Posted on Mi 10 August 2005

Linking pyblosxom to SVN

If you run a pyblosxom blog with auto-copied stories from SVN you are probably interested in getting stable story dates that don't change every time you update a story. The date of the initial SVN log entry of a story is something like the "day of birth" of a story, so it's a good value to use. Christopher Baus implemented a plugin for pyblosxom, which looks overly complicated to me: it depends on memcached and comes in two large python scripts.

To simplify things I wrote this minimal replacement:

import pysvn, os, sys, anydbm

from config import py

def get_mtime(fname):
        cache_fname = os.path.join(py['datadir'], 'SVNDATES')
        cache = anydbm.open(cache_fname, "c")

        if cache.has_key(fname):
                d = float(cache[fname])
        else:
                client = pysvn.Client(fname)
                l = client.log(fname)

                if len(l) > 0:
                        d = l[0]['date']
                        cache[fname] = str(d)
                else:
                        d = -1

                del client

        del cache
        return d

def cb_filestat(args):
        args["mtime"] = list(args["mtime"])
        d = get_mtime(args["filename"])
        if d >= 0:
                args["mtime"][8] = d
        return args

Since accessing SVN logs is quite slow the script caches the "date of birth" in a dbm file. Make sure that your web server has enough priviliges to access that database file which is stored in $datadir/SVNDATES by default.

© Lennart Poettering. Built using Pelican. Theme by Giulio Fidente on github. .