def _validate_book_id(self):
# TODO: add check constraints
# Validate book identifier
if self.book_id_type == 'ISBN':
if pyisbn.validate(self.book_id):
if len(self.book_id) == 10:
self.isbn10 = self.book_id
self.isbn13 = pyisbn.convert(self.book_id)
else:
self.isbn13 = self.book_id
self.isbn10 = pyisbn.convert(self.book_id)
else:
raise ValidationError(
{'book_id': '{} is an invalid ISBN'.format(self.book_id)})
elif self.book_id_type == 'ASIN':
regex = r"[A-Z0-9]{10}"
# Remove whitespaces
self.book_id = self.book_id.strip()
if len(self.book_id) == 10 and re.fullmatch(regex, self.book_id):
self.asin = self.book_id
else:
raise ValidationError(
{'book_id': '{} is an invalid ASIN'.format(self.book_id)})
else:
raise ValidationError({
'book_id_type':
'Allowed Book Id types: {}'.format(self.allowed_book_id_types)
})
def clean(self, value):
value = super(ISBNField, self).clean(value)
value = ''.join(value.split('-'))
try:
if not pyisbn.validate(value):
raise ValidationError('Invalid ISBN')
except (TypeError, ValueError), e:
raise ValidationError, str(e)
def clean(self, value):
value = super(ISBNField, self).clean(value)
value = ''.join(value.split('-'))
try:
if not pyisbn.validate(value):
raise ValidationError('Invalid ISBN')
except (TypeError, ValueError), e:
raise ValidationError, str(e)
def validate_identifiers(self, data={}):
""" Identifier verification: types can only be used once """
if 'label' in data and data['label'] == 'ISBN':
isbn = data.get('value')
if not isbn:
return
try:
isbn_ok = pyisbn.validate(isbn)
except ValueError, e:
isbn_ok = False
if not isbn_ok:
return 'Invalid ISBN %s' % isbn
def validate_identifiers(self, data={}):
""" Identifier verification: types can only be used once """
if 'label' in data and data['label'] == 'ISBN':
isbn = data.get('value')
if not isbn:
return
try:
isbn_ok = pyisbn.validate(isbn)
except ValueError, e:
isbn_ok = False
if not isbn_ok:
return 'Invalid ISBN %s' % isbn
def search(request):
results = {
'sources': {},
'error': {},
}
'''
more: false,
results: [
{ text: "Western", children: [
{ id: "CA", text: "California" },
{ id: "AZ", text: "Arizona" }
] },
{ text: "Eastern", children: [
{ id: "FL", text: "Florida" }
] }
]
'''
title = request.QUERY_PARAMS.get('title', False)
if title:
sources = Source.objects.filter(title__icontains=title)
source_serializer = SourceSerializer(sources, many=True)
results['sources']['spuqi'] = source_serializer.data
isbn_str = request.QUERY_PARAMS.get('isbn', False)
if isbn_str:
try:
pyisbn.Isbn(isbn_str)
except pyisbn.IsbnError:
results['error'] = {
'message': _('ISBN number must contain only digit-numbers'),
}
else:
isbn_number = pyisbn.convert(isbn_str)
if not pyisbn.validate(isbn_number):
results['error'] = {
'message': _('A valid ISBN number is required')
}
#try:
# if googlebooks_api.list('isbn:%s' % isbn)['totalItems'] > 0:
# results['sources']['googlebooks'] = googlebooks_api.list(
# 'isbn:%s' % isbn)['items']
results['sources']['googlebooks'] = googlebooks_api.list(
'isbn:%s' % isbn_number)
# except ConnectionError:
# results.errors.append(_('Could not connect to Google Book API'))
return Response({"results": results})
def get_by_isbn(cls, isbn, create=True, defaults=None):
if len(isbn) == 10:
# NOTE: always use ISBN-13
# this should also raise an exception if isbn is invalid
isbn = pyisbn.convert(isbn)
if not pyisbn.validate(isbn):
raise ValueError("Invalid ISBN: %s" % isbn)
memkey = "book:%s" % isbn
book = memcache.get(memkey)
if book:
return book
if not create:
book = Book.get_by_key_name(memkey)
else:
defaults = defaults or {}
defaults["isbn13"] = isbn
book = Book.get_or_insert(memkey, **defaults)
if book:
memcache.set(book.memkey, book, Book.MEMCACHE_TTL)
return book
def filter_language():
import pyisbn
with open('isbn2ind.pickle', 'rb') as f:
isbn2ind = pickle.load(f)
with open('isbn2lang.pickle', 'rb') as f:
isbn2lang = pickle.load(f)
with open('isbn2ratings_count4.pickle', 'rb') as f:
isbn2ratings_count = pickle.load(f)
with open('isbn2rating4.pickle', 'rb') as f:
isbn2ratings = pickle.load(f)
with open('isbn2title4.pickle', 'rb') as f:
isbn2title = pickle.load(f)
with open('isbn2authors4.pickle', 'rb') as f:
isbn2authors = pickle.load(f)
isbns = []
# rating_counts = list(isbn2ratings.values()) # list(isbn2ratings_count.values())
# print('max rating', max(rating_counts))
# plt.hist(rating_counts, bins=np.arange(0, 20, 1))
# plt.show()
for isbn in tqdm(isbn2ind):
lang = isbn2lang[isbn]
rating_count = isbn2ratings_count[isbn]
rating = isbn2ratings[isbn]
title = isbn2title[isbn]
# print(lang)
if isbn in ['ordernumber85'] or 'http:' in isbn:
continue
try:
if not pyisbn.validate(isbn):
continue
except:
continue
if rating_count > 230 and (
lang == 'en') and 3.8 <= rating <= 20 and len(title) > 1:
isbns.append(isbn)
print('num_left', len(isbns))
titles = set()
new_isbns = []
for isbn in tqdm(isbns):
title = isbn2title[isbn].lower()
# if a replace ind is None because rating is smaller
# than similar book, then add is false
if title in titles:
continue
else:
titles.add(title)
new_isbns.append(isbn)
print(titles)
# for isbn in tqdm(isbns):
# title = isbn2title[isbn].lower()
# rating_count = isbn2ratings_count[isbn]
# author = isbn2authors[isbn]
# replace_ind = None
# # if a replace ind is None because rating is smaller
# # than similar book, then add is false
# add = True
# for i, isbn1 in enumerate(new_isbns):
# title1 = isbn2title[isbn1].lower()
# author1 = isbn2authors[isbn1]
# if author1 != author:
# continue
# if fuzz.ratio(title, title1) > 93:
# rating_count1 = isbn2ratings_count[isbn1]
# if rating_count > rating_count1:
# replace_ind = i
# add = False
# # print(title1, ',', title)
# break
# if replace_ind is not None:
# new_isbns[replace_ind] = isbn
# elif add:
# new_isbns.append(isbn)
isbns = sorted(new_isbns)
# isbns = sorted(isbns)
isbn2ind = {isbn: ind for ind, isbn in enumerate(isbns)}
# print(isbn2ind.keys())
print('num_left', len(isbns))
with open('isbn2ind.pickle', 'wb') as f:
pickle.dump(isbn2ind, f, protocol=pickle.HIGHEST_PROTOCOL)
def validate_isbn(isbn):
if not pyisbn.validate(isbn):
raise ValidationError(u'{} is not a valid ISBN number'.format(isbn))
def test_validate_invalid(isbn):
assert validate(isbn) is False
def boolvalidate(isbn):
try:
p = pyisbn.validate(isbn)
except:
return False
return p
def validate(isbn):
try:
return pyisbn.validate(isbn)
except:
return False
def validate_isbn(isbn):
if not pyisbn.validate(isbn):
raise ValidationError('Nieprawidłowy numer ISBN!')
def test_validate(isbn):
expect(validate(isbn)) == True
def isbn_validator(isbn):
try:
pyisbn.validate(isbn)
except pyisbn.IsbnError as err:
raise ValidationError(err)
def test_validate_invalid(isbn):
expect(validate(isbn)) == False
def test_validate(isbn):
assert validate(isbn)
