Subversion Repositories SmartDukaan

Rev

Rev 4198 | Rev 5639 | Go to most recent revision | Details | Compare with Previous | Last modification | View Log | RSS feed

Rev Author Line No. Line
4198 varun.gupt 1 
'''
2 
Created on 05-Dec-2011
3 
@author: Varun Gupta
4 
'''
5 
 
6 
class URL:
7 
 
8 
    def __init__(self, url):
9 
        self.url = url
10 
        self.source = URL.getSource(url)
11 
 
12 
    @staticmethod
13 
    def getSource(url):
14 
        try:
15 
            return str(url.split('.')[1].strip())
16 
        except Exception:
17 
            return None
18 
 
19 
    def __str__(self):
20 
        return '%s (%s)' % (self.url, self.source)
5291 varun.gupt 21 
 
22 
    def __unicode__(self):
23 
        return '%s (%s)' % (self.url, self.source)
4198 varun.gupt 24 
 
25 
class URLQueue:
26 
 
27 
    def __init__(self):
28 
        self.urls = [
29 
            URL('http://www.flipkart.com/mobiles/all/'),
5291 varun.gupt 30 
            URL('http://www.snapdeal.com/json/product/get/search/175/0/20?q=&sort=plrty&keyword='),
4198 varun.gupt 31 
            URL('http://www.infibeam.com/Mobiles/search'),
32 
            URL('http://www.homeshop18.com/gsm-handsets/category:3027/'),
33 
            URL('http://www.flipkart.com/mobiles/tablet-20278'),
5291 varun.gupt 34 
            URL('http://www.snapdeal.com/json/product/get/search/133/0/20?q=&sort=plrty&keyword='),
4198 varun.gupt 35 
            URL('http://www.homeshop18.com/ipads-2f-tablets/category:8937/'),
5291 varun.gupt 36 
            #URL('http://www.adexmart.com/modules/coremanager/modules/filtersearch/filtersearch.json.php?act=filter&ident=16&page=1&perpage=1000&orderby=newest&orderway=desc')
4198 varun.gupt 37 
        ]
38 
 
39 
    def enqueue(self, url):
40 
 
41 
        if url is not None:
42 
            new_url = url if url.__class__ == URL else URL(url)
43 
 
44 
            print 'Enqueue', new_url
45 
            self.urls.append(new_url)
46 
            print 'New URL set:', self.urls
47 
 
48 
    def get(self):
49 
        print 'Count of URLs in queue:', len(self.urls)
50 
        try:
51 
            url = self.urls.pop(0) if len(self.urls) > 0 else None
52 
        except IndexError:
53 
            url = None
54 
 
55 
        print 'Poping', url
56 
        print 'New URL set:', self.urls
57 
        return url