1. Christophe Combelles
  2. atomisator




The big picture

Atomisator is a generic data processing framework that works with plugins. Its purpose is to provide an engine to build a data processing task.

  1. You provide the plugins that does the work.
  2. You describe how to combine them in a configuration file.
  3. Atomisator does the work.

Atomisator works in two steps:

  • Collecting: data is acquired and filtered, then stored.
  • Rendering: stored data is enhanced and rendered.
The big picture

The big picture.

Quick Start

Let's build a Planet with Atomisator !

To do it, create a configuration file somewhere, using the -c option:

$ atomisator -c atomisator.cfg

A default configuration file is generated, that looks like this (comments where removed):


sources =
    rss http://tarekziade.wordpress.com/atom
    rss http://digg.com/rss/index.xml

filters =

enhancers =

outputs =
    rss atomisator.xml http://atomisator.ziade.org/example "Meta feed" "Automatic feed created by Atomisator."

database = sqlite:///atomisator.db

This file can be interpreted by Atomisator, with the -f option:

$ atomisator -f /path/to/atomisator.cfg
Reading source http://tarekziade.wordpress.com/atom
10 entries read.
Reading source http://digg.com/rss/index.xml
40 entries read.
50 total.
Writing feed in atomisator.xml
Feed ready.

You can specify the path as a free argument as well:

$ atomisator /path/to/atomisator.cfg

Atomisator will read the sources, then generate an atomisator.xml file.

Two phases process

You can also call separately the process that fills the database:

$ atomisator -r -f /path/to/atomisator.cfg

In this case the feed generation will not occur.

The -g option can be used to generate the xml file on its own:

$ atomisator -g -f /path/to/atomisator.cfg

This is useful to read sources and generate the feed within different process and different time basis.

For other useful options, run:

$ atomisator --help

More details

As said earlier, Atomisator works in two steps:

  • Collecting: data is acquired and filtered, then stored.
  • Rendering: stored data is enhanced and rendered.

Collecting data

Collecting data is done by reading (1) and filtering (2) data. These two steps are done by invoking plugins.

For example, if you want to get some data from an RSS feed, you can use the rss plugin that is provided by default, and use it in an Atomisator configuration file:

sources =
    rss     http://tarekziade.wordpress.com/atom
    rss     http://digg.com/rss2.xml

This tells Atomisator to look for the rss reader plugin, and invoke it with the url as an argument. The plugin has to return entries.

The next step is to filter collected entries. Filtering means either changing the data on the fly, either discarding it.

For example, if you want to remove all entries that contains particular expressions, you can use the stopwords plugin that is provided by default and use it in an Atomisator configuration file:

filters =
    stopwords   word_list.txt

This tells atomisator to look for the stopwords plugin and to invoke it with collected entries. The plugin uses the word list in the pointed file, and look of the data match the given regular expressions. If so, the data is discarded.

Rendering data

After the data has been collected, it is stored in a database. You can now render it, using the enhancers (5) and the outputs (6) filters.

For example, if you want to see if an entry correspond to a web page that has been digged, you can use the digg plugin:

enhancers =

This plugin will be invoked by Atomisator, and it will inject Digg comments in the data by querying Digg.com via RPC.

Last the output plugins are called for the final step. For example the rss output plugin will render the data in an RSS 2.0 XML file:

outputs =
    rss atomisator.xml http://atomisator.ziade.org/example meta "Automatic feed created by Atomisator."

All the arguments following the plugin name will be sent to it, besides the entries.

Available plugins

XXX to be written

How to write a plugin

XXX to be written