thoth-dbmanager 0.4.8__tar.gz → 0.4.10__tar.gz

{thoth_dbmanager-0.4.8/thoth_dbmanager.egg-info → thoth_dbmanager-0.4.10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: thoth_dbmanager
-Version: 0.4.8
+Version: 0.4.10
 Summary: A Python library for managing SQL databases with support for multiple database types, LSH-based similarity search, and a modern plugin architecture.
 Author-email: Marco Pancotti <mp@tylconsulting.it>
 Project-URL: Homepage, https://github.com/mptyl/thoth_dbmanager

{thoth_dbmanager-0.4.8 → thoth_dbmanager-0.4.10}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "thoth_dbmanager"
-version = "0.4.8"
+version = "0.4.10"
 authors = [
   { name="Marco Pancotti", email="mp@tylconsulting.it" },
 ]

{thoth_dbmanager-0.4.8 → thoth_dbmanager-0.4.10}/thoth_dbmanager/core/interfaces.py

@@ -143,7 +143,7 @@ class DbPlugin(ABC):
     def __init__(self, db_root_path: str, db_mode: str = "dev", **kwargs):
         """
         Initialize the database plugin.
-        
+
         Args:
             db_root_path: Path to the database root directory
             db_mode: Database mode (dev, prod, etc.)
@@ -153,6 +153,16 @@ class DbPlugin(ABC):
         self.db_mode = db_mode
         self.adapter: Optional[DbAdapter] = None
         self._initialized = False
+
+    @property
+    def db_type(self) -> str:
+        """
+        Get the primary database type for this plugin.
+
+        Returns:
+            The first supported database type, or "unknown" if none specified
+        """
+        return self.supported_db_types[0] if self.supported_db_types else "unknown"
     
     @abstractmethod
     def create_adapter(self, **kwargs) -> DbAdapter:

{thoth_dbmanager-0.4.8 → thoth_dbmanager-0.4.10}/thoth_dbmanager/lsh/manager.py

@@ -100,10 +100,14 @@ class LshManager:
             if not self.storage_strategy.exists(self.lsh_base_path):
                 # Show the actual file paths being checked for better debugging
                 lsh_file = self.lsh_base_path.with_suffix('.pkl')
-                minhashes_file = self.lsh_base_path.parent / f"{self.lsh_base_path.stem}_minhashes.pkl"
+                # Use the old naming convention for minhashes file
+                db_id = self.lsh_base_path.stem.replace('_lsh', '') if self.lsh_base_path.stem.endswith('_lsh') else self.lsh_base_path.stem
+                minhashes_file = self.lsh_base_path.parent / f"{db_id}_minhashes.pkl"
+                unique_values_file = self.lsh_base_path.parent / f"{db_id}_unique_values.pkl"
                 logging.warning(f"LSH files not found. Expected files:")
                 logging.warning(f"  LSH file: {lsh_file}")
                 logging.warning(f"  Minhashes file: {minhashes_file}")
+                logging.warning(f"  Unique values file: {unique_values_file}")
                 return False
             
             lsh_data, minhashes_data = self.storage_strategy.load(self.lsh_base_path)

{thoth_dbmanager-0.4.8 → thoth_dbmanager-0.4.10}/thoth_dbmanager/lsh/storage.py

@@ -56,15 +56,19 @@ class PickleStorage(LshStorageStrategy):
     def save(self, lsh_data: Any, minhashes_data: Any, base_path: Path) -> None:
         """Save LSH data using pickle format."""
         lsh_path = base_path.with_suffix('.pkl')
-        minhashes_path = base_path.parent / f"{base_path.stem}_minhashes.pkl"
-        
+
+        # Use the old naming convention: {db_id}_minhashes.pkl instead of {db_id}_lsh_minhashes.pkl
+        # Extract db_id from base_path stem (remove _lsh suffix if present)
+        db_id = base_path.stem.replace('_lsh', '') if base_path.stem.endswith('_lsh') else base_path.stem
+        minhashes_path = base_path.parent / f"{db_id}_minhashes.pkl"
+
         # Ensure directory exists
         base_path.parent.mkdir(parents=True, exist_ok=True)
-        
+
         # Save LSH data
         with open(lsh_path, 'wb') as f:
             pickle.dump(lsh_data, f)
-            
+
         # Save minhashes data
         with open(minhashes_path, 'wb') as f:
             pickle.dump(minhashes_data, f)
@@ -72,32 +76,41 @@ class PickleStorage(LshStorageStrategy):
     def load(self, base_path: Path) -> Tuple[Optional[Any], Optional[Any]]:
         """Load LSH data from pickle files."""
         lsh_path = base_path.with_suffix('.pkl')
-        minhashes_path = base_path.parent / f"{base_path.stem}_minhashes.pkl"
-        
+
+        # Use the old naming convention: {db_id}_minhashes.pkl instead of {db_id}_lsh_minhashes.pkl
+        # Extract db_id from base_path stem (remove _lsh suffix if present)
+        db_id = base_path.stem.replace('_lsh', '') if base_path.stem.endswith('_lsh') else base_path.stem
+        minhashes_path = base_path.parent / f"{db_id}_minhashes.pkl"
+
         try:
             # Load LSH data
             with open(lsh_path, 'rb') as f:
                 lsh_data = pickle.load(f)
-                
+
             # Load minhashes data
             with open(minhashes_path, 'rb') as f:
                 minhashes_data = pickle.load(f)
-                
+
             return lsh_data, minhashes_data
-            
+
         except (FileNotFoundError, pickle.PickleError):
             return None, None
     
     def exists(self, base_path: Path) -> bool:
         """Check if both LSH and minhashes pickle files exist."""
         lsh_path = base_path.with_suffix('.pkl')
-        minhashes_path = base_path.parent / f"{base_path.stem}_minhashes.pkl"
+
+        # Use the old naming convention: {db_id}_minhashes.pkl instead of {db_id}_lsh_minhashes.pkl
+        # Extract db_id from base_path stem (remove _lsh suffix if present)
+        db_id = base_path.stem.replace('_lsh', '') if base_path.stem.endswith('_lsh') else base_path.stem
+        minhashes_path = base_path.parent / f"{db_id}_minhashes.pkl"
 
         # Add debug logging to help diagnose path issues
         import logging
         logger = logging.getLogger(__name__)
         logger.debug(f"Checking LSH files existence:")
         logger.debug(f"  Base path: {base_path}")
+        logger.debug(f"  DB ID: {db_id}")
         logger.debug(f"  LSH path: {lsh_path} (exists: {lsh_path.exists()})")
         logger.debug(f"  Minhashes path: {minhashes_path} (exists: {minhashes_path.exists()})")
 

{thoth_dbmanager-0.4.8 → thoth_dbmanager-0.4.10}/thoth_dbmanager/plugins/sqlite.py

@@ -75,7 +75,7 @@ class SQLitePlugin(DbPlugin):
         # Handle database path resolution
         database_path = kwargs.get('database_path')
         database_name = kwargs.get('database_name')
-        
+
         if not database_path and database_name:
             # Create database path from name and root path
             db_root = Path(self.db_root_path)
@@ -83,22 +83,23 @@ class SQLitePlugin(DbPlugin):
             db_dir.mkdir(parents=True, exist_ok=True)
             database_path = str(db_dir / f"{database_name}.db")
             kwargs['database_path'] = database_path
-        
-        # Set database path for adapter
+
+        # Set database path for adapter - ensure we use the provided database_path
         self.database_path = database_path
-        
-        # Initialize with updated kwargs
-        super().initialize(**kwargs)
-        
-        # Set up database directory path and ID
+
+        # Set up database directory path and ID before calling super().initialize()
         if database_name:
             self.db_id = database_name
         else:
             # Extract database name from path
-            self.db_id = Path(database_path).stem
-        
-        self._setup_directory_path(self.db_id)
-        
+            self.db_id = Path(database_path).stem if database_path else None
+
+        if self.db_id:
+            self._setup_directory_path(self.db_id)
+
+        # Initialize with updated kwargs - this will create and connect the adapter
+        super().initialize(**kwargs)
+
         logger.info(f"SQLite plugin initialized for database: {self.db_id} at {self.database_path}")
     
     def _setup_directory_path(self, db_id: str) -> None:

{thoth_dbmanager-0.4.8 → thoth_dbmanager-0.4.10/thoth_dbmanager.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: thoth_dbmanager
-Version: 0.4.8
+Version: 0.4.10
 Summary: A Python library for managing SQL databases with support for multiple database types, LSH-based similarity search, and a modern plugin architecture.
 Author-email: Marco Pancotti <mp@tylconsulting.it>
 Project-URL: Homepage, https://github.com/mptyl/thoth_dbmanager