Snowpark Python DB-APIの使用¶

def create_sql_server_connection():
    import pyodbc
    SERVER = "<your host name>"
    PORT = <your port>
    UID = "<your user name>"
    PWD = "<your password>"
    DATABASE = "<your database name>"
    connection_str = (
        f"DRIVER={{ODBC Driver 18 for SQL Server}};"
        f"SERVER={SERVER}:{PORT};"
        f"UID={UID};"
        f"PWD={PWD};"
        f"DATABASE={DATABASE};"
        "TrustServerCertificate=yes"
        "Encrypt=yes"
        # Optional to identify source of queries
        "APP=snowflake-snowpark-python;"
    )
    connection = pyodbc.connect(connection_str)
    return connection

# Feel free to combine local/udtf ingestion and partition column/predicates as
# stated in the understanding parallelism section

# Call dbapi to pull data from target table

df = session.read.dbapi(
    create_sql_server_connection,
    table="target_table"
)

# Call dbapi to pull data from target query

df_query = session.read.dbapi(
    create_sql_server_connection,
    query="select * from target_table"
)

# Pull data from target table with parallelism using partition column

df_local_par_column = session.read.dbapi(
    create_sql_server_connection,
    table="target_table",
    fetch_size=100000,
    num_partitions=4,
    column="ID",  # Swap with the column you want your partition based on
    upper_bound=10000,
    lower_bound=0
)

udtf_configs = {
    "external_access_integration": "<your external access integration>"
}

# Pull data from target table with udtf ingestion with parallelism using predicates

df_udtf_predicates = session.read.dbapi(
    create_sql_server_connection,
    table="target_table",
    udtf_configs=udtf_configs,
    fetch_size=100000,
    predicates=[
        "ID < 3",
        "ID >= 3"
    ]
)

# Get user name and password from mssql_secret

import _snowflake
username_password_object = _snowflake.get_username_password('snowflake-secret-object')
USER = username_password_object.username
PASSWORD = username_password_object.password

import snowflake.snowpark.context
session = snowflake.snowpark.context.get_active_session()

def create_sql_server_connection():
    import pyodbc
    SERVER = SQL_SERVER_CONNECTION_PARAMETERS["SERVER"]
    UID = SQL_SERVER_CONNECTION_PARAMETERS["UID"]
    PWD = SQL_SERVER_CONNECTION_PARAMETERS["PWD"]
    DATABASE = "test_query_history"
    connection_str = (
        f"DRIVER={{ODBC Driver 18 for SQL Server}};"
        f"SERVER={SERVER};"
        f"UID={UID};"
        f"PWD={PWD};"
        f"DATABASE={DATABASE};"
        "TrustServerCertificate=yes;"
        "Encrypt=yes;"
        # Optional to identify source of queries
        "APP=snowflake-snowpark-python;"
    )
    connection = pyodbc.connect(connection_str)
    return connection

# Feel free to combine local/udtf ingestion and partition column/predicates as
# stated in the understanding parallelism section

# Call dbapi to pull data from target table

df = session.read.dbapi(
    create_sql_server_connection,
    table="target_table"
)

# Call dbapi to pull data from target query

df_query = session.read.dbapi(
    create_sql_server_connection,
    query="select * from target_table"
)

# Pull data from target table with parallelism using partition column

df_local_par_column = session.read.dbapi(
    create_sql_server_connection,
    table="target_table",
    fetch_size=100000,
    num_partitions=4,
    column="ID",  # swap with the column you want your partition based on
    upper_bound=10000,
    lower_bound=0
)

udtf_configs = {
    "external_access_integration": "<your external access integration>"
}

# Pull data from target table with udtf ingestion with parallelism using predicates

df_udtf_predicates = session.read.dbapi(
    create_sql_server_connection,
    table="target_table",
    udtf_configs=udtf_configs,
    fetch_size=100000,
    predicates=[
        "ID < 3",
        "ID >= 3"
    ]
)

# Save data into sf_table
df.write.mode("overwrite").save_as_table('sf_table')

def create_sql_server_connection():
    import pyodbc
    SERVER = "<your host name>"
    PORT = <your port>
    UID = "<your user name>"
    PWD = "<your password>"
    DATABASE = "<your database name>"
    connection_str = (
        f"DRIVER={{ODBC Driver 18 for SQL Server}};"
        f"SERVER={SERVER}:{PORT};"
        f"UID={UID};"
        f"PWD={PWD};"
        f"DATABASE={DATABASE};"
        "TrustServerCertificate=yes"
        "Encrypt=yes"
        # include this parameter for source tracing
        "APP=snowflake-snowpark-python;"
    )
    connection = pyodbc.connect(connection_str)
    return connection

def create_oracle_db_connection():
    import oracledb
    HOST = "<your host>"
    PORT = <your port>
    SERVICE_NAME = "<your service name>"
    USER = "<your user name>"
    PASSWORD = "your password"
    DSN = f"{HOST}:{PORT}/{SERVICE_NAME}"
    connection = oracledb.connect(
        user=USER,
        password=PASSWORD,
        dsn=DSN
    )
    # Optional: include this parameter for source tracing
    connection.clientinfo = "snowflake-snowpark-python"
    return connection

# Feel free to combine local/udtf ingestion and partition column/predicates as
# stated in the understanding parallelism section

# Call dbapi to pull data from target table

df = session.read.dbapi(
    create_oracle_db_connection,
    table="target_table"
)

# Call dbapi to pull data from target query

df_query = session.read.dbapi(
    create_oracle_db_connection,
    query="select * from target_table"
)

# Pull data from target table with parallelism using partition column

df_local_par_column = session.read.dbapi(
    create_oracle_db_connection,
    table="target_table",
    fetch_size=100000,
    num_partitions=4,
    column="ID",  # swap with the column you want your partition based on
    upper_bound=10000,
    lower_bound=0
)

udtf_configs = {
    "external_access_integration": "<your external access integration>"
}

# Pull data from target table with udtf ingestion with parallelism using predicates

df_udtf_predicates = session.read.dbapi(
    create_oracle_db_connection,
    table="target_table",
    udtf_configs=udtf_configs,
    fetch_size=100000,
    predicates=[
        "ID < 3",
        "ID >= 3"
    ]
)

# Get user name and password from ora_secret

import _snowflake
username_password_object = _snowflake.get_username_password('snowflake-secret-object')
USER = username_password_object.username
PASSWORD = username_password_object.password

import snowflake.snowpark.context
session = snowflake.snowpark.context.get_active_session()

# Define the factory method for creating a connection to Oracle

def create_oracle_db_connection():
    import oracledb
    host = "ora_host"
    port = "ora_port"
    service_name = "ora_service"
    user = USER
    password = PASSWORD
    DSN = f"{host}:{port}/{service_name}"
    connection = oracledb.connect(
        user=USER,
        password=PASSWORD,
        dsn=DSN,
    )
    # Optional: include this parameter for source tracing
    connection.clientinfo = "snowflake-snowpark-python"
    return connection

# Feel free to combine local/udtf ingestion and partition column/predicates as
# stated in the understanding parallelism section

# Call dbapi to pull data from target table

df = session.read.dbapi(
    create_oracle_db_connection,
    table="target_table"
)

# Call dbapi to pull data from target query

df_query = session.read.dbapi(
    create_oracle_db_connection,
    query="select * from target_table"
)

# Pull data from target table with parallelism using partition column

df_local_par_column = session.read.dbapi(
    create_oracle_db_connection,
    table="target_table",
    fetch_size=100000,
    num_partitions=4,
    column="ID",  # swap with the column you want your partition based on
    upper_bound=10000,
    lower_bound=0
)

udtf_configs = {
    "external_access_integration": "<your external access integration>"
}

# Pull data from target table with udtf ingestion with parallelism using predicates

df_udtf_predicates = session.read.dbapi(
    create_oracle_db_connection,
    table="target_table",
    udtf_configs=udtf_configs,
    fetch_size=100000,
    predicates=[
        "ID < 3",
        "ID >= 3"
    ]
)

# Save data into sf_table

df_ora.write.mode("overwrite").save_as_table('sf_table')

def create_oracle_db_connection():
    import oracledb
    HOST = "myhost"
    PORT = "myport"
    SERVICE_NAME = "myservice"
    USER = "myuser"
    PASSWORD = "mypassword"
    DSN = f"{HOST}:{PORT}/{SERVICE_NAME}"
    connection = oracledb.connect(
        user=USER,
        password=PASSWORD,
        dsn=DSN,
    )
    # include this parameter for source tracing
    connection.clientinfo = "snowflake-snowpark-python"
    return connection

def create_pg_connection():
    import psycopg2
    connection = psycopg2.connect(
        host="pg_host",
        port=pg_port,
        dbname="pg_dbname",
        user="pg_user",
        password="pg_password",
        # Optional: include this parameter for source tracing
        application_name="snowflake-snowpark-python"
    )
    return connection

# Feel free to combine local/udtf ingestion and partition column/predicates as
# stated in the understanding parallelism section

# Call dbapi to pull data from target table

df = session.read.dbapi(
    create_pg_connection,
    table="target_table"
)

# Call dbapi to pull data from target query

df_query = session.read.dbapi(
    create_pg_connection,
    query="select * from target_table"
)

# Pull data from target table with parallelism using partition column

df_local_par_column = session.read.dbapi(
    create_pg_connection,
    table="target_table",
    fetch_size=100000,
    num_partitions=4,
    column="ID",  # Swap with the column you want your partition based on
    upper_bound=10000,
    lower_bound=0
)

udtf_configs = {
    "external_access_integration": "<your external access integration>"
}

# Pull data from target table with udtf ingestion with parallelism using predicates

df_udtf_predicates = session.read.dbapi(
    create_pg_connection,
    table="target_table",
    udtf_configs=udtf_configs,
    fetch_size=100000,
    predicates=[
        "ID < 3",
        "ID >= 3"
    ]
)

# Get the user name and password from :code:`pg_secret`

import _snowflake
username_password_object = _snowflake.get_username_password('snowflake-secret-object')
USER = username_password_object.username
PASSWORD = username_password_object.password

import snowflake.snowpark.context
session = snowflake.snowpark.context.get_active_session()

# Define the factory method for creating a connection to PostgreSQL

def create_pg_connection():
    import psycopg2
    connection = psycopg2.connect(
        host="pg_host",
        port=pg_port,
        dbname="pg_dbname",
        user=USER,
        password=PASSWORD,
        # Optional: include this parameter for source tracing
        application_name="snowflake-snowpark-python"
    )
    return connection

# Feel free to combine local/udtf ingestion and partition column/predicates as
# stated in the understanding parallelism section

# Call dbapi to pull data from target table

df = session.read.dbapi(
    create_pg_connection,
    table="target_table"
)

# Call dbapi to pull data from target query

df_query = session.read.dbapi(
    create_pg_connection,
    query="select * from target_table"
)

# Pull data from target table with parallelism using partition column

df_local_par_column = session.read.dbapi(
    create_pg_connection,
    table="target_table",
    fetch_size=100000,
    num_partitions=4,
    column="ID",  # swap with the column you want your partition based on
    upper_bound=10000,
    lower_bound=0
)

udtf_configs = {
    "external_access_integration": "<your external access integration>"
}

# Pull data from target table with udtf ingestion with parallelism using predicates

df_udtf_predicates = session.read.dbapi(
    create_pg_connection,
    table="target_table",
    udtf_configs=udtf_configs,
    fetch_size=100000,
    predicates=[
        "ID < 3",
        "ID >= 3"
    ]
)

# Save data into sf_table

df.write.mode("overwrite").save_as_table('sf_table')
# Get the user name and password from :code:`pg_secret`

def create_pg_connection():
    import psycopg2
    connection = psycopg2.connect(
        host="pg_host",
        port=pg_port,
        dbname="pg_dbname",
        user="pg_user",
        password="pg_password",
        # Include this parameter for source tracing
        application_name="snowflake-snowpark-python"
    )
    return connection

def create_mysql_connection():
    import pymysql
    connection = pymysql.connect(
        host="mysql_host",
        port=mysql_port,
        database="mysql_db",
        user="mysql_user",
        password="mysql_password",
        # Optional: include this parameter for source tracing
        init_command="SET @program_name='snowflake-snowpark-python';"
    )
    return connection

# Feel free to combine local/udtf ingestion and partition column/predicates as
# stated in the understanding parallelism section

# Call dbapi to pull data from target table

df = session.read.dbapi(
    create_mysql_connection,
    table="target_table"
)

# Call dbapi to pull data from target query

df_query = session.read.dbapi(
    create_mysql_connection,
    query="select * from target_table"
)

# Pull data from target table with parallelism using partition column

df_local_par_column = session.read.dbapi(
    create_mysql_connection,
    table="target_table",
    fetch_size=100000,
    num_partitions=4,
    column="ID",  # swap with the column you want your partition based on
    upper_bound=10000,
    lower_bound=0
)

udtf_configs = {
    "external_access_integration": "<your external access integration>"
}

# Pull data from target table with udtf ingestion with parallelism using predicates

df_udtf_predicates = session.read.dbapi(
    create_mysql_connection,
    table="target_table",
    udtf_configs=udtf_configs,
    fetch_size=100000,
    predicates=[
        "ID < 3",
        "ID >= 3"
    ]
)

# Get user name and password from mysql_secret
import _snowflake
username_password_object = _snowflake.get_username_password('snowflake-secret-object')
USER = username_password_object.username
PASSWORD = username_password_object.password

import snowflake.snowpark.context
session = snowflake.snowpark.context.get_active_session()

# Define the factory method for creating a connection to MySQL

def create_mysql_connection():
    import pymysql
    connection = pymysql.connect(
        host="mysql_host",
        port=mysql_port,
        dbname="mysql_dbname",
        user=USER,
        password=PASSWORD,
        # Optional: include this parameter for source tracing
        init_command="SET @program_name='snowflake-snowpark-python';"
    )
    return connection

# Feel free to combine local/udtf ingestion and partition column/predicates as
# stated in the understanding parallelism section

# Call dbapi to pull data from target table

df = session.read.dbapi(
    create_mysql_connection,
    table="target_table"
)

# Call dbapi to pull data from target query

df_query = session.read.dbapi(
    create_mysql_connection,
    query="select * from target_table"
)

# Pull data from target table with parallelism using partition column

df_local_par_column = session.read.dbapi(
    create_mysql_connection,
    table="target_table",
    fetch_size=100000,
    num_partitions=4,
    column="ID",  # swap with the column you want your partition based on
    upper_bound=10000,
    lower_bound=0
)

udtf_configs = {
    "external_access_integration": "<your external access integration>"
}

# Pull data from target table with udtf ingestion with parallelism using predicates

df_udtf_predicates = session.read.dbapi(
    create_mysql_connection,
    table="target_table",
    udtf_configs=udtf_configs,
    fetch_size=100000,
    predicates=[
        "ID < 3",
        "ID >= 3"
    ]
)

# Save data into sf_table

df.write.mode("overwrite").save_as_table('sf_table')

def create_mysql_connection():
    import pymysql
    connection = pymysql.connect(
        host="mysql_host",
        port=mysql_port,
        database="mysql_db",
        user="mysql_user",
        password="mysql_password",
        # include this parameter for source tracing
        init_command="SET @program_name='snowflake-snowpark-python';"
    )
    return connection

def create_dbx_connection():
    import databricks.sql
    connection = databricks.sql.connect(
        server_hostname=HOST,
        http_path=PATH,
        access_token=ACCESS_TOKEN
    )
    return connection

# Feel free to combine local/udtf ingestion and partition column/predicates as
# stated in the understanding parallelism section

# Call dbapi to pull data from target table

df = session.read.dbapi(
    create_dbx_connection,
    table="target_table"
)

# Call dbapi to pull data from target query

df_query = session.read.dbapi(
    create_dbx_connection,
    query="select * from target_table"
)

# Pull data from target table with parallelism using partition column

df_local_par_column = session.read.dbapi(
    create_dbx_connection,
    table="target_table",
    fetch_size=100000,
    num_partitions=4,
    column="ID",  # swap with the column you want your partition based on
    upper_bound=10000,
    lower_bound=0
)

udtf_configs = {
    "external_access_integration": "<your external access integration>"
}

# Pull data from target table with udtf ingestion with parallelism using predicates

df_udtf_predicates = session.read.dbapi(
    create_dbx_connection,
    table="target_table",
    udtf_configs=udtf_configs,
    fetch_size=100000,
    predicates=[
        "ID < 3",
        "ID >= 3"
    ]
)

# Get user name and password from dbx_secret

import _snowflake
ACCESS_TOKEN = _snowflake.get_generic_secret_string('cred')

import snowflake.snowpark.context
session = snowflake.snowpark.context.get_active_session()

# Define the factory method for creating a connection to Databricks

def create_dbx_connection():
    import databricks.sql
    connection = databricks.sql.connect(
        server_hostname="dbx_host",
        http_path="dbx_path",
        access_token=ACCESS_TOKEN,
    )
    return connection

# Feel free to combine local/udtf ingestion and partition column/predicates as
# stated in the understanding parallelism section

# Call dbapi to pull data from target table

df = session.read.dbapi(
    create_dbx_connection,
    table="target_table"
)

# Call dbapi to pull data from target query

df_query = session.read.dbapi(
    create_dbx_connection,
    query="select * from target_table"
)

# Pull data from target table with parallelism using partition column

df_local_par_column = session.read.dbapi(
    create_dbx_connection,
    table="target_table",
    fetch_size=100000,
    num_partitions=4,
    column="ID",  # swap with the column you want your partition based on
    upper_bound=10000,
    lower_bound=0
)

udtf_configs = {
    "external_access_integration": "<your external access integration>"
}

# Pull data from target table with udtf ingestion with parallelism using predicates

df_udtf_predicates = session.read.dbapi(
    create_dbx_connection,
    table="target_table",
    udtf_configs=udtf_configs,
    fetch_size=100000,
    predicates=[
        "ID < 3",
        "ID >= 3"
    ]
)

# Save data into sf_table

df.write.mode("overwrite").save_as_table('sf_table')

def create_dbx_connection():
    import databricks.sql
    connection = databricks.sql.connect(
        server_hostname=HOST,
        http_path=PATH,
        access_token=ACCESS_TOKEN,
        # include this parameter for source tracing
        user_agent_entry="snowflake-snowpark-python"
    )
    return connection