http proxy server程序tinyhttpproxy.py

tinyhttpproxy.py使用方法：

1）很好用，下载然后在后台运行。只依赖于基本的python modules，运行的时候不需要root权限。

proxy [-p port] [-l logfile] [-dh] [allowed_client_name ...]]
 
   -p   - Port to bind to
   -l   - Path to logfile. If not specified, STDOUT is used
   -d   - Run in the background

TinyHTTPProxy.py的源代码:

1. #!/usr/bin/python
2.  
3. __doc__ = """Tiny HTTP Proxy.
4.  
5. This module implements GET, HEAD, POST, PUT and DELETE methods
6. on BaseHTTPServer, and behaves as an HTTP proxy. The CONNECT
7. method is also implemented experimentally, but has not been
8. tested yet.
9.  
10. Any help will be greatly appreciated. SUZUKI Hisao
11.  
12. 2009/11/23 - Modified by Mitko Haralanov
13. * Added very simple FTP file retrieval
14. * Added custom logging methods
15. * Added code to make this a standalone application
16. """
17.  
18. __version__ = "0.3.1"
19.  
20. import BaseHTTPServer, select, socket, SocketServer, urlparse
21. import logging
22. import logging.handlers
23. import getopt
24. import sys
25. import os
26. import signal
27. import threading
28. from types import FrameType, CodeType
29. from time import sleep
30. import ftplib
31.  
32. DEFAULT_LOG_FILENAME = "proxy.log"
33.  
34. class ProxyHandler (BaseHTTPServer.BaseHTTPRequestHandler):
35. __base = BaseHTTPServer.BaseHTTPRequestHandler
36. __base_handle = __base.handle
37.  
38. server_version = "TinyHTTPProxy/" + __version__
39. rbufsize = 0 # self.rfile Be unbuffered
40.  
41. def handle(self):
42. (ip, port) = self.client_address
43. self.server.logger.log (logging.INFO, "Request from '%s'", ip)
44. if hasattr(self, 'allowed_clients') and ip not in self.allowed_clients:
45. self.raw_requestline = self.rfile.readline()
46. if self.parse_request(): self.send_error(403)
47. else:
48. self.__base_handle()
49.  
50. def _connect_to(self, netloc, soc):
51. i = netloc.find(':')
52. if i >= 0:
53. host_port = netloc[:i], int(netloc[i+1:])
54. else:
55. host_port = netloc, 80
56. self.server.logger.log (logging.INFO, "connect to %s:%d", host_port[0], host_port[1])
57. try: soc.connect(host_port)
58. except socket.error, arg:
59. try: msg = arg[1]
60. except: msg = arg
61. self.send_error(404, msg)
62. return 0
63. return 1
64.  
65. def do_CONNECT(self):
66. soc = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
67. try:
68. if self._connect_to(self.path, soc):
69. self.log_request(200)
70. self.wfile.write(self.protocol_version +
71. " 200 Connection established\r\n")
72. self.wfile.write("Proxy-agent: %s\r\n" % self.version_string())
73. self.wfile.write("\r\n")
74. self._read_write(soc, 300)
75. finally:
76. soc.close()
77. self.connection.close()
78.  
79. def do_GET(self):
80. (scm, netloc, path, params, query, fragment) = urlparse.urlparse(
81. self.path, 'http')
82. if scm not in ('http', 'ftp') or fragment or not netloc:
83. self.send_error(400, "bad url %s" % self.path)
84. return
85. soc = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
86. try:
87. if scm == 'http':
88. if self._connect_to(netloc, soc):
89. self.log_request()
90. soc.send("%s %s %s\r\n" % (self.command,
91. urlparse.urlunparse(('', '', path,
92. params, query,
93. '')),
94. self.request_version))
95. self.headers['Connection'] = 'close'
96. del self.headers['Proxy-Connection']
97. for key_val in self.headers.items():
98. soc.send("%s: %s\r\n" % key_val)
99. soc.send("\r\n")
100. self._read_write(soc)
101. elif scm == 'ftp':
102. # fish out user and password information
103. i = netloc.find ('@')
104. if i >= 0:
105. login_info, netloc = netloc[:i], netloc[i+1:]
106. try: user, passwd = login_info.split (':', 1)
107. except ValueError: user, passwd = "anonymous", None
108. else: user, passwd ="anonymous", None
109. self.log_request ()
110. try:
111. ftp = ftplib.FTP (netloc)
112. ftp.login (user, passwd)
113. if self.command == "GET":
114. ftp.retrbinary ("RETR %s"%path, self.connection.send)
115. ftp.quit ()
116. except Exception, e:
117. self.server.logger.log (logging.WARNING, "FTP Exception: %s",
118. e)
119. finally:
120. soc.close()
121. self.connection.close()
122.  
123. def _read_write(self, soc, max_idling=20, local=False):
124. iw = [self.connection, soc]
125. local_data = ""
126. ow = []
127. count = 0
128. while 1:
129. count += 1
130. (ins, _, exs) = select.select(iw, ow, iw, 1)
131. if exs: break
132. if ins:
133. for i in ins:
134. if i is soc: out = self.connection
135. else: out = soc
136. data = i.recv(8192)
137. if data:
138. if local: local_data += data
139. else: out.send(data)
140. count = 0
141. if count == max_idling: break
142. if local: return local_data
143. return None
144.  
145. do_HEAD = do_GET
146. do_POST = do_GET
147. do_PUT = do_GET
148. do_DELETE=do_GET
149.  
150. def log_message (self, format, *args):
151. self.server.logger.log (logging.INFO, "%s %s", self.address_string (),
152. format % args)
153.  
154. def log_error (self, format, *args):
155. self.server.logger.log (logging.ERROR, "%s %s", self.address_string (),
156. format % args)
157.  
158. class ThreadingHTTPServer (SocketServer.ThreadingMixIn,
159. BaseHTTPServer.HTTPServer):
160. def __init__ (self, server_address, RequestHandlerClass, logger=None):
161. BaseHTTPServer.HTTPServer.__init__ (self, server_address,
162. RequestHandlerClass)
163. self.logger = logger
164.  
165. def logSetup (filename, log_size, daemon):
166. logger = logging.getLogger ("TinyHTTPProxy")
167. logger.setLevel (logging.INFO)
168. if not filename:
169. if not daemon:
170. # display to the screen
171. handler = logging.StreamHandler ()
172. else:
173. handler = logging.handlers.RotatingFileHandler (DEFAULT_LOG_FILENAME,
174. maxBytes=(log_size*(1<<20)),
175. backupCount=5)
176. else:
177. handler = logging.handlers.RotatingFileHandler (filename,
178. maxBytes=(log_size*(1<<20)),
179. backupCount=5)
180. fmt = logging.Formatter ("[%(asctime)-12s.%(msecs)03d] "
181. "%(levelname)-8s {%(name)s %(threadName)s}"
182. " %(message)s",
183. "%Y-%m-%d %H:%M:%S")
184. handler.setFormatter (fmt)
185.  
186. logger.addHandler (handler)
187. return logger
188.  
189. def usage (msg=None):
190. if msg: print msg
191. print sys.argv[0], "[-p port] [-l logfile] [-dh] [allowed_client_name ...]]"
192. print
193. print " -p - Port to bind to"
194. print " -l - Path to logfile. If not specified, STDOUT is used"
195. print " -d - Run in the background"
196. print
197.  
198. def handler (signo, frame):
199. while frame and isinstance (frame, FrameType):
200. if frame.f_code and isinstance (frame.f_code, CodeType):
201. if "run_event" in frame.f_code.co_varnames:
202. frame.f_locals["run_event"].set ()
203. return
204. frame = frame.f_back
205.  
206. def daemonize (logger):
207. class DevNull (object):
208. def __init__ (self): self.fd = os.open ("/dev/null", os.O_WRONLY)
209. def write (self, *args, **kwargs): return 0
210. def read (self, *args, **kwargs): return 0
211. def fileno (self): return self.fd
212. def close (self): os.close (self.fd)
213. class ErrorLog:
214. def __init__ (self, obj): self.obj = obj
215. def write (self, string): self.obj.log (logging.ERROR, string)
216. def read (self, *args, **kwargs): return 0
217. def close (self): pass
218.  
219. if os.fork () != 0:
220. ## allow the child pid to instanciate the server
221. ## class
222. sleep (1)
223. sys.exit (0)
224. os.setsid ()
225. fd = os.open ('/dev/null', os.O_RDONLY)
226. if fd != 0:
227. os.dup2 (fd, 0)
228. os.close (fd)
229. null = DevNull ()
230. log = ErrorLog (logger)
231. sys.stdout = null
232. sys.stderr = log
233. sys.stdin = null
234. fd = os.open ('/dev/null', os.O_WRONLY)
235. #if fd != 1: os.dup2 (fd, 1)
236. os.dup2 (sys.stdout.fileno (), 1)
237. if fd != 2: os.dup2 (fd, 2)
238. if fd not in (1, 2): os.close (fd)
239.  
240. def main ():
241. logfile = None
242. daemon = False
243. max_log_size = 20
244. port = 8000
245. allowed = []
246. run_event = threading.Event ()
247. local_hostname = socket.gethostname ()
248.  
249. try: opts, args = getopt.getopt (sys.argv[1:], "l:dhp:", [])
250. except getopt.GetoptError, e:
251. usage (str (e))
252. return 1
253.  
254. for opt, value in opts:
255. if opt == "-p": port = int (value)
256. if opt == "-l": logfile = value
257. if opt == "-d": daemon = not daemon
258. if opt == "-h":
259. usage ()
260. return 0
261.  
262. # setup the log file
263. logger = logSetup (logfile, max_log_size, daemon)
264.  
265. if daemon:
266. daemonize (logger)
267. signal.signal (signal.SIGINT, handler)
268.  
269. if args:
270. allowed = []
271. for name in args:
272. client = socket.gethostbyname(name)
273. allowed.append(client)
274. logger.log (logging.INFO, "Accept: %s (%s)" % (client, name))
275. ProxyHandler.allowed_clients = allowed
276. else:
277. logger.log (logging.INFO, "Any clients will be served...")
278.  
279. server_address = (socket.gethostbyname (local_hostname), port)
280. ProxyHandler.protocol = "HTTP/1.0"
281. httpd = ThreadingHTTPServer (server_address, ProxyHandler, logger)
282. sa = httpd.socket.getsockname ()
283. print "Servering HTTP on", sa[0], "port", sa[1]
284. req_count = 0
285. while not run_event.isSet ():
286. try:
287. httpd.handle_request ()
288. req_count += 1
289. if req_count == 1000:
290. logger.log (logging.INFO, "Number of active threads: %s",
291. threading.activeCount ())
292. req_count = 0
293. except select.error, e:
294. if e[0] == 4 and run_event.isSet (): pass
295. else:
296. logger.log (logging.CRITICAL, "Errno: %d - %s", e[0], e[1])
297. logger.log (logging.INFO, "Server shutdown")
298. return 0
299.  
300. if __name__ == '__main__':
301. sys.exit (main ())
302. ---------------------------

以下都支持 http/https proxy，
trafficserver :http://trafficserver.apache.org/

mitmproxy http://mitmproxy.org/

还有跨平台且有GUI的 Charles proxy: 'Charles Web Debugging Proxy • HTTP Monitor / HTTP Proxy / HTTPS & SSL Proxy / Reverse Proxy', https://www.charlesproxy.com/
Charles 实在是太棒了！但还是不如fiddler方便，前者还是更侧重抓包而不是代理，还是 fiddler 更强大些。如果有条件还是在 windows 上开 fiddler 的代理吧。

如果就想用 nginx 做正向代理，可以参考:
https://github.com/chobits/ngx_http_proxy_connect_module

HTTP代理存在两种。
一种是基于RFC 7230 - Hypertext Transfer Protocol (HTTP/1.1): Message Syntax and Routing以及其他相关RFC文档中描述的普通代理，代理服务器充当的是一个中间的组件，通过在请求URI设置的对应authority来指定目标服务器。

另一种是基于RFC 2817 - Upgrading to TLS Within HTTP/1.1的5.2 Requesting a Tunnel with CONNECT（draft-luotonen-web-proxy-tunneling-01原草案）实现的TCP隧道代理。与上面的区别就在于“TCP隧道”，这种代理可以以HTTP协议的方式来实现理论上任意TCP之上的应用层协议代理，两端之间的通讯都是在HTTP的body部分完成的。因此就可以完成基于TLS的HTTPS通讯，被加密过的数据直接在body中传输。

Nginx不支持第二种代理，因此不能完成https代理。Nginx的作者对于这个功能有明确的回复（http://forum.nginx.org/read.php?2,15124,15256），表示不会在近期实现，并且推荐使用Squid: "Not in near future: there is alreay good forward proxy Squid."