none

ml/Cut.py

@@ -0,0 +1,24 @@
+#encoding:utf8
+import json
+import sys
+import Global
+
+class Cut:
+	def __init__(self):
+#every 30 news in a flie
+		self.size = 30
+		self.file_data = open(Global.data_dir)
+
+	def cutfile(self,file):
+		
+		num = 0
+		While(True):
+			line = self.file_data.readline()
+			if not line:
+				break
+
+			num+=1
+			filename = str(num/self.size)+'.txt'
+			if num%self.size != 0:	
+				
+				

ml/Global.py

@@ -1,3 +1,4 @@
 data_dir = "../news_spider/title.json"
 db_dir = "../news_spider/news.db"
 stopword_dir="./stopword.txt"
+inverse_dir="./ii.txt"

ml/InverseIndex.py

@@ -12,6 +12,7 @@ class InverseIndex:
 	def __init__(self):
 		self.file_data= open(Global.data_dir)
 		self.file_sw = open(Global.stopword_dir)
+		self.ii = open(Global.inverse_dir,'wb')
 		self.stopword=[]
 		self.worddict = dict()
 
@@ -24,7 +25,6 @@ class InverseIndex:
 			print line,
 
 	def loaddata(self):
-
 		self.loadsw()
 		count=0
 		while True:
@@ -38,8 +38,13 @@ class InverseIndex:
 				if w not in self.worddict:
 					self.worddict[w] = []	
 				if w not in self.stopword:
+					print w,
 					self.worddict[w].append(count)
 
+	def write2file(self):
+		for w in self.worddict:
+			ii.write(w+' '+str(worddict[w])+'\n')
 
 ii = InverseIndex()
 ii.loaddata()
+ii.write2file()