Subversion Repositories SmartDukaan

import urllib2

from BeautifulSoup import BeautifulSoup

import re

from dtr.utils.utils import to_java_date

import optparse

from datetime import datetime

import smtplib

from email.mime.text import MIMEText

from email.mime.multipart import MIMEMultipart

from dtr.utils.utils import fetchResponseUsingProxy, get_mongo_connection, ungzipResponse

import json

import urllib

import chardet

from shop2020.utils.EmailAttachmentSender import get_attachment_part

from shop2020.utils import EmailAttachmentSender

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()

headers = {

            'User-Agent':'Mozilla/5.0 (X11; Linux i686) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/41.0.2272.118 Safari/537.36',

            'Accept' : 'text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8',      

            'Accept-Language' : 'en-US,en;q=0.8',                     

            'Accept-Encoding' : 'gzip,deflate,sdch',

            'Host' : 'www.homeshop18.com',

            'Referer': 'http://www.homeshop18.com/mobile-phones/categoryid:14569/search:/sort:Popularity/inStock:%28%22true%22%29/start:0/'     

}

exceptionList = []

bestSellers = []

now = datetime.now()

csrfValue = None

class __RankInfo:

    def __init__(self, identifier, rank, category, available_price, in_stock, thumbnail, source_product_name, marketPlaceUrl):

        self.identifier = identifier

        self.rank  = rank

        self.available_price = available_price

        self.in_stock = in_stock

        self.category = category

        self.thumbnail = thumbnail

        self.source_product_name = source_product_name

        self.marketPlaceUrl = marketPlaceUrl    

def commitBestSellers(category):

    global exceptionList

    print "Rank",

    print '\t',

    print 'Identifier'

    for x in bestSellers:

        print x.rank,

        print '\t',

        print x.identifier,

        print '\t',

        col = list(get_mongo_connection(host=options.mongoHost).Catalog.MasterData.find({'identifier':x.identifier, 'source_id':7}))

        print "count sku",len(col)

        print '\n'

        if len(col) == 0:

            x.category = category

            exceptionList.append(x)

        else:

            get_mongo_connection(host=options.mongoHost).Catalog.MasterData.update({'identifier':x.identifier, 'source_id':7 }, {'$set' : {'rank':x.rank,'updatedOn':to_java_date(now)}})

def scrapeBestSellerMobiles():

    global bestSellers

    rank = 0

    bestSellers = []

    print "Homeshop18 Best Sellers Mobiles..."

    for i in range(0,5):

        mobileCategoryUrl = 'http://www.homeshop18.com/mobile-phones/categoryid:14569/search:/sort:Popularity/inStock:%28%22true%22%29/start:'+str(i*24)+'/?lazy=true'

        #mobileCategoryUrl = "http://m.homeshop18.com/search.mobi?categoryId=14569&isAjax=true&page="+str(i)+"&csrf="+csrfValue

        data = fetchResponseUsingProxy(mobileCategoryUrl, proxy=False )

        soup = BeautifulSoup(data)

        tags = soup.findAll("div", {'class':'inside'})

        for tag in tags:

            if not tag.has_key('id'):

                continue

            rank = rank +1

            if rank >100:

                break

            titleTag = tag.find('p', {'class' : 'product_title'})

            source_product_name = titleTag.text

            '''

            encoding =  chardet.detect(source_product_name)

            try:

                source_product_name = source_product_name.decode(encoding.get('encoding'))

            except:

                source_product_name = source_product_name.decode('latin-1')

            '''

            source_product_name = source_product_name.encode('utf8')

            productUrl = titleTag.find('a').get('href')

            productUrl = 'http://www.homeshop18.com'+str(productUrl)

            inStock = 1

            identifier = tag['id'].split('_')[1]

            available_price = long(tag.find('p',{'class':'price clearfix'}).find('b').text.split(' ')[1])

            thumbnail = tag.find('p',{'class':'product_image'}).find('img').get('data-original')

            print productUrl, source_product_name, thumbnail, rank, available_price, identifier, inStock

            r_info = __RankInfo(identifier, rank, None, available_price, inStock, thumbnail, source_product_name, "http://m.homeshop18.com/product.mobi?productId=%s"%str(identifier))

            bestSellers.append(r_info)   

def scrapeBestSellerTablets():

    global bestSellers

    rank = 0

    bestSellers = []

    print "Homeshop18 Best Sellers Tablets..."

    for i in range(0,5):

        tabletCategoryUrl = 'http://www.homeshop18.com/tablets/categoryid:8937/search:/sort:Popularity/inStock:%28%22true%22%29/start:'+str(i*24)+'/?lazy=true'

        data = fetchResponseUsingProxy(tabletCategoryUrl, proxy=False)

        soup = BeautifulSoup(data)

        tags = soup.findAll("div", {'class':'inside'})

        for tag in tags:

            if not tag.has_key('id'):

                continue

            rank = rank +1

            if rank >100:

                break

            titleTag = tag.find('p', {'class' : 'product_title'})

            source_product_name = titleTag.text

            '''

            encoding =  chardet.detect(source_product_name)

            try:

                source_product_name = source_product_name.decode(encoding.get('encoding'))

            except:

                source_product_name = source_product_name.decode('latin-1')

            '''

            source_product_name = source_product_name.encode('utf8')

            productUrl = titleTag.find('a').get('href')

            productUrl = 'http://www.homeshop18.com'+str(productUrl)

            inStock = 1

            identifier = tag['id'].split('_')[1]

            available_price = long(tag.find('p',{'class':'price clearfix'}).find('b').text.split(' ')[1])

            thumbnail = tag.find('p',{'class':'product_image'}).find('img').get('data-original')

            print productUrl, source_product_name, thumbnail, rank, available_price, identifier, inStock

            r_info = __RankInfo(identifier, rank, None , available_price, inStock, thumbnail, source_product_name, "http://m.homeshop18.com/product.mobi?productId=%s"%str(identifier))

            bestSellers.append(r_info)

def resetRanks(category_id):

    get_mongo_connection(host=options.mongoHost).Catalog.MasterData.update({'rank':{'$gt':0},'source_id':7,'category_id':category_id}, {'$set':{'rank':0}}, upsert=False, multi=True)

def sendEmail():

    message="""<html>

            <body>

            <h3>HomeShop18 Best Sellers not in master</h3>

            <table border="1" style="width:100%;">

            <thead>

            <tr><th>Identifier</th>

            <th>Category</th>

            <th>Rank</th>

            <th>Available_price</th>

            <th>In_stock</th>

            <th>Thumbnail</th>

            <th>Source_product_name</th>

            <th>MarketPlaceUrl</th>

            </tr></thead>

            <tbody>"""

    for item in exceptionList:

        try:

            message+="""<tr>

            <td style="text-align:center">"""+str(item.identifier)+"""</td>

            <td style="text-align:center">"""+str(item.category)+"""</td>

            <td style="text-align:center">"""+str(item.rank)+"""</td>

            <td style="text-align:center">"""+str(item.available_price)+"""</td>

            <td style="text-align:center">"""+str(item.in_stock)+"""</td>

            <td style="text-align:center">"""+str(item.thumbnail)+"""</td>

            <td style="text-align:center">"""+str(item.source_product_name)+"""</td>

            <td style="text-align:center">"""+str(item.marketPlaceUrl)+"""</td>

            </tr>"""

        except:

            continue

    message+="""</tbody></table></body></html>"""

    print message

    #recipients = ['amit.gupta@saholic.com']

    recipients = ['kshitij.sood@saholic.com','ritesh.chauhan@saholic.com','aishwarya.singh@saholic.com']

    EmailAttachmentSender.mail_send_grid("dtr@smartdukaan.com","apikey", "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", recipients, "Homeshop18 Best Sellers",message ,[],[],[])              

def main():

    #getCsrfValue()

    scrapeBestSellerMobiles()

    if len(bestSellers) > 0:

        resetRanks(3)

        commitBestSellers("MOBILE")

    scrapeBestSellerTablets()

    if len(bestSellers) > 0:

        resetRanks(5)

        commitBestSellers("TABLET")

    sendEmail()

if __name__=='__main__':

    main()