def __init__(self, name, tree, url):
self.name = name
self.url = url
self.blocks = get_textblocks(tree)
self.weight = sum( b.weight for b in self.blocks )
#self.weight_noanchor = sum( b.weight_noanchor for b in self.blocks )
self.anchor_strs = []
self.add_anchor_strs()
return
def identify_layout(self, tree, pat_threshold=0.9, strict=True):
top = (None, None)
blocks = get_textblocks(tree)
if 2 <= self.debug:
tree.dump()
max_weight = sum( b.weight for b in blocks ) * pat_threshold
for pat1 in self.pats:
layout = pat1.match_blocks(blocks, strict=strict)
if layout:
weight = sum( sect.weight for sect in layout )
if max_weight < weight:
top = (pat1, layout)
max_weight = weight
return top
