def test_mobileclick_download_training_data(self): ''' Download a subset of the full training data ''' import os from mobileclick.scripts.mobileclick_download_training_data import main files = [ './data/MC2-training/en/1C2-E-iunits.tsv', './data/MC2-training-documents/1C2-E.HTML/1C2-E-0001-1.html' ] # check if files have been downloaded if not all([os.path.exists(f) for f in files]): main(istest=True) # files to be downloaded for f in files: self.assertTrue(os.path.exists(f))
def test_mobileclick_random_ranking_method(self): ''' Random ranking method ''' from mobileclick.scripts.mobileclick_random_ranking_method import main queryfilepath = create_query_subset( './data/MC2-training/en/1C2-E-queries.tsv', './data/MC2-training-documents/1C2-E.INDX/') main(['--runname', 'test', '--query', queryfilepath, '--iunit', './data/MC2-training/en/1C2-E-iunits.tsv', '--index', './data/MC2-training-documents/1C2-E.INDX/', '--pagedir', './data/MC2-training-documents/1C2-E.HTML/', '--outputdir', './runs']) drop_tmp_files()
def test_mobileclick_lang_model_summarization_method(self): ''' Lang Model summarization method ''' from mobileclick.scripts.mobileclick_lang_model_summarization_method import main queryfilepath = create_query_subset( './data/MC2-training/en/1C2-E-queries.tsv', './data/MC2-training-documents/1C2-E.INDX/') intentfilepath = create_tmp_intent_file(queryfilepath) main(['--runname', 'test', '--query', queryfilepath, '--iunit', './data/MC2-training/en/1C2-E-iunits.tsv', '--index', './data/MC2-training-documents/1C2-E.INDX/', '--intent', intentfilepath, '--pagedir', './data/MC2-training-documents/1C2-E.HTML/', '--outputdir', './runs', '--language', 'english']) drop_tmp_files()