Pump a table from one database to another

pump_plugin.py

# MIT License

# Copyright (c) 2019 Bellhops Inc.

# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:

# The above copyright notice and this permission notice shall be included in all
# copies or substantial portions of the Software.

# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.

import os
import datetime
import logging
import sys
import botocore

from sqlalchemy import MetaData
from sqlalchemy import Table
from geoalchemy2 import Table as GeoTable
from geoalchemy2.types import Geography
from sqlalchemy import Column, NVARCHAR, NUMERIC
from sqlalchemy.schema import CreateTable
from sqlalchemy.sql.sqltypes import TEXT, NullType, ARRAY
from sqlalchemy.dialects import postgresql
from sqlalchemy.dialects.postgresql.base import DOUBLE_PRECISION

from airflow.hooks.postgres_hook import PostgresHook
from airflow.hooks.S3_hook import S3Hook
from airflow.models import Variable
from airflow.operators.subdag_operator import SubDagOperator
from airflow.operators.python_operator import PythonOperator

from airflow import utils as airflow_utils


class Pump(object):

    def __init__(self, table_name, origin_conn_id, origin_schema_name,
                 destination_conn_id, destination_schema_name, destination_db_type, s3_conn_id,
                 s3_bucket, s3_directory, drop_destination_table=False):
        self.table_name = table_name
        self.origin_hook = PostgresHook(postgres_conn_id=origin_conn_id)
        self.origin_conn_uri = self.origin_hook.get_uri()
        self.origin_schema_name = origin_schema_name
        self.destination_hook = PostgresHook(postgres_conn_id=destination_conn_id)
        self.destination_schema_name = destination_schema_name
        self.destination_db_type = destination_db_type
        self.s3_hook = S3Hook(aws_conn_id=s3_conn_id)
        self.drop_destination_table = drop_destination_table
        self.s3_bucket = s3_bucket
        self.s3_directory = s3_directory

        self.origin_engine = self.origin_hook.get_sqlalchemy_engine()
        self.origin_metadata = MetaData(schema=self.origin_schema_name)
        self.origin_table_name_with_schema = '.'.join([self.origin_schema_name, self.table_name])
        self.destination_table_name_with_schema = '.'.join([self.destination_schema_name, self.table_name])

        self.bucket_name = self.s3_bucket
        self.directory_name = self.s3_directory
        self.s3_file_name = '''{directory_name}/{table_name}'''.format(directory_name=self.directory_name,
                                                                              table_name=table_name) + '_{:%Y-%m-%dT%H%M}'.format(datetime.datetime.now()) + '.csv'
        self.destination_conn_uri = self.destination_hook.get_uri()
        self.destination_engine = self.destination_hook.get_sqlalchemy_engine()
        self.aws_key, self.aws_pass, _, _ = self.s3_hook._get_credentials(None)

        self.local_file = None

    @property
    def origin_has_table(self):
        return self.origin_engine.has_table(self.table_name, schema=self.origin_schema_name)

    @property
    def destination_has_table(self):
        return self.destination_engine.has_table(self.table_name, schema=self.destination_schema_name)

    @property
    def is_destination_type_redshift(self):
        return self.destination_db_type == 'redshift'

    def get_columns_from_table(self, table):
        logging.info("Getting columns for {table}.".format(table=table.name))
        columns = []
        for column in table.columns:
            if self.is_destination_type_redshift:
                logging.info("Changing column constraints for {table}.".format(table=table.name))
                if column.name == 'id':
                    columns.append(Column(column.name, column.type, primary_key=True, autoincrement=False))
                elif type(column.type) is TEXT:
                    columns.append(Column(column.name, NVARCHAR(65535)))
                elif type(column.type) is DOUBLE_PRECISION:
                    columns.append(Column(column.name, NUMERIC(20, 5)))
                elif type(column.type) is NullType:
                    columns.append(Column(column.name, NVARCHAR(65535)))
                elif type(column.type) is Geography:
                    columns.append(Column(column.name, NVARCHAR(65535)))
                else:
                    columns.append(Column(column.name, column.type))
            else:
                if type(column.type) is DOUBLE_PRECISION:
                    columns.append(Column(column.name, NUMERIC(20, 5)))
                else:
                    columns.append(Column(column.name, column.type))
        return columns

    def execute_psql_command(self, conn_uri, command, password=None):
        if password:
            psql_query = '''export PGPASSWORD='{password}'; psql "{conn_url}" -c "{command}" '''.format(
                conn_url=conn_uri, command=command, password=password)
            print(psql_query)
            os.system(psql_query)
        else:
            psql_query = '''psql "{conn_url}" -c "{command}" '''.format(
                conn_url=conn_uri, command=command)
            print(psql_query)
            res = os.system(psql_query)
            if res != 0:
                sys.exit(1)

    def create_directory(self, directory):
        os.system('''mkdir -p {directory}'''.format(directory=directory))
        return directory

    def remove_local_file(self):
        command = '''rm -f {file_path}'''.format(file_path=self.local_file)
        logging.info("Removing file with command: {command}".format(command=command))
        os.system(command)

    def copy_table_to_local_csv_command(self, table_name):
        pump_tmp_directory = Variable.get('PUMP_TMP_DIRECTORY', default_var='/tmp/pump')
        directory = self.create_directory("{pump_tmp_directory}/{destination_schema_name}".format(
            destination_schema_name=self.destination_schema_name,
            pump_tmp_directory=pump_tmp_directory
        ))
        file_ = "{directory}/{table_name}".format(directory=directory, table_name=table_name) + '_{:%Y-%m-%dT%H%M}'.format(datetime.datetime.now()) + '.csv'
        command = '''\COPY (SELECT * FROM {table_name}) TO '{file}' HEADER CSV;'''.format(
            table_name=table_name,
            file=file_
        )
        self.local_file = file_
        logging.info("Dumped table {table_name} to file {file_name}".format(table_name=self.table_name, file_name=self.local_file))
        return command

    def copy_s3_file_to_table_command(self, table_name, schema_name, bucket_name, file_name, aws_key, aws_pass):
        table_name = '.'.join([schema_name, table_name])
        command = '''COPY {table_name} FROM 's3://{bucket_name}/{file}' credentials 'aws_access_key_id={aws_key};aws_secret_access_key={aws_pass}' IGNOREHEADER 1 CSV;'''.format(
            table_name=table_name,
            bucket_name=bucket_name,
            file=file_name,
            aws_key=aws_key,
            aws_pass=aws_pass
        )
        return command

    def copy_local_file_to_table_command(self, table_name, schema_name, file_name):
        table_name = '.'.join([schema_name, table_name])
        command = '''\COPY {table_name} FROM '{file}' HEADER CSV;'''.format(
            table_name=table_name,
            file=file_name,
        )
        return command

    def copy_table_to_local_file(self):
        command = self.copy_table_to_local_csv_command(self.origin_table_name_with_schema)
        conn = self.origin_hook.get_connection(self.origin_hook.postgres_conn_id)
        conn_uri = 'postgres://{user}@{host}:{port}/{schema}'.format(user=conn.login,
                                                                     host=conn.host,
                                                                     port=conn.port,
                                                                     schema=conn.schema)
        self.execute_psql_command(conn_uri, command, conn.password)

    def copy_table_to_s3(self):
        logging.info("Copying local file to s3 bucket:{bucket} file: {file}".format(bucket=self.bucket_name,
                                                                                    file=self.s3_file_name))
        self.s3_hook.load_file(
            self.local_file,
            self.s3_file_name,
            bucket_name=self.bucket_name,
            replace=True,
        )

    def remove_s3_file(self):
        logging.info("Removing file from s3 bucket:{bucket} file: {file}".format(bucket=self.bucket_name,
                                                                                 file=self.s3_file_name))
        boto_client = self.s3_hook.get_conn()
        try:
            boto_client.head_object(Bucket=self.s3_bucket, Key=self.s3_file_name)
            boto_client.delete_object(Bucket=self.s3_bucket, Key=self.s3_file_name)
        except botocore.exceptions.ClientError as e:
            logging.info("Could not find file. Error {}".format(e))

    def copy_s3_file_to_table(self):
        logging.info("Copying file from s3 bucket:{bucket} file: {file} to table {table_name}".format(
            bucket=self.bucket_name,
            file=self.s3_file_name,
            table_name=self.table_name))
        command = self.copy_s3_file_to_table_command(table_name=self.table_name,
                                                     schema_name=self.destination_schema_name,
                                                     bucket_name=self.bucket_name,
                                                     file_name=self.s3_file_name,
                                                     aws_key=self.aws_key,
                                                     aws_pass=self.aws_pass)
        self.destination_hook.run(command)

    def copy_local_file_to_table(self):
        logging.info("Copying local file {local_file} to {table_name}".format(
            table_name=self.table_name,
            local_file=self.local_file
        ))
        command = self.copy_local_file_to_table_command(table_name=self.table_name,
                                                        schema_name=self.destination_schema_name,
                                                        file_name=self.local_file)
        self.execute_psql_command(self.destination_conn_uri, command)

    def ddl_statement_create_table(self):
        if self.is_destination_type_redshift:
            table = Table(self.table_name, self.origin_metadata, autoload=True, autoload_with=self.origin_engine)
        else:
            table = GeoTable(self.table_name, self.origin_metadata, autoload=True, autoload_with=self.origin_engine)
        columns = self.get_columns_from_table(table)
        destination_table = Table(self.table_name, MetaData(schema=self.destination_schema_name), *columns)
        created_table = CreateTable(destination_table).compile(dialect=postgresql.dialect())
        created_table_ddl = str(created_table)
        logging.info("Created table: {created_table_ddl}".format(created_table_ddl=created_table_ddl))
        return created_table_ddl

    def ddl_statement_truncate_table(self):
        truncate_table_ddl = '''TRUNCATE TABLE {table_name}'''.format(table_name=self.destination_table_name_with_schema)
        return truncate_table_ddl

    def ddl_statement_drop_table(self):
        drop_table_ddl = '''DROP TABLE IF EXISTS {table_name} CASCADE'''.format(table_name=self.destination_table_name_with_schema)
        return drop_table_ddl

    def create_table(self):
        create_ddl_statement = self.ddl_statement_create_table()
        truncate_ddl_statement = self.ddl_statement_truncate_table()
        drop_ddl_statement = self.ddl_statement_drop_table()

        if self.drop_destination_table:
            self.destination_hook.run(drop_ddl_statement)
        elif self.destination_has_table:
            self.destination_hook.run(truncate_ddl_statement)

        if not self.destination_has_table:
            self.destination_hook.run(create_ddl_statement)

    def create_and_load_table(self):
        if self.origin_has_table:
            self.copy_table_to_local_file()
            self.create_table()
            if self.is_destination_type_redshift:
                self.copy_table_to_s3()
                self.copy_s3_file_to_table()
                self.remove_s3_file()
            else:
                self.copy_local_file_to_table()
            self.remove_local_file()
        else:
            logging.error("Origin hook {origin_hook} does not have table {table} in schema {schema}".format(
                origin_hook=self.origin_hook,
                table=self.table_name,
                schema=self.origin_schema_name
            ))
            sys.exit(1)


class PumpSubDagOperator(SubDagOperator):

    @airflow_utils.apply_defaults
    def __init__(self, dag, task_id, start_date, schedule_interval, default_args, table_names, pump_config, **kwargs):

        self.start_date = start_date
        self.dag_schedule_interval = schedule_interval
        self.default_args = default_args

        self.table_names = table_names
        self.origin_conn_id = pump_config['origin_conn_id']
        self.destination_conn_id = pump_config['destination_conn_id']
        self.origin_schema_name = pump_config['origin_schema_name']
        self.destination_schema_name = pump_config['destination_schema_name']
        self.destination_db_type = pump_config['destination_db_type']
        self.s3_conn_id = pump_config['s3_conn_id']
        self.s3_bucket = pump_config['s3_bucket']
        self.s3_directory = pump_config['s3_directory']
        if 'drop_destination_table' in pump_config:
            self.drop_destination_table = bool(pump_config['drop_destination_table'])
        else:
            self.drop_destination_table = False

        from airflow import DAG  # circular import

        self.sub_dag_name = dag.dag_id + '.' + task_id
        self.subdag = DAG(
            self.sub_dag_name,
            start_date=self.start_date,
            schedule_interval=self.dag_schedule_interval,
            default_args=self.default_args
        )

        self.init_tasks()

        super(PumpSubDagOperator, self).__init__(
            dag=dag,
            subdag=self.subdag,
            task_id=task_id,
            trigger_rule='all_done'
        )

    @property
    def task_type(self):
        return 'SubDagOperator'

    def pump_table(self, table_name, **kwargs):
        pump = Pump(
            table_name=table_name,
            origin_conn_id=self.origin_conn_id,
            destination_conn_id=self.destination_conn_id,
            origin_schema_name=self.origin_schema_name,
            destination_schema_name=self.destination_schema_name,
            destination_db_type=self.destination_db_type,
            s3_conn_id=self.s3_conn_id,
            s3_bucket=self.s3_bucket,
            s3_directory=self.s3_directory,
            drop_destination_table=self.drop_destination_table
        )
        pump.create_and_load_table()

    def create_task(self, table_name):
        task = PythonOperator(
            task_id="pump_{table_name}".format(table_name=table_name),
            python_callable=self.pump_table,
            op_kwargs={
                "table_name": table_name
            },
            dag=self.subdag
        )

    def init_tasks(self):
        for table_name in self.table_names:
            self.create_task(table_name)


if __name__ == '__main__':
    pump = Pump(
        table_name='test',
        origin_conn_id='airflow_origin_conn_name',
        origin_schema_name='origin_schema',
        destination_conn_id='airflow_destination_conn_name',
        destination_schema_name='destination_schema',
        destination_db_type='postgres',
        s3_conn_id='airflow_s3_conn_name',
        s3_bucket='bucket_name',
        s3_directory='directory_name_in_bucket',
        drop_destination_table=True
    )