更新url工具方法

zhanghe06 · zhanghe06 · commit 3df72c7af15a · 2015-11-12T14:10:48.000+08:00
diff --git a/tools/url.py b/tools/url.py
@@ -6,31 +6,47 @@
 from urlparse import urljoin
 
 
-def get_next_url(current_url, path):
+def get_next_url(current_url, next_url):
     """
     &#32452;&#35013;url
+    protocol :// hostname[:port] / path / [;parameters][?query]#fragment
     :param current_url:
-    :param path:
+    :param next_url:
     :return:
     """
-    if path is None or path == '':
+    if next_url is None or next_url == '':
         return ''
-    if path.startswith('http'):
-        return path
-    if path.startswith('/'):
-        url = urlparse(current_url)
-        return urlunparse((url.scheme, url.netloc, path, url.params, url.query, url.fragment))
-    return urljoin(current_url, path)
+    if next_url.startswith('http'):
+        return next_url
+    if next_url.startswith('/'):
+        current_url_parse = urlparse(current_url)
+        next_url_parse = urlparse(next_url)
+        return urlunparse((current_url_parse.scheme, current_url_parse.netloc, next_url_parse.path, next_url_parse.params, next_url_parse.query, next_url_parse.fragment))
+    return urljoin(current_url, next_url)
 
 
 def test():
     print get_next_url('http://www.163.com/mail/index.htm', 'http://www.163.com/about.htm')
+    print urljoin('http://www.163.com/mail/index.htm', 'http://www.163.com/about.htm')
+    print '\n',
     print get_next_url('http://www.163.com/mail/index.htm', '/about.htm')
+    print urljoin('http://www.163.com/mail/index.htm', '/about.htm')
+    print '\n',
     print get_next_url('http://www.163.com/mail/index.htm', 'about.htm')
+    print urljoin('http://www.163.com/mail/index.htm', 'about.htm')
+    print '\n',
+    print get_next_url('http://sh.58.com/banjia/?sort=pingfen', '/banjia/pn2/?sort=pingfen')
+    print urljoin('http://sh.58.com/banjia/?sort=pingfen', '/banjia/pn2/?sort=pingfen')
 
 
 if __name__ == '__main__':
     test()
     test_url = 'http://suining.58.com/zhongdiangong/?sort=pingfen'
     print urlparse(test_url).hostname.rstrip('.58.com')
-    print urlparse(test_url).path.strip('/')
+    print urlparse(test_url).path.strip('/')
+
+
+"""
+&#20197;&#19978;&#27979;&#35797;&#32467;&#26524;&#21487;&#20197;&#30475;&#20986;
+&#19968;&#20010;urljoin&#23601;&#25630;&#23450;&#20102;
+"""
