Skip to content

Latest commit

 

History

History
85 lines (66 loc) · 3.06 KB

README.md

File metadata and controls

85 lines (66 loc) · 3.06 KB
Raw

DataProcessor

Build Status

A data processing library.

Data processing

This library collects and processes information of "run"s.

node

If you are a scientist, you may have huge data which are almost same but slightly different. We call each of these data as "run", and series of them as "project". Since these runs and projects will be related each others, we can consider a network of them. Thus we call each of runs and projects as node. We assume that each of runs and projects has their directory, and the path of the directory is unique. So the identifier of node is the path of its directory. In this library node denotes meta data for each run or project. Typically, node contains path, name, and paths of connected nodes.

pipe

This library introduce pipe corresponding to a single manipulation for nodes; since this library manages nodes by a list of node (called node_list), pipe do a single manipulation for node_list. For example,

  • "scan_directory" pipe gathers runs or projects and appends it into the node_list.
  • "add_comment" pipe add some comments to a specified node.
  • "configure" pipe collects information of parameters from each run directory.

You can easily combine them to satisfy your purpose. We call combined pipes as manipulations, and specified in JSON format. Details of pipes is documented in pipes list.

manipulations

You can execute manipulations with an executable script bin/dataprocessor. You must specify manipulations by a JSON file. When it is written in manipulations.json, you can do manipulations by the following command:

$ bin/dataprocessor manipulations.json

The following JSON executes first pipe-name1, next pipe-name2. pipe's arguments are specified in args key.(THE ORDER OF ARGS IS IMPORTANT) pipe's optional arguments(keywords) are specified in kwds key. Currently supported pipe are listed in here.

[
    {"name": "pipe-name1", "args": ["argument1", "argument2"], "kwds": {"keywords1": "some-value"}},
    {"name": "pipe-name2", "args": ["argument1"]}
]

This JSON file is in sample dir.

If you add a comment in node_list, you should want to save it. We recommend that you save it by JSON format. There exist JSON saver and loader pipes.

WebApp

You can browse your data managed with DataProcessor by a DataProcessor webapp. In order to edit data through this webapp, a HTTP server is necessary. This project contains a script bin/server.py which start/stop a simple HTTP server. The usage of this script is also written in sample.

Requirements

  • python-daemon (>= 1.6) (for stand a HTTP server as a daemon)
  • jinja2 (for converting template)

For Developer

If you want to develop this tools, please read Developer Guide.

Lisence

GPLv3