Guaka! - Python

code

Guaka — Mon, 01 Oct 2007 22:28:36 +0000

See my profile at github.

Some projects that I've contributed code to, available under free software licenses:

trustlet, analysis of trust metrics, Python, NetworkX, numpy
wik2dict, converter of MediaWiki dumps into the DICT format, currently in the trustlet codebase, Python
BeWelcome, hospitality exchange software, PHP, MySQL, XHTML, CSS
Google Calendar plugin for MediaWiki, PHP

Small patches:

CheeseTracker, audio software, C++, JACK
bio2jack

I've been involved in finding bugs, usability issues and new features for the following projects:

I've also worked on code for the following companies and organizations:

CouchSurfing.com (PHP, MySQL, CSS), I started developing in August 2006, right after the crash, I was the CS tech team coordinator from autumn 2006 till January 2007 and I remained core developer until May 2007.
structural analysis of audio from musical pieces (Python, numarray), for IRCAM
custom GNU Pascal libraries
ActionApps data migration (PHP on Windows and GNU/Linux) for CEPES in Peru
accessible.rotterdam.nl and more (ASP, MS SQL Server), for PPC

Recursively create lowercase symlinks to filenames with uppercase letters - in Python

Guaka — Tue, 04 May 2010 17:51:54 +0000

tag:

I'm working on a project to convert a big ASP website into Drupal. On Windows OSes there is not really a distinction between upper and lower case characters in filenames. At first I thought to just leave the capitals, but on web pages links were sometimes with capitals and sometimes lowercase. So I added some stuff to the Apache configuration, usually inside the VirtualHost directive (this does not work in .htaccess):

        
        RewriteEngine on
        RewriteMap lc int:tolower
        RewriteCond %{REQUEST_URI} [A-Z]
        RewriteRule (.*) ${lc:$1} [R=301,L]

But that wasn't all. For the project I'm parsing thousands of ASP files and the plan is to do this everyday while the existing website is updated. We're rsyncing files to be parsed, and the recursively rename to lowercase script I found is nice but it's not very efficient to copy and rename almost 3 GB everyday. I wrote a little Python script in the train today to create lowercase symbolic links to file and directory names with uppercase letters. The hardest part was to properly change directories - it's important to change into the directory the symlink is created but at the end of an os.walk it's necessary to get back to the current working directory.

I'm sure it can be useful to other people dealing with similar issues...

lowlink.py

#!/usr/bin/env python
"""
lowlink.py

Recursively creates lower case symlinks to filenames with uppercase letters.

Created by Guaka in 2010, consider this public domain, copy and re-use freely.
"""

import os

def lowlink(path = '.'):
    cwd = os.getcwd()  # os.walk can't handle path changes very well 
    for root, dirs, files in os.walk(path):
        # change directory for symlink creation
        os.chdir(os.path.join(cwd, root))
        print root
        for name in files + dirs:
            lowname = name.lower()
            if name != lowname and not os.path.exists(lowname):
                os.symlink(name, lowname)
        os.chdir(cwd) # restore path for os.walk

lowlink()

wik2dict

Guaka — Sat, 06 Mar 2010 00:54:52 +0000

wik2dict is a tool I wrote in Python. It converts MediaWiki SQL dumps into the DICT format.

I'm not actively maintaining this anymore, Aard Dictionary seems the way to go.

Google's issue tracker is not great

Guaka — Wed, 10 Oct 2007 19:55:43 +0000

tag:

We're using Google Code for our trust metrics project, out of sheer convenience, and because of lack of political correctness. Well, I've given up on worrying about privacy a while ago anyway. I just tried to use the issue tracker that comes with Google Code, but first I got a time-out, there were two identical issues because I refreshed. And on top of that, even though I am administrator of the project, I can not delete it. So I guess I'm going to install Trac in the near future and forget about Google Code.

numpy documentation

Guaka — Thu, 04 Oct 2007 00:08:39 +0000

tag:

With trustlet, I'm finally getting to a point where we can do cool stuff, evaluating prediction graphs of trust metrics. Now I need to refresh my Python array stuff. Since last time I was using that stuff, in 2004, they have refactored it (once more) and called it numpy. That's fine. But, in order to read an important part of the documentation you need to pay 39.99 US$. Now, that's less and less by the day, but still, it's very inconvenient, and probably holding back wider acceptation of Python in the scientific community. Like software, documentation should be free. I might even pay if the author would have chosen to use the Creative Commons ShareAlike NonCommercial Attribution license - but then I would be able to pass on the information to my friends (even they are all over the internet). I think it would have been a wiser choice to find a job at some institute or company to further the development of numpy and scipy, and its documentation in a freer way. That shouldn't be too hard these days.