Plugin development

For extensive customizability, all the different stages of the document generation are implemented as plugins with a common interface. Usually, new plugins will only require the implementation of one or more methods.

Besides this implementation, the plugins must be registered. This library uses python entry points to discover new plugins: basically, each plugin (or group of plugins) should be packaged using setuptools and call the setup function with the appropriate entry_points arguments.

Entry points are defined by three values: group, name and object reference.

The groups currently defined represent each one of the different pipeline stages:

• reportcompiler.data_fetchers
• reportcompiler.source_parsers
• reportcompiler.template_renderers
• reportcompiler.postprocessors

Each group can have several entry points defined as a list. The entry point name chosen for the plugin will be the one later referenced on each stage declaration (‘type’ parameter in Plugin modules).

Finally, the object reference should be the module path and the class within the module implementing the interface, both parts separated by a colon.

For example, the HPV Information Centre MySQL plugin has the following parameter in the setup() function:

entry_points={
    'reportcompiler.data_fetchers': [
        'mysql_ic=reportcompiler_ic_fetcher.mysql_ic:MySQLICFetcher',
    ],
}

, and, when used as a data fetcher in a fragment, has the following declaration:

data_fetchers = [
        {
            'name': 'nw',
            'credentials': 'information_center',
            'type': 'mysql_ic',
            'missing_string': '-',
            'table': 'data_m2_cervical_incidence_ci5c',
            'fields': odict[
                'r.area_name': 'country_name',
                'M2ICI1009': 'NUM',
                'M2ICI1006': 'CRUDE',
                'M2ICI1011': 'ASR',
            ],
            'condition': {
                'M2ICI1000': 'iso'
            }
        }
]