def __init__(self, name, tree, encoder=None): self.name = name self.blocks = get_textblocks(tree, encoder) 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 = [] return
def __init__(self, name, tree, encoder=None):
self.name = name
self.blocks = get_textblocks(tree, encoder)
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 = []
return
def identify_layout(self, tree, pat_threshold, strict=True):
top = (None, None)
blocks = get_textblocks(tree, encoder=self.encoder)
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
def identify_layout(self, tree, pat_threshold, strict=True):
top = (None, None)
blocks = get_textblocks(tree, encoder=self.encoder)
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