Subversion Repositories SmartDukaan

import urllib2

from BeautifulSoup import BeautifulSoup

import pymongo

import re

from dtr.utils.utils import to_java_date

from datetime import datetime

import optparse

con = None

parser = optparse.OptionParser()

parser.add_option("-m", "--m", dest="mongoHost",

                      default="localhost",

                      type="string", help="The HOST where the mongo server is running",

                      metavar="mongo_host")

(options, args) = parser.parse_args()

asin_regex = r'/([A-Z0-9]{10})'

bestSellers = []

now = datetime.now()

class __RankInfo:

    def __init__(self, identifier, rank):

        self.identifier = identifier

        self.rank  = rank

def get_mongo_connection(host=options.mongoHost, port=27017):

    global con

    if con is None:

        print "Establishing connection %s host and port %d" %(host,port)

        try:

            con = pymongo.MongoClient(host, port)

        except Exception, e:

            print e

            return None

    return con

def getSoupObject(url):

    print "Getting soup object for"

    print url

    global RETRY_COUNT

    RETRY_COUNT = 1 

    while RETRY_COUNT < 10:

        try:

            soup = None

            request = urllib2.Request(url)

            request.add_header('Accept','text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8')

            request.add_header('Accept-Language','en-US,en;q=0.8,hi;q=0.6')

            request.add_header('Connection','keep-alive')

            request.add_header('User-Agent','Mozilla/5.0 (X11; Linux i686) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/37.0.2062.120 Safari/537.36')

            response = urllib2.urlopen(request)   

            response_data = response.read()

            response.close()

            try:

                page=response_data.decode("utf-8")

                soup = BeautifulSoup(page,convertEntities=BeautifulSoup.HTML_ENTITIES)

            except:

                soup = BeautifulSoup(response_data,convertEntities=BeautifulSoup.HTML_ENTITIES)

            if soup is None:

                raise

            return soup

        except Exception as e:

            print e

            print "Retrying"

            RETRY_COUNT = RETRY_COUNT + 1

def scrapeBestSellerMobiles():

    global bestSellers

    rank = 0

    for i in range(1,6):

        aboveFoldUrl = "http://www.amazon.in/gp/bestsellers/electronics/1389432031/ref=zg_bs_1389432031_pg_%d?ie=UTF8&pg=%d&ajax=1&isAboveTheFold=1" %(i,i) 

        belowFoldUrl = "http://www.amazon.in/gp/bestsellers/electronics/1389432031/ref=zg_bs_1389432031_pg_%d?ie=UTF8&pg=%d&ajax=1&isAboveTheFold=0" %(i,i)

        above_soup = getSoupObject(aboveFoldUrl)

        below_soup = getSoupObject(belowFoldUrl)

        for x in above_soup.findAll('div',{'class':'zg_itemImmersion'}):

            am_url =  x.find('div',{'class':'zg_title'}).find('a')['href']

            identifier =  (re.search(asin_regex, am_url)).group(1)

            rank = rank + 1

            r_info = __RankInfo(identifier,rank)

            bestSellers.append(r_info)

        for x in below_soup.findAll('div',{'class':'zg_itemImmersion'}):

            am_url =  x.find('div',{'class':'zg_title'}).find('a')['href']

            identifier =  (re.search(asin_regex, am_url)).group(1)

            rank = rank + 1

            r_info = __RankInfo(identifier,rank)

            bestSellers.append(r_info)

def commitBestSellers():

    print "Rank",

    print '\t',

    print 'Identifier'

    for x in bestSellers:

        print x.rank,

        print '\t',

        print x.identifier,

        col = get_mongo_connection().Catalog.MasterData.find({'identifier':x.identifier.strip()})

        print "count sku",

        print '\t',

        print len(list(col))

        get_mongo_connection().Catalog.MasterData.update({'identifier':x.identifier.strip() }, {'$set' : {'rank':x.rank,'updatedOn':to_java_date(now)}}, multi=True)

def scrapeBestSellerTablets():

    global bestSellers

    bestSellers = []

    rank = 0

    for i in range(1,6):

        aboveFoldUrl = "http://www.amazon.in/gp/bestsellers/electronics/1375458031/ref=zg_bs_1375458031_pg_%d?ie=UTF8&pg=%d&ajax=1" %(i,i) 

        belowFoldUrl = "http://www.amazon.in/gp/bestsellers/electronics/1375458031/ref=zg_bs_1375458031_pg_%d?ie=UTF8&pg=%d&ajax=1&isAboveTheFold=0" %(i,i)

        above_soup = getSoupObject(aboveFoldUrl)

        below_soup = getSoupObject(belowFoldUrl)

        for x in above_soup.findAll('div',{'class':'zg_itemImmersion'}):

            am_url =  x.find('div',{'class':'zg_title'}).find('a')['href']

            identifier =  (re.search(asin_regex, am_url)).group(1)

            rank = rank + 1

            print identifier,

            print '\t',

            print rank

            r_info = __RankInfo(identifier,rank)

            bestSellers.append(r_info)

        for x in below_soup.findAll('div',{'class':'zg_itemImmersion'}):

            am_url =  x.find('div',{'class':'zg_title'}).find('a')['href']

            identifier =  (re.search(asin_regex, am_url)).group(1)

            rank = rank + 1

            print identifier,

            print '\t',

            print rank

            r_info = __RankInfo(identifier,rank)

            bestSellers.append(r_info)

def resetRanks(category_id):

    oldRankedItems = get_mongo_connection().Catalog.MasterData.find({'rank':{'$gt':0},'source_id':1,'category_id':category_id})

    for item in oldRankedItems:

        get_mongo_connection().Catalog.MasterData.update({'_id':item['_id']}, {'$set' : {'rank':0,'updatedOn':to_java_date(now)}}, multi=True)

def main():

    scrapeBestSellerMobiles()

    if len(bestSellers) > 0:

        resetRanks(3)

        commitBestSellers()

    scrapeBestSellerTablets()

    if len(bestSellers) > 0:

        resetRanks(5)

        commitBestSellers()

if __name__=='__main__':

    main()