thoth-dbmanager 0.4.7__py3-none-any.whl → 0.4.9__py3-none-any.whl

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/lsh/manager.py

@@ -98,7 +98,16 @@ class LshManager:
         """
         try:
             if not self.storage_strategy.exists(self.lsh_base_path):
-                logging.warning(f"LSH files not found at {self.lsh_base_path}")
+                # Show the actual file paths being checked for better debugging
+                lsh_file = self.lsh_base_path.with_suffix('.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/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)
@@ -72,25 +76,48 @@ 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"
-        
-        return lsh_path.exists() and minhashes_path.exists()
+
+        # 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()})")
+
+        lsh_exists = lsh_path.exists()
+        minhashes_exists = minhashes_path.exists()
+
+        if not lsh_exists or not minhashes_exists:
+            logger.warning(f"LSH files missing - LSH: {lsh_path} (exists: {lsh_exists}), Minhashes: {minhashes_path} (exists: {minhashes_exists})")
+
+        return lsh_exists and minhashes_exists

{thoth_dbmanager-0.4.7.dist-info → thoth_dbmanager-0.4.9.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: thoth_dbmanager
-Version: 0.4.7
+Version: 0.4.9
 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-0.4.7.dist-info → thoth_dbmanager-0.4.9.dist-info}/RECORD

@@ -12,7 +12,7 @@ thoth_dbmanager/adapters/sqlserver.py,sha256=V555kUH54Fb1Atow0BfvbSHmoSwGnrB_RJG
 thoth_dbmanager/adapters/supabase.py,sha256=bl2C6LpOpykPF3vIbdNRDk43aXLADzSk0wQuwTcEHZA,10348
 thoth_dbmanager/core/__init__.py,sha256=FlqNW0GZNv1rnwNgyXGzveLqaw0Z90y5AKhR_1DvHBE,269
 thoth_dbmanager/core/factory.py,sha256=sLj8tKI1ADqSlnU5F0NRtHHtpSimnATuwAqUYx5dfxI,9694
-thoth_dbmanager/core/interfaces.py,sha256=x2eHhIIQmrDEnt_0Ll3GicVWVI_rQip_x5k5sNwirFM,9280
+thoth_dbmanager/core/interfaces.py,sha256=wZpKVQJdwMlAsHTQMB7yVviD2-N_dlOe19F-GhgEoGE,9576
 thoth_dbmanager/core/registry.py,sha256=2E9btPrPVnfI4z4JauKVEkN8mp8qy5gIVm3kKxUUD5E,7960
 thoth_dbmanager/helpers/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 thoth_dbmanager/helpers/multi_db_generator.py,sha256=LOrWEqh-RTVZXPF7vrlF6BmBGh1XC09kM6XkqkFNPQ4,23232
@@ -22,8 +22,8 @@ thoth_dbmanager/helpers/search.py,sha256=k04L7clSPfeQOlq_ifsH3PZ21ZK-rujh_Qy4hWr
 thoth_dbmanager/lsh/__init__.py,sha256=zAUTRuRX4MNc2nU93tTBkT2_IVL7OBkBL0WZPPgO3Ec,555
 thoth_dbmanager/lsh/core.py,sha256=171FqHW7ItAqAPk6g_AoayKTE3Bs1rRZxnt55MJVzjY,6813
 thoth_dbmanager/lsh/factory.py,sha256=2Bpkk-OygjaptZAw1yysxO1cxG3QTxmJ1yFGcXHqX3w,2411
-thoth_dbmanager/lsh/manager.py,sha256=NDKU2re6Xp7cU6pB5EtucZqQgPiT5nyugoUCCIAG29s,5777
-thoth_dbmanager/lsh/storage.py,sha256=Wp1fjVBVE3inRFTAZCxlc6vwZdlKDwY7wCfwfu7P5Xw,3085
+thoth_dbmanager/lsh/manager.py,sha256=LGrKbGKiBuISlNXaU4Yxfc_BqJfN27MaXapJbzEAjJQ,6513
+thoth_dbmanager/lsh/storage.py,sha256=qei6fwpmRCBSS8CRtDlnZCuWEmyuOK9gVSTkEJdX0eI,4543
 thoth_dbmanager/plugins/__init__.py,sha256=98iKwNiKnFFeJfw0qLGVOrP2Mo2wGPAolsx7R2B3emA,592
 thoth_dbmanager/plugins/mariadb.py,sha256=ElYa4Rexwrofcjcs0UQKan8fZpbU6-n9zghYR9SgRb4,17975
 thoth_dbmanager/plugins/mysql.py,sha256=mbDsIDV2H_BWYANU4JHMsUkxLQICuGtjKTTPbig2Ngs,16546
@@ -32,8 +32,8 @@ thoth_dbmanager/plugins/postgresql.py,sha256=pI1W9oHpQty8tHMoEDcsOT-Msv6S4aoFcAr
 thoth_dbmanager/plugins/sqlite.py,sha256=iMlOH4XMEY-Ac3yrs1HzjKaR4km1DxLDwAEjGZzM0Cc,6504
 thoth_dbmanager/plugins/sqlserver.py,sha256=mMb3F5FmSWV02FZwj-Ult-2TjuyeVA4Fl1iME1dbgLU,5289
 thoth_dbmanager/plugins/supabase.py,sha256=mWlaGAdpywx4-pU4Ffpmn24ze8sg0sM5kc6bFDoeYRg,8645
-thoth_dbmanager-0.4.7.dist-info/licenses/LICENSE,sha256=81-BOzGgwtY1XdYfkwMQB87AkOGXI9OMq0kjNcZA4UE,1071
-thoth_dbmanager-0.4.7.dist-info/METADATA,sha256=koLm6fBoM2uMRWQSg0t9ncSN-kWbUwZw2BnCa4jYVhI,12246
-thoth_dbmanager-0.4.7.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-thoth_dbmanager-0.4.7.dist-info/top_level.txt,sha256=b9ttxm9RUc0KUCASEKRx6FqoREYJ1-KZWSpNuaM0uQ4,16
-thoth_dbmanager-0.4.7.dist-info/RECORD,,
+thoth_dbmanager-0.4.9.dist-info/licenses/LICENSE,sha256=81-BOzGgwtY1XdYfkwMQB87AkOGXI9OMq0kjNcZA4UE,1071
+thoth_dbmanager-0.4.9.dist-info/METADATA,sha256=tdNXYo8QmI9els7dgE97lrb9X1-kHacvW3cl4MI4uZo,12246
+thoth_dbmanager-0.4.9.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+thoth_dbmanager-0.4.9.dist-info/top_level.txt,sha256=b9ttxm9RUc0KUCASEKRx6FqoREYJ1-KZWSpNuaM0uQ4,16
+thoth_dbmanager-0.4.9.dist-info/RECORD,,