def test_with_json(self):
sample = {
"firstName": "jayne",
"lastName": "cobb",
"isMale": True,
"age": 38
}
verify_as_json(sample)
def test_parsing_of_robust04_without_transformation(self):
transformed_document_sample = transform_documents_and_select_entries_by_id(
conf=self.parameters(collection_directory=
'collection_to_doc_vectors/test/data/robust',
pyserini_collection='TrecCollection',
pyserini_generator='JsoupGenerator',
sample=LA_TIMES_SAMPLE))
verify_as_json(transformed_document_sample)
def test_parsing_of_clueweb09_without_transformation(self):
transformed_document_sample = transform_documents_and_select_entries_by_id(
conf=self.parameters(
collection_directory=
'collection_to_doc_vectors/test/data/clueweb09-sample',
pyserini_collection='ClueWeb09Collection',
pyserini_generator='JsoupGenerator',
sample=CLUEWEB_SAMPLE))
verify_as_json(transformed_document_sample)
def test_reconstruction_with_single_input_line(self):
run_file = [
'''[
{"system":"s1", "query":690, "docid": "FT944-12903", "rank": 1, "score": 10.0},
{"system":"s1", "query":690, "docid": "LA110789-0114", "rank": 2, "score": 9.0},
{"system":"s1", "query":690, "docid": "FT931-16546", "rank": 3, "score": 8.0}]'''
]
reduce_output = open('test/BuildTrainingSetTest.test_reducer_on_map_output_with_single_input_line.approved.txt')\
.read().split('\n')
verify_as_json(
self.bla(self.system_ranks(run_file),
self.parse_relevance('test/test-qrels.txt'),
self.parse_bm25_run('test/test-run.txt'), reduce_output))
def test_with_main_content_extraction_and_word_vectors(self):
transformed_document_sample = transform_documents_and_select_entries_by_id(
conf=self.parameters(extract_main_content=True,
transform_to_word_vectors=True))
verify_as_json(transformed_document_sample)
def test_transformation_to_word_vectors(self):
transformed_document_sample = transform_documents_and_select_entries_by_id(
conf=self.parameters(transform_to_word_vectors=True))
verify_as_json(transformed_document_sample)
def test_no_transformation(self):
transformed_document_sample = transform_documents_and_select_entries_by_id(
conf=self.parameters())
verify_as_json(transformed_document_sample)
def test_web_qrel_251_300_run_file_count(cls):
actual = list_run_files_of_qrel_file(
'artificial-qrels/qrels.inofficial.duplicate-free.web.251-300.txt')
verify_as_json(sorted(actual))
def test_web_qrel_1_50_run_file_count(cls):
actual = list_run_files_of_qrel_file(
'resources/artificial-qrels/qrels.inofficial.duplicate-free.web.1-50.txt'
)
verify_as_json(sorted(actual))
def test_web_2012_run_file_count(cls):
actual = list_run_files_of_qrel_file('qrels-web-2012')
verify_as_json(sorted(actual))
def test_generation_for_small_sample_qrels(self):
actual = generate_pairs_from_qrels('test/sample-small-qrels.txt')
verify_as_json([i for i in actual])