dbrequests
dbrequests is a python package built for easy use of raw SQL within python and pandas projects.
It uses ideas from records and is built using sqlalchemy-engines, but is more heavily integrated with pandas. It aims to reproduce the pilosophy behind the R-package dbtools.
Database support includes RedShift, Postgres, MySQL, SQLite, Oracle, and MS-SQL (drivers not included).
Usage
Send queries and bulk queries
Easy sending of raw sql and output as pandas DataFrames, with credentials given as dictionary (for an example see creds_example.json) or the url of the database:
from dbrequests import Database
db = Database(creds=creds)
df = db.send_query("""
SELECT * FROM test;
""")
df # table test as pandas DataFrame
You can put the sql query in a file and direct send_query
to the file. The sql-file may be parametrized:
SELECT {col1}, {col2} FROM test;
from dbrequests import Database
db = Database(creds=creds, sql_dir = '/path/to/dir/')
db.send_query('select', col1='id', col2='name')
df # table test, including columns 'id', 'name' as pandas DataFrame
You can also pass arguments to pandas read_sql
-Function:
from dbrequests import Database
db = Database(creds=creds, sql_dir = '/path/to/dir/')
db.send_query('select', col1='id', col2='name', index_col='id')
df # table test, including column 'name' as pandas DataFrame with index 'id'
You may also send queries with no table as output to the database via send_bulk_query
, which exhibits the same behavior as send_query
:
db.send_bulk_query('drop test from test;')
Send data
Easy sending of pandas Dataframes in multiple modes:
db.send_data(df, 'table', mode='insert')
Supported modes are:
- 'insert': Appending new records. Duplicate primary keys will result in errors (sql insert into).
- 'truncate': Delete the table and completely rewrite it (sql truncate and insert into).
- 'replace': Replace records with duplicate primary keys (sql replace into).
- 'update': Update records with duplicate primary keys (sql insert into duplicate key update).
Utilities
- Comments can be automatically removed from SQL code by adding
remove_comments=True
either to the Database defintion or send_query. This is especially useful if outcommenting code blocks including parametized variables and thus{}
. The default of this behavior isFalse
. - Percentage signs can be transfered to a Python readable way by adding
escape_percentage=True
either to the Database definition or send_query. This means percentage signs dont have to be escaped manually when sending them via Python. The default isFalse
. - Database.get_table_names will give existing tables in the database
- Parameters such es
chunksize
forpandas.to_sql
may be given to the wrapper functionsend_data
and are handed over to pandas. The same is true forsend_query
. - For transactions the context manager
transaction
may be of use.
Installation
The package can be installed via pip:
pip install dbrequests
Extensibility
dbrequests is designed to easily accommodate different needs in the form of drivers / dialects. For examples of how to extend the capabilities of the Connection class, see connection_subclass.py under examples.
Existing extensions
- MySQL / MariaDB: use
from dbrequests.mysql import Database
for using the MySQL specific extension as Database connector. The extension provides MySQL specific functionalities, like using load data infile
for writing data to tables.