PM-1322 Idempotent create database

README.md

@@ -89,7 +89,7 @@ Create a database and relevant tables before first-time program execution using
    invoke create-database
    ```
 
-   > :warning: This script drops all existing tables and creates new ones.
+   > **Note:** The script creates `POSTGRES_DB` if not exists and applies `./database/create_tables.sql` script.
 
 2. **Initialize `bot_logs` table**:
 
@@ -106,6 +106,12 @@ Create a database and relevant tables before first-time program execution using
    invoke init-database --mins-ago 20
    ```
 
+It is possible to drop database using:
+
+```
+invoke drop-database
+```
+
 ### AWS Keyspaces/Cassandra Configuration
 
 To connect to AWS Keyspaces/Cassandra, the following environment variables need to be set:

tasks.py

@@ -23,9 +23,9 @@ def create_database(ctx):
     # Creating the database
     try:
         cursor.execute(sql.SQL("CREATE DATABASE {}").format(sql.Identifier(db_name)))
-        print(f"Database {db_name} created successfully")
+        print(f"Database '{db_name}' created successfully")
     except psycopg2.errors.DuplicateDatabase:
-        print(f"Database {db_name} already exists")
+        print(f"Database '{db_name}' already exists, not creating")
 
     cursor.close()
     conn.close()
@@ -38,13 +38,13 @@ def create_database(ctx):
     cursor = conn.cursor()
 
     # Path to the SQL script relative to tasks.py
-    sql_script_path = "uptime_service_validation/database/createDB.sql"
+    sql_script_path = "uptime_service_validation/database/create_tables.sql"
 
     # Running the SQL script file
     with open(sql_script_path, "r") as file:
         sql_script = file.read()
         cursor.execute(sql_script)
-        print("Database setup completed successfully")
+        print("'create_tables.sql' script completed successfully")
 
     cursor.close()
     conn.close()
@@ -106,3 +106,29 @@ def init_database(ctx, batch_end_epoch=None, mins_ago=None, override_empty=False
     conn.commit()
     cursor.close()
     conn.close()
+
+
+@task
+def drop_database(ctx):
+    db_host = os.environ.get("POSTGRES_HOST")
+    db_port = os.environ.get("POSTGRES_PORT")
+    db_name = os.environ.get("POSTGRES_DB")
+    db_user = os.environ.get("POSTGRES_USER")
+    db_password = os.environ.get("POSTGRES_PASSWORD")
+
+    # Establishing connection to PostgreSQL server
+    conn = psycopg2.connect(
+        host=db_host, port=db_port, user=db_user, password=db_password
+    )
+    conn.autocommit = True
+    cursor = conn.cursor()
+
+    # Dropping the database
+    try:
+        cursor.execute(sql.SQL("DROP DATABASE {}").format(sql.Identifier(db_name)))
+        print(f"Database '{db_name}' dropped!")
+    except Exception as e:
+        print(f"Error dropping database '{db_name}'! Error: {e}")
+
+    cursor.close()
+    conn.close()

uptime_service_validation/database/createDB.sql → uptime_service_validation/database/create_tables.sql

@@ -1,6 +1,4 @@
---Need to check on this structure. Feels like some of these should be non-null.
-DROP TABLE IF EXISTS bot_logs CASCADE;
-CREATE TABLE bot_logs (
+CREATE TABLE IF NOT EXISTS bot_logs (
 	id SERIAL PRIMARY KEY, 
 	processing_time DOUBLE PRECISION, 
 	files_processed INT, 
@@ -9,15 +7,13 @@ CREATE TABLE bot_logs (
 	batch_end_epoch BIGINT
 );
 
-DROP TABLE IF EXISTS statehash CASCADE;
-CREATE TABLE statehash (
+CREATE TABLE IF NOT EXISTS statehash (
 	id SERIAL PRIMARY KEY,
 	value TEXT
 );
-CREATE UNIQUE INDEX uq_state_hash ON statehash USING btree (value);
+CREATE UNIQUE INDEX IF NOT EXISTS uq_state_hash ON statehash USING btree (value);
 
-DROP TABLE IF EXISTS bot_logs_statehash CASCADE;
-CREATE TABLE bot_logs_statehash (
+CREATE TABLE IF NOT EXISTS bot_logs_statehash (
 	id SERIAL PRIMARY KEY,
 	parent_statehash_id INT, 
 	statehash_id INT, 
@@ -34,8 +30,7 @@ CREATE TABLE bot_logs_statehash (
 		REFERENCES bot_logs(id)
 );
 
-DROP TABLE IF EXISTS nodes CASCADE;
-CREATE TABLE nodes (
+CREATE TABLE IF NOT EXISTS nodes (
 	id SERIAL PRIMARY KEY,
 	block_producer_key TEXT,
 	updated_at TIMESTAMPTZ(6),
@@ -46,8 +41,7 @@ CREATE TABLE nodes (
 	application_status BOOLEAN
 );
 
-DROP TABLE IF EXISTS points CASCADE;
-CREATE TABLE points (
+CREATE TABLE IF NOT EXISTS points (
 	id SERIAL PRIMARY KEY,
 	file_name TEXT,
 	file_timestamps TIMESTAMPTZ(6), 
@@ -69,8 +63,7 @@ CREATE TABLE points (
 		REFERENCES statehash(id)
 );
 
-DROP TABLE IF EXISTS score_history CASCADE;
-CREATE TABLE score_history (
+CREATE TABLE IF NOT EXISTS score_history (
 	id SERIAL PRIMARY KEY,
 	node_id INT,
 	score_at TIMESTAMP(6), 
@@ -80,6 +73,4 @@ CREATE TABLE score_history (
 		FOREIGN KEY(node_id) 
 		REFERENCES nodes(id)
 );
-CREATE UNIQUE INDEX uq_sh_node_score_at ON score_history USING btree (node_id, score_at);
-
--- Point Summary table that is auto-gen?
+CREATE UNIQUE INDEX IF NOT EXISTS uq_sh_node_score_at ON score_history USING btree (node_id, score_at);