User:Jasper van Loenen/Prototyping/wikistats

Wikistats

This script keeps track of all changes on the wiki and generates a list of most active users. The script uses python to read the wiki rss feed, saves the changes to a SQLite database and writes the stats to a .txt file which can be found here. This file is automatically updated.

import urllib2
import lxml.etree
from sqlite3 import *

#initate mysqlite database
connection = connect('/home/jvloenen/public_html/cgi-bin/feed/wikitracker.db')
cursor = connection.cursor()
#create tables - run once
#cursor.execute('''CREATE TABLE wiki_users (id integer primary key, user_name text, update_times integer, update_amount integer)''')
#cursor.execute('''CREATE TABLE wiki_updates (id integer primary key, user_id integer, update_date text)''')
#connection.commit()

rssFeedUrl = "http://pzwart3.wdka.hro.nl/mediawiki/index.php?title=Special:RecentChanges&feed=rss"

#open rss feed
print "Getting feed"
f = urllib2.urlopen(rssFeedUrl)
print "Parsing"
doc = lxml.etree.parse(f)    

for item in doc.xpath("/rss/channel/item"):
    title = item.xpath("title/text()")[0]
    description = item.xpath("description/text()")[0]
    descriptionLength = len(description)
    pubDate = item.xpath("pubDate/text()")[0]
    userName = item.xpath("*[local-name()='creator']/text()")[0]
    cursor.execute("SELECT id FROM wiki_users WHERE user_name = ?", [userName])
    data=cursor.fetchall()
    if len(data)==0:
        print('There is no user %s, so he/she will be added'%userName)
        cursor.execute("INSERT INTO wiki_users (user_name,  update_times, update_amount) values (?, ?, ?)", (userName, 1, descriptionLength))
        cursor.execute("SELECT last_insert_rowid() FROM wiki_users")
        data=cursor.fetchall() #overrules the previous value of 'data'
        userId = data[0][0]
        new_update_time = 1
        new_update_amount = descriptionLength
    else:
        userId = data[0][0]
        cursor.execute("SELECT update_times, update_amount FROM wiki_users WHERE id = ?", ([userId]))
        data=cursor.fetchall()
        new_update_time = data[0][0]+1
        new_update_amount = data[0][1]+descriptionLength    

    cursor.execute("SELECT id FROM wiki_updates WHERE user_id = ? AND update_date = ?", (userId, pubDate))
    data=cursor.fetchall()
    if len(data)==0:
        print "Saving new entry for user "+str(userId)
        cursor.execute("INSERT INTO wiki_updates (user_id, update_date) values (?, ?)", (userId, pubDate))
        cursor.execute("UPDATE wiki_users SET update_times =?, update_amount = ? WHERE id = ?", (new_update_time, new_update_amount, userId))
        connection.commit()

print "Done"