Skip to content

Latest commit

 

History

History
167 lines (104 loc) · 7.6 KB

README.rst

File metadata and controls

167 lines (104 loc) · 7.6 KB
Raw

SQL Connectors

Documentation Status Updates

A simple wrapper for SQL connections using SQLAlchemy and Pandas read_sql to standardize SQL workflow. The main goals of this project is to reduce boilerplate code when working with SQL based data sources and to enable interactive exploration of data sources in Python.

Features

  • Standardized client for working with different SQL datasources, including a standardized format for defining your connection configurations
  • A SqlClient interface based off the SQLAlchemy Engine with some helpful functions like Pandas' read_sql and functions to leverage reflection from SQLAlchemy

Installation

Stable release

To install SQL Connectors, run this command in your terminal:

$ pip install --process-dependency-links sql_connectors

This is the preferred method to install SQL Connectors, as it will always install the most recent stable release.

If you don't have pip installed, this Python installation guide can guide you through the process.

Dev install

The sources for SQL Connectors can be downloaded from the Github repo.

You can clone the public repository and install in development mode:

$ git clone git://github.com/aiguofer/sql_connectors
$ cd sql_connectors
$ pip install --process-dependency-links -e .[dev]

Configurations

Configurations can be stored wherever you want by implementing your own Storage. However, the default is LocalStorage reading in configuration files from ~/.config/sql_connectors.

You can change the Storage class using the SQL_CONNECTORS_STORAGE environment variable (for example sql_connectors.storage.LocalStorage), and you can specify a different configuration directory or URI with SQL_CONNECTORS_PATH_OR_URI.

The example_connection.json file is provided as a template; feel free to replace this with your own connection details and re-name the file.

The contents of the example file are:

{
    "drivername": "sqlite",
    "relative_paths": ["database"],
    "default_env": "default",
    "default": {
        "database": "example_connection.db"
    }
}

The fields mean the following:

drivername (string)
This required field is a SQLAlchemy dialect or dialect+driver. See the SQLAlchemy Engine documentation for more details. You may first have to install the required python modules for your dialect+driver to work if it's a third party plug-in.
relative_paths (list of strings)
This optional field lets you specify if an option for your connection needs to load a file relative to your config directory. For example, if you had a connection that needed to use a cert, you could add query.sslrootcert to this list, set "query": { "sslmode": "verify-ca", "sslrootcert": "certs/root.crt"}, and drop the cert in $SQL_CONNECTORS_CONFIG_DIR/certs/root.crt.
default_env (string)
This optional field lets you specify which environment should be used by default. If not included, it will use default.
default_schema (string)
This optional field lets you specify which schema should be used by default. If not included, it will use None.
default_reflect (boolean)
This optional field lets you specify whether it should reflect the data source by default. If not included, it will use False.
env.username (string)
This optional field specifies the username for the connection. If it's left out or set to null and the driver is not 'sqlite', the user will be prompte when they try to create the client. If the connection doesn't have credentials, set this to an empty string. Should not be set for 'sqlite'.
env.password (string)
This optional field specifies the password for the connection. If it's left out or set to null and the driver is not 'sqlite', the user will be prompte when they try to create the client. If the connection doesn't have credentials, set this to an empty string. Should not be set for 'sqlite'.
env.host (string)
This optional field specifies the host for the connection. Should not be set for 'sqlite'.
env.port (string or integer)
This optional field specifies the port for the connection. Should not be set for 'sqlite'.
env.database (string)
This optional field specifies the database name for the connection. If it's a 'sqlite' connection and left empty, it will use :memory:. Otherwise, you can specify a relative path or an absolute path; if you want the file in your config directory, you can use the relative_paths property.
env.query (object)
This optional field is a json object with options to pass onto the dialect and/or DBAPI upon connect.
env.allowed_hosts (list of strings)
This optional field is a list of strings containing hostnames where the given credentials are accepted. If the hostname is not in the list, it will prompt the user for credentials. This was added due to some specific usecase where we share service credentials but they're only allowed on our common servers.

How-To

The module will check your available connection configurations and create variables within the top level module for each of them. It will create 2 variables for each config, connection_name and connection_name_envs; these are both functions, the first will return a get_client function with some defaults set based on the config, and the second will return a get_available_envs function that when called returns available environments for the given data source. When reflection is enabled, the client will hold metadata about the available tables.

Here's a basic usage example assuming the example config file exists:

from sql_connectors import connections
client = connections.example_connection()
client.read_sql('select 1')

Here's a more complex example that's pretty redundant but shows more functionality

from sql_connectors import connections

available_envs = connections.example_connection_envs()
client = connections.example_connection(env=available_envs[0], reflect=True)

client.read_sql('select 1').to_sql('example_table', client, if_exists='replace')
available_tables = client.table_names()
table1 = client.get_table(available_tables[0])
df = client.read_sql(table1.select())

Credits

This package was created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.