From 9da1dac6e8585602706fcf4d46e6e51f9dfca7a4 Mon Sep 17 00:00:00 2001
From: lzjqsdd <lzj7179@163.com>
Date: Wed, 27 Apr 2016 23:21:28 +0800
Subject: [PATCH] none

---
 ml/Cut.py          |  24 ++++++++++++++++++++++++
 ml/Global.py       |   1 +
 ml/Global.pyc      | Bin 270 -> 307 bytes
 ml/InverseIndex.py |   7 ++++++-
 4 files changed, 31 insertions(+), 1 deletion(-)
 create mode 100644 ml/Cut.py

diff --git a/ml/Cut.py b/ml/Cut.py
new file mode 100644
index 0000000..630f2dc
--- /dev/null
+++ b/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:	
+				
+				
diff --git a/ml/Global.py b/ml/Global.py
index aa10081..f13f78e 100644
--- a/ml/Global.py
+++ b/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"
diff --git a/ml/Global.pyc b/ml/Global.pyc
index 6a914e7f1ae3af4673586db18578734587258566..7bb9ddaeab82d342eebbab10cf5f2fc0d4a0e6ab 100644
GIT binary patch
delta 115
zcmeBU+RS9b{F#@_<+@yWG6NJa0%->zE|vijDGUrzK$4Lmijg6Oi6M%KA%&SCikTsW
zg&|mjb)uQSKrshUnVx=Trd~-!iJu0`#10*v5^kVaW?orpQE}?T4PjhtKrS022m%0e
CLKL0=

delta 81
zcmdnY)W>AT{F#^Qx`$kNG6NJa0%->zE*1k4DGUrzK$4Lmijg6Oi6M%KA%&SCSc7Gv
ZjX$%W2J^((Iuox4aj*bYvN5tT0st-M4Kx4%

diff --git a/ml/InverseIndex.py b/ml/InverseIndex.py
index b985422..847ddfc 100644
--- a/ml/InverseIndex.py
+++ b/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()