def get_neighbors_below(obj, objs, page):
neighbors = []
for c_obj in objs:
if c_obj is obj or utils.is_center_aligned(c_obj, page):
continue
if (utils.is_in_direction(obj, c_obj, BELOW_ANGLE) and
(utils.is_center_aligned(obj, c_obj) or
utils.is_hor_aligned(obj, c_obj)) and
not utils.has_elements_between(obj, c_obj, objs)):
neighbors.append(c_obj)
return neighbors
def merge_layouts(layout):
BELOW_ANGLE = math.radians(270)
def get_neighbors_below(obj, objs, page):
neighbors = []
for c_obj in objs:
if c_obj is obj or utils.is_center_aligned(c_obj, page):
continue
if (utils.is_in_direction(obj, c_obj, BELOW_ANGLE) and
(utils.is_center_aligned(obj, c_obj) or
utils.is_hor_aligned(obj, c_obj)) and
not utils.has_elements_between(obj, c_obj, objs)):
neighbors.append(c_obj)
return neighbors
layout_objs = layout._objs
#layout_objs = filter(lambda x: isinstance(x, LTTextContainer), layout._objs)
has_updated = True
while has_updated:
has_updated = False
layout_map = {}
for l in layout_objs:
if l not in layout_map:
g = LTTextGroupLRTB([l])
layout_map[l] = g
if utils.is_center_aligned(l, layout):
continue
else:
g = layout_map[l]
neighbors = get_neighbors_below(l, layout_objs, layout)
x_ordered_neighbors = sorted(neighbors, key=lambda x: x.x0)
for neighbor in x_ordered_neighbors:
if neighbor not in g._objs:
has_updated = True
g.add(neighbor)
layout_map[neighbor] = g
layout_objs = sorted(set(layout_map.values()), key=lambda x: -x.y1)
return layout_objs