更新sohu静态文件保存

zhanghe06 · zhanghe06 · commit 6f52cb78a555 · 2015-10-29T11:00:42.000+08:00
diff --git a/fuck/sohu.py b/fuck/sohu.py
@@ -14,6 +14,9 @@
 url_list = [root_url]  # &#29228;&#34411;&#24453;&#35775;&#38382;url&#21015;&#34920;
 url_visited_list = []  # &#29228;&#34411;&#24050;&#35775;&#38382;url&#21015;&#34920;
 
+file_path = 'sohu.txt'
+url_file = open(file_path, 'a')
+
 
 def read_src(src_url=None):
     """
@@ -46,25 +49,36 @@ def print_static(src):
     if src.startswith('http://') and src.startswith('/') and (src.endswith('.jpg') or src.endswith('.gif') or src.endswith('.png')):
         if src.startswith('/'):
             src = ''.join([root_url, src])
-        print '&#22270;&#29255;&#65306;%s' % src
+            url_file.write(src)
+            url_file.write('\n')
+            print '&#22270;&#29255;&#65306;%s' % src
     if src.endswith('.js'):
         if src.startswith('/'):
             src = ''.join([root_url, src])
-        print 'js&#25991;&#20214;&#65306;%s' % src
+            url_file.write(src)
+            url_file.write('\n')
+            print 'js&#25991;&#20214;&#65306;%s' % src
     if src.endswith('.css'):
         if src.startswith('/'):
             src = ''.join([root_url, src])
-        print 'css&#25991;&#20214;&#65306;%s' % src
+            url_file.write(src)
+            url_file.write('\n')
+            print 'css&#25991;&#20214;&#65306;%s' % src
 
 
 def run():
     """
     &#20027;&#31243;&#24207;
     """
     try:
+        count = 0
         while len(url_list) > 0:
             read_src(url_list.pop(0))
+            count += 1
+            if count % 10 == 0:
+                url_file.flush()
     except KeyboardInterrupt:
+        url_file.close()
         print '&#31243;&#24207;&#36864;&#20986;'
 
 
