Update version

Author: ngrislain

qrlew_datasets/databases/mariadb.py

@@ -18,7 +18,7 @@ def __init__(self, name=NAME, user=USER, password=PASSWORD, port=PORT) -> None:
         self.port = port
         self.engine()
 
-    def try_get_existing(self) -> Optional[Engine]:
+    def existing_engine(self) -> Optional[Engine]:
         """Try to connect to mariadb"""
         if subprocess.run(['docker', 'exec', self.name, 'mariadb-admin',
                               f'--password={self.password}',
@@ -34,7 +34,7 @@ def try_get_existing(self) -> Optional[Engine]:
         else:
             return None
 
-    def try_get_container(self) -> Optional[Engine]:
+    def container_engine(self) -> Optional[Engine]:
         """Try to start or run a mariadb container"""
         # Try to start an existing container
         if subprocess.run(['docker', 'start', self.name]).returncode != 0:
@@ -57,25 +57,25 @@ def try_get_container(self) -> Optional[Engine]:
             print("Waiting mariadb to be ready...")
             sleep(1)
             attempts += 1
-        return self.try_get_existing()
+        return self.existing_engine()
 
     # Implements engine
     def engine(self) -> Engine:
         """Create a mariadb engine"""
-        engine = self.try_get_existing()
+        engine = self.existing_engine()
         if engine is None:
-            engine = self.try_get_container()
+            engine = self.container_engine()
         assert(engine is not None)
         return engine
 
-    # Dump psql files
+    # Dump mariadb files
     def dump(self, path: str) -> None:
-        """Dump psql"""
+        """Dump mariadb"""
         self._dump('localhost', self.user, self.password, self.schema(), path)
 
-    # Dump psql files
+    # Dump mariadb files
     def _dump(self, host: str, user: str, password: str, db: str, path: str) -> None:
-        """Dump psql"""
+        """mariadb-dump"""
         subprocess.run(['docker', 'exec', self.name, 'mariadb-dump',
                         f'--host={host}',
                         f'--user={user}',

qrlew_datasets/file.py renamed to qrlew_datasets/files.py

@@ -0,0 +1,55 @@
+import subprocess
+from sqlalchemy import create_engine
+from sqlalchemy.engine import Engine
+from qrlew_datasets.database import Database
+from qrlew_datasets.network import Network
+
+class RelationalDatasetRepository(Database):
+    '''Sources from relational.fit.cvut.cz'''
+    def __init__(self, name: str) -> None:
+        super().__init__()
+        self._schema = name
+
+    def engine(self) -> Engine:
+        return create_engine(f'mysql+pymysql://guest:[email protected]:3306/{self.schema()}')
+    
+    def url(self) -> str:
+        return f'mysql://guest:[email protected]:3306/{self.schema()}'
+    
+    def schema(self) -> str:
+        return self._schema
+
+class Financial(RelationalDatasetRepository):
+    def __init__(self) -> None:
+        super().__init__('financial')
+
+class Hepatitis(RelationalDatasetRepository):
+    def __init__(self) -> None:
+        super().__init__('Hepatitis_std')
+
+class IMDB(RelationalDatasetRepository):
+    def __init__(self) -> None:
+        super().__init__('imdb_ijs')
+
+NAME: str = "qrlew-pgloader"
+
+class Loader:
+    '''Load different sources to destination'''
+    def __init__(self, destination: Database) -> None:
+        self.destination = destination
+        self.net = Network().name
+
+    def load(self, source: Database) -> bool:
+        """Try to run pgloader to load a DB into the target DB"""
+        # Try to start an existing container
+        subprocess.run([
+                'docker',
+                'run',
+                '--rm',
+                '-it',
+                '--net', self.net,
+                'dimitri/pgloader:latest',
+                'pgloader',
+                 source.url(),
+                 self.destination.url()])
+        return True

qrlew_datasets/sources.py

@@ -1,6 +1,6 @@
 import os.path
 import importlib.resources as pkg_resources
-from qrlew_datasets.file import SQL, CSV
+from qrlew_datasets.files import SQL, CSV
 
 def test_files():
     for sql_source in ['extract', 'financial', 'hepatitis', 'imdb', 'retail']: