def Group(self, visited):
self.grouped.clear()
for link in self.links:
parentLang = GroupLang(link.parentNode.lang, self.params.langIds)
numSiblings = len(link.parentNode.links)
numVisitedSiblings = GetVistedSiblings(link.childNode.urlId,
link.parentNode, visited)
numVisitedSiblings = len(numVisitedSiblings)
matchedSiblings = GetMatchedSiblings(link.childNode.urlId,
link.parentNode, visited)
numMatchedSiblings = len(matchedSiblings)
parentMatched = GetNodeMatched(link.parentNode, visited)
linkLang = GroupLang(link.textLang, self.params.langIds)
#print("numSiblings", numSiblings, numMatchedSiblings, link.childNode.url)
#for sibling in link.parentNode.links:
# print(" sibling", sibling.childNode.url)
key = (parentLang, numSiblings, numVisitedSiblings,
numMatchedSiblings, parentMatched, linkLang)
if key not in self.grouped:
self.grouped[key] = []
self.grouped[key].append(link)
def LinkToKey(self, link, visited):
parentLang = GroupLang(link.parentNode.lang, self.params.langIds)
matchedSiblings = GetMatchedSiblings(link.childNode.urlId, link.parentNode, visited)
numMatchedSiblings = len(matchedSiblings)
key = (parentLang, numMatchedSiblings)
#print("key", key)
return key
def LinkToKey(self, link, visited):
matchedSiblings = GetMatchedSiblings(link.childNode.urlId,
link.parentNode, visited)
numMatchedSiblings = len(matchedSiblings)
if numMatchedSiblings >= self.params.NUM_ACTIONS:
numMatchedSiblings = self.params.NUM_ACTIONS - 1
key = (numMatchedSiblings, )
#print("key", key)
return key
def LinkToKey(self, link, visited):
parentLang = GroupLang(link.parentNode.lang, self.params.langIds)
linkLang = GroupLang(link.textLang, self.params.langIds)
numSiblings = len(link.parentNode.links)
numVisitedSiblings = GetVistedSiblings(link.childNode.urlId, link.parentNode, visited)
numVisitedSiblings = len(numVisitedSiblings)
matchedSiblings = GetMatchedSiblings(link.childNode.urlId, link.parentNode, visited)
numMatchedSiblings = len(matchedSiblings)
parentMatched = GetNodeMatched(link.parentNode, visited)
#key = (parentLang,linkLang)
key = (parentLang, linkLang, numSiblings, numVisitedSiblings, numMatchedSiblings, parentMatched)
#print("key", key)
return key
def AddLink(self, link, visited):
langId = link.parentNode.lang
numSiblings = len(link.parentNode.links)
numVisitedSiblings = GetVistedSiblings(link.childNode.urlId, link.parentNode, visited)
numVisitedSiblings = len(numVisitedSiblings)
matchedSiblings = GetMatchedSiblings(link.childNode.urlId, link.parentNode, visited)
numMatchedSiblings = len(matchedSiblings)
#print("numSiblings", numSiblings, numMatchedSiblings, link.childNode.url)
#for sibling in link.parentNode.links:
# print(" sibling", sibling.childNode.url)
key = (langId,numSiblings, numVisitedSiblings, numMatchedSiblings)
if key not in self.dict:
self.dict[key] = []
self.dict[key].append(link)