notebooks/parameterize_script/util/notebook_constants.py (115 lines of code) (raw):

""" Notebook constants """ # Common ## Command Line Arguments OUTPUT_NOTEBOOK_ARG = "output.notebook" MAX_PARALLELISM_ARG = "max.parallelism" LOG_LEVEL_ARG = "log_level" ## Notebook Arguments GCP_PROJECT = "GCP_PROJECT" PROJECT = "PROJECT" REGION = "REGION" GCS_STAGING_LOCATION = "GCS_STAGING_LOCATION" SUBNET = "SUBNET" IS_PARAMETERIZED = "IS_PARAMETERIZED" MAX_PARALLELISM = "MAX_PARALLELISM" SERVICE_ACCOUNT = "SERVICE_ACCOUNT" JDBCTOJDBC_OUTPUT_MODE_ARG="jdbctojdbc.output.mode" BATCH_SIZE_ARG="jdbctojdbc.output.batch.size" # Write modes OUTPUT_MODE_OVERWRITE = "overwrite" OUTPUT_MODE_APPEND = "append" # HIVE TO BIGQUERY ## Command Line Arguments HIVE_METASTORE_ARG = "hive.metastore" INPUT_HIVE_DATABASE_ARG = "input.hive.database" INPUT_HIVE_TABLES_ARG = "input.hive.tables" OUTPUT_BIGQUERY_DATASET_ARG = "output.bigquery.dataset" TEMP_BUCKET_ARG = "temp.bucket" HIVE_OUTPUT_MODE_ARG = "hive.output.mode" BQ_DATASET_REGION_ARG = "bq.dataset.region" ## Notebook Arguments HIVE_METASTORE = "HIVE_METASTORE" INPUT_HIVE_DATABASE = "INPUT_HIVE_DATABASE" INPUT_HIVE_TABLES = "INPUT_HIVE_TABLES" OUTPUT_BIGQUERY_DATASET = "OUTPUT_BIGQUERY_DATASET" TEMP_BUCKET = "TEMP_BUCKET" HIVE_OUTPUT_MODE = "HIVE_OUTPUT_MODE" BQ_DATASET_REGION = "BQ_DATASET_REGION" # MYSQL TO SPANNER ## Command Line Arguments MYSQL_HOST_ARG = "mysql.host" MYSQL_PORT_ARG = "mysql.port" MYSQL_USERNAME_ARG = "mysql.username" MYSQL_PASSWORD_ARG = "mysql.password" MYSQL_DATABASE_ARG = "mysql.database" MYSQL_TABLE_LIST_ARG = "mysql.table.list" MYSQL_OUTPUT_SPANNER_MODE_ARG = "mysql.output.spanner.mode" SPANNER_INSTANCE_ARG = "spanner.instance" SPANNER_DATABASE_ARG = "spanner.database" # provide table & pk column which do not have PK in MYSQL "{"table_name":"primary_key"}" SPANNER_TABLE_PRIMARY_KEYS_ARG = "spanner.table.primary.keys" ## Notebook Arguments MYSQL_HOST = "MYSQL_HOST" MYSQL_PORT = "MYSQL_PORT" MYSQL_USERNAME = "MYSQL_USERNAME" MYSQL_PASSWORD = "MYSQL_PASSWORD" MYSQL_DATABASE = "MYSQL_DATABASE" MYSQL_TABLE_LIST = "MYSQL_TABLE_LIST" MYSQL_OUTPUT_SPANNER_MODE = "MYSQL_OUTPUT_SPANNER_MODE" SPANNER_INSTANCE = "SPANNER_INSTANCE" SPANNER_DATABASE = "SPANNER_DATABASE" SPANNER_TABLE_PRIMARY_KEYS = "SPANNER_TABLE_PRIMARY_KEYS" # ORACLE TO BIGQUERY ## Command Line Arguments ORACLE_HOST_ARG = "oracle.host" ORACLE_PORT_ARG = "oracle.port" ORACLE_USERNAME_ARG = "oracle.username" ORACLE_PASSWORD_ARG = "oracle.password" ORACLE_DATABASE_ARG = "oracle.database" ORACLE_SCHEMA_ARG = "oracle.schema" ORACLE_TABLE_LIST_ARG = "oracle.table.list" BIGQUERY_MODE_ARG = "bigquery.mode" BIGQUERY_DATASET_ARG = "bigquery.dataset" TEMP_GCS_BUCKET_ARG = "temp.gcs.bucket" ## Notebook Arguments ORACLE_HOST = "ORACLE_HOST" ORACLE_PORT = "ORACLE_PORT" ORACLE_USERNAME = "ORACLE_USERNAME" ORACLE_PASSWORD = "ORACLE_PASSWORD" ORACLE_DATABASE = "ORACLE_DATABASE" ORACLE_SCHEMA = "ORACLE_SCHEMA" ORACLE_TABLE_LIST = "ORACLE_TABLE_LIST" BIGQUERY_DATASET = "BIGQUERY_DATASET" BIGQUERY_MODE = "BIGQUERY_MODE" TEMP_GCS_BUCKET = "TEMP_GCS_BUCKET" # POSTGRESQL TO BIGQUERY ## Command Line Arguments OUTPUT_NOTEBOOK_ARG = "output.notebook" MAX_PARALLELISM_ARG = "max.parallelism" POSTGRESQL_HOST_ARG = "postgresql.host" POSTGRESQL_PORT_ARG = "postgresql.port" POSTGRESQL_USERNAME_ARG = "postgresql.username" POSTGRESQL_PASSWORD_ARG = "postgresql.password" POSTGRESQL_DATABASE_ARG = "postgresql.database" POSTGRESQL_TABLE_LIST_ARG = "postgresql.table.list" POSTGRESQL_SCHEMA_LIST_ARG = "postgresql.schema.list" BIGQUERY_DATASET_ARG = "bigquery.dataset" BIGQUERY_MODE_ARG = "bigquery.mode" ## Notebook Arguments POSTGRESQL_HOST = "POSTGRESQL_HOST" POSTGRESQL_PORT = "POSTGRESQL_PORT" POSTGRESQL_USERNAME = "POSTGRESQL_USERNAME" POSTGRESQL_PASSWORD = "POSTGRESQL_PASSWORD" POSTGRESQL_DATABASE = "POSTGRESQL_DATABASE" POSTGRESQL_TABLE_LIST = "POSTGRESQL_TABLE_LIST" POSTGRESQL_SCHEMA_LIST = "POSTGRESQL_SCHEMA_LIST" BIGQUERY_DATASET = "BIGQUERY_DATASET" BIGQUERY_MODE = "BIGQUERY_MODE" #ORACLE TO POSTGRES ## Command Line Arguments ORACLE_HOST_ARG = "oracle.host" ORACLE_PORT_ARG = "oracle.port" ORACLE_USERNAME_ARG = "oracle.username" ORACLE_PASSWORD_ARG = "oracle.password" ORACLE_DATABASE_ARG = "oracle.database" ORACLETABLE_LIST_ARG = "oracle.table.list" POSTGRES_HOST_ARG="postgres.host" POSTGRES_PORT_ARG="postgres.port" POSTGRES_USERNAME_ARG="postgres.username" POSTGRES_PASSWORD_ARG="postgres.password" POSTGRES_DATABASE_ARG="postgres.database" POSTGRES_SCHEMA_ARG="postgres.schema" ## Notebook Arguments ORACLE_HOST = "ORACLE_HOST" ORACLE_PORT = "ORACLE_PORT" ORACLE_USERNAME = "ORACLE_USERNAME" ORACLE_PASSWORD = "ORACLE_PASSWORD" ORACLE_DATABASE = "ORACLE_DATABASE" ORACLETABLE_LIST = "ORACLETABLE_LIST" POSTGRES_HOST="POSTGRES_HOST" POSTGRES_PORT="POSTGRES_PORT" POSTGRES_USERNAME="POSTGRES_USERNAME" POSTGRES_PASSWORD="POSTGRES_PASSWORD" POSTGRES_DATABASE="POSTGRES_DATABASE" POSTGRES_SCHEMA="POSTGRES_SCHEMA" JDBCTOJDBC_OUTPUT_MODE="JDBCTOJDBC_OUTPUT_MODE" BATCH_SIZE="JDBCTOJDBC_OUTPUT_BATCH_SIZE"