dailycrawler.py

#!/usr/bin/env python

'''
Autor: Cristian Carpenter G.
'''

from BeautifulSoup import BeautifulSoup
import urllib, re
import MySQLdb
from datetime import datetime

def number(string):
    foo = string.replace('.', '')
    foo = foo.replace(',', '.')
    foo = foo.replace(' ', '')
    return foo

def clean(string):
    foo = string.replace('.', '')
    foo = foo.replace (' ', '')
    return foo

db = MySQLdb.Connect(user="root", passwd="******", db="ipsa")
stockdata = db.cursor()
html = urllib.urlopen("http://www.bolsadesantiago.com/web/bcs/preciosacciones").read()
soup = BeautifulSoup(html)
prices = soup.findAll(attrs={'class': re.compile("bcs-portlet-section")})

for i in range(1, 41):
    symbol = prices[i].findAll(attrs={"class": "col-1"})[0]('a')[0].find(text=True)
    price  = prices[i].findAll(attrs={"class": "col-2"})[0].find(text=True)
    volume = prices[i].findAll(attrs={"class": "col-4"})[0].find(text=True)
    price  = number(price)
    volume = clean(volume)
    date   = datetime.today()
    date   = date.date()
    stockdata.execute("""INSERT INTO acciones (symbol, price, volume, date) VALUES (%s, %s, %s, %s)""", (symbol, price, volume, date))
    db.commit()