def __init__(self, fatherElement, centralDic):
#print '>> construct DataNode'
self.doter = DataNodeDoter()
self.doter.init(self)
#text content of a tag
self.type = 'datanode'
self.setName('datas')
#data dic
self.dic = centralDic
self.datadic = Datas(centralDic)
#data container for each page
self.pagedatas = []
self._imp = 0
self.fatherElement = fatherElement
class DataNode:
'''
DataNode is a special StyleNode
a container for nodes like b p img a and text
'''
def __init__(self, fatherElement, centralDic):
#print '>> construct DataNode'
self.doter = DataNodeDoter()
self.doter.init(self)
#text content of a tag
self.type = 'datanode'
self.setName('datas')
#data dic
self.dic = centralDic
self.datadic = Datas(centralDic)
#data container for each page
self.pagedatas = []
self._imp = 0
self.fatherElement = fatherElement
def addFeatures(self, features):
features = [f for f in features]
self.datadic.addFeatures(features)
_dic = Datas(self.dic)
_dic.addFeatures(features)
#print 'pagedic', _dic.list.datas
self.pagedatas.append(_dic)
def setName(self, data):
self._name = str(data)
def hasData(self):
return self.datadic.hasData()
def getName(self):
return self._name
def getP(self):
'''
get frequency
'''
#return self.fatherElement.getP()
return 1
def getCompImp(self):
if self._imp: return self._imp
m = len(self.pagedatas)
l = self.datadic.size()
if not l: return 0
'''
print '-' * 50
print 'm: dicsize: ', m
print 'nodedic: ', self.datadic.list.datas
print 'pagedatas:'
'''
'''
for p in self.pagedatas:
print p.list.datas
'''
def P(i):
'''
print '-' * 50
print 'P(i): ' + '-'*30
print 'm: ', m
'''
n = 0
data_index = self.datadic[i]
print 'data_index: ', data_index
li = []
for page in self.pagedatas:
#print 'find pageindex in page', data_index, page.list.datas
res = np.where(page.list.datas == data_index )
#print 'find res:', res
try:
i = res[0][0]
li.append(1)
except:
li.append(0)
pass
n = sum(li)
if not n: n=1
print 'n, m : %d, %d' % (n, m)
return [i/n for i in li]
def H(i):
if m == 1: return 0
res = 0
for p in P(i):
if not p: continue
res -= p * math.log(p, m)
return res
if m ==1: return 1
res = sum(
[H(i) for i in range(l)]
)
#res = 1 - res / l
res = 1 - res/l
self._imp = res
print 'H(i): ', self._imp
return res
def _addData(self, data):
self.datas.append(data)
self.nums.append(1)
def _incNum(self, pos):
self.nums[pos] += 1
def __str__(self):
res = ''
res += self.doter.initDotNode() + '\n'
self.doter.incIndex()
return res