[pyblosxom-users] New tumblelog plugin

[chombee <ch...@la...>]

I rewrote my tumblelog plugin:

http://github.com/seanh/PyBlosxom-tumblelog

I wasn't happy with the YAML-based approach for several reasons. It now
registers several entry parser functions (one for each type of entry).
The parser functions define custom file formats for each type of entry
and do custom processing (so far passing bodies, captions and the like
through markdown if it's available, but other types of processing for
specific entry types are possible). The file formats are simple and try
to emulate pyblosxom's default format. The idea was to make the file
format as close as possible to how you'd write if you're were just
noting something down to yourself in a text file, or emailing it to
someone, it should be natural and with as little as possible to
remember, avoid having to look up an example of the file format when you
want to write an entry. For example, the file format of a link entry is
like this:

    The link title
    http://www.example.com
    The rest of the file constitutes a multi-line
    caption for the entry.

When this is rendered with the link template you get something like:

    <div class="entry link">
        <a href="http://example.com" alt="The link title" title="The link title>
            The link title
        </a>
        <p>The rest of the file constitutes a multi-line
        caption for the entry.</p>
        <div class="date">Tue 23 Feb 2010</div>
    </div>

As with the previous version, each entry type is rendered using its own
template instead of the default story template.

I'm much happier with this approach. When you look at your entries dir
you can see what type of entry each file is from the filename extension:
link, .quote, .picture, .text etc. instead of them all being .yaml.
Entries should be easier to write in these custom file formats than in
YAML, which was an order of magnitude more awkward than standard
pyblosxom entries. And with a separate parse function in tumblelog.py
for each entry type, flexible per-type processing is possible.