def all(self, dom, marker):
tables = dom.getElementsByTagName('table')
tree = Node().loadNodeTree(dom,0)
itables = []
self.tDfs(tree,itables)
for table in itables:
c = 0
table.result = match(table, self.maxDist,self.height, self.tags)
tr, td, t = self.count_table(table)
#print table.result
d = {}
for o in table.result:
if o:
d.setdefault(o,0)
d[o]+=1
for i in d:
c += d[i]
pred = 0
if c >= 2:
if tr > 0 and td/float(tr) > 1:
pred = 1
self.c(table, pred)
marker.mark(table.dom, 'table')
def dfs(self, node, maxDist, height, tags):
"""
Navega em profundidade na árvore buscando nós adjacentes com distancia
de edição menor que maxDist e os agrupa na mesma componente.
@param Node node: No atual da dfs
@param int esp: Nível da árvore
@param float maxDist: Proporção máxima de diferença para ser agrupado
na mesma componente
"""
node.result = match(node, maxDist, height, tags)
for x in xrange(0,len(node.childNodes)):
if not node.result[x]:
self.dfs(node.childNodes[x], maxDist, height, tags)
def _mark2(self, dom, marker, postProcess=False):
tables = dom.getElementsByTagName('table')
tree = Node().loadNodeTree(dom,0)
itables = []
self.tDfs(tree,itables)
#print 'tables', len(tables)
#print 'itables', len(itables)
for table in itables:
p = False
if postProcess:
(tr,td, t) = self.count_tr_td(table)
else:
(tr, td) = 0,1
if tr > 0 and td/float(tr) > 1:
p = True
if p or not postProcess:
table.result = match(table, self.maxDist,self.height, self.tags)
#print table.result
d = {}
for o in table.result:
if o:
d.setdefault(o,0)
d[o]+=1
c = 0
for i in d:
c += d[i]
if postProcess:
if c >= 2:
print 'mark', td/float(tr)
marker.mark(table.dom,'table')
else:
if c >= 2:
marker.mark(table.dom,'table')
