Let me announce 'metatag', a simple metadata extraction utility.

The Design Idea is to make it completely event driven. There are string matching rules over added metadata, who invoke engines when matched, which in turn add more metadata and so on. Thus the whole metadata extraction process is controlled by those easily configurable rules. Processing a file or directory just starts by adding "/=filename" to the metadata, everything else bootstraps from that. After metadata got extracted there are exporters which implement different backends for storing this metadata (currently only a 'print' and a 'gitannex' exporter are implemented)

While still in a infancy state it already works for me. It now needs more rules and engines for metadata extraction and some more efforts to 'standardize' generated metadata. I'd like to welcome comments and contributions.

A README about it can be found at


The code is available under git from

git clone git://git.pipapo.org/metatag

To make the contribution barrier as low as possible there is a public pushable 'mob' repository where everyone can send changes too at git://git.pipapo.org/mob/metatag

after installing it, using it on a annexed directory is like

metatag -r -O gitannex,gitexclude -o gitannex:-stat ./

There is a mailinglist for the project, you can subscribe at