def getUserBookTags(self):
results = self.service.GetUserBookTags(self.api.url)
if results:
book_tags, count = results
else:
return False
if not book_tags:
return False
user = None
users = User.objects(uid=self.api.api_id)
if users:
user = users[0]
else:
return False
try:
user_book_related = None
user_book_related, created = UserBookRelated.objects.get_or_create(
uid=user.uid, auto_save=False,
defaults={'uid':user.uid, 'book_tags':book_tags})
if not created:
user_book_related.book_tags.extend(book_tags)
try:
user_book_related.save()
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s when save %s at api:%s" %(error, \
error.to_dict(), user_book_related, self.api.url))
except Exception, error:
logger.debug("[ERROR]%s: UserBookRelated.objects.get_or_create(uid=%s ..."\
%(error, self.api.api_id))
def getUser(self):
user = self.service.GetPeople(self.api.url)
if user:
exist_user = User.objects(uid=user.uid)
if not exist_user:
try:
user.save()
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s at uid:%s" %(error, error.to_dict(), user.uid))
return False
else:
return exist_user[0]
def getProfile(self):
user_api = Api(url='http://api.douban.com/people/'+self.uid,
api_id=self.uid, api_type='USER')
user = UserProcessor(user_api, 'default').getUser()
logger.debug('%s'%(user))
if user:
profile = {}
profile['uid'] = user.uid
profile['nickname'] = user.nickname
profile['signature'] = user.signature
profile['image_link'] = user.image_link
profile['website'] = user.link
profile['introduction'] = user.content
return profile
else:
return None
def getBook(self, oldbook=None):
book = self.service.GetBook(self.api.url)
if book:
if oldbook:
oldbook.summary = book.summary
oldbook.common_tags = book.common_tags
oldbook.rating = book.rating
try:
oldbook.save()
return oldbook
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s at bid:%s" %(error, error.to_dict(), oldbook.bid))
try:
book.save()
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s at bid:%s" %(error, error.to_dict(), book.bid))
def getUserContacts(self, recursion=False):
results = self.service.GetContacts(self.api.url)
if results:
users, count = results
else:
return False
new_users = []
new_users_data = []
indexs = []
for index, user in enumerate(users):
exist_user = User.objects(uid=user.uid)
if exist_user:
users[index] = exist_user[0]
else:
new_users.append(user)
new_user = user.__dict__.get('_data')
for key, value in new_user.items():
if value is None:
new_user.pop(key)
new_user.pop('image')
new_users_data.append(new_user)
indexs.append(index)
if new_users:
ids = self.db.user.insert(new_users_data)
for i, index in enumerate(indexs):
users[index].id = ids[i]
try:
uid = self.api.api_id
user_related = None
user_related, created = UserRelated.objects.get_or_create(
uid=uid, auto_save=False,
defaults = {'uid':uid, 'contacts':users})
if not created:
user_related.contacts.extend(users)
try:
user_related.save()
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s when save %s at api:%s" \
%(error, error.to_dict(), user_related, self.api.url))
except Exception, error:
logger.debug("[ERROR]%s: UserRelated.objects.get_or_create(uid=%s ..."\
%(error, self.api.api_id))
def getUserFriends(self, recursion=False):
users = self.service.GetFriends(self.api.url)
if results:
users, count = results
else:
return False
new_users = []
for index, user in enumerate(users):
exist_user = User.objects(uid=user.uid)
if exist_user:
users[index] = exist_user[0]
else:
new_users.append(user)
try:
user.save()
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s when save uid:%s at api:%s" \
%(error, error.to_dict(), user.uid, self.api.url))
def Get(self, uri, converter, extra_headers={}, *args, **kwargs):
auth_header = self.client.get_auth_header('GET', uri)
if auth_header:
extra_headers.update(auth_header)
elif self.api_key:
param = 'apikey=' + self.api_key
if '?' in uri:
uri += '&' + param
else:
uri += '?' + param
#print 'extra_headers:', extra_headers #TODO remove!
request = Request(uri)
if extra_headers:
for key, value in extra_headers.items():
request.add_header(key, value)
try:
response = urlopen(request)
return converter(response.read())
except URLError, e:
if hasattr(e, 'reason'):
logger.debug('[ERROR]Failed to reach a server. Reason: %s' %(e.reason))
elif hasattr(e, 'code'):
logger.debug('[ERROR]The server couldn\'t fulfill the request. Error code: %s' %(e.code))
logger.debug('-------Error Url: ' + uri)
return None
def processData(op):
stop = False
while(not stop):
data_thread = DataProcessThread(op)
data_thread.start()
time.sleep(1)
logger.debug("Now at:" + str(DataProcessThread.index))
if op == 'FIX_BOOK_COLLECTION' and \
DataProcessThread.index >= DataProcessThread.book_collections_size:
stop = True
elif op == 'book_related' and \
DataProcessThread.index >= DataProcessThread.book_related_size:
stop = True
elif op == 'FIX_BOOK_REVIEW' and \
DataProcessThread.index >= DataProcessThread.book_review_size:
stop = True
elif op == 'FIX_NOTE' and \
DataProcessThread.index >= DataProcessThread.note_size:
stop = True
elif op == 'FIX_RECOMMENDATION' and \
DataProcessThread.index >= DataProcessThread.recommendation_size:
stop = True
def getBookTags(self):
results = self.service.GetBookTags(self.api.url)
if results:
book_tags, count = results
else:
return False
book = None
books = Book.objects(bid=self.api.api_id)
if books:
book = books[0]
else:
return False
if not book_tags:
return False
else:
db = MongoDB.getConnection('mining')
for tag in book_tags:
db.book_tag.update({'title':tag.title},
{'$addToSet':{'books':book.bid},
'$inc':{'count':tag.count}},
upsert=True)
try:
book_related = None
book_related, created = BookRelated.objects.get_or_create(\
bid=book.bid, auto_save=False, defaults=\
{'bid':book.bid, 'book':book, 'tags':book_tags})
if not created:
book_related.tags.extend(book_tags)
try:
book_related.save()
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s when save %s at api:%s" %(error, \
error.to_dict(), book_related, self.api.url))
except Exception, error:
logger.debug("[ERROR]%s: BookRelated.objects.get_or_create(uid=%s ..."\
%(error, self.api.api_id))
def getUserRecommendations(self):
results = self.service.GetUserRecommendations(self.api.url)
if results:
recommendations, count, user = results
else:
return False
if not recommendations or not user:
return False
recommendation_datas = []
for recommendation in recommendations:
recommendation_data = recommendation.__dict__.get('_data')
for key, value in recommendation_data.items():
if value is None:
recommendation_data.pop(key)
recommendation_data['author'] = DBRef('user', user.id)
recommendation_datas.append(recommendation_data)
if recommendation_datas:
ids = self.db.recommendation.insert(recommendation_datas, check_keys=False)
for index, object_id in enumerate(ids):
recommendations[index].id = object_id
try:
user_related = None
user_related, created = UserRelated.objects.get_or_create(
uid=user.uid, auto_save=False,
defaults={'uid':user.uid, 'recommendations':recommendations})
if not created:
user_related.recommendations.extend(recommendations)
try:
user_related.save()
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s when save %s at api:%s" \
%(error, error.to_dict(), user_related, self.api.url))
except Exception, error:
logger.debug("[ERROR]%s: UserRelated.objects.get_or_create(uid=%s ..."\
%(error, self.api.api_id))
def getBookCollections(self):
results = self.service.GetBookCollections(self.api.url)
if results:
collections, count, user = results
else:
return False
if not user or not collections:
return False
collection_datas = []
for collection in collections:
collection_data = collection.__dict__.get('_data')
for key, value in collection_data.items():
if value is None:
collection_data.pop(key)
collection_data['author'] = DBRef('user', user.id)
if collection.book:
#添加该书到book_tag集合
db = MongoDB.getConnection('mining')
for tag in collection.tags:
db.book_tag.update({'title':tag},
{'$addToSet':{'books':collection.book.bid},
'$inc':{'count':1}},
upsert=True)
collection_data['book'] = DBRef('book', collection.book.id)
collection_datas.append(collection_data)
if collection_datas:
ids = self.db.book_collection.insert(collection_datas, check_keys=False)
for index, object_id in enumerate(ids):
collections[index].id = object_id
try:
user_book_related = None
user_book_related, created = UserBookRelated.objects.get_or_create(\
uid=self.api.api_id, auto_save=False, defaults=\
{'uid':self.api.api_id, 'book_collections':collections})
if not created:
user_book_related.book_collections.extend(collections)
try:
user_book_related.save()
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s when save %s at api:%s" %(error, \
error.to_dict(), user_book_related, self.api.url))
except Exception, error:
logger.debug("[ERROR]%s: UserBookRelated.objects.get_or_create(uid=%s ..."\
%(error, self.api.api_id))
if user_book_related and created:
try:
user.user_book_related = user_book_related
user.save()
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s when save %s at api:%s" \
%(error, error.to_dict(), user, self.api.url))
def getBookReviews(self):
results = self.service.GetBookReviews(self.api.url)
if results:
book_reviews, count, book = results
else:
return False
if not book or not book_reviews:
return False
book_review_datas = []
indexs = []
for index, book_review in enumerate(book_reviews):
exist = BookReview.objects(rid=book_review.rid)
if exist:
book_reviews[index] = exist[0]
else:
book_review_data = book_review.__dict__.get('_data')
for key, value in book_review_data.items():
if value is None:
book_review_data.pop(key)
if book_review.author:
book_review_data['author'] = DBRef('user', book_review.author.id)
book_review_data['book'] = DBRef('book', book.id)
book_review_datas.append(book_review_data)
indexs.append(index)
if book_review_datas:
ids = self.db.book_review.insert(book_review_datas, check_keys=False)
for i, index in enumerate(indexs):
book_reviews[index].id = ids[i]
try:
book_related = None
book_related, created = BookRelated.objects.get_or_create(\
bid=book.bid, auto_save=False, defaults=\
{'bid':book.bid, 'reviews':book_reviews})
if not created:
book_related.reviews.extend(book_reviews)
try:
book_related.save()
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s when save %s at api:%s" %(error, \
error.to_dict(), book_related, self.api.url))
except Exception, error:
logger.debug("[ERROR]%s: BookRelated.objects.get_or_create(uid=%s ..."\
%(error, self.api.api_id))
if book_related and created:
try:
book.book_related = book_related
book.save()
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s when save %s at api:%s" \
%(error, error.to_dict(), book, self.api.url))
'$inc':{'count':tag.count}},
upsert=True)
try:
book_related = None
book_related, created = BookRelated.objects.get_or_create(\
bid=book.bid, auto_save=False, defaults=\
{'bid':book.bid, 'book':book, 'tags':book_tags})
if not created:
book_related.tags.extend(book_tags)
try:
book_related.save()
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s when save %s at api:%s" %(error, \
error.to_dict(), book_related, self.api.url))
except Exception, error:
logger.debug("[ERROR]%s: BookRelated.objects.get_or_create(uid=%s ..."\
%(error, self.api.api_id))
if book_related and created:
try:
book.book_related = book_related
book.save()
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s when save %s at api:%s"
%(error, error.to_dict(), book, self.api.url))
#logger.debug('[%s]Api:%s is processed!' %(datetime.now(), self.api.url))
return book_tags
logger.debug("[ERROR]%s: %s when save uid:%s at api:%s" \
%(error, error.to_dict(), user.uid, self.api.url))
try:
uid = self.api.api_id
user_related = None
user_related, created = UserRelated.objects.get_or_create(
uid=uid, auto_save=False,
defaults = {'uid':uid, 'friends':users})
if not created:
user_related.friends.extend(users)
try:
user_related.save()
except mongoengine.base.ValidationError, error:
logger.debug("[ERROR]%s: %s when save %s at api:%s" \
%(error, error.to_dict(), user_related, self.api.url))
except Exception, error:
logger.debug("[ERROR]%s: UserRelated.objects.get_or_create(uid=%s ..."\
%(error, self.api.api_id))
if count:
start_index = int(count.get('startIndex'))
total_results = int(count.get('totalResults'))
items_per_page = int(count.get('itemsPerPage'))
if start_index + items_per_page < total_results:
user_contacts_api = self.api_operation.apiGenerator(\
'USER_CONTACTS', self.api.api_id,\
start_index=start_index+items_per_page)
api = Api(url=user_contacts_api, api_id=self.api.api_id,\
api_type='USER_CONTACTS')
api.save()