Rework the database migration system

This commit is contained in:
2025-09-27 21:51:31 +03:00
Unverified
parent 333c5e9ec3
commit 2c80e31a24
12 changed files with 701 additions and 167 deletions
+435 -81
View File
@@ -1,92 +1,446 @@
from __future__ import annotations
"""
Database migration utility using Alembic.
This module handles running database migrations on startup.
"""
import os
from pathlib import Path
from typing import Optional
from traceback import format_exc
import hashlib
from sqlalchemy.engine import Engine
import sys
from alembic import command
from alembic.config import Config
from models import Base
from alembic.runtime.migration import MigrationContext
from sqlalchemy import create_engine
from constants import DATABASE_URL
MIGRATIONS_DIR = Path(__file__).resolve().parent / "migrations"
LOCK_FILE = MIGRATIONS_DIR / ".autogen.lock"
SCHEMA_HASH_FILE = MIGRATIONS_DIR / ".schema.hash"
def _ensure_alembic_layout() -> None:
"""Create a minimal Alembic environment if missing."""
versions = MIGRATIONS_DIR / "versions"
versions.mkdir(parents=True, exist_ok=True)
def _alembic_config() -> Config:
cfg = Config()
cfg.set_main_option("script_location", str(MIGRATIONS_DIR))
cfg.set_main_option("sqlalchemy.url", DATABASE_URL)
# Provide a minimal ini section so env.py can read config_ini_section
cfg.config_file_name = "alembic.ini"
cfg.set_section_option("alembic", "sqlalchemy.url", DATABASE_URL)
return cfg
def _model_schema_fingerprint() -> str:
"""Compute a deterministic fingerprint of the current SQLAlchemy model schema."""
parts: list[str] = []
md = Base.metadata
for table in sorted(md.tables.values(), key=lambda t: t.name):
parts.append(f"T:{table.name}")
for col in sorted(table.columns, key=lambda c: c.name):
col_type = str(col.type)
parts.append(f"C:{col.name}:{col_type}:N{int(bool(col.nullable))}")
digest = hashlib.sha256("|".join(parts).encode("utf-8")).hexdigest()
return digest
def run_auto_migration(engine: Engine) -> None:
"""Use Alembic to autogenerate and apply migrations automatically on startup."""
# Ensure env present
_ensure_alembic_layout()
cfg = _alembic_config()
def run_migrations():
"""
Run database migrations using Alembic.
This function will upgrade the database to the latest migration.
Fully automated - handles all scenarios automatically.
"""
try:
# Upgrade existing migrations (if any) first
command.upgrade(cfg, "head")
except Exception:
print("[alembic] upgrade to head failed:\n" + format_exc())
# Get the directory where this script is located
current_dir = os.path.dirname(os.path.abspath(__file__))
# Create Alembic configuration
alembic_cfg = Config(os.path.join(current_dir, "alembic.ini"))
# Set the database URL in the config
alembic_cfg.set_main_option("sqlalchemy.url", DATABASE_URL)
# Check if any migration files exist
versions_dir = os.path.join(current_dir, "alembic", "versions")
# Always attempt autogenerate only when model schema fingerprint changed
try:
# Avoid concurrent autogenerate on dev server reloads
if not os.path.exists(versions_dir):
os.makedirs(versions_dir)
migration_files = [f for f in os.listdir(versions_dir) if f.endswith('.py') and not f.startswith('__')]
if not migration_files:
print("No migration files found. Creating initial migration...")
# Check if database exists and has tables
engine = create_engine(DATABASE_URL)
with engine.connect() as connection:
from sqlalchemy import text
result = connection.execute(text("SELECT name FROM sqlite_master WHERE type='table' AND name != 'alembic_version'"))
existing_tables = result.fetchall()
if existing_tables:
print("Found existing database with tables. Creating migration to match current schema...")
# Create migration with autogenerate to detect differences
command.revision(alembic_cfg, autogenerate=True, message="Initial migration from existing database")
# Check if the generated migration is empty (common with existing databases)
versions_dir = os.path.join(current_dir, "alembic", "versions")
migration_files = [f for f in os.listdir(versions_dir) if f.endswith('.py') and not f.startswith('__')]
if migration_files:
latest_migration = max(migration_files)
migration_path = os.path.join(versions_dir, latest_migration)
# Check if migration is empty
with open(migration_path, 'r') as f:
content = f.read()
if 'pass' in content and 'op.create_table' not in content and 'op.add_column' not in content:
print("Generated migration is empty. Creating complete schema migration...")
# Remove the empty migration
os.remove(migration_path)
# Create a complete migration
_create_complete_migration(alembic_cfg)
else:
print("No existing tables found. Creating fresh migration...")
# Create fresh migration
command.revision(alembic_cfg, autogenerate=True, message="Initial migration")
print("Initial migration created successfully.")
# Run the upgrade command
print("Running database migrations...")
command.upgrade(alembic_cfg, "head")
print("Database migrations completed successfully.")
except Exception as e:
print(f"Error running database migrations: {e}")
# Fully automated recovery - handle ALL error scenarios
print("Attempting automated recovery...")
try:
LOCK_FILE.parent.mkdir(parents=True, exist_ok=True)
fd = os.open(str(LOCK_FILE), os.O_CREAT | os.O_EXCL | os.O_RDWR)
os.close(fd)
have_lock = True
except FileExistsError:
have_lock = False
if have_lock:
try:
new_hash = _model_schema_fingerprint()
old_hash = SCHEMA_HASH_FILE.read_text(encoding="utf-8").strip() if SCHEMA_HASH_FILE.exists() else ""
if new_hash != old_hash:
command.revision(cfg, message="auto", autogenerate=True)
command.upgrade(cfg, "head")
# Update stored fingerprint
SCHEMA_HASH_FILE.write_text(new_hash, encoding="utf-8")
finally:
try:
LOCK_FILE.unlink(missing_ok=True)
except Exception:
pass
except Exception:
print("[alembic] autogenerate failed:\n" + format_exc())
# Clear the alembic_version table to reset state
engine = create_engine(DATABASE_URL)
with engine.connect() as connection:
from sqlalchemy import text
connection.execute(text("DROP TABLE IF EXISTS alembic_version"))
connection.commit()
# Remove any existing migration files to start fresh
versions_dir = os.path.join(current_dir, "alembic", "versions")
for file in os.listdir(versions_dir):
if file.endswith('.py') and not file.startswith('__'):
os.remove(os.path.join(versions_dir, file))
# Create a completely fresh migration with full schema
print("Creating fresh migration with complete schema...")
_create_complete_migration(alembic_cfg)
# Run the migration
command.upgrade(alembic_cfg, "head")
print("Automated recovery completed successfully.")
except Exception as recovery_error:
print(f"Automated recovery failed: {recovery_error}")
# Last resort: create database using SQLAlchemy directly
print("Using fallback: creating database directly...")
_create_database_directly()
print("Database created successfully using fallback method.")
def _create_complete_migration(alembic_cfg):
"""Create a complete migration file with all database schema."""
# Create a new migration file
command.revision(alembic_cfg, message="Complete schema migration")
# Get the latest migration file
versions_dir = os.path.join(os.path.dirname(os.path.abspath(__file__)), "alembic", "versions")
migration_files = [f for f in os.listdir(versions_dir) if f.endswith('.py') and not f.startswith('__')]
latest_migration = max(migration_files) if migration_files else None
if latest_migration:
migration_path = os.path.join(versions_dir, latest_migration)
_populate_migration_file(migration_path)
def _populate_migration_file(migration_path):
"""Populate a migration file with the complete database schema from models."""
# Generate the migration content dynamically from models
migration_content = _generate_migration_from_models()
# Read the current migration file
with open(migration_path, 'r') as f:
content = f.read()
# Replace the empty upgrade/downgrade functions
import re
# More flexible regex to match the actual content
content = re.sub(
r'def upgrade\(\) -> None:.*?pass.*?(?=\n\ndef downgrade|\n\nif __name__|\Z)',
migration_content,
content,
flags=re.DOTALL
)
# Write the updated content back
with open(migration_path, 'w') as f:
f.write(content)
def _generate_migration_from_models():
"""Generate migration content dynamically from SQLAlchemy models."""
from models import Base
import sqlalchemy as sa
# Generate migration content using Alembic's op functions
upgrade_statements = []
downgrade_statements = []
# Get all tables from Base metadata
for table_name, table in Base.metadata.tables.items():
if table_name != 'alembic_version': # Skip alembic_version table
# Check if table exists and compare schema
schema_diff = _detect_schema_differences(table_name, table)
if schema_diff['table_exists']:
if schema_diff['needs_update']:
# Generate ALTER TABLE statements for existing table
upgrade_statements.append(f" # Update {table_name} table schema")
for statement in schema_diff['alter_statements']:
upgrade_statements.append(f" {statement}")
else:
# Table exists and is up to date
upgrade_statements.append(f" # Table {table_name} is already up to date")
else:
# Generate CREATE TABLE for new table
table_code = _generate_table_creation_code(table_name, table)
upgrade_statements.append(f" # Create {table_name} table")
upgrade_statements.append(table_code)
# Generate DROP TABLE statement for downgrade
downgrade_statements.append(f" op.drop_table('{table_name}')")
# Combine all statements
upgrade_content = "def upgrade() -> None:\n \"\"\"Upgrade schema.\"\"\"\n" + "\n".join(upgrade_statements)
downgrade_content = "def downgrade() -> None:\n \"\"\"Downgrade schema.\"\"\"\n" + "\n".join(downgrade_statements)
return upgrade_content + "\n\n" + downgrade_content
def _detect_schema_differences(table_name, expected_table):
"""Detect differences between existing table and expected schema."""
engine = create_engine(DATABASE_URL)
with engine.connect() as connection:
from sqlalchemy import text, inspect
# Check if table exists
inspector = inspect(connection)
if table_name not in inspector.get_table_names():
return {
'table_exists': False,
'needs_update': False,
'alter_statements': []
}
# Get existing columns
existing_columns = inspector.get_columns(table_name)
existing_column_names = {col['name'] for col in existing_columns}
# Get expected columns
expected_column_names = {col.name for col in expected_table.columns}
# Check for missing columns
missing_columns = expected_column_names - existing_column_names
extra_columns = existing_column_names - expected_column_names
alter_statements = []
# Add missing columns
for column in expected_table.columns:
if column.name in missing_columns:
column_def = _generate_column_definition(column)
alter_statements.append(f"op.add_column('{table_name}', {column_def})")
# Add missing indexes
for index in expected_table.indexes:
if not index.unique:
cols = "', '".join([col.name for col in index.columns])
alter_statements.append(f"op.create_index(op.f('ix_{table_name}_{index.name}'), '{table_name}', ['{cols}'], unique=False)")
return {
'table_exists': True,
'needs_update': len(alter_statements) > 0,
'alter_statements': alter_statements
}
def _generate_column_definition(column):
"""Generate column definition for ALTER TABLE."""
type_def = _get_column_type(column)
nullable = "nullable=True" if column.nullable else "nullable=False"
definition = f"sa.Column('{column.name}', {type_def}, {nullable}"
# Handle default values properly
if column.default is not None:
if hasattr(column.default, 'arg'):
# Handle callable defaults
if callable(column.default.arg):
definition += f", default=datetime.now"
else:
definition += f", default={repr(column.default.arg)}"
else:
definition += f", default={repr(column.default)}"
definition += ")"
return definition
def _generate_table_creation_code(table_name, table):
"""Generate op.create_table code for a SQLAlchemy table."""
lines = [f" op.create_table('{table_name}',"]
# Add columns
for column in table.columns:
column_def = f" sa.Column('{column.name}', {_get_column_type(column)}, nullable={column.nullable}"
if column.default is not None:
column_def += f", default={repr(column.default)}"
column_def += ")"
lines.append(column_def)
# Add constraints
for constraint in table.constraints:
if hasattr(constraint, 'columns'):
if constraint.__class__.__name__ == 'PrimaryKeyConstraint':
lines.append(f" sa.PrimaryKeyConstraint('{constraint.columns.keys()[0]}')")
elif constraint.__class__.__name__ == 'UniqueConstraint':
cols = "', '".join(constraint.columns.keys())
lines.append(f" sa.UniqueConstraint('{cols}')")
# Add foreign key constraints
for fk in table.foreign_keys:
lines.append(f" sa.ForeignKeyConstraint(['{fk.parent.name}'], ['{fk.column.table.name}.{fk.column.name}'], )")
lines.append(" )")
# Add indexes
for index in table.indexes:
if not index.unique:
cols = "', '".join([col.name for col in index.columns])
lines.append(f" op.create_index(op.f('ix_{table_name}_{index.name}'), '{table_name}', ['{cols}'], unique=False)")
return "\n".join(lines)
def _get_column_type(column):
"""Get SQLAlchemy column type string."""
type_name = column.type.__class__.__name__
if type_name == 'String':
return f"sa.String(length={column.type.length})"
elif type_name == 'Integer':
return "sa.Integer()"
elif type_name == 'Text':
return "sa.Text()"
elif type_name == 'Boolean':
return "sa.Boolean()"
elif type_name == 'DateTime':
return "sa.DateTime()"
else:
return f"sa.{type_name}()"
def _create_database_directly():
"""Fallback method: create database directly using SQLAlchemy."""
from models import Base
from db import engine
from sqlalchemy import text, inspect
# Check existing tables and update schema
with engine.connect() as connection:
inspector = inspect(connection)
existing_tables = inspector.get_table_names()
# For each model table, check if it needs updates
for table_name, table in Base.metadata.tables.items():
if table_name != 'alembic_version':
if table_name in existing_tables:
# Table exists, check for missing columns
existing_columns = {col['name'] for col in inspector.get_columns(table_name)}
expected_columns = {col.name for col in table.columns}
missing_columns = expected_columns - existing_columns
# Add missing columns
for column in table.columns:
if column.name in missing_columns:
# Convert to raw SQL for direct execution
sql_type = _get_sql_type(column)
nullable = "NULL" if column.nullable else "NOT NULL"
# Handle datetime columns without default (SQLite limitation)
if column.type.__class__.__name__ == 'DateTime':
# Add column without default, then update existing rows
alter_sql = f"ALTER TABLE {table_name} ADD COLUMN {column.name} {sql_type} {nullable}"
try:
connection.execute(text(alter_sql))
print(f"Added column {column.name} to {table_name}")
# Update existing rows with current timestamp
update_sql = f"UPDATE {table_name} SET {column.name} = CURRENT_TIMESTAMP WHERE {column.name} IS NULL"
connection.execute(text(update_sql))
print(f"Updated {column.name} with current timestamp")
except Exception as e:
print(f"Could not add column {column.name}: {e}")
else:
# Handle other column types with defaults
default_clause = ""
if column.default is not None:
if hasattr(column.default, 'arg') and callable(column.default.arg):
# Skip callable defaults for SQLite compatibility
pass
elif hasattr(column.default, 'arg'):
default_clause = f" DEFAULT {repr(column.default.arg)}"
alter_sql = f"ALTER TABLE {table_name} ADD COLUMN {column.name} {sql_type} {nullable}{default_clause}"
try:
connection.execute(text(alter_sql))
print(f"Added column {column.name} to {table_name}")
except Exception as e:
print(f"Could not add column {column.name}: {e}")
else:
# Table doesn't exist, create it
print(f"Creating table {table_name}")
# Create alembic_version table manually
connection.execute(text("""
CREATE TABLE IF NOT EXISTS alembic_version (
version_num VARCHAR(32) NOT NULL,
CONSTRAINT alembic_version_pkc PRIMARY KEY (version_num)
)
"""))
connection.execute(text("INSERT OR IGNORE INTO alembic_version (version_num) VALUES ('direct_creation')"))
connection.commit()
def _get_sql_type(column):
"""Get SQL type for direct SQL execution."""
type_name = column.type.__class__.__name__
if type_name == 'String':
return f"VARCHAR({column.type.length})"
elif type_name == 'Integer':
return "INTEGER"
elif type_name == 'Text':
return "TEXT"
elif type_name == 'Boolean':
return "BOOLEAN"
elif type_name == 'DateTime':
return "DATETIME"
else:
return "TEXT" # fallback
def check_migration_status():
"""
Check if the database needs migrations.
Returns True if migrations are needed, False otherwise.
"""
try:
# Create engine
engine = create_engine(DATABASE_URL)
# Check if alembic_version table exists
with engine.connect() as connection:
# Check if alembic_version table exists
from sqlalchemy import text
result = connection.execute(
text("SELECT name FROM sqlite_master WHERE type='table' AND name='alembic_version'")
)
alembic_table_exists = result.fetchone() is not None
if not alembic_table_exists:
return True
# Get current migration context
context = MigrationContext.configure(connection)
current_rev = context.get_current_revision()
# Get the latest revision from alembic
current_dir = os.path.dirname(os.path.abspath(__file__))
alembic_cfg = Config(os.path.join(current_dir, "alembic.ini"))
script_dir = command.ScriptDirectory.from_config(alembic_cfg)
head_rev = script_dir.get_current_head()
return current_rev != head_rev
except Exception as e:
print(f"Error checking migration status: {e}")
return True # Assume migrations are needed if we can't check
if __name__ == "__main__":
# This allows running migrations directly
run_migrations()