def __init__(self, api, ItemId=None, URL=None):
self._asin = None
if ItemId and not URL:
# check for http://www.amazon.com
if 'amazon' in ItemId:
raise ValueError('URL passed as ASIN')
URL = reviews_url(ItemId)
elif not URL and not ItemId:
raise ValueError('Invalid review page parameters. Input a URL or a valid ASIN!')
elif URL and 'product-reviews' not in URL: # If product-reviews is in the url it's probably a valid product review page. Let it be.
# cleanup the url
URL = reviews_url(extract_reviews_id(URL))
self.api = api
self._URL = URL
self._soup = None
def __init__(self, api, ItemId=None, URL=None):
self._asin = None
if ItemId and not URL:
# check for http://www.amazon.com
if 'amazon' in ItemId:
raise ValueError('URL passed as ASIN')
URL = reviews_url(ItemId)
elif not URL and not ItemId:
raise ValueError('Invalid review page parameters. Input a URL or a valid ASIN!')
elif URL and 'product-reviews' not in URL: # If product-reviews is in the url it's probably a valid product review page. Let it be.
# cleanup the url
URL = reviews_url(extract_reviews_id(URL))
self.api = api
self._URL = URL
self._soup = None
def reviews_url(self):
# we could use the asin to directly make a review url
# but some products actually use the ISBN for the review url
# and the ASIN version would fail
# so we'll get the url given to us, and get the asin/isbn from that
try:
# the variable has changed in python simple product api, sigh
item = getattr(self.product, 'item', None)
if not item:
item = getattr(self.product, 'parsed_response', None)
url = unicode(item['CustomerReviews']['IFrameURL'])
p = urlparse.urlparse(url)
q = urlparse.parse_qs(p.query)
asin = q['asin'][0]
except Exception as e:
asin = self.asin
reviews = reviews_url(asin)
return reviews
def reviews_url(self):
# we could use the asin to directly make a review url
# but some products actually use the ISBN for the review url
# and the ASIN version would fail
# so we'll get the url given to us, and get the asin/isbn from that
try:
# the variable has changed in python simple product api, sigh
item = getattr(self.product, 'item', None)
if not item:
item = getattr(self.product, 'parsed_response', None)
url = unicode(item['CustomerReviews']['IFrameURL'])
p = urlparse.urlparse(url)
q = urlparse.parse_qs(p.query)
asin = q['asin'][0]
except Exception as e:
asin = self.asin
reviews = reviews_url(asin)
return reviews
def test_reviews_url(self):
p = self.amzn.lookup(ItemId="B00FLIJJSA")
assert p.reviews_url == amazon_scraper.reviews_url("B00FLIJJSA"), p.reviews_url
def test_reviews_url(self):
p = self.amzn.lookup(ItemId='B00FLIJJSA')
assert p.reviews_url == amazon_scraper.reviews_url('B00FLIJJSA'), p.reviews_url
