Source code for airflow.providers.google.cloud.transfers.presto_to_gcs

#
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements.  See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership.  The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License.  You may obtain a copy of the License at
#
#   http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied.  See the License for the
# specific language governing permissions and limitations
# under the License.
from typing import Any, Dict, List, Tuple

from prestodb.client import PrestoResult
from prestodb.dbapi import Cursor as PrestoCursor

from airflow.providers.google.cloud.transfers.sql_to_gcs import BaseSQLToGCSOperator
from airflow.providers.presto.hooks.presto import PrestoHook
from airflow.utils.decorators import apply_defaults


[docs]class _PrestoToGCSPrestoCursorAdapter: """ An adapter that adds additional feature to the Presto cursor. The implementation of cursor in the prestodb library is not sufficient. The following changes have been made: * The poke mechanism for row. You can look at the next row without consuming it. * The description attribute is available before reading the first row. Thanks to the poke mechanism. * the iterator interface has been implemented. A detailed description of the class methods is available in `PEP-249 <https://www.python.org/dev/peps/pep-0249/>`__. """ def __init__(self, cursor: PrestoCursor): self.cursor: PrestoCursor = cursor self.rows: List[Any] = [] self.initialized: bool = False @property
[docs] def description(self) -> List[Tuple]: """ This read-only attribute is a sequence of 7-item sequences. Each of these sequences contains information describing one result column: * ``name`` * ``type_code`` * ``display_size`` * ``internal_size`` * ``precision`` * ``scale`` * ``null_ok`` The first two items (``name`` and ``type_code``) are mandatory, the other five are optional and are set to None if no meaningful values can be provided. """ if not self.initialized: # Peek for first row to load description. self.peekone() return self.cursor.description
@property
[docs] def rowcount(self) -> int: """The read-only attribute specifies the number of rows""" return self.cursor.rowcount
[docs] def close(self) -> None: """Close the cursor now""" self.cursor.close()
[docs] def execute(self, *args, **kwargs) -> PrestoResult: """Prepare and execute a database operation (query or command).""" self.initialized = False self.rows = [] return self.cursor.execute(*args, **kwargs)
[docs] def executemany(self, *args, **kwargs): """ Prepare a database operation (query or command) and then execute it against all parameter sequences or mappings found in the sequence seq_of_parameters. """ self.initialized = False self.rows = [] return self.cursor.executemany(*args, **kwargs)
[docs] def peekone(self) -> Any: """Return the next row without consuming it.""" self.initialized = True element = self.cursor.fetchone() self.rows.insert(0, element) return element
[docs] def fetchone(self) -> Any: """ Fetch the next row of a query result set, returning a single sequence, or ``None`` when no more data is available. """ if self.rows: return self.rows.pop(0) return self.cursor.fetchone()
[docs] def fetchmany(self, size=None) -> list: """ Fetch the next set of rows of a query result, returning a sequence of sequences (e.g. a list of tuples). An empty sequence is returned when no more rows are available. """ if size is None: size = self.cursor.arraysize result = [] for _ in range(size): row = self.fetchone() if row is None: break result.append(row) return result
[docs] def __next__(self) -> Any: """ Return the next row from the currently executing SQL statement using the same semantics as ``.fetchone()``. A ``StopIteration`` exception is raised when the result set is exhausted. :return: """ result = self.fetchone() if result is None: raise StopIteration() return result
[docs] def __iter__(self) -> "_PrestoToGCSPrestoCursorAdapter": """Return self to make cursors compatible to the iteration protocol""" return self
[docs]class PrestoToGCSOperator(BaseSQLToGCSOperator): """Copy data from PrestoDB to Google Cloud Storage in JSON or CSV format. :param presto_conn_id: Reference to a specific Presto hook. :type presto_conn_id: str """
[docs] ui_color = "#a0e08c"
[docs] type_map = { "BOOLEAN": "BOOL", "TINYINT": "INT64", "SMALLINT": "INT64", "INTEGER": "INT64", "BIGINT": "INT64", "REAL": "FLOAT64", "DOUBLE": "FLOAT64", "DECIMAL": "NUMERIC", "VARCHAR": "STRING", "CHAR": "STRING", "VARBINARY": "BYTES", "JSON": "STRING", "DATE": "DATE", "TIME": "TIME", # BigQuery don't time with timezone native. "TIME WITH TIME ZONE": "STRING", "TIMESTAMP": "TIMESTAMP", # BigQuery supports a narrow range of time zones during import. # You should use TIMESTAMP function, if you want have TIMESTAMP type "TIMESTAMP WITH TIME ZONE": "STRING", "IPADDRESS": "STRING", "UUID": "STRING",
} @apply_defaults def __init__(self, *, presto_conn_id: str = "presto_default", **kwargs): super().__init__(**kwargs) self.presto_conn_id = presto_conn_id
[docs] def query(self): """Queries presto and returns a cursor to the results.""" presto = PrestoHook(presto_conn_id=self.presto_conn_id) conn = presto.get_conn() cursor = conn.cursor() self.log.info("Executing: %s", self.sql) cursor.execute(self.sql) return _PrestoToGCSPrestoCursorAdapter(cursor)
[docs] def field_to_bigquery(self, field) -> Dict[str, str]: """Convert presto field type to BigQuery field type.""" clear_field_type = field[1].upper() # remove type argument e.g. DECIMAL(2, 10) => DECIMAL clear_field_type, _, _ = clear_field_type.partition("(") new_field_type = self.type_map.get(clear_field_type, "STRING") return {"name": field[0], "type": new_field_type}
[docs] def convert_type(self, value, schema_type): """ Do nothing. Presto uses JSON on the transport layer, so types are simple. :param value: Presto column value :type value: Any :param schema_type: BigQuery data type :type schema_type: str """ return value

Was this entry helpful?