class Tagger():
def __init__(self):
self.repository = Repository()
def tag(self, path_to_document_root):
content = Content(path_to_document_root)
self.repository.add_all(content.all_documents())
document_tags = defaultdict(list)
for document in self.repository.all_docs():
tags = self.significant_tags_per_doc(document)
document_tags[document.id()] = tags
return document_tags
def significant_tags_per_doc(self, document_to_tag):
tokens = document_to_tag.all_tokens()
tf_for_given_document = document_to_tag.term_frequencies()
tf_across_documents = self.repository.term_frequencies_across_docs()
document_frequencies = self.repository.document_frequency()
document_tf_vectors = self.repository.term_freqency_vectors()
token_weight = defaultdict(float)
for token in tokens:
median_weight = tf_across_documents[token] / float(document_frequencies[token]) # TFIDF
list_of_deviation_square = map(lambda tf_hash: math.pow((tf_hash[token] - median_weight), 2), document_tf_vectors)
token_weight[token] = tf_for_given_document[token] * math.sqrt(reduce(lambda x, y: x + y, list_of_deviation_square))
sorted_tag_tuples = map(lambda key: (key, token_weight[key]), sorted(token_weight, key=token_weight.get, reverse=True))
return sorted_tag_tuples[:10]
def __init__(self):
self.repository = Repository()
def setUp(self):
self.rep = Repository()
class RepositoryTest(unittest.TestCase):
def setUp(self):
self.rep = Repository()
def test_add_document(self):
doc1 = Document("d1", "doc 1 text")
self.rep.add(doc1)
self.assertEqual(len(self.rep.all_docs()), 1)
self.rep.add(Document("d2", "doc 2 text"))
self.assertEqual(len(self.rep.all_docs()), 2)
def test_add_only_if_does_not_exist(self):
self.rep.add(Document("d1", "doc 1 text"))
with self.assertRaises(RuntimeError):
self.rep.add(Document("d1", "doc 1 text changed"))
def test_find_doc_by_id(self):
self.rep.add(Document("d1", "doc 1 text"))
self.rep.add(Document("d2", "doc 2 text"))
doc1 = self.rep.find_doc_by_id("d1")
doc2 = self.rep.find_doc_by_id("d2")
self.assertEqual(doc1.id(), "d1")
self.assertEqual(doc2.text(), "doc 2 text")
