def test_multimodal_document_fail_bad_doctype(visual_embedding):
# the multimodal document don't have any chunks
with pytest.raises(BadDocType):
md = MultimodalDocument()
md.tags['id'] = 1
md.embedding = visual_embedding
md.modality_content_mapping
def test_from_content_category_mapping(modality_content_mapping):
md = MultimodalDocument.from_modality_content_mapping(
modality_content_mapping=modality_content_mapping)
assert len(md.modalities) == 2
assert 'visual' and 'textual' in md.modalities
assert len(md.chunks) == 2
assert md.granularity == md.chunks[0].granularity - 1
def test_from_chunks_granularity_2(chunk_5, chunk_6):
md = MultimodalDocument(chunks=[chunk_5, chunk_6])
assert len(md.modalities) == 2
assert 'visual' and 'textual' in md.modalities
assert len(md.chunks) == 2
assert md.granularity == md.chunks[0].granularity - 1
assert md.chunks[0].granularity == 3
def multimodal_generator():
for i in range(0, 5):
document = MultimodalDocument(modality_content_map={
'1': f'aaa {i}',
'2': f'bbb {i}'
})
yield document
def test_from_chunks_success(chunk_1, chunk_2):
md = MultimodalDocument(chunks=[chunk_1, chunk_2])
assert len(md.modalities) == 2
assert 'visual' and 'textual' in md.modalities
assert len(md.chunks) == 2
assert md.granularity == md.chunks[0].granularity - 1
assert md.chunks[0].granularity == 1
def query_generator(image_paths, text_queries):
for image_path, text in zip(image_paths, text_queries):
with open(image_path, 'rb') as fp:
buffer = fp.read()
yield MultimodalDocument(modality_content_map={
'image': buffer,
'text': text
})
def multimodal_document(chunk_1, chunk_2):
with MultimodalDocument() as md:
md.tags['id'] = 1
md.text = b'hello world'
md.embedding = np.random.random([10 + np.random.randint(0, 1)])
md.chunks.add(
chunk_1) # attach a document with embedding and without content
md.chunks.add(
chunk_2) # attach a document with content and without embedding
yield md
def doc_with_multimodal_chunks(embeddings):
doc = MultimodalDocument()
chunk1 = Document()
chunk2 = Document()
chunk3 = Document()
chunk1.modality = 'visual1'
chunk2.modality = 'visual2'
chunk3.modality = 'textual'
chunk1.embedding = embeddings[0]
chunk2.embedding = embeddings[1]
chunk3.embedding = embeddings[2]
chunk1.update_id()
chunk2.update_id()
chunk3.update_id()
doc.update_id()
doc.chunks.append(chunk1)
doc.chunks.append(chunk2)
doc.chunks.append(chunk3)
return doc
def test_from_documents_set():
docs = []
for i in range(0, 3):
doc = MultimodalDocument.from_modality_content_mapping({'modA': f'textA {i}', 'modB': f'textB {i}'})
docs.append(doc)
for doc in MultimodalDocumentSet(docs):
assert len(doc.chunks) == 2
for doc in MultimodalDocumentSet(DocumentSet(docs)):
assert len(doc.chunks) == 2
def test_from_chunks_fail_multiple_granularity(chunk_1, chunk_2, chunk_4):
"""Initialize a :class:`MultimodalDocument` expect to fail which has 3 chunks with different
granularity value, expect all chunks has the same granularity value.
"""
with pytest.raises(BadDocType):
MultimodalDocument.from_chunks(chunks=[chunk_1, chunk_2, chunk_4])
def test_from_chunks_fail_length_mismatch(chunk_1, chunk_2, chunk_3):
"""Initialize a :class:`MultimodalDocument` expect to fail which has 3 chunks
with 2 modalities.
"""
with pytest.raises(LengthMismatchException):
MultimodalDocument.from_chunks(chunks=[chunk_1, chunk_2, chunk_3])
def test_from_chunks_fail_length_mismatch(chunk_1, chunk_2, chunk_3):
"""Initialize a :class:`MultimodalDocument` expect to fail which has 3 chunks
with 2 modalities.
"""
assert not MultimodalDocument(chunks=[chunk_1, chunk_2, chunk_3]).is_valid
def test_assert_granularity(chunk_1, chunk_6):
with pytest.raises(BadDocType):
md = MultimodalDocument(chunks=[chunk_1, chunk_6])
def test_multimodal_document_fail_bad_doctype(visual_embedding):
# the multimodal document don't have any chunks
md = MultimodalDocument()
md.tags['id'] = 1
md.embedding = visual_embedding
assert not md.is_valid
