bpk_candidates.py

#! /usr/bin/python -u

# @author apit - http://fizdoonk.multiply.com/
# output: http://fizdoonk.blogspot.com/2009/12/not-oldest-one.html

import re
from lxml.html import parse
from GChartWrapper import *

def display_pie(stat, title=None):
    G = Pie3D(stat.values())
    G.label(*stat.keys())
    if title:
        G.title(title, 'cccccc', 36)
    G.color('3072F3','red','00aaaa')
    G.size((500, 300))
    print "<img src='%s' />" % G

def parse_page(url):
    """
    """
    try:
        doc = parse(url).getroot()
    except IOError:
        print "Failed to fetch the page. Check your connection."
        exit(-1)
    except:
        raise

    ages = {}
    sexes = {'P': 0, 'W': 0}
    for row in doc.cssselect('#AutoNumber10 table tr')[1:]:
        row = row.cssselect('td')
        id = re.sub(r'[\.\,]', '', row[2].text_content().strip())
        name = row[3].text_content().strip().title()
        sex = row[4].text_content().strip()
        m = re.findall(r'[7-8][0-9]', id[9:12])
        if m:
            age =  2009 - (1900 + int(m[0]))
            ages[age] = (ages[age] + 1) if ages.has_key(age) else 1
        else:
            pass #print id
        #print "%s: %d [%s]" % (name, age, sex)
        sexes[sex] += 1

    return sexes, ages


sexes, ages = parse_page('http://www.cpns.bpk.go.id/home/detpesertalolos.php?j=S1%20Teknik%20Informatika/Komputer&kd=TK')
display_pie(sexes, 'Sex')
display_pie(ages, 'Age')