res-manager 2.0.0

This tool helps you to manage your experimental results in a more elegant way, saving your experiment data by pickle with some high-level interfaces to save, load and preview.

This project helps you to handle your experimental results efficiently when doing your research. Basically it uses pickle and sqlite3 to save, load and manage the data, now we provide you with some high-level interfaces so that you can get rid of pickle dumping and loading from disk all the time.

Since it is based on pickle, in theory you can use it to manage all types of data in python, you can even save an object as shown in the following examples.

Usage

Just start with from res_manager import ResultManager, now you can create an object with rm = ResultManager(path) and then use it to save, load, clear and preview your data.

Install ResultManager with pip install res-manager

>>> from res_manager import ResultManager
>>> rm = ResultManager('data')
>>> rm.save_data([1, 2, 3], topic='test saving', name='data1', commit_comment='Test saving a list')
>>> rm.save_data(65535, topic='test saving', commit_comment='Test saving a number without a name')
>>> rm.save_data(rm, topic='topic 2', commit_comment='Test saving an object')
>>> rm.save_data({0:1,1:'string'}, name='dict_without_topic')
>>> rm.print_meta_info()
...
>>> rm.load_data_by_id(5)
<res_manager.manager.ResultManager object at 0x7f29c8587470>
>>> rm.load_data_by_name('data1')
[1, 2, 3]
>>> rm.load_data_by_name('dict_without_topic')
{0: 1, 1: 'string'}
>>> rm.delete_data_by_id(3)

ResultManager

This project mainly provides:

• A class named ResultManager that provides functions to save, load, clear and preview data.

• ResultManager.save_data: Save your data.

• ResultManager.load_data_by_name and ResultManager.load_data_by_id: Loading data by name or ID.

• ResultManager.print_meta_info Print all meta info of saved data, including name, path, topic, comments etc.

• ResultManager.delete_data_by_id Delete data from yor storage.

Future Plan

• Adding support to save and load figures for each piece of data

• Add more interfaces

Authors

• Author and lead developer: Hongshan Li

Documentation

This project is still simple, therefore no docs…

Requirements

• prettytable>=0.7.2

Note

ResultManager is developed by Python 3.5 and only tested on Python 3.5 and Python 2.7