1:mod:`http.server` --- HTTP servers
2===================================
3
4.. module:: http.server
5   :synopsis: HTTP server and request handlers.
6
7**Source code:** :source:`Lib/http/server.py`
8
9.. index::
10   pair: WWW; server
11   pair: HTTP; protocol
12   single: URL
13   single: httpd
14
15--------------
16
17This module defines classes for implementing HTTP servers (Web servers).
18
19One class, :class:`HTTPServer`, is a :class:`socketserver.TCPServer` subclass.
20It creates and listens at the HTTP socket, dispatching the requests to a
21handler.  Code to create and run the server looks like this::
22
23   def run(server_class=HTTPServer, handler_class=BaseHTTPRequestHandler):
24       server_address = ('', 8000)
25       httpd = server_class(server_address, handler_class)
26       httpd.serve_forever()
27
28
29.. class:: HTTPServer(server_address, RequestHandlerClass)
30
31   This class builds on the :class:`~socketserver.TCPServer` class by storing
32   the server address as instance variables named :attr:`server_name` and
33   :attr:`server_port`. The server is accessible by the handler, typically
34   through the handler's :attr:`server` instance variable.
35
36
37The :class:`HTTPServer` must be given a *RequestHandlerClass* on instantiation,
38of which this module provides three different variants:
39
40.. class:: BaseHTTPRequestHandler(request, client_address, server)
41
42   This class is used to handle the HTTP requests that arrive at the server.  By
43   itself, it cannot respond to any actual HTTP requests; it must be subclassed
44   to handle each request method (e.g. GET or POST).
45   :class:`BaseHTTPRequestHandler` provides a number of class and instance
46   variables, and methods for use by subclasses.
47
48   The handler will parse the request and the headers, then call a method
49   specific to the request type. The method name is constructed from the
50   request. For example, for the request method ``SPAM``, the :meth:`do_SPAM`
51   method will be called with no arguments. All of the relevant information is
52   stored in instance variables of the handler.  Subclasses should not need to
53   override or extend the :meth:`__init__` method.
54
55   :class:`BaseHTTPRequestHandler` has the following instance variables:
56
57   .. attribute:: client_address
58
59      Contains a tuple of the form ``(host, port)`` referring to the client's
60      address.
61
62   .. attribute:: server
63
64      Contains the server instance.
65
66   .. attribute:: close_connection
67
68      Boolean that should be set before :meth:`handle_one_request` returns,
69      indicating if another request may be expected, or if the connection should
70      be shut down.
71
72   .. attribute:: requestline
73
74      Contains the string representation of the HTTP request line. The
75      terminating CRLF is stripped. This attribute should be set by
76      :meth:`handle_one_request`. If no valid request line was processed, it
77      should be set to the empty string.
78
79   .. attribute:: command
80
81      Contains the command (request type). For example, ``'GET'``.
82
83   .. attribute:: path
84
85      Contains the request path.
86
87   .. attribute:: request_version
88
89      Contains the version string from the request. For example, ``'HTTP/1.0'``.
90
91   .. attribute:: headers
92
93      Holds an instance of the class specified by the :attr:`MessageClass` class
94      variable. This instance parses and manages the headers in the HTTP
95      request. The :func:`~http.client.parse_headers` function from
96      :mod:`http.client` is used to parse the headers and it requires that the
97      HTTP request provide a valid :rfc:`2822` style header.
98
99   .. attribute:: rfile
100
101      An :class:`io.BufferedIOBase` input stream, ready to read from
102      the start of the optional input data.
103
104   .. attribute:: wfile
105
106      Contains the output stream for writing a response back to the
107      client. Proper adherence to the HTTP protocol must be used when writing to
108      this stream.
109
110      .. versionchanged:: 3.6
111         This is an :class:`io.BufferedIOBase` stream.
112
113   :class:`BaseHTTPRequestHandler` has the following attributes:
114
115   .. attribute:: server_version
116
117      Specifies the server software version.  You may want to override this. The
118      format is multiple whitespace-separated strings, where each string is of
119      the form name[/version]. For example, ``'BaseHTTP/0.2'``.
120
121   .. attribute:: sys_version
122
123      Contains the Python system version, in a form usable by the
124      :attr:`version_string` method and the :attr:`server_version` class
125      variable. For example, ``'Python/1.4'``.
126
127   .. attribute:: error_message_format
128
129      Specifies a format string that should be used by :meth:`send_error` method
130      for building an error response to the client. The string is filled by
131      default with variables from :attr:`responses` based on the status code
132      that passed to :meth:`send_error`.
133
134   .. attribute:: error_content_type
135
136      Specifies the Content-Type HTTP header of error responses sent to the
137      client.  The default value is ``'text/html'``.
138
139   .. attribute:: protocol_version
140
141      This specifies the HTTP protocol version used in responses.  If set to
142      ``'HTTP/1.1'``, the server will permit HTTP persistent connections;
143      however, your server *must* then include an accurate ``Content-Length``
144      header (using :meth:`send_header`) in all of its responses to clients.
145      For backwards compatibility, the setting defaults to ``'HTTP/1.0'``.
146
147   .. attribute:: MessageClass
148
149      Specifies an :class:`email.message.Message`\ -like class to parse HTTP
150      headers.  Typically, this is not overridden, and it defaults to
151      :class:`http.client.HTTPMessage`.
152
153   .. attribute:: responses
154
155      This attribute contains a mapping of error code integers to two-element tuples
156      containing a short and long message. For example, ``{code: (shortmessage,
157      longmessage)}``. The *shortmessage* is usually used as the *message* key in an
158      error response, and *longmessage* as the *explain* key.  It is used by
159      :meth:`send_response_only` and :meth:`send_error` methods.
160
161   A :class:`BaseHTTPRequestHandler` instance has the following methods:
162
163   .. method:: handle()
164
165      Calls :meth:`handle_one_request` once (or, if persistent connections are
166      enabled, multiple times) to handle incoming HTTP requests. You should
167      never need to override it; instead, implement appropriate :meth:`do_\*`
168      methods.
169
170   .. method:: handle_one_request()
171
172      This method will parse and dispatch the request to the appropriate
173      :meth:`do_\*` method.  You should never need to override it.
174
175   .. method:: handle_expect_100()
176
177      When a HTTP/1.1 compliant server receives an ``Expect: 100-continue``
178      request header it responds back with a ``100 Continue`` followed by ``200
179      OK`` headers.
180      This method can be overridden to raise an error if the server does not
181      want the client to continue.  For e.g. server can chose to send ``417
182      Expectation Failed`` as a response header and ``return False``.
183
184      .. versionadded:: 3.2
185
186   .. method:: send_error(code, message=None, explain=None)
187
188      Sends and logs a complete error reply to the client. The numeric *code*
189      specifies the HTTP error code, with *message* as an optional, short, human
190      readable description of the error.  The *explain* argument can be used to
191      provide more detailed information about the error; it will be formatted
192      using the :attr:`error_message_format` attribute and emitted, after
193      a complete set of headers, as the response body.  The :attr:`responses`
194      attribute holds the default values for *message* and *explain* that
195      will be used if no value is provided; for unknown codes the default value
196      for both is the string ``???``. The body will be empty if the method is
197      HEAD or the response code is one of the following: ``1xx``,
198      ``204 No Content``, ``205 Reset Content``, ``304 Not Modified``.
199
200      .. versionchanged:: 3.4
201         The error response includes a Content-Length header.
202         Added the *explain* argument.
203
204   .. method:: send_response(code, message=None)
205
206      Adds a response header to the headers buffer and logs the accepted
207      request. The HTTP response line is written to the internal buffer,
208      followed by *Server* and *Date* headers. The values for these two headers
209      are picked up from the :meth:`version_string` and
210      :meth:`date_time_string` methods, respectively. If the server does not
211      intend to send any other headers using the :meth:`send_header` method,
212      then :meth:`send_response` should be followed by an :meth:`end_headers`
213      call.
214
215      .. versionchanged:: 3.3
216         Headers are stored to an internal buffer and :meth:`end_headers`
217         needs to be called explicitly.
218
219   .. method:: send_header(keyword, value)
220
221      Adds the HTTP header to an internal buffer which will be written to the
222      output stream when either :meth:`end_headers` or :meth:`flush_headers` is
223      invoked. *keyword* should specify the header keyword, with *value*
224      specifying its value. Note that, after the send_header calls are done,
225      :meth:`end_headers` MUST BE called in order to complete the operation.
226
227      .. versionchanged:: 3.2
228         Headers are stored in an internal buffer.
229
230   .. method:: send_response_only(code, message=None)
231
232      Sends the response header only, used for the purposes when ``100
233      Continue`` response is sent by the server to the client. The headers not
234      buffered and sent directly the output stream.If the *message* is not
235      specified, the HTTP message corresponding the response *code*  is sent.
236
237      .. versionadded:: 3.2
238
239   .. method:: end_headers()
240
241      Adds a blank line
242      (indicating the end of the HTTP headers in the response)
243      to the headers buffer and calls :meth:`flush_headers()`.
244
245      .. versionchanged:: 3.2
246         The buffered headers are written to the output stream.
247
248   .. method:: flush_headers()
249
250      Finally send the headers to the output stream and flush the internal
251      headers buffer.
252
253      .. versionadded:: 3.3
254
255   .. method:: log_request(code='-', size='-')
256
257      Logs an accepted (successful) request. *code* should specify the numeric
258      HTTP code associated with the response. If a size of the response is
259      available, then it should be passed as the *size* parameter.
260
261   .. method:: log_error(...)
262
263      Logs an error when a request cannot be fulfilled. By default, it passes
264      the message to :meth:`log_message`, so it takes the same arguments
265      (*format* and additional values).
266
267
268   .. method:: log_message(format, ...)
269
270      Logs an arbitrary message to ``sys.stderr``. This is typically overridden
271      to create custom error logging mechanisms. The *format* argument is a
272      standard printf-style format string, where the additional arguments to
273      :meth:`log_message` are applied as inputs to the formatting. The client
274      ip address and current date and time are prefixed to every message logged.
275
276   .. method:: version_string()
277
278      Returns the server software's version string. This is a combination of the
279      :attr:`server_version` and :attr:`sys_version` attributes.
280
281   .. method:: date_time_string(timestamp=None)
282
283      Returns the date and time given by *timestamp* (which must be ``None`` or in
284      the format returned by :func:`time.time`), formatted for a message
285      header. If *timestamp* is omitted, it uses the current date and time.
286
287      The result looks like ``'Sun, 06 Nov 1994 08:49:37 GMT'``.
288
289   .. method:: log_date_time_string()
290
291      Returns the current date and time, formatted for logging.
292
293   .. method:: address_string()
294
295      Returns the client address.
296
297      .. versionchanged:: 3.3
298         Previously, a name lookup was performed. To avoid name resolution
299         delays, it now always returns the IP address.
300
301
302.. class:: SimpleHTTPRequestHandler(request, client_address, server)
303
304   This class serves files from the current directory and below, directly
305   mapping the directory structure to HTTP requests.
306
307   A lot of the work, such as parsing the request, is done by the base class
308   :class:`BaseHTTPRequestHandler`.  This class implements the :func:`do_GET`
309   and :func:`do_HEAD` functions.
310
311   The following are defined as class-level attributes of
312   :class:`SimpleHTTPRequestHandler`:
313
314   .. attribute:: server_version
315
316      This will be ``"SimpleHTTP/" + __version__``, where ``__version__`` is
317      defined at the module level.
318
319   .. attribute:: extensions_map
320
321      A dictionary mapping suffixes into MIME types. The default is
322      signified by an empty string, and is considered to be
323      ``application/octet-stream``. The mapping is used case-insensitively,
324      and so should contain only lower-cased keys.
325
326   The :class:`SimpleHTTPRequestHandler` class defines the following methods:
327
328   .. method:: do_HEAD()
329
330      This method serves the ``'HEAD'`` request type: it sends the headers it
331      would send for the equivalent ``GET`` request. See the :meth:`do_GET`
332      method for a more complete explanation of the possible headers.
333
334   .. method:: do_GET()
335
336      The request is mapped to a local file by interpreting the request as a
337      path relative to the current working directory.
338
339      If the request was mapped to a directory, the directory is checked for a
340      file named ``index.html`` or ``index.htm`` (in that order). If found, the
341      file's contents are returned; otherwise a directory listing is generated
342      by calling the :meth:`list_directory` method. This method uses
343      :func:`os.listdir` to scan the directory, and returns a ``404`` error
344      response if the :func:`~os.listdir` fails.
345
346      If the request was mapped to a file, it is opened and the contents are
347      returned.  Any :exc:`OSError` exception in opening the requested file is
348      mapped to a ``404``, ``'File not found'`` error. Otherwise, the content
349      type is guessed by calling the :meth:`guess_type` method, which in turn
350      uses the *extensions_map* variable.
351
352      A ``'Content-type:'`` header with the guessed content type is output,
353      followed by a ``'Content-Length:'`` header with the file's size and a
354      ``'Last-Modified:'`` header with the file's modification time.
355
356      Then follows a blank line signifying the end of the headers, and then the
357      contents of the file are output. If the file's MIME type starts with
358      ``text/`` the file is opened in text mode; otherwise binary mode is used.
359
360      For example usage, see the implementation of the :func:`test` function
361      invocation in the :mod:`http.server` module.
362
363
364The :class:`SimpleHTTPRequestHandler` class can be used in the following
365manner in order to create a very basic webserver serving files relative to
366the current directory::
367
368   import http.server
369   import socketserver
370
371   PORT = 8000
372
373   Handler = http.server.SimpleHTTPRequestHandler
374
375   with socketserver.TCPServer(("", PORT), Handler) as httpd:
376       print("serving at port", PORT)
377       httpd.serve_forever()
378
379.. _http-server-cli:
380
381:mod:`http.server` can also be invoked directly using the :option:`-m`
382switch of the interpreter with a ``port number`` argument.  Similar to
383the previous example, this serves files relative to the current directory::
384
385        python -m http.server 8000
386
387By default, server binds itself to all interfaces.  The option ``-b/--bind``
388specifies a specific address to which it should bind.  For example, the
389following command causes the server to bind to localhost only::
390
391        python -m http.server 8000 --bind 127.0.0.1
392
393.. versionadded:: 3.4
394    ``--bind`` argument was introduced.
395
396
397.. class:: CGIHTTPRequestHandler(request, client_address, server)
398
399   This class is used to serve either files or output of CGI scripts from the
400   current directory and below. Note that mapping HTTP hierarchic structure to
401   local directory structure is exactly as in :class:`SimpleHTTPRequestHandler`.
402
403   .. note::
404
405      CGI scripts run by the :class:`CGIHTTPRequestHandler` class cannot execute
406      redirects (HTTP code 302), because code 200 (script output follows) is
407      sent prior to execution of the CGI script.  This pre-empts the status
408      code.
409
410   The class will however, run the CGI script, instead of serving it as a file,
411   if it guesses it to be a CGI script.  Only directory-based CGI are used ---
412   the other common server configuration is to treat special extensions as
413   denoting CGI scripts.
414
415   The :func:`do_GET` and :func:`do_HEAD` functions are modified to run CGI scripts
416   and serve the output, instead of serving files, if the request leads to
417   somewhere below the ``cgi_directories`` path.
418
419   The :class:`CGIHTTPRequestHandler` defines the following data member:
420
421   .. attribute:: cgi_directories
422
423      This defaults to ``['/cgi-bin', '/htbin']`` and describes directories to
424      treat as containing CGI scripts.
425
426   The :class:`CGIHTTPRequestHandler` defines the following method:
427
428   .. method:: do_POST()
429
430      This method serves the ``'POST'`` request type, only allowed for CGI
431      scripts.  Error 501, "Can only POST to CGI scripts", is output when trying
432      to POST to a non-CGI url.
433
434   Note that CGI scripts will be run with UID of user nobody, for security
435   reasons.  Problems with the CGI script will be translated to error 403.
436
437:class:`CGIHTTPRequestHandler` can be enabled in the command line by passing
438the ``--cgi`` option::
439
440        python -m http.server --cgi 8000
441
442