pddb 0.3.3

Introduction

PandasDatabase is a RESTful database engine application built on top of Pandas. Essentially, it is an abstraction layer that projects the database-table-column model into a very simple set of API’s. As a database engine, it has some useful features that make it a good candidate for prototype work:

• Inherits all the performance and robustness of Pandas.

• Very simple and intuitive API set.

• Tables support dynamic schema, so every time columns are changed during development there is no need to alter tables or CREATE statements.

• All data is persisted in plaintext, human-readable CSV format.

Some of those features come at a cost that probably makes PandasDatabase less than ideal for production environments:

• Security. At the server, the security model is based on file permissions. For the API’s, production environments should very likely never expose database API’s of any form.

• Performance. While low latency production environments might not run into an issue, performance-critical applications will probably run into a bottleneck when writing to disk.

• Data types. Exposing all the table data in CSV format means that complex data types such as date cannot be supported.

The Problem

A very large number of small projects have fairly simple requirements for data storage. For all those projects, interfacing with a database engine is boilerplate work that adds unnecessary overhead during development. This project provides a very simple yet powerful solution that is great for prototype work to enable those small projects to hit the ground running. Once the critical components of the project are finished and a proof of concept version is running, projects can transition to a more mature database engine that can better suit the needs of a production environment.

Getting Started

This project is entirely Python based. To be able to use it, first install the dependencies:

$ pip install pandas bottle

The easiest way to install PandasDatabase is using pip:

$ pip install pddb

To fire up the database engine, simply run:

$ python -m pddb.pddb dbname --permissions w

By default, the database is started in read-only mode, which is why we need to pass the --permissions w flag. This should start a bottle application with the following endpoints available:

• /pddb

• /pddb/find/<table>

• /pddb/insert/<table>

• /pddb/upsert/<table>

• /pddb/delete/<table>

The parameters to those endpoints can be passed as a GET query string, or via POST. For example, to insert a new record, the user can simply visit the following URL once the database engine is running:

http://127.0.0.1:8080/pddb/insert/table_name?Name=John

Likewise, the user can find the inserted record by visiting:

http://127.0.0.1:8080/pddb/find/table_name

Matching conditions can also be added:

http://127.0.0.1:8080/pddb/find/table_name?Name=John

Performing an update is a little more complicated. Rather than exposing multiple API’s, a single API is used and the parameters are parsed to understand the user’s desired operation. So, instead of using /pddb/upsert/table_name?column_name=column_value, the user must use /pddb/upsert/table_name?record__column_name=column_value&where__condition_name=condition_value. Essentially, prepend record__ or where__ to let the database engine know which pair of key-value corresponds to what parameter. For example, to change the name John to Jane in our record, we can simply visit:

http://127.0.0.1:8080/pddb/upsert/table_name?record__Name=Jane&where__Name=John

Note that this also applies to the rest of the API’s, even though the parameters being parsed default to the most obvious choice. For example, /pddb/find assumes that the parameters in the query string correspond to the equivalent of a WHERE in SQL. However, the find query can also be written as:

http://127.0.0.1:8080/pddb/find/table_name?where__Name=John

The usefulness of this does not appear evident until more complex queries are used, such as WHERE-NOT:

http://127.0.0.1:8080/pddb/find/table_name?where_not__Name=John

While admittedly a bit quirky, these very simple API’s allow for any application in any language to interface with the database engine by performing very simple GET requests. The user does not need to worry about exposing an API or interfacing with a database in another process or server, which gives more time to developing the critical parts of the project first.

License

Copyright (c) 2016 Oscar Martinez All rights reserved.

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the “Software”), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

0.1.0 (2016-01-10)

• First release on PyPI.

0.1.1 (2016-01-10)

• Fix case sensitive tests

• Disable testing for all versions except 2.7, 3.3 and 3.4

• Update docs

0.2.0 (2016-01-12)

• Do not attempt to drop table if it does not exist

• Improved Windows compatibility

0.3.0 (2016-01-17)

• Delete database folder only if empty after dropping all tables

• Improved Windows compatibility

0.3.1 (2016-01-18)

• Added auto_cast option

• String type enforcing enabled

• Unicode and byte type support

0.3.2 (2016-02-14)

• Added ability to use custom bottle route methods

0.3.3 (2016-02-15)

• Delete CSV files if dataframe is empty when saving

• Expose tostr() method as a public method