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__make_dump(self):
#create the dump file
article_title=["Rain"]
articles_expected_set = set(article_title)
wiki_knowledge.make_dump(self.tmp_dump_file,article_title)
actual_titles_set = {wikidoc.title for wikidoc in parse_tools.iterate_wiki_pages(self.tmp_dump_file)}
self.assertEqual(actual_titles_set, articles_expected_set, "title mismatch")
def test__make_dump(self):
#create the dump file
article_title = ["Rain"]
articles_expected_set = set(article_title)
wiki_knowledge.make_dump(self.tmp_dump_file, article_title)
actual_titles_set = {
wikidoc.title
for wikidoc in parse_tools.iterate_wiki_pages(self.tmp_dump_file)
}
self.assertEqual(actual_titles_set, articles_expected_set,
"title mismatch")
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__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 makedump(args):
_log.debug("run makedump with args={}".format(args))
wiki_knowledge.make_dump(args.dumpfile, args.article)