collective.feedfiller 0.1dev-r77073

Feedfiller

What is it?

Feedfiller is intended to work alongside collective.feedfeeder, and fill all the news feed items with the clean body content of the page they refer to. If it does not ‘know’ the page format of the target page, all it can do is include the whole page. We will improve this as the project develops.

Clearly there are potential copyright issues with the re-publishing of copyrighted works. But for research and analysis purposes, these may not be an issue for your organization. Our own purpose is to use collected text for classification and analysis for internal use. You should seek your own legal advice on this topic.

How does it work?

Feedfiller subscribes to the event created after storage of each news feed item created by FeedFeeder and fetches the target page of that item. This means that all items will be be filled with the content of the page they refer to. Fetched pages are flayed (“Flay: Verb: to strip off the skin or surface of”) by a Flayer looked up in a FlayerRegistry by URL.

Flayers may be easily written to accomodate new pages. Flayers can be created and registered for different sections of a site, in case HTML structure varies in sub-trees of the site.

If no flayer is registered for the URL, a default flayer is used that returns the whole body of the page.

Currently site-specific flayers try to reveal author, copyright, and body, but the default flayer

The flayer base-class currently stores the original page fetched from the server, to facilitate further development and refinemement of flayers without repeatedly fetching content.

TODO

The next step is to develop a table-driven flayer, for which table entries can be generated interactively by clicking on an enhanced version of the default flay, a bit like a basic firebug view of the structure of a page with buttons to manually select the body area of a page. This will rmoyrequire a new view for this purpose, available to managers.

Eventually the table-drive flayer should be able to handle the complexity of the BBC news page.

CREDITS

The project was started by Russ Ferriday, Topia Systems Ltd, in November, 2008.

Thanks to Zest Software and the van Rees brothers for FeedFeeder.

Contributions are welcome, and contributors are listed below:

Changelog

0.1 - Unreleased

• Initial release