def test_should_not_merge_right_adjacent_block_with_same_different_tag(
self):
block1 = AnnotationBlock(TAG1, DEFAULT_BOUNDING_BOX)
block2 = AnnotationBlock(
TAG2, block1.bounding_box.move_by(block1.bounding_box.width, 0))
merged_blocks = merge_blocks([block1, block2])
assert [b.tag for b in merged_blocks] == [TAG1, TAG2]
def test_should_merge_right_adjacent_block_with_same_tag(self):
block1 = AnnotationBlock(TAG1, DEFAULT_BOUNDING_BOX)
block2 = AnnotationBlock(
TAG1, block1.bounding_box.move_by(block1.bounding_box.width, 0))
merged_blocks = merge_blocks([block1, block2])
assert [b.tag for b in merged_blocks] == [TAG1]
assert merged_blocks[0].bounding_box == block1.bounding_box.include(
block2.bounding_box)
def test_should_not_merge_too_far_away_block_with_same_tag(self):
block1 = AnnotationBlock(TAG1, DEFAULT_BOUNDING_BOX)
block2 = AnnotationBlock(
TAG1,
block1.bounding_box.move_by(
block1.bounding_box.width + DEFAULT_NEARBY_TOLERANCE + 1, 0))
merged_blocks = merge_blocks([block1, block2],
nearby_tolerance=DEFAULT_NEARBY_TOLERANCE)
assert [b.tag for b in merged_blocks] == [TAG1, TAG1]
def test_should_merge_right_nearby_block_with_same_tag(self):
block1 = AnnotationBlock(TAG1, DEFAULT_BOUNDING_BOX)
block2 = AnnotationBlock(
TAG1,
block1.bounding_box.move_by(
block1.bounding_box.width + DEFAULT_NEARBY_TOLERANCE, 0))
merged_blocks = merge_blocks([block1, block2],
nearby_tolerance=DEFAULT_NEARBY_TOLERANCE)
assert [b.tag for b in merged_blocks] == [TAG1]
assert merged_blocks[0].bounding_box == block1.bounding_box.include(
block2.bounding_box)
def test_should_merge_multiple_sequential_right_adjacent_blocks(self):
block1 = AnnotationBlock(TAG1, DEFAULT_BOUNDING_BOX)
block2 = AnnotationBlock(
TAG1, block1.bounding_box.move_by(block1.bounding_box.width, 0))
block3 = AnnotationBlock(
TAG1, block2.bounding_box.move_by(block2.bounding_box.width, 0))
merged_blocks = merge_blocks([block1, block2, block3])
assert [b.tag for b in merged_blocks] == [TAG1]
assert merged_blocks[0].bounding_box == (block1.bounding_box.include(
block2.bounding_box).include(block3.bounding_box))
def svg_page_to_blockified_png_bytes(svg_page, color_map, image_size=None):
structured_document = SvgStructuredDocument(svg_page)
blocks = expand_blocks(
merge_blocks(
annotation_document_page_to_annotation_blocks(
structured_document,
structured_document.get_pages()[0]
)
)
)
viewbox = svg_page.attrib.get('viewBox')
if not viewbox:
raise RuntimeError(
'viewbox missing on svg, available attributes: %s' % svg_page.attrib.keys()
)
_, _, width, height = viewbox.split()
image = annotated_blocks_to_image(
blocks, color_map,
width=float(width), height=float(height), background='white',
scale_to_size=image_size
)
out = BytesIO()
image.save(out, 'png')
return out.getvalue()
def test_should_return_same_single_blocks(self):
block = AnnotationBlock(TAG1, DEFAULT_BOUNDING_BOX)
merged_blocks = merge_blocks([block])
assert merged_blocks == [block]