def test_chunks_fail(document_factory, document, groundtruth):
# document and groundtruth not the same length
groundtruth.chunks.append(document_factory.create('test chunk 4'))
pair = DocGroundtruthPair(doc=document, groundtruth=groundtruth)
with pytest.raises(AssertionError):
for _ in pair.chunks:
pass
def test_docgroundtruth_pair():
def add_matches(doc: jina_pb2.DocumentProto, num_matches):
for idx in range(num_matches):
match = doc.matches.add()
match.adjacency = doc.adjacency + 1
def add_chunks(doc: jina_pb2.DocumentProto, num_chunks):
for idx in range(num_chunks):
chunk = doc.chunks.add()
chunk.granularity = doc.granularity + 1
doc = jina_pb2.DocumentProto()
gt = jina_pb2.DocumentProto()
add_matches(doc, 3)
add_matches(gt, 3)
add_chunks(doc, 3)
add_chunks(gt, 3)
pair = DocGroundtruthPair(doc, gt)
j = 0
for chunk_pair in pair.chunks:
assert chunk_pair.doc.granularity == 1
assert chunk_pair.groundtruth.granularity == 1
j += 1
k = 0
for match_pair in pair.matches:
assert match_pair.doc.adjacency == 1
assert match_pair.groundtruth.adjacency == 1
k += 1
assert j == 3
assert k == 3
def ground_truth_pairs():
num_docs = 10
pairs = []
for idx in range(num_docs):
doc = Document(embedding=np.array([1, 1]))
gt = Document(embedding=np.array([2, 2]))
pairs.append(DocGroundtruthPair(doc=doc, groundtruth=gt))
return pairs
def ground_truth_pairs():
num_docs = 10
def add_matches(doc: jina_pb2.DocumentProto, num_matches):
for idx in range(num_matches):
match = doc.matches.add()
match.tags['id'] = idx
match.score.value = idx
pairs = []
for idx in range(num_docs):
doc = jina_pb2.DocumentProto()
gt = jina_pb2.DocumentProto()
add_matches(doc, num_docs)
add_matches(gt, num_docs)
pairs.append(DocGroundtruthPair(doc=doc, groundtruth=gt))
return pairs
def test_chunks_success(document, groundtruth):
pair = DocGroundtruthPair(doc=document, groundtruth=groundtruth)
assert isinstance(pair.chunks, types.GeneratorType)
for _ in pair.chunks:
pass
def test_init(document, groundtruth):
assert DocGroundtruthPair(doc=document, groundtruth=groundtruth)
def create(self):
return DocGroundtruthPair(
doc=doc_with_field_type.create(),
groundtruth=groundtruth_with_field_type.create()
)