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 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)