| 1 | """CGI-savvy HTTP Server.
|
|---|
| 2 |
|
|---|
| 3 | This module builds on SimpleHTTPServer by implementing GET and POST
|
|---|
| 4 | requests to cgi-bin scripts.
|
|---|
| 5 |
|
|---|
| 6 | If the os.fork() function is not present (e.g. on Windows),
|
|---|
| 7 | os.popen2() is used as a fallback, with slightly altered semantics; if
|
|---|
| 8 | that function is not present either (e.g. on Macintosh), only Python
|
|---|
| 9 | scripts are supported, and they are executed by the current process.
|
|---|
| 10 |
|
|---|
| 11 | In all cases, the implementation is intentionally naive -- all
|
|---|
| 12 | requests are executed sychronously.
|
|---|
| 13 |
|
|---|
| 14 | SECURITY WARNING: DON'T USE THIS CODE UNLESS YOU ARE INSIDE A FIREWALL
|
|---|
| 15 | -- it may execute arbitrary Python code or external programs.
|
|---|
| 16 |
|
|---|
| 17 | Note that status code 200 is sent prior to execution of a CGI script, so
|
|---|
| 18 | scripts cannot send other status codes such as 302 (redirect).
|
|---|
| 19 | """
|
|---|
| 20 |
|
|---|
| 21 |
|
|---|
| 22 | __version__ = "0.4"
|
|---|
| 23 |
|
|---|
| 24 | __all__ = ["CGIHTTPRequestHandler"]
|
|---|
| 25 |
|
|---|
| 26 | import os
|
|---|
| 27 | import sys
|
|---|
| 28 | import urllib
|
|---|
| 29 | import BaseHTTPServer
|
|---|
| 30 | import SimpleHTTPServer
|
|---|
| 31 | import select
|
|---|
| 32 | import copy
|
|---|
| 33 |
|
|---|
| 34 |
|
|---|
| 35 | class CGIHTTPRequestHandler(SimpleHTTPServer.SimpleHTTPRequestHandler):
|
|---|
| 36 |
|
|---|
| 37 | """Complete HTTP server with GET, HEAD and POST commands.
|
|---|
| 38 |
|
|---|
| 39 | GET and HEAD also support running CGI scripts.
|
|---|
| 40 |
|
|---|
| 41 | The POST command is *only* implemented for CGI scripts.
|
|---|
| 42 |
|
|---|
| 43 | """
|
|---|
| 44 |
|
|---|
| 45 | # Determine platform specifics
|
|---|
| 46 | have_fork = hasattr(os, 'fork')
|
|---|
| 47 | have_popen2 = hasattr(os, 'popen2')
|
|---|
| 48 | have_popen3 = hasattr(os, 'popen3')
|
|---|
| 49 |
|
|---|
| 50 | # Make rfile unbuffered -- we need to read one line and then pass
|
|---|
| 51 | # the rest to a subprocess, so we can't use buffered input.
|
|---|
| 52 | rbufsize = 0
|
|---|
| 53 |
|
|---|
| 54 | def do_POST(self):
|
|---|
| 55 | """Serve a POST request.
|
|---|
| 56 |
|
|---|
| 57 | This is only implemented for CGI scripts.
|
|---|
| 58 |
|
|---|
| 59 | """
|
|---|
| 60 |
|
|---|
| 61 | if self.is_cgi():
|
|---|
| 62 | self.run_cgi()
|
|---|
| 63 | else:
|
|---|
| 64 | self.send_error(501, "Can only POST to CGI scripts")
|
|---|
| 65 |
|
|---|
| 66 | def send_head(self):
|
|---|
| 67 | """Version of send_head that support CGI scripts"""
|
|---|
| 68 | if self.is_cgi():
|
|---|
| 69 | return self.run_cgi()
|
|---|
| 70 | else:
|
|---|
| 71 | return SimpleHTTPServer.SimpleHTTPRequestHandler.send_head(self)
|
|---|
| 72 |
|
|---|
| 73 | def is_cgi(self):
|
|---|
| 74 | """Test whether self.path corresponds to a CGI script.
|
|---|
| 75 |
|
|---|
| 76 | Returns True and updates the cgi_info attribute to the tuple
|
|---|
| 77 | (dir, rest) if self.path requires running a CGI script.
|
|---|
| 78 | Returns False otherwise.
|
|---|
| 79 |
|
|---|
| 80 | If any exception is raised, the caller should assume that
|
|---|
| 81 | self.path was rejected as invalid and act accordingly.
|
|---|
| 82 |
|
|---|
| 83 | The default implementation tests whether the normalized url
|
|---|
| 84 | path begins with one of the strings in self.cgi_directories
|
|---|
| 85 | (and the next character is a '/' or the end of the string).
|
|---|
| 86 | """
|
|---|
| 87 | collapsed_path = _url_collapse_path(self.path)
|
|---|
| 88 | dir_sep = collapsed_path.find('/', 1)
|
|---|
| 89 | head, tail = collapsed_path[:dir_sep], collapsed_path[dir_sep+1:]
|
|---|
| 90 | if head in self.cgi_directories:
|
|---|
| 91 | self.cgi_info = head, tail
|
|---|
| 92 | return True
|
|---|
| 93 | return False
|
|---|
| 94 |
|
|---|
| 95 | cgi_directories = ['/cgi-bin', '/htbin']
|
|---|
| 96 |
|
|---|
| 97 | def is_executable(self, path):
|
|---|
| 98 | """Test whether argument path is an executable file."""
|
|---|
| 99 | return executable(path)
|
|---|
| 100 |
|
|---|
| 101 | def is_python(self, path):
|
|---|
| 102 | """Test whether argument path is a Python script."""
|
|---|
| 103 | head, tail = os.path.splitext(path)
|
|---|
| 104 | return tail.lower() in (".py", ".pyw")
|
|---|
| 105 |
|
|---|
| 106 | def run_cgi(self):
|
|---|
| 107 | """Execute a CGI script."""
|
|---|
| 108 | dir, rest = self.cgi_info
|
|---|
| 109 |
|
|---|
| 110 | i = rest.find('/')
|
|---|
| 111 | while i >= 0:
|
|---|
| 112 | nextdir = rest[:i]
|
|---|
| 113 | nextrest = rest[i+1:]
|
|---|
| 114 |
|
|---|
| 115 | scriptdir = self.translate_path(nextdir)
|
|---|
| 116 | if os.path.isdir(scriptdir):
|
|---|
| 117 | dir, rest = nextdir, nextrest
|
|---|
| 118 | i = rest.find('/')
|
|---|
| 119 | else:
|
|---|
| 120 | break
|
|---|
| 121 |
|
|---|
| 122 | # find an explicit query string, if present.
|
|---|
| 123 | i = rest.rfind('?')
|
|---|
| 124 | if i >= 0:
|
|---|
| 125 | rest, query = rest[:i], rest[i+1:]
|
|---|
| 126 | else:
|
|---|
| 127 | query = ''
|
|---|
| 128 |
|
|---|
| 129 | # dissect the part after the directory name into a script name &
|
|---|
| 130 | # a possible additional path, to be stored in PATH_INFO.
|
|---|
| 131 | i = rest.find('/')
|
|---|
| 132 | if i >= 0:
|
|---|
| 133 | script, rest = rest[:i], rest[i:]
|
|---|
| 134 | else:
|
|---|
| 135 | script, rest = rest, ''
|
|---|
| 136 |
|
|---|
| 137 | scriptname = dir + '/' + script
|
|---|
| 138 | scriptfile = self.translate_path(scriptname)
|
|---|
| 139 | if not os.path.exists(scriptfile):
|
|---|
| 140 | self.send_error(404, "No such CGI script (%r)" % scriptname)
|
|---|
| 141 | return
|
|---|
| 142 | if not os.path.isfile(scriptfile):
|
|---|
| 143 | self.send_error(403, "CGI script is not a plain file (%r)" %
|
|---|
| 144 | scriptname)
|
|---|
| 145 | return
|
|---|
| 146 | ispy = self.is_python(scriptname)
|
|---|
| 147 | if not ispy:
|
|---|
| 148 | if not (self.have_fork or self.have_popen2 or self.have_popen3):
|
|---|
| 149 | self.send_error(403, "CGI script is not a Python script (%r)" %
|
|---|
| 150 | scriptname)
|
|---|
| 151 | return
|
|---|
| 152 | if not self.is_executable(scriptfile):
|
|---|
| 153 | self.send_error(403, "CGI script is not executable (%r)" %
|
|---|
| 154 | scriptname)
|
|---|
| 155 | return
|
|---|
| 156 |
|
|---|
| 157 | # Reference: http://hoohoo.ncsa.uiuc.edu/cgi/env.html
|
|---|
| 158 | # XXX Much of the following could be prepared ahead of time!
|
|---|
| 159 | env = copy.deepcopy(os.environ)
|
|---|
| 160 | env['SERVER_SOFTWARE'] = self.version_string()
|
|---|
| 161 | env['SERVER_NAME'] = self.server.server_name
|
|---|
| 162 | env['GATEWAY_INTERFACE'] = 'CGI/1.1'
|
|---|
| 163 | env['SERVER_PROTOCOL'] = self.protocol_version
|
|---|
| 164 | env['SERVER_PORT'] = str(self.server.server_port)
|
|---|
| 165 | env['REQUEST_METHOD'] = self.command
|
|---|
| 166 | uqrest = urllib.unquote(rest)
|
|---|
| 167 | env['PATH_INFO'] = uqrest
|
|---|
| 168 | env['PATH_TRANSLATED'] = self.translate_path(uqrest)
|
|---|
| 169 | env['SCRIPT_NAME'] = scriptname
|
|---|
| 170 | if query:
|
|---|
| 171 | env['QUERY_STRING'] = query
|
|---|
| 172 | host = self.address_string()
|
|---|
| 173 | if host != self.client_address[0]:
|
|---|
| 174 | env['REMOTE_HOST'] = host
|
|---|
| 175 | env['REMOTE_ADDR'] = self.client_address[0]
|
|---|
| 176 | authorization = self.headers.getheader("authorization")
|
|---|
| 177 | if authorization:
|
|---|
| 178 | authorization = authorization.split()
|
|---|
| 179 | if len(authorization) == 2:
|
|---|
| 180 | import base64, binascii
|
|---|
| 181 | env['AUTH_TYPE'] = authorization[0]
|
|---|
| 182 | if authorization[0].lower() == "basic":
|
|---|
| 183 | try:
|
|---|
| 184 | authorization = base64.decodestring(authorization[1])
|
|---|
| 185 | except binascii.Error:
|
|---|
| 186 | pass
|
|---|
| 187 | else:
|
|---|
| 188 | authorization = authorization.split(':')
|
|---|
| 189 | if len(authorization) == 2:
|
|---|
| 190 | env['REMOTE_USER'] = authorization[0]
|
|---|
| 191 | # XXX REMOTE_IDENT
|
|---|
| 192 | if self.headers.typeheader is None:
|
|---|
| 193 | env['CONTENT_TYPE'] = self.headers.type
|
|---|
| 194 | else:
|
|---|
| 195 | env['CONTENT_TYPE'] = self.headers.typeheader
|
|---|
| 196 | length = self.headers.getheader('content-length')
|
|---|
| 197 | if length:
|
|---|
| 198 | env['CONTENT_LENGTH'] = length
|
|---|
| 199 | referer = self.headers.getheader('referer')
|
|---|
| 200 | if referer:
|
|---|
| 201 | env['HTTP_REFERER'] = referer
|
|---|
| 202 | accept = []
|
|---|
| 203 | for line in self.headers.getallmatchingheaders('accept'):
|
|---|
| 204 | if line[:1] in "\t\n\r ":
|
|---|
| 205 | accept.append(line.strip())
|
|---|
| 206 | else:
|
|---|
| 207 | accept = accept + line[7:].split(',')
|
|---|
| 208 | env['HTTP_ACCEPT'] = ','.join(accept)
|
|---|
| 209 | ua = self.headers.getheader('user-agent')
|
|---|
| 210 | if ua:
|
|---|
| 211 | env['HTTP_USER_AGENT'] = ua
|
|---|
| 212 | co = filter(None, self.headers.getheaders('cookie'))
|
|---|
| 213 | if co:
|
|---|
| 214 | env['HTTP_COOKIE'] = ', '.join(co)
|
|---|
| 215 | # XXX Other HTTP_* headers
|
|---|
| 216 | # Since we're setting the env in the parent, provide empty
|
|---|
| 217 | # values to override previously set values
|
|---|
| 218 | for k in ('QUERY_STRING', 'REMOTE_HOST', 'CONTENT_LENGTH',
|
|---|
| 219 | 'HTTP_USER_AGENT', 'HTTP_COOKIE', 'HTTP_REFERER'):
|
|---|
| 220 | env.setdefault(k, "")
|
|---|
| 221 |
|
|---|
| 222 | self.send_response(200, "Script output follows")
|
|---|
| 223 |
|
|---|
| 224 | decoded_query = query.replace('+', ' ')
|
|---|
| 225 |
|
|---|
| 226 | if self.have_fork:
|
|---|
| 227 | # Unix -- fork as we should
|
|---|
| 228 | args = [script]
|
|---|
| 229 | if '=' not in decoded_query:
|
|---|
| 230 | args.append(decoded_query)
|
|---|
| 231 | nobody = nobody_uid()
|
|---|
| 232 | self.wfile.flush() # Always flush before forking
|
|---|
| 233 | pid = os.fork()
|
|---|
| 234 | if pid != 0:
|
|---|
| 235 | # Parent
|
|---|
| 236 | pid, sts = os.waitpid(pid, 0)
|
|---|
| 237 | # throw away additional data [see bug #427345]
|
|---|
| 238 | while select.select([self.rfile], [], [], 0)[0]:
|
|---|
| 239 | if not self.rfile.read(1):
|
|---|
| 240 | break
|
|---|
| 241 | if sts:
|
|---|
| 242 | self.log_error("CGI script exit status %#x", sts)
|
|---|
| 243 | return
|
|---|
| 244 | # Child
|
|---|
| 245 | try:
|
|---|
| 246 | try:
|
|---|
| 247 | os.setuid(nobody)
|
|---|
| 248 | except os.error:
|
|---|
| 249 | pass
|
|---|
| 250 | os.dup2(self.rfile.fileno(), 0)
|
|---|
| 251 | os.dup2(self.wfile.fileno(), 1)
|
|---|
| 252 | os.execve(scriptfile, args, env)
|
|---|
| 253 | except:
|
|---|
| 254 | self.server.handle_error(self.request, self.client_address)
|
|---|
| 255 | os._exit(127)
|
|---|
| 256 |
|
|---|
| 257 | else:
|
|---|
| 258 | # Non Unix - use subprocess
|
|---|
| 259 | import subprocess
|
|---|
| 260 | cmdline = [scriptfile]
|
|---|
| 261 | if self.is_python(scriptfile):
|
|---|
| 262 | interp = sys.executable
|
|---|
| 263 | if interp.lower().endswith("w.exe"):
|
|---|
| 264 | # On Windows, use python.exe, not pythonw.exe
|
|---|
| 265 | interp = interp[:-5] + interp[-4:]
|
|---|
| 266 | cmdline = [interp, '-u'] + cmdline
|
|---|
| 267 | if '=' not in query:
|
|---|
| 268 | cmdline.append(query)
|
|---|
| 269 |
|
|---|
| 270 | self.log_message("command: %s", subprocess.list2cmdline(cmdline))
|
|---|
| 271 | try:
|
|---|
| 272 | nbytes = int(length)
|
|---|
| 273 | except (TypeError, ValueError):
|
|---|
| 274 | nbytes = 0
|
|---|
| 275 | p = subprocess.Popen(cmdline,
|
|---|
| 276 | stdin = subprocess.PIPE,
|
|---|
| 277 | stdout = subprocess.PIPE,
|
|---|
| 278 | stderr = subprocess.PIPE,
|
|---|
| 279 | env = env
|
|---|
| 280 | )
|
|---|
| 281 | if self.command.lower() == "post" and nbytes > 0:
|
|---|
| 282 | data = self.rfile.read(nbytes)
|
|---|
| 283 | else:
|
|---|
| 284 | data = None
|
|---|
| 285 | # throw away additional data [see bug #427345]
|
|---|
| 286 | while select.select([self.rfile._sock], [], [], 0)[0]:
|
|---|
| 287 | if not self.rfile._sock.recv(1):
|
|---|
| 288 | break
|
|---|
| 289 | stdout, stderr = p.communicate(data)
|
|---|
| 290 | self.wfile.write(stdout)
|
|---|
| 291 | if stderr:
|
|---|
| 292 | self.log_error('%s', stderr)
|
|---|
| 293 | p.stderr.close()
|
|---|
| 294 | p.stdout.close()
|
|---|
| 295 | status = p.returncode
|
|---|
| 296 | if status:
|
|---|
| 297 | self.log_error("CGI script exit status %#x", status)
|
|---|
| 298 | else:
|
|---|
| 299 | self.log_message("CGI script exited OK")
|
|---|
| 300 |
|
|---|
| 301 |
|
|---|
| 302 | def _url_collapse_path(path):
|
|---|
| 303 | """
|
|---|
| 304 | Given a URL path, remove extra '/'s and '.' path elements and collapse
|
|---|
| 305 | any '..' references and returns a colllapsed path.
|
|---|
| 306 |
|
|---|
| 307 | Implements something akin to RFC-2396 5.2 step 6 to parse relative paths.
|
|---|
| 308 | The utility of this function is limited to is_cgi method and helps
|
|---|
| 309 | preventing some security attacks.
|
|---|
| 310 |
|
|---|
| 311 | Returns: A tuple of (head, tail) where tail is everything after the final /
|
|---|
| 312 | and head is everything before it. Head will always start with a '/' and,
|
|---|
| 313 | if it contains anything else, never have a trailing '/'.
|
|---|
| 314 |
|
|---|
| 315 | Raises: IndexError if too many '..' occur within the path.
|
|---|
| 316 |
|
|---|
| 317 | """
|
|---|
| 318 | # Similar to os.path.split(os.path.normpath(path)) but specific to URL
|
|---|
| 319 | # path semantics rather than local operating system semantics.
|
|---|
| 320 | path_parts = path.split('/')
|
|---|
| 321 | head_parts = []
|
|---|
| 322 | for part in path_parts[:-1]:
|
|---|
| 323 | if part == '..':
|
|---|
| 324 | head_parts.pop() # IndexError if more '..' than prior parts
|
|---|
| 325 | elif part and part != '.':
|
|---|
| 326 | head_parts.append( part )
|
|---|
| 327 | if path_parts:
|
|---|
| 328 | tail_part = path_parts.pop()
|
|---|
| 329 | if tail_part:
|
|---|
| 330 | if tail_part == '..':
|
|---|
| 331 | head_parts.pop()
|
|---|
| 332 | tail_part = ''
|
|---|
| 333 | elif tail_part == '.':
|
|---|
| 334 | tail_part = ''
|
|---|
| 335 | else:
|
|---|
| 336 | tail_part = ''
|
|---|
| 337 |
|
|---|
| 338 | splitpath = ('/' + '/'.join(head_parts), tail_part)
|
|---|
| 339 | collapsed_path = "/".join(splitpath)
|
|---|
| 340 |
|
|---|
| 341 | return collapsed_path
|
|---|
| 342 |
|
|---|
| 343 |
|
|---|
| 344 | nobody = None
|
|---|
| 345 |
|
|---|
| 346 | def nobody_uid():
|
|---|
| 347 | """Internal routine to get nobody's uid"""
|
|---|
| 348 | global nobody
|
|---|
| 349 | if nobody:
|
|---|
| 350 | return nobody
|
|---|
| 351 | try:
|
|---|
| 352 | import pwd
|
|---|
| 353 | except ImportError:
|
|---|
| 354 | return -1
|
|---|
| 355 | try:
|
|---|
| 356 | nobody = pwd.getpwnam('nobody')[2]
|
|---|
| 357 | except KeyError:
|
|---|
| 358 | nobody = 1 + max(map(lambda x: x[2], pwd.getpwall()))
|
|---|
| 359 | return nobody
|
|---|
| 360 |
|
|---|
| 361 |
|
|---|
| 362 | def executable(path):
|
|---|
| 363 | """Test for executable file."""
|
|---|
| 364 | try:
|
|---|
| 365 | st = os.stat(path)
|
|---|
| 366 | except os.error:
|
|---|
| 367 | return False
|
|---|
| 368 | return st.st_mode & 0111 != 0
|
|---|
| 369 |
|
|---|
| 370 |
|
|---|
| 371 | def test(HandlerClass = CGIHTTPRequestHandler,
|
|---|
| 372 | ServerClass = BaseHTTPServer.HTTPServer):
|
|---|
| 373 | SimpleHTTPServer.test(HandlerClass, ServerClass)
|
|---|
| 374 |
|
|---|
| 375 |
|
|---|
| 376 | if __name__ == '__main__':
|
|---|
| 377 | test()
|
|---|
