class CompanyDuckDuckClassifier:
def __init__(self):
self.duckduckgo_search = DuckduckgoSearch(True)
self.company_postfix = ['corp', 'corporation', 'company', 'inc', 'headquarters']
self.cache = Cache(DUCKDUCK_COMPANY_CACHE, INPUT_LANGUAGE)
self.min_hits_to_match = DUCKDUCK_COMPANY_MIN_HITS_TO_MATCH
self.cache.load()
def classify(self, term):
cache_result = self.cache.search_cache(term)
if self.cache.search_cache(term) is not None:
return self.normalize_results(ClassificationResult(term, cache_result.Matches))
result = ClassificationResult(term)
for company_word in self.company_postfix:
term_to_search = term + ' ' + company_word
try:
if self.duckduckgo_search.general_search(term_to_search) != '':
result.Matches['company'] += 1
except Exception, e:
return ClassificationResult(term, {key: -1 for key in categories})
self.cache.update_cache(term, result)
return self.normalize_results(result)
class DuckDuckGoWordOccurrenceClassifier:
def __init__(self):
self.duckduckgo_search = DuckduckgoSearch(False)
self.word_occurrence_classifier = WordOccurrenceClassifier()
self.cache = Cache(DUCKDUCK_WORD_OCCURRENCE_CACHE, INPUT_LANGUAGE)
self.cache.load()
def classify(self, term):
cache_result = self.cache.search_cache(term)
if cache_result is not None:
return self.word_occurrence_classifier.normalize_results(ClassificationResult(term, cache_result.Matches))
try:
search_result = self.duckduckgo_search.general_search(term)
except Exception, e:
return ClassificationResult(term, {key: -1 for key in categories})
result = self.word_occurrence_classifier.calculate_score(term, search_result)
self.cache.update_cache(term, result)
return self.word_occurrence_classifier.normalize_results(result)
class FacebookSearch:
company = {"company", "non-profit organization","organization", "professional services", "product/service", "government organization", "church/religious organization"}
place = {"country", "city", "landmark", "public places", "historical place", "tours/sightseeing", "travel/leisure",
"national park", "neighborhood"}
person = {"politician", "public figure", "government official", "writer", "athlete", "artist", "musician/band",
"news personality", "entertainer", "actor/director", "author", "comedian"}
def __init__(self):
##https://developers.facebook.com/tools/explorer/
user_token = "CAACEdEose0cBAO9m1ZChz3qqjqcS3HBlerOSd3wteZC7EqQYlcbngZCmQtvhKQmMmE1sORZAzZA07PboTUXgskIcZBFKeA05FpASH2hEoZCW4im9ZCuNLAlgnOHc00YM5tZByxLCZBo33JkGmq4aPFDNil7FnJGCI7dxgJVwd1ZApkurkCB8WGdZBSJTdNc6zFoCUXFHAKZCZC8OAF6RSV0ljZAw9TA"
extended_access_token = "CAABfbgndG3ABAOGqO5oQ1HwqVOYrlZB6CofIOciVPgMFC4zIRRk7wJvjrZBTIpFlJ3eTZA72fs4UKmyPgasMZA6MEtAaCSegvAju2zsUXAgaTRCxAfFjwrh9x8ZBLJ4lRlEHVWg0m6ZAWk9mMWdYpAVc27cIZCwi5IXO4t0U2fWkVzbAn0UrWJS"
self.graph = GraphAPI(extended_access_token)
self.fb_cache = Cache(FACEBOOK_CACHE,INPUT_LANGUAGE)
self.fb_cache.load()
self.sleep_count = 0
def shutdown(self):
self.fb_cache.save()
def search_Facebook(self, term):
cache_result = self.fb_cache.search_cache(term)
if cache_result is None:
self.sleep_count +=1
if self.sleep_count == 5:
time.sleep(random.randint(1,10))
self.sleep_count = 0
cnt = [0, 0, 0, 0]
try:
cnt = self.search_user(term, cnt)
cnt = self.search_page(term, cnt)
cnt = self.search_place(term, cnt)
except Exception,e:
#print e
cnt = [-1, -1, -1, -1]
self.fb_cache.update_cache_from_list(term,cnt)
else: