def addExternDic(sentimentDic, filename, anls):
lines = fileutil.readFile(filename)
for l in lines:
l = l.strip("\n")
if l not in sentimentDic:
sentimentDic[l] = anls
return sentimentDic
def getThemeDic():
filename = "./data/themedic.in"
rows = fileutil.readFile(filename)
themes = set([])
for r in rows:
r = r.strip("\n")
themes.add(r)
return themes
def getTestData(filename):
rows = fileutil.readFile(filename)
rawdata = []
for r in rows:
r = r.strip("\n")
l = r.split(" ")
rawdata.append(l)
return rawdata
def getDegreeDic():
filename = "./data/degree.in"
degree = set([])
if fileutil.checkFileIfExist(filename):
rows = fileutil.readFile(filename)
for r in rows:
r = r.strip("\n")
degree.add(r)
return degree
def getPreDic():
filename = "./data/predic.in"
constPreDic = [
'没有', '不是', '别', '不', '不能', '不如', '不想', '没', '不敢', '本来', '不大', '不要',
'没什么', '无法', '不用', '不然', '非', '不会', '无', '未', '不怎么', '不够', '不算', '减少',
'从不', '不再', '不让', '不见得', '省了', '不服', '不正', '不可', '没法', '不比'
]
pre = set([])
if fileutil.checkFileIfExist(filename):
rows = fileutil.readFile(filename)
for r in rows:
r = r.strip("\n")
pre.add(r)
for r in constPreDic:
pre.add(r)
return pre
def getSentimentDic():
filename = "./data/sentimentdic.in"
sentimentDic = {}
rows = fileutil.readFile(filename)
for row in rows:
row = row.strip("\n")
row = row.split(" ")
word = row[0]
sentiment = row[1]
if word not in sentimentDic:
sentimentDic[word] = sentiment
filenameExNe = "./data/ne.in"
filenameExPo = "./data/po.in"
# sentimentDic = addExternDic(sentimentDic, filenameExNe, str(-1))
# sentimentDic = addExternDic(sentimentDic, filenameExPo, str(1))
return sentimentDic
def getSqlFormat(filename):
return fileutil.readFile(filename)