def test_no_saved_files(self):
generator = WMTRepresentationGenerator(self.wmt_target, self.wmt_source)
data_obj = generator.generate()
self.assertTrue('target' in data_obj)
self.assertTrue('source' in data_obj)
self.assertTrue('tags' in data_obj)
self.assertTrue(len(data_obj['target']) == len(data_obj['source']))
self.assertTrue(len(data_obj['target']) == len(data_obj['tags']))
def test_no_saved_files(self):
generator = WMTRepresentationGenerator(self.wmt_target,
self.wmt_source)
data_obj = generator.generate()
self.assertTrue('target' in data_obj)
self.assertTrue('source' in data_obj)
self.assertTrue('tags' in data_obj)
self.assertTrue(len(data_obj['target']) == len(data_obj['source']))
self.assertTrue(len(data_obj['target']) == len(data_obj['tags']))
def test_save_files(self):
generator = WMTRepresentationGenerator(self.wmt_target, self.wmt_source, tmp_dir=self.tmp_dir, persist=True)
data_obj = generator.generate()
target = os.path.join(self.tmp_dir, 'EN_ES.tgt_ann.train.target')
tags = os.path.join(self.tmp_dir, 'EN_ES.tgt_ann.train.tags')
source = os.path.join(self.tmp_dir, 'EN_ES.source.train.txt')
self.assertTrue(os.path.exists(self.tmp_dir) and os.path.isdir(self.tmp_dir))
self.assertTrue(os.path.exists(target) and os.path.isfile(target))
self.assertTrue(os.path.exists(tags) and os.path.isfile(tags))
self.assertTrue(os.path.exists(source) and os.path.isfile(source))
def test_save_files(self):
generator = WMTRepresentationGenerator(self.wmt_target,
self.wmt_source,
tmp_dir=self.tmp_dir,
persist=True)
data_obj = generator.generate()
target = os.path.join(self.tmp_dir, 'EN_ES.tgt_ann.train.target')
tags = os.path.join(self.tmp_dir, 'EN_ES.tgt_ann.train.tags')
source = os.path.join(self.tmp_dir, 'EN_ES.source.train.txt')
self.assertTrue(
os.path.exists(self.tmp_dir) and os.path.isdir(self.tmp_dir))
self.assertTrue(os.path.exists(target) and os.path.isfile(target))
self.assertTrue(os.path.exists(tags) and os.path.isfile(tags))
self.assertTrue(os.path.exists(source) and os.path.isfile(source))