| 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 |
|
|
|
9 |
pageCount = 67
|
|
|
10 |
productCountPerScraping = 20
|
|
|
11 |
currentPage = 0
|
|
|
12 |
|
|
|
13 |
|
|
|
14 |
def __init__(self):
|
|
|
15 |
BaseScraper.__init__(self)
|
|
|
16 |
self.url = None
|
|
|
17 |
self.id = None
|
|
|
18 |
|
|
|
19 |
def setUrl(self, url):
|
|
|
20 |
self.url = url
|
|
|
21 |
|
|
|
22 |
def scrape(self):
|
|
|
23 |
html = BaseScraper.read(self, self.url)
|
|
|
24 |
self.soup = BeautifulSoup(html)
|
|
|
25 |
self.phones = None
|
|
|
26 |
#self.setPageCount()
|
|
|
27 |
|
|
|
28 |
def getPhones(self):
|
|
|
29 |
phones = []
|
|
|
30 |
for div in self.soup.findAll('div', {'class': 'mainresult-show-right'}):
|
|
|
31 |
try:
|
|
|
32 |
anchor = div.find('a')
|
|
|
33 |
name = anchor.contents[2].strip()
|
|
|
34 |
if(len(name)== 0):
|
|
|
35 |
name = anchor.contents[0].strip()
|
|
|
36 |
product_url = anchor['href'].strip()
|
|
|
37 |
price = div.find('div', {'class': 'mainresult-show-right-startrate'})('span')[0]('span')[0].contents[0].strip()[3:]
|
|
|
38 |
in_stock = 1
|
|
|
39 |
|
|
|
40 |
try:
|
|
|
41 |
if price is None:
|
|
|
42 |
continue
|
|
|
43 |
else:
|
|
|
44 |
phones.append({
|
|
|
45 |
'name': str(name),
|
|
|
46 |
'price': removePriceFormatting(price),
|
|
|
47 |
'source': 'tradus',
|
|
|
48 |
'product_url': str(product_url),
|
|
|
49 |
'in_stock': in_stock
|
|
|
50 |
})
|
|
|
51 |
except Exception as e:
|
|
|
52 |
print e
|
|
|
53 |
pass
|
|
|
54 |
|
|
|
55 |
except IndexError as iex:
|
|
|
56 |
try:
|
|
|
57 |
price = div.find('div', {'class': 'mainresult-show-right-startrate'})('span')[0].contents[0].strip()[18:]
|
|
|
58 |
in_stock = 1
|
|
|
59 |
|
|
|
60 |
phones.append({
|
|
|
61 |
'name': str(name),
|
|
|
62 |
'price': removePriceFormatting(price),
|
|
|
63 |
'source': 'tradus',
|
|
|
64 |
'product_url': str(product_url),
|
|
|
65 |
'in_stock': in_stock
|
|
|
66 |
})
|
|
|
67 |
|
|
|
68 |
except Exception as ex:
|
|
|
69 |
print ex
|
|
|
70 |
pass
|
|
|
71 |
except Exception as e:
|
|
|
72 |
print e
|
|
|
73 |
pass
|
|
|
74 |
self.phones = phones
|
|
|
75 |
return phones
|
|
|
76 |
|
|
|
77 |
def setPageCount(self):
|
|
|
78 |
self.currentPage = self.currentPage + 1
|
|
|
79 |
|
|
|
80 |
def getNextUrl(self):
|
|
|
81 |
time.sleep(1)
|
|
|
82 |
if TradusScraper.currentPage < TradusScraper.pageCount:
|
|
|
83 |
TradusScraper.currentPage += 1
|
|
|
84 |
return 'http://www.tradus.com/search/tradus_search/?query=mobile&filters=cat:7756&page=%s' % (TradusScraper.currentPage)
|
|
|
85 |
else:
|
|
|
86 |
return None
|
|
|
87 |
|
|
|
88 |
def getDataFromProductPage(self, url):
|
|
|
89 |
html = BaseScraper.read(self, url)
|
|
|
90 |
soup = BeautifulSoup(html)
|
|
|
91 |
name = soup.find('h1',{'class': 'left-content-product-heading'}).string.strip()
|
|
|
92 |
price = soup.find('b', {'id': 'tPrice'}).string.strip()
|
|
|
93 |
in_stock = 1
|
|
|
94 |
|
|
|
95 |
data = {
|
|
|
96 |
"product_url": str(url),
|
|
|
97 |
"source": "tradus",
|
|
|
98 |
"price": price,
|
|
|
99 |
"in_stock": 1,
|
|
|
100 |
"name": name
|
|
|
101 |
}
|
|
|
102 |
return data
|
|
|
103 |
|
|
|
104 |
if __name__ == '__main__':
|
|
|
105 |
s = TradusScraper()
|
|
|
106 |
'''html = BaseScraper.read(s,'http://www.tradus.com/search/tradus_search/?query=mobile&filters=cat:7756')
|
|
|
107 |
soup = BeautifulSoup(html)
|
|
|
108 |
|
|
|
109 |
phones = []
|
|
|
110 |
for div in soup.findAll('div', {'class': 'mainresult-show-right'}):
|
|
|
111 |
try:
|
|
|
112 |
anchor = div.find('a')
|
|
|
113 |
name = anchor.contents[2].strip()
|
|
|
114 |
if(len(name)== 0):
|
|
|
115 |
name = anchor.contents[0].strip()
|
|
|
116 |
product_url = anchor['href'].strip()
|
|
|
117 |
price = div.find('div', {'class': 'mainresult-show-right-startrate'})('span')[0]('span')[0].contents[0].strip()[3:]
|
|
|
118 |
in_stock = 1
|
|
|
119 |
|
|
|
120 |
try:
|
|
|
121 |
if price is None:
|
|
|
122 |
continue
|
|
|
123 |
else:
|
|
|
124 |
phones.append({
|
|
|
125 |
'name': str(name),
|
|
|
126 |
'price': removePriceFormatting(price),
|
|
|
127 |
'source': 'tradus',
|
|
|
128 |
'product_url': str(product_url),
|
|
|
129 |
'in_stock': in_stock
|
|
|
130 |
})
|
|
|
131 |
except Exception as e:
|
|
|
132 |
print e
|
|
|
133 |
pass
|
|
|
134 |
|
|
|
135 |
except IndexError as iex:
|
|
|
136 |
try:
|
|
|
137 |
price = div.find('div', {'class': 'mainresult-show-right-startrate'})('span')[0].contents[0].strip()[18:]
|
|
|
138 |
in_stock = 1
|
|
|
139 |
|
|
|
140 |
phones.append({
|
|
|
141 |
'name': str(name),
|
|
|
142 |
'price': removePriceFormatting(price),
|
|
|
143 |
'source': 'tradus',
|
|
|
144 |
'product_url': str(product_url),
|
|
|
145 |
'in_stock': in_stock
|
|
|
146 |
})
|
|
|
147 |
|
|
|
148 |
except Exception as ex:
|
|
|
149 |
print ex
|
|
|
150 |
pass
|
|
|
151 |
except Exception as e:
|
|
|
152 |
print e
|
|
|
153 |
pass
|
|
|
154 |
print phones'''
|
|
|
155 |
|
|
|
156 |
|
|
|
157 |
data = s.getDataFromProductPage('http://www.tradus.com/samsung-galaxy-y-pro-duos-b5512-mobile-phone/p/MOB0000004549294')
|
|
|
158 |
print data
|