def __drawTree__(self, treeroot, graph, rootId, names):
childId = len(names)
graph.add_vertices(1)
names.append(util.getTag(treeroot))
if rootId >= 0:
graph.add_edges([(rootId, childId)])
children = list(treeroot)
for child in children:
self.__drawTree__(child, graph, childId, names)
def handle(self, xmltree, xmlnode, dstree, dsnode):
# Determine whether to ignore
if xmlnode.get(IGNORE) == IGNORE_TRUE:
return dsnode
if xmlnode.getparent() != None:
parentTag = util.getTag(xmlnode.getparent())
if self.peudodsnode != None and parentTag != 'p':
self.peudoTrigger = True
dsnode = self.peudodsnode
# Process tag
tag = util.getTag(xmlnode)
if tag == 'em':
return self.__handle_em__(xmlnode, dstree, dsnode)
elif tag == 'html':
return self.__handle_html__(xmlnode, dstree, dsnode)
elif tag == 'title':
return self.__handle_title__(xmlnode, dstree, dsnode)
elif tag == 'abbr':
return self.__handle_abbr__(xmlnode, dstree, dsnode)
elif tag == 'acronym':
return self.__handle_acronym__(xmlnode, dstree, dsnode)
elif tag == 'address':
return self.__handle_address__(xmlnode, dstree, dsnode)
elif tag == 'blockquote':
return self.__handle_blockquote__(xmlnode, dstree, dsnode)
elif tag == 'br':
return self.__handle_br__(xmlnode, dstree, dsnode)
elif tag == 'cite':
return self.__handle_cite__(xmlnode, dstree, dsnode)
elif tag == 'dfn':
return self.__handle_dfn__(xmlnode, dstree, dsnode)
elif tag == 'em':
return self.__handle_em__(xmlnode, dstree, dsnode)
elif tag == 'h1':
return self.__handle_h1__(xmlnode, dstree, dsnode)
elif tag == 'h2':
return self.__handle_h2__(xmlnode, dstree, dsnode)
elif tag == 'h3':
return self.__handle_h3__(xmlnode, dstree, dsnode)
elif tag == 'h4':
return self.__handle_h4__(xmlnode, dstree, dsnode)
elif tag == 'h5':
return self.__handle_h5__(xmlnode, dstree, dsnode)
elif tag == 'h6':
return self.__handle_h6__(xmlnode, dstree, dsnode)
elif tag == 'kbd':
return self.__handle_kbd__(xmlnode, dstree, dsnode)
elif tag == 'p':
return self.__handle_p__(xmlnode, dstree, dsnode)
elif tag == 'pre':
return self.__handle_pre__(xmlnode, dstree, dsnode)
elif tag == 'q':
return self.__handle_q__(xmlnode, dstree, dsnode)
elif tag == 'samp':
return self.__handle_samp__(xmlnode, dstree, dsnode)
elif tag == 'span':
return self.__handle_span__(xmlnode, dstree, dsnode)
elif tag == 'strong':
return self.__handle_strong__(xmlnode, dstree, dsnode)
elif tag == 'var':
return self.__handle_var__(xmlnode, dstree, dsnode)
elif tag == 'a':
return self.__handle_a__(xmlnode, dstree, dsnode)
elif tag == 'dl':
return self.__handle_dl__(xmlnode, dstree, dsnode)
elif tag == 'dt':
return self.__handle_dt__(xmlnode, dstree, dsnode)
elif tag == 'dd':
return self.__handle_dd__(xmlnode, dstree, dsnode)
elif tag == 'ol':
return self.__handle_ol__(xmlnode, dstree, dsnode)
elif tag == 'ul':
return self.__handle_ul__(xmlnode, dstree, dsnode)
elif tag == 'li':
return self.__handle_li__(xmlnode, dstree, dsnode)
elif tag == 'b':
return self.__handle_b__(xmlnode, dstree, dsnode)
elif tag == 'big':
return self.__handle_big__(xmlnode, dstree, dsnode)
elif tag == 'i':
return self.__handle_i__(xmlnode, dstree, dsnode)
elif tag == 'small':
return self.__handle_small__(xmlnode, dstree, dsnode)
elif tag == 'sub':
return self.__handle_sub__(xmlnode, dstree, dsnode)
elif tag == 'sup':
return self.__handle_sup__(xmlnode, dstree, dsnode)
elif tag == 'tt':
return self.__handle_tt__(xmlnode, dstree, dsnode)
elif tag == 'del':
return self.__handle_del__(xmlnode, dstree, dsnode)
elif tag == 'ins':
return self.__handle_ins__(xmlnode, dstree, dsnode)
elif tag == 'caption':
return self.__handle_caption__(xmlnode, dstree, dsnode)
elif tag == 'table':
return self.__handle_table__(xmlnode, dstree, dsnode)
elif tag == 'img':
return self.__handle_img__(xmlnode, dstree, dsnode)
elif tag == 'tr':
# Ignore subtree
xmlchildtree = etree.ElementTree(xmlnode)
for xmlchildnode in xmlchildtree.iter(tag=etree.Element):
xmlchildnode.set(IGNORE, IGNORE_TRUE)
return dsnode
else:
return dsnode
def isIndent(self, xmlnode):
tag = util.getTag(xmlnode)
return tag == 'blockquote'
def getLevel(node):
return util.getTag(node)