def process_item(self, item, spider):
if not item['ISBN']: # if item do not have isbn then drop it
raise DropItem('Duplicate item found: %s' % item)
if item['platform'] == 3: # if item is comments return to detail_pipeline
return item
new_book = Book()
if item["name"]:
new_book.title = item["name"][0]
if item["price"]:
new_book.price = item["price"][0]
if item["author"]:
new_book.author = item["author"][0]
if item["press"]:
new_book.press = item["press"][0]
if item["instant"]:
new_book.instant_price = item["instant"][0]
if item["img"]:
new_book.cover = item["img"][0]
if item["description"]:
new_book.description = item["description"][0]
new_book.isbn = item["ISBN"][0]
new_book.link = item["url"]
new_book.platform = item['platform']
#new_book.time = ?
new_book.platform = item['platform']
book_dao.insert(new_book)
def process_item(item, spider):
if item['platform'] == -1: # not a book, drop it
return item
if not item['ISBN']: # not a book, drop it
item['platform'] = -1
return item
if item['platform'] == 3: # if item is comments return to detail_pipeline
return item
new_book = Book()
if item["name"]:
new_book.title = item["name"][0]
if item["price"]:
new_book.price = item["price"][0]
if item["author"]:
new_book.author = item["author"][0]
if item["press"]:
new_book.press = item["press"][0]
if item["instant"]:
new_book.instant_price = item["instant"][0]
if item["img"]:
new_book.cover = item["img"][0]
if item["description"]:
new_book.description = item["description"][0]
new_book.isbn = item["ISBN"][0]
new_book.link = item["url"]
new_book.platform = item['platform']
#new_book.time = ?
new_book.platform = item['platform']
book_dao.insert(new_book)
return item
def process_item(self, item, spider):
if(len(item['ISBN']) == 0):
return item
newbook = Book()
newbook.title = item["name"][0]
newbook.price = item["instant"][0]
newbook.isbn = item["ISBN"][0]
newbook.author = item["author"][0]
newbook.press = item["press"][0]
newbook.instant_price = item["price"][0]
newbook.link = item["url"]
newbook.cover = item["img"][0]
if len(item["description"]) != 0 :
newbook.description = item["description"][0]
newbook.platform = item['platform']
#newbook.time = ?
newbook.platform = item['platform']
book_dao.insert(newbook)
def test_insert():
for i in range(1, 101):
book = Book()
index = (i + 1) / 2
book.isbn = '%05d001' % (index)
book.price = (index) * 1.0
book.title = 'title-%d' % (index)
book.author = 'author-%s' % (index)
book.press = 'press-%s' % (index / 10)
book.description = 'description for book-%d' % (index)
book.cover = 'cover-%d' % (index)
book.link = 'http://www.oricinus_price/book-%d-%d' % (index, i % 2)
book.platform = i % 2
book.instant_price = book.price * 0.8
book.crawling_time = i
book_dao.insert(book)
def test_insert():
for i in range(1,101):
book = Book()
index = (i+1) / 2
book.isbn = '%05d001'%(index)
book.price = (index)*1.0
book.title = 'title-%d'%(index)
book.author = 'author-%s'%(index)
book.press = 'press-%s'%(index / 10)
book.description = 'description for book-%d'%(index)
book.cover = 'cover-%d'%(index)
book.link = 'http://www.oricinus_price/book-%d-%d'%(index, i%2)
book.platform = i % 2
book.instant_price = book.price * 0.8
book.crawling_time = i
book_dao.insert(book)
def __parse_po_to_book(self, bookpo, goodspo):
book = Book()
# bookpo->book
book.isbn = bookpo.isbn
book.price = bookpo.price
book.title = bookpo.title
book.author = bookpo.author
book.press = bookpo.press
book.description = bookpo.description
book.cover = bookpo.cover
#goodspo->book
book.link = goodspo.link
book.platform = goodspo.platform
book.instant_price = goodspo.instant_price
book.crawling_time = goodspo.crawling_time
return book
def process_item(self, item, spider):
if item['ISBN']:
return item
new_book = Book()
if item["name"]:
new_book.title = item["name"][0]
if item["price"]:
new_book.price = item["price"][0]
if item["author"]:
new_book.author = item["author"][0]
if item["press"]:
new_book.press = item["press"][0]
if item["instant"]:
new_book.instant_price = item["instant"][0]
if item["img"]:
new_book.cover = item["img"][0]
if item["description"]:
new_book.description = item["description"][0]
new_book.isbn = item["ISBN"][0]
new_book.link = item["url"]
new_book.platform = item['platform']
#new_book.time = ?
new_book.platform = item['platform']
book_dao.insert(new_book)