def test__same_text_correlation(self):
""" Test that for same text correlation is 1"""
_log.info('-' * 80)
# arrange
text1 = "love is rain as long story short"
text2 = text1
dump_file = getInputFile("swiki_knowledge_output.xml")
parsed_file = getOutputFile("swiki_knowledge_output.parsed.xml")
#wdb_file = getOutputFile("swiki_knowledge_output.wdb")
articles = ['Rain', 'Love', 'Tree']
# act
wn.make_dump(dump_file, articles, compress=False)
wn.parse_dump(dump_file, parsed_file)
db_wrapper = wn.build_database_wrapper(parsed_file,
StopWordsStemmer([]))
#self.addCleanup(os.remove, self.tmp_dump_file)
comparer = SemanticComparer(db_wrapper)
correlation = comparer.compare(text1, text2)
_log.info(
test_utils.get_texts_correlation_message(text1, text2,
correlation))
self.assertAlmostEqual(correlation,
1.0,
msg="for same text correlation should be 1")
def test__same_text_correlation(self):
""" Test that for same text correlation is 1"""
_log.info('-'*80)
# arrange
text1 = "love is rain as long story short"
text2 = text1
dump_file = getInputFile("swiki_knowledge_output.xml")
parsed_file = getOutputFile("swiki_knowledge_output.parsed.xml")
#wdb_file = getOutputFile("swiki_knowledge_output.wdb")
articles = ['Rain', 'Love', 'Tree']
# act
wn.make_dump(dump_file, articles, compress=False)
wn.parse_dump(dump_file, parsed_file)
db_wrapper = wn.build_database_wrapper(parsed_file, StopWordsStemmer([]))
#self.addCleanup(os.remove, self.tmp_dump_file)
comparer = SemanticComparer(db_wrapper)
correlation = comparer.compare(text1, text2)
_log.info(test_utils.get_texts_correlation_message(text1, text2, correlation))
self.assertAlmostEqual(correlation, 1.0, msg="for same text correlation should be 1")
def test__save_and_load(self):
wiki_knowledge.parse_dump(self.expected_xml_path, self.tmp_parse_file)
expected_db = wiki_knowledge.build_database_wrapper(self.tmp_parse_file, StopWordsStemmer([]))
wiki_knowledge.save_db_wrapper_to_wdb(expected_db, self.tmp_wdb_file)
actual = wiki_knowledge.load_db_wrapper_from_wdb(self.tmp_wdb_file)
self.assertEqual(expected_db.words_num, actual.words_num, "Mismatch WikiKnowledges number of words")
self.assertEqual(expected_db.title_index, actual.title_index, "Mismatch WikiKnowledges titles")
def test__many_articles(self):
wiki_dump_path = getInputFile("many_articles_dump.xml")
parsed_xml_path = getOutputFile("many_articles_dump.parsed.xml")
wiki_knowledge.parse_dump(wiki_dump_path, parsed_xml_path)
db_wrapper = wiki_knowledge.build_database_wrapper(parsed_xml_path, PorterStemmer())
c = db_wrapper.get_readable_centroid(ibm_licence_text)
print c
def test__save_and_load(self):
wiki_knowledge.parse_dump(self.expected_xml_path, self.tmp_parse_file)
expected_db = wiki_knowledge.build_database_wrapper(
self.tmp_parse_file, StopWordsStemmer([]))
wiki_knowledge.save_db_wrapper_to_wdb(expected_db, self.tmp_wdb_file)
actual = wiki_knowledge.load_db_wrapper_from_wdb(self.tmp_wdb_file)
self.assertEqual(expected_db.words_num, actual.words_num,
"Mismatch WikiKnowledges number of words")
self.assertEqual(expected_db.title_index, actual.title_index,
"Mismatch WikiKnowledges titles")
def test__many_articles(self):
wiki_dump_path = getInputFile("many_articles_dump.xml")
parsed_xml_path = getOutputFile("many_articles_dump.parsed.xml")
wiki_knowledge.parse_dump(wiki_dump_path, parsed_xml_path)
db_wrapper = wiki_knowledge.build_database_wrapper(
parsed_xml_path, PorterStemmer())
c = db_wrapper.get_readable_centroid(ibm_licence_text)
print c
def test__execution(self):
""" This is not exactly a test, but a program execution..."""
text1 = "i love to learn"
text2 = "the world we know"
dump_file = self.tmp_dump_file
wiki_knowledge.make_dump(dump_file, self.expected_articles, compress=False)
wiki_knowledge.parse_dump(dump_file, self.tmp_parse_file)
db_wrapper = wiki_knowledge.build_database_wrapper(self.tmp_parse_file, StopWordsStemmer([]))
#wiki_knowledge = test_utils.Factory.build_wiki_knowledge()
#clean up file created by factory at end
self.addCleanup(os.remove, self.tmp_dump_file)
comparer = semantic_interpreter.SemanticComparer(db_wrapper)
correlation = comparer.compare(text1, text2)
_log.info(test_utils.get_texts_correlation_message(text1, text2, correlation))
def test_number_of_concepts(self):
""" db builder reads parsed xml properly"""
_log.info('-'*80)
# arrange
dump_file = getInputFile("wikidump_Knowledge_Love_War.xml")
parsed_file = getOutputFile("wikidump_Knowledge_Love_War.parsed.xml")
# act
wn.parse_dump(dump_file, parsed_file)
db_wrapper = wn.build_database_wrapper(parsed_file, StopWordsStemmer([]))
titles_count =len(db_wrapper.title_index)
concepts_count =len(db_wrapper.concepts_index)
# assert
self.assertEqual(titles_count, 3, "number of tiltes should be 3, got {0}".format(titles_count))
self.assertEqual(concepts_count, 3, "number of tiltes should be 3, got {0}".format(concepts_count))
def test__execution(self):
""" This is not exactly a test, but a program execution..."""
text1 = "i love to learn"
text2 = "the world we know"
dump_file = self.tmp_dump_file
wiki_knowledge.make_dump(dump_file,
self.expected_articles,
compress=False)
wiki_knowledge.parse_dump(dump_file, self.tmp_parse_file)
db_wrapper = wiki_knowledge.build_database_wrapper(
self.tmp_parse_file, StopWordsStemmer([]))
#wiki_knowledge = test_utils.Factory.build_wiki_knowledge()
#clean up file created by factory at end
self.addCleanup(os.remove, self.tmp_dump_file)
comparer = semantic_interpreter.SemanticComparer(db_wrapper)
correlation = comparer.compare(text1, text2)
_log.info(
test_utils.get_texts_correlation_message(text1, text2,
correlation))
def test_number_of_concepts(self):
""" db builder reads parsed xml properly"""
_log.info('-' * 80)
# arrange
dump_file = getInputFile("wikidump_Knowledge_Love_War.xml")
parsed_file = getOutputFile("wikidump_Knowledge_Love_War.parsed.xml")
# act
wn.parse_dump(dump_file, parsed_file)
db_wrapper = wn.build_database_wrapper(parsed_file,
StopWordsStemmer([]))
titles_count = len(db_wrapper.title_index)
concepts_count = len(db_wrapper.concepts_index)
# assert
self.assertEqual(
titles_count, 3,
"number of tiltes should be 3, got {0}".format(titles_count))
self.assertEqual(
concepts_count, 3,
"number of tiltes should be 3, got {0}".format(concepts_count))
def test__parse_dump(self):
wiki_dump_path = io_tu.getInputFile(io_tu.FilesList.test__parse_tools)
wiki_parsed_dump_path = io_tu.getOutputFile(
io_tu.FilesList.test__parse_tools)
wn.parse_dump(wiki_dump_path, wiki_parsed_dump_path)
def parse(args):
_log.debug("run parse with args={}".format(args))
wiki_knowledge.parse_dump(args.dump, args.output)
def test__parse_dump(self):
wiki_dump_path = io_tu.getInputFile(io_tu.FilesList.test__parse_tools)
wiki_parsed_dump_path = io_tu.getOutputFile(io_tu.FilesList.test__parse_tools)
wn.parse_dump(wiki_dump_path, wiki_parsed_dump_path)