def test_should_tag_single_token_within_partial_prediction_at_same_scale(self):
token_1 = SimpleToken(TOKEN_TEXT_1)
structured_document = SimpleStructuredDocument(lines=[SimpleLine([token_1])])
structured_document.set_bounding_box(
structured_document.get_pages()[0],
DEFAULT_BOUNDING_BOX
)
structured_document.set_bounding_box(
structured_document.get_pages()[0],
BoundingBox(0, 0, DEFAULT_WIDTH * 10, DEFAULT_HEIGHT * 10)
)
structured_document.set_bounding_box(
token_1,
BoundingBox(0, 0, DEFAULT_WIDTH, DEFAULT_HEIGHT)
)
annotated_image = filled_image(
BG_COLOR, {TAG_1: COLOR_1},
width=DEFAULT_WIDTH * 10,
height=DEFAULT_HEIGHT * 10
)
fill_rect(
annotated_image,
BoundingBox(0, 0, DEFAULT_WIDTH, DEFAULT_HEIGHT),
COLOR_1
)
annotate_structured_document_using_predicted_images(
structured_document,
[annotated_image]
)
assert structured_document.get_tag(token_1, scope=CV_TAG_SCOPE) == TAG_1
def test_should_ignore_block_without_bounding_box(self):
token = SimpleToken('test')
structured_document = SimpleStructuredDocument(
lines=[SimpleLine([token])])
structured_document.set_tag(token, TAG1)
blocks = annotation_document_page_to_annotation_blocks(
structured_document,
structured_document.get_pages()[0])
assert len(blocks) == 0
def test_should_not_tag_single_token_not_within_prediction(self):
token_1 = SimpleToken(TOKEN_TEXT_1)
structured_document = SimpleStructuredDocument(
lines=[SimpleLine([token_1])])
structured_document.set_bounding_box(
structured_document.get_pages()[0], DEFAULT_BOUNDING_BOX)
structured_document.set_bounding_box(token_1, DEFAULT_BOUNDING_BOX)
annotate_structured_document_using_predicted_images(
structured_document, [filled_image(BG_COLOR, {TAG_1: COLOR_1})])
assert structured_document.get_tag(token_1, scope=CV_TAG_SCOPE) is None
def test_should_strip_tag_prefix(self):
token = SimpleToken('test',
tag=TAG1,
tag_prefix=B_TAG_PREFIX,
bounding_box=DEFAULT_BOUNDING_BOX)
assert token.get_tag() == B_TAG_PREFIX + TAG1
structured_document = SimpleStructuredDocument(
lines=[SimpleLine([token])])
blocks = annotation_document_page_to_annotation_blocks(
structured_document,
structured_document.get_pages()[0])
assert [b.tag for b in blocks] == [TAG1]
def test_should_convert_single_token_to_block_with_same_bounding_box(self):
token = SimpleToken('test',
tag=TAG1,
bounding_box=DEFAULT_BOUNDING_BOX)
structured_document = SimpleStructuredDocument(
lines=[SimpleLine([token])])
blocks = annotation_document_page_to_merged_blocks(
structured_document,
structured_document.get_pages()[0])
assert len(blocks) == 1
block = blocks[0]
assert block.tag == TAG1
assert block.bounding_box == DEFAULT_BOUNDING_BOX