| Line 88... |
Line 88... |
| 88 |
titleTag = tag.find('p', {'class' : 'product_title'})
|
88 |
titleTag = tag.find('p', {'class' : 'product_title'})
|
| 89 |
source_product_name = titleTag.text
|
89 |
source_product_name = titleTag.text
|
| 90 |
productUrl = titleTag.find('a').get('href')
|
90 |
productUrl = titleTag.find('a').get('href')
|
| 91 |
productUrl = 'http://www.homeshop18.com'+str(productUrl)
|
91 |
productUrl = 'http://www.homeshop18.com'+str(productUrl)
|
| 92 |
inStock = 1
|
92 |
inStock = 1
|
| 93 |
identfier = tag['id'].split('_')[1]
|
93 |
identifier = tag['id'].split('_')[1]
|
| 94 |
available_price = long(tag.find('p',{'class':'price clearfix'}).find('b').text.split(' ')[1])
|
94 |
available_price = long(tag.find('p',{'class':'price clearfix'}).find('b').text.split(' ')[1])
|
| 95 |
thumbnail = tag.find('p',{'class':'product_image'}).find('img').get('data-original')
|
95 |
thumbnail = tag.find('p',{'class':'product_image'}).find('img').get('data-original')
|
| 96 |
print productUrl, source_product_name, thumbnail, rank, available_price, identfier, inStock
|
96 |
print productUrl, source_product_name, thumbnail, rank, available_price, identifier, inStock
|
| 97 |
r_info = __RankInfo(identfier, rank, None, available_price, inStock, thumbnail, source_product_name, productUrl)
|
97 |
r_info = __RankInfo(identifier, rank, None, available_price, inStock, thumbnail, source_product_name, "http://m.homeshop18.com/product.mobi?productId=%s"%str(identifier))
|
| 98 |
bestSellers.append(r_info)
|
98 |
bestSellers.append(r_info)
|
| 99 |
|
99 |
|
| 100 |
def scrapeBestSellerTablets():
|
100 |
def scrapeBestSellerTablets():
|
| 101 |
global bestSellers
|
101 |
global bestSellers
|
| 102 |
rank = 0
|
102 |
rank = 0
|
| Line 116... |
Line 116... |
| 116 |
titleTag = tag.find('p', {'class' : 'product_title'})
|
116 |
titleTag = tag.find('p', {'class' : 'product_title'})
|
| 117 |
source_product_name = titleTag.text
|
117 |
source_product_name = titleTag.text
|
| 118 |
productUrl = titleTag.find('a').get('href')
|
118 |
productUrl = titleTag.find('a').get('href')
|
| 119 |
productUrl = 'http://www.homeshop18.com'+str(productUrl)
|
119 |
productUrl = 'http://www.homeshop18.com'+str(productUrl)
|
| 120 |
inStock = 1
|
120 |
inStock = 1
|
| 121 |
identfier = tag['id'].split('_')[1]
|
121 |
identifier = tag['id'].split('_')[1]
|
| 122 |
available_price = long(tag.find('p',{'class':'price clearfix'}).find('b').text.split(' ')[1])
|
122 |
available_price = long(tag.find('p',{'class':'price clearfix'}).find('b').text.split(' ')[1])
|
| 123 |
thumbnail = tag.find('p',{'class':'product_image'}).find('img').get('data-original')
|
123 |
thumbnail = tag.find('p',{'class':'product_image'}).find('img').get('data-original')
|
| 124 |
print productUrl, source_product_name, thumbnail, rank, available_price, identfier, inStock
|
124 |
print productUrl, source_product_name, thumbnail, rank, available_price, identifier, inStock
|
| 125 |
r_info = __RankInfo(identfier, rank, None , available_price, inStock, thumbnail, source_product_name, productUrl)
|
125 |
r_info = __RankInfo(identifier, rank, None , available_price, inStock, thumbnail, source_product_name, "http://m.homeshop18.com/product.mobi?productId=%s"%str(identifier))
|
| 126 |
bestSellers.append(r_info)
|
126 |
bestSellers.append(r_info)
|
| 127 |
|
127 |
|
| 128 |
def resetRanks(category_id):
|
128 |
def resetRanks(category_id):
|
| 129 |
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)
|
129 |
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)
|
| 130 |
|
130 |
|