Subversion Repositories SmartDukaan

Rev

Rev 6024 | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
5639 amar.kumar 1 
from BeautifulSoup import BeautifulSoup
2 
from BaseScraper import BaseScraper
3 
from Utils import removePriceFormatting
4 
 
5 
import time
6 
 
7 
class TradusScraper(BaseScraper):
8 
 
5761 amar.kumar 9 
    mobilePageCount = 32
10 
    tabletPageCount = 21
5639 amar.kumar 11 
    productCountPerScraping = 20
5761 amar.kumar 12 
    mobileCurrentPage = 0
13 
    tabletCurrentPage = 0
5639 amar.kumar 14 
 
15 
 
16 
    def __init__(self):
17 
        BaseScraper.__init__(self)
18 
        self.url = None
19 
        self.id = None
20 
 
21 
    def setUrl(self, url):
22 
        self.url = url
23 
 
24 
    def scrape(self):
25 
        html = BaseScraper.read(self, self.url)
26 
        self.soup = BeautifulSoup(html)
27 
        self.phones = None
28 
        #self.setPageCount()
29 
 
30 
    def getPhones(self):
31 
        phones = []
6022 amar.kumar 32 
        for div in self.soup.findAll('div', {'class': 'prod_main_div'}):
5639 amar.kumar 33 
            try:
6022 amar.kumar 34 
                productUrlContainer = div.find('div', {'class': 'product_name search-product-block'})
35 
                name = productUrlContainer.contents[1].string
6024 amar.kumar 36 
                product_url = "http://www.tradus.com" +productUrlContainer.contents[1]['href']
6022 amar.kumar 37 
                price = div.find('span', {'class':'numDiv_left'}).string.strip()
38 
                price = removePriceFormatting(price)
5639 amar.kumar 39 
                in_stock = 1
40 
 
41 
                try:
42 
                    if price is None:
43 
                        continue
44 
                    else:
45 
                        phones.append({
46 
                                'name': str(name),
47 
                                'price': removePriceFormatting(price),
48 
                                'source': 'tradus',
49 
                                'product_url': str(product_url),
50 
                                'in_stock': in_stock
51 
                            })
52 
                except Exception as e:
53 
                    print e
54 
                    pass
55 
 
56 
            except IndexError as iex:
57 
                try:
58 
                    price = div.find('div', {'class': 'mainresult-show-right-startrate'})('span')[0].contents[0].strip()[18:]
59 
                    in_stock = 1
60 
 
61 
                    phones.append({
62 
                                'name': str(name),
63 
                                'price': removePriceFormatting(price),
64 
                                'source': 'tradus',
65 
                                'product_url': str(product_url),
66 
                                'in_stock': in_stock
67 
                            })
68 
 
69 
                except Exception as ex:
70 
                    print ex
71 
                    pass
72 
            except Exception as e:
73 
                print e
74 
                pass
5761 amar.kumar 75 
        self.phones = phones
76 
        return phones
5639 amar.kumar 77 
 
78 
    def setPageCount(self):
79 
        self.currentPage = self.currentPage + 1
80 
 
81 
    def getNextUrl(self):
82 
        time.sleep(1)
6166 amar.kumar 83 
        nextLink = self.soup.find('a', {'class': 'filter-link more'})
84 
        if(nextLink is not None):
85 
            urlDivider = self.url.find('page=')
86 
            if urlDivider is not -1:
87 
                urlDivider +=5
88 
                pageNumber = int(self.url[urlDivider:]) +1
89 
                return self.url[:urlDivider] + str(pageNumber)
5761 amar.kumar 90 
            else:
91 
                return None
5639 amar.kumar 92 
        else:
93 
            return None
5761 amar.kumar 94 
 
5639 amar.kumar 95 
 
96 
    def getDataFromProductPage(self, url):
97 
        html = BaseScraper.read(self, url)
98 
        soup = BeautifulSoup(html)
6022 amar.kumar 99 
        name = soup.find('span',{'itemprop':'name'}).string.strip()
100 
        price= soup.find('span',{'class':'mrp3'}).contents[0].strip()
101 
        price = removePriceFormatting(price)
5639 amar.kumar 102 
        in_stock = 1
103 
 
104 
        data = {
105 
            "product_url": str(url),
106 
            "source": "tradus",
107 
            "price": price,
108 
            "in_stock": 1,
109 
            "name": name
110 
        }
111 
        return data
112 
 
6022 amar.kumar 113 
def removePriceFormatting(price_string):
114 
    return price_string.strip().replace('Rs.', '').replace('Rs', '').replace(',', '').replace(' ', '').replace(' ', '').split('.')[0]
115 
 
5639 amar.kumar 116 
if __name__ == '__main__':
117 
    s = TradusScraper()
6022 amar.kumar 118 
    data = s.getDataFromProductPage('http://www.tradus.com/zing-q800-dual-sim-mobile-phone/p/MOB0000004506663')
5639 amar.kumar 119 
    print data