User:Bukzor
My Favorite Languages | |
Language | Proficiency |
Python | Expert |
UNIX Shell | Very Active |
SQL | Active |
JavaScript | Semi-Active |
C | Rusty |
Perl | Hacker/Hater |
C++ | Rusty |
Right now I'm just having fun improving other people's python. That's probably not what this is all about, but I like it.
Automatic pylint
The current state of this project can be found here: rosetta_pylint.py
- use the MediaWiki API to grab a list of the pages in Category:Python
- The mediawiki API is pretty straightforward. I feel done with that part.
- grab the HTML for those pages, put them into a DOM
- I'm having trouble getting any of the builtin html or xml parsers to give me a DOM. htmlparser is just a ghetto little state machine, and the xml parsers are too strict ( is an 'unknown entity').
- I've posted a stackoverflow question on this subject here. --Bukzor 16:31, 20 April 2010 (UTC)
- Despite everyone agreeing that Python doesn't have a builtin HTML->DOM parser, I've parsed the site A-Z with ElementTree with minimal effort. I had to fix a bunch of inavalid HTML though. Look at my edits for the previous couple days for details.
- select for "python" as a CSS class, and get lumps of Python code.
- Now I have ~700 python snippets that I'm working on pylint'ing and analyzing. --Bukzor 01:29, 24 April 2010 (UTC)
- I note that plenty of the Python solutions are using plain <pre> tags which are being skipped in my current scheme. I'll have to add some code to detect this...
- automate feeding that code through pylint
- The current Ubunut pylint (0.18) throws up on 'import curses' for unknown reasons, but installing the latest version (0.20) allows me to pylint all of the scraped snippets as a whole. Collectively, they're rated at -1.58/10 (that's negative). I hope to get that up to 10/10 someday. --Bukzor 05:02, 26 April 2010 (UTC)
- save a report of pages->scores
--Bukzor 05:02, 26 April 2010 (UTC)