Subversion Repositories SmartDukaan

Rev

Rev 5770 | Rev 6024 | Go to most recent revision | Show entire file | Ignore whitespace | Details | Blame | Last modification | View Log | RSS feed

Rev 5770 Rev 6022
Line 27... Line 27...
27 
        self.phones = None
 27 
        self.phones = None
28 
        #self.setPageCount()
 28 
        #self.setPageCount()
29
29
30 
    def getPhones(self):
 30 
    def getPhones(self):
31 
        phones = []
 31 
        phones = []
32 
        for div in self.soup.findAll('div', {'class': 'mainresult-show-right'}):
 32 
        for div in self.soup.findAll('div', {'class': 'prod_main_div'}):
33 
            try:
 33 
            try:
34 
                anchor = div.find('a')
 34 
                productUrlContainer = div.find('div', {'class': 'product_name search-product-block'})
35 
                if(len(anchor.contents)==1):
 - 
 
36 
                    name = anchor.contents[0].strip()
 35 
                name = productUrlContainer.contents[1].string
37 
                elif(anchor.contents[1].string =="Tablet"):
 - 
 
38 
                    name = anchor.contents[0] + "Tablet"
 - 
 
39 
                    if(len(anchor.contents)>2):
 - 
 
40 
                        name = name + anchor.contents[2];
 36 
                product_url = "www.tradus.com" +productUrlContainer.contents[1]['href']
41 
                else:
 - 
 
42 
                    name = anchor.contents[2].strip()
 - 
 
43 
                    if(len(name)== 0):
 - 
 
44 
                        name = anchor.contents[0].strip()
 37 
                price = div.find('span', {'class':'numDiv_left'}).string.strip()
45 
                product_url = anchor['href'].strip()
 38 
                price = removePriceFormatting(price)
46 
                price = div.find('div', {'class': 'mainresult-show-right-startrate'})('span')[0]('span')[0].contents[0].strip()[3:]
 - 
 
47 
                in_stock = 1
 39 
                in_stock = 1
48
40
49 
                try:
 41 
                try:
50 
                    if price is None:
 42 
                    if price is None:
51 
                        continue
 43 
                        continue
Line 105... Line 97...
105
97
106 
 
 98 
 
107 
    def getDataFromProductPage(self, url):
 99 
    def getDataFromProductPage(self, url):
108 
        html = BaseScraper.read(self, url)
 100 
        html = BaseScraper.read(self, url)
109 
        soup = BeautifulSoup(html)
 101 
        soup = BeautifulSoup(html)
110 
        name = soup.find('h1',{'class': 'left-content-product-heading'}).string.strip()
 102 
        name = soup.find('span',{'itemprop':'name'}).string.strip()
111 
        price = soup.find('b', {'id': 'tPrice'}).string.strip()
 103 
        price= soup.find('span',{'class':'mrp3'}).contents[0].strip()
112 
        if("Rs." in price):
 - 
 
113 
            price = price[4:]
 104 
        price = removePriceFormatting(price)
114 
        in_stock = 1
 105 
        in_stock = 1
115
106
116 
        data = {
 107 
        data = {
117 
            "product_url": str(url),
108 
            "product_url": str(url),
118 
            "source": "tradus",
109 
            "source": "tradus",
Line 120... Line 111...
120 
            "in_stock": 1,
111 
            "in_stock": 1,
121 
            "name": name
 112 
            "name": name
122 
        }
 113 
        }
123 
        return data
 114 
        return data
124 
 
 115 
 
- 
 
 116 
def removePriceFormatting(price_string):
- 
 
 117 
    return price_string.strip().replace('Rs.', '').replace('Rs', '').replace(',', '').replace(' ', '').replace(' ', '').split('.')[0]
- 
 
 118 
 
125 
if __name__ == '__main__':
 119 
if __name__ == '__main__':
126 
    s = TradusScraper()
 120 
    s = TradusScraper()
127 
    '''html = BaseScraper.read(s,'http://www.tradus.com/search/tradus_search/?query=mobile&filters=cat:7756')
 - 
 
128 
    soup = BeautifulSoup(html)
 - 
 
129
- 
 
130 
    phones = []
 - 
 
131 
    for div in soup.findAll('div', {'class': 'mainresult-show-right'}):
 - 
 
132 
        try:
 - 
 
133 
            anchor = div.find('a')
 - 
 
134 
            name = anchor.contents[2].strip()
 - 
 
135 
            if(len(name)== 0):
 - 
 
136 
                name = anchor.contents[0].strip()
 - 
 
137 
            product_url = anchor['href'].strip()
 - 
 
138 
            price = div.find('div', {'class': 'mainresult-show-right-startrate'})('span')[0]('span')[0].contents[0].strip()[3:]
 - 
 
139 
            in_stock = 1
 - 
 
140
- 
 
141 
            try:
 - 
 
142 
                if price is None:
 - 
 
143 
                    continue
 - 
 
144 
                else:
 - 
 
145 
                    phones.append({
 - 
 
146 
                            'name': str(name),
- 
 
147 
                            'price': removePriceFormatting(price),
 - 
 
148 
                            'source': 'tradus',
- 
 
149 
                            'product_url': str(product_url),
- 
 
150 
                            'in_stock': in_stock
 - 
 
151 
                        })
 - 
 
152 
            except Exception as e:
 - 
 
153 
                print e
 - 
 
154 
                pass
 - 
 
155
- 
 
156 
        except IndexError as iex:
 - 
 
157 
            try:
 - 
 
158 
                price = div.find('div', {'class': 'mainresult-show-right-startrate'})('span')[0].contents[0].strip()[18:]
 - 
 
159 
                in_stock = 1
 - 
 
160
- 
 
161 
                phones.append({
 - 
 
162 
                            'name': str(name),
- 
 
163 
                            'price': removePriceFormatting(price),
 - 
 
164 
                            'source': 'tradus',
- 
 
165 
                            'product_url': str(product_url),
- 
 
166 
                            'in_stock': in_stock
 - 
 
167 
                        })
 - 
 
168
- 
 
169 
            except Exception as ex:
 - 
 
170 
                print ex
 - 
 
171 
                pass
 - 
 
172 
        except Exception as e:
 - 
 
173 
            print e
 - 
 
174 
            pass
 - 
 
175 
    print phones'''
 - 
 
176
- 
 
177
- 
 
178 
    data = s.getDataFromProductPage('http://www.tradus.com/samsung-galaxy-y-pro-duos-b5512-mobile-phone/p/MOB0000004549294')
 121 
    data = s.getDataFromProductPage('http://www.tradus.com/zing-q800-dual-sim-mobile-phone/p/MOB0000004506663')
179 
    print data
 122 
    print data
180 

            123