration
+
+    dsn = os.getenv('SENTRY_DSN')
+    if dsn is None:
+        return
+
+    sentry_sdk.init(
+        dsn=dsn,
+        environment=os.getenv('DEPLOYMENT_TARGET', 'development'),
+        integrations=[RayIntegration()],
+        send_default_pii=True,
+    )
+
+
+def worker_process_setup_hook(*_, **__):
+    init_sentry()

{synapse_sdk-2025.9.3 → synapse_sdk-2025.9.5/synapse_sdk.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: synapse-sdk
-Version: 2025.9.3
+Version: 2025.9.5
 Summary: synapse sdk
 Author-email: datamaker <developer@datamaker.io>
 License: MIT