Subversion Repositories SmartDukaan

'''

Created on 24-Aug-2011

@author: Varun Gupta

'''

from BeautifulSoup import BeautifulSoup

from BaseScraper import BaseScraper

class InfibeamScraper(BaseScraper):

    def __init__(self):

        self.url = None

        self.id = None

    def setUrl(self, url):

        self.url = url

    def scrape(self):

        html = BaseScraper.read(self, self.url)

        self.soup = BeautifulSoup(html)

    def getPhones(self):

        phone_prices = []

        for li in self.soup.findAll('ul', {'class': 'srch_result portrait'})[0]('li'):

            name = li.findAll('span', {'class': 'title'})[0].string

            try:

                price = li.findAll('div', {'class': 'price'})[0].findAll('span', {'class': 'normal'})[0].string

            except IndexError:

                price = li.findAll('span', {'class': 'price'})[0].contents[-1].strip()

            url = li.findAll('a')[0]['href']

            try:

                phone_prices.append({'name': str(name), 'price': str(price), 'in_stock': 1, 'product_url': str(url)})

            except UnicodeEncodeError as e:

                print 'Unicode Error', e, name

                name_ascii = "".join([char if ord(char) < 128 else " " for char in name])

                print name_ascii

                phone_prices.append({"name": str(name_ascii), "price": str(price), "in_stock": 1, "product_url": str(url)})

        return phone_prices

    def getNextUrl(self):

        b = self.soup.findAll('div', {'class': 'resultsSummary'})[0].findAll('b')

        current_max = int(b[0].string.split('-')[1])

        total_products = int(b[1].string)

        return 'http://www.infibeam.com/Mobiles/search?page=%d' % (1 + current_max / 20) if current_max < total_products else None

if __name__ == '__main__':

    s = InfibeamScraper()

    s.setUrl('http://www.infibeam.com/Mobiles/search?page=17')

    s.scrape()

    products = s.getPhones()

    print products