idnits 2.17.1 

draft-ietf-hybi-thewebsocketprotocol-07.txt:

  Checking boilerplate required by RFC 5378 and the IETF Trust (see
  https://trustee.ietf.org/license-info):
  ----------------------------------------------------------------------------

     No issues found here.

  Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt:
  ----------------------------------------------------------------------------

     No issues found here.

  Checking nits according to https://www.ietf.org/id-info/checklist :
  ----------------------------------------------------------------------------

  ** There are 7 instances of too long lines in the document, the longest one
     being 9 characters in excess of 72.


  Miscellaneous warnings:
  ----------------------------------------------------------------------------

  == The copyright year in the IETF Trust and authors Copyright Line does not
     match the current year

  == The document seems to lack the recommended RFC 2119 boilerplate, even if
     it appears to use RFC 2119 keywords -- however, there's a paragraph with
     a matching beginning. Boilerplate error?

     (The document does seem to have the reference to RFC 2119 which the
     ID-Checklist requires).
  -- The document date (April 22, 2011) is 4746 days in the past.  Is this
     intentional?


  Checking references for intended status: Proposed Standard
  ----------------------------------------------------------------------------

     (See RFCs 3967 and 4897 for information about using normative references
     to lower-maturity documents in RFCs)

  ** Downref: Normative reference to an Informational RFC: RFC 1951

  ** Obsolete normative reference: RFC 2616 (Obsoleted by RFC 7230, RFC 7231,
     RFC 7232, RFC 7233, RFC 7234, RFC 7235)

  ** Obsolete normative reference: RFC 2818 (Obsoleted by RFC 9110)

  ** Obsolete normative reference: RFC 3490 (Obsoleted by RFC 5890, RFC 5891)

  ** Obsolete normative reference: RFC 3548 (Obsoleted by RFC 4648)

  ** Obsolete normative reference: RFC 5246 (Obsoleted by RFC 8446)

  == Outdated reference: A later version (-23) exists of
     draft-ietf-httpstate-cookie-20

  == Outdated reference: A later version (-06) exists of
     draft-ietf-websec-origin-00


     Summary: 7 errors (**), 0 flaws (~~), 4 warnings (==), 1 comment (--).

     Run idnits with the --verbose option for more detailed information about
     the items above.

--------------------------------------------------------------------------------


2	HyBi Working Group                                              I. Fette
3	Internet-Draft                                              Google, Inc.
4	Intended status: Standards Track                          April 22, 2011
5	Expires: October 24, 2011

7	                         The WebSocket protocol
8	                draft-ietf-hybi-thewebsocketprotocol-07

10	Abstract

12	   The WebSocket protocol enables two-way communication between a user
13	   agent running untrusted code running in a controlled environment to a
14	   remote host that has opted-in to communications from that code.  The
15	   security model used for this is the Origin-based security model
16	   commonly used by Web browsers.  The protocol consists of an initial
17	   handshake followed by basic message framing, layered over TCP.  The
18	   goal of this technology is to provide a mechanism for browser-based
19	   applications that need two-way communication with servers that does
20	   not rely on opening multiple HTTP connections (e.g. using
21	   XMLHttpRequest or <iframe>s and long polling).

23	   Please send feedback to the hybi@ietf.org mailing list.

25	Status of this Memo

27	   This Internet-Draft is submitted in full conformance with the
28	   provisions of BCP 78 and BCP 79.

30	   Internet-Drafts are working documents of the Internet Engineering
31	   Task Force (IETF).  Note that other groups may also distribute
32	   working documents as Internet-Drafts.  The list of current Internet-
33	   Drafts is at http://datatracker.ietf.org/drafts/current/.

35	   Internet-Drafts are draft documents valid for a maximum of six months
36	   and may be updated, replaced, or obsoleted by other documents at any
37	   time.  It is inappropriate to use Internet-Drafts as reference
38	   material or to cite them other than as "work in progress."

40	   This Internet-Draft will expire on October 24, 2011.

42	Copyright Notice

44	   Copyright (c) 2011 IETF Trust and the persons identified as the
45	   document authors.  All rights reserved.

47	   This document is subject to BCP 78 and the IETF Trust's Legal
48	   Provisions Relating to IETF Documents
49	   (http://trustee.ietf.org/license-info) in effect on the date of
50	   publication of this document.  Please review these documents
51	   carefully, as they describe your rights and restrictions with respect
52	   to this document.  Code Components extracted from this document must
53	   include Simplified BSD License text as described in Section 4.e of
54	   the Trust Legal Provisions and are provided without warranty as
55	   described in the Simplified BSD License.

57	Table of Contents

59	   1.  Introduction . . . . . . . . . . . . . . . . . . . . . . . . .  4
60	     1.1.  Background . . . . . . . . . . . . . . . . . . . . . . . .  4
61	     1.2.  Protocol overview  . . . . . . . . . . . . . . . . . . . .  4
62	     1.3.  Opening handshake  . . . . . . . . . . . . . . . . . . . .  6
63	     1.4.  Closing handshake  . . . . . . . . . . . . . . . . . . . .  8
64	     1.5.  Design philosophy  . . . . . . . . . . . . . . . . . . . .  9
65	     1.6.  Security model . . . . . . . . . . . . . . . . . . . . . . 10
66	     1.7.  Relationship to TCP and HTTP . . . . . . . . . . . . . . . 10
67	     1.8.  Establishing a connection  . . . . . . . . . . . . . . . . 11
68	     1.9.  Subprotocols using the WebSocket protocol  . . . . . . . . 11
69	   2.  Conformance requirements . . . . . . . . . . . . . . . . . . . 13
70	     2.1.  Terminology  . . . . . . . . . . . . . . . . . . . . . . . 13
71	   3.  WebSocket URIs . . . . . . . . . . . . . . . . . . . . . . . . 15
72	     3.1.  Parsing WebSocket URIs . . . . . . . . . . . . . . . . . . 15
73	     3.2.  Constructing WebSocket URIs  . . . . . . . . . . . . . . . 16
74	     3.3.  Valid WebSocket URIs . . . . . . . . . . . . . . . . . . . 16
75	   4.  Data Framing . . . . . . . . . . . . . . . . . . . . . . . . . 17
76	     4.1.  Overview . . . . . . . . . . . . . . . . . . . . . . . . . 17
77	     4.2.  Base Framing Protocol  . . . . . . . . . . . . . . . . . . 17
78	     4.3.  Client-to-Server Masking . . . . . . . . . . . . . . . . . 20
79	     4.4.  Fragmentation  . . . . . . . . . . . . . . . . . . . . . . 21
80	     4.5.  Control Frames . . . . . . . . . . . . . . . . . . . . . . 23
81	       4.5.1.  Close  . . . . . . . . . . . . . . . . . . . . . . . . 23
82	       4.5.2.  Ping . . . . . . . . . . . . . . . . . . . . . . . . . 24
83	       4.5.3.  Pong . . . . . . . . . . . . . . . . . . . . . . . . . 24
84	     4.6.  Data Frames  . . . . . . . . . . . . . . . . . . . . . . . 24
85	     4.7.  Examples . . . . . . . . . . . . . . . . . . . . . . . . . 25
86	     4.8.  Extensibility  . . . . . . . . . . . . . . . . . . . . . . 25
87	   5.  Opening Handshake  . . . . . . . . . . . . . . . . . . . . . . 27
88	     5.1.  Client Requirements  . . . . . . . . . . . . . . . . . . . 27
89	     5.2.  Server-side requirements . . . . . . . . . . . . . . . . . 31
90	       5.2.1.  Reading the client's opening handshake . . . . . . . . 32
91	       5.2.2.  Sending the server's opening handshake . . . . . . . . 32
92	   6.  Error Handling . . . . . . . . . . . . . . . . . . . . . . . . 36
93	     6.1.  Handling errors in UTF-8 from the server . . . . . . . . . 36
94	     6.2.  Handling errors in UTF-8 from the client . . . . . . . . . 36
95	   7.  Closing the connection . . . . . . . . . . . . . . . . . . . . 37
96	     7.1.  Definitions  . . . . . . . . . . . . . . . . . . . . . . . 37
97	       7.1.1.  Close the WebSocket Connection . . . . . . . . . . . . 37
98	       7.1.2.  Start the WebSocket Closing Handshake  . . . . . . . . 37
99	       7.1.3.  The WebSocket Connection Is Closed . . . . . . . . . . 37
100	       7.1.4.  Fail the WebSocket Connection  . . . . . . . . . . . . 37
101	     7.2.  Abnormal closures  . . . . . . . . . . . . . . . . . . . . 37
102	       7.2.1.  Client-initiated closure . . . . . . . . . . . . . . . 38
103	       7.2.2.  Server-initiated closure . . . . . . . . . . . . . . . 38
104	     7.3.  Normal closure of connections  . . . . . . . . . . . . . . 38
105	     7.4.  Status codes . . . . . . . . . . . . . . . . . . . . . . . 38
106	       7.4.1.  Defined Status Codes . . . . . . . . . . . . . . . . . 38
107	       7.4.2.  Reserved status code ranges  . . . . . . . . . . . . . 39
108	   8.  Extensions . . . . . . . . . . . . . . . . . . . . . . . . . . 41
109	     8.1.  Negotiating extensions . . . . . . . . . . . . . . . . . . 41
110	     8.2.  Known extensions . . . . . . . . . . . . . . . . . . . . . 42
111	       8.2.1.  Compression  . . . . . . . . . . . . . . . . . . . . . 42
112	   9.  Security considerations  . . . . . . . . . . . . . . . . . . . 44
113	   10. IANA considerations  . . . . . . . . . . . . . . . . . . . . . 46
114	     10.1. Registration of ws: scheme . . . . . . . . . . . . . . . . 46
115	     10.2. Registration of wss: scheme  . . . . . . . . . . . . . . . 47
116	     10.3. Registration of the "WebSocket" HTTP Upgrade keyword . . . 48
117	     10.4. Sec-WebSocket-Key  . . . . . . . . . . . . . . . . . . . . 48
118	     10.5. Sec-WebSocket-Extensions . . . . . . . . . . . . . . . . . 49
119	     10.6. Sec-WebSocket-Accept . . . . . . . . . . . . . . . . . . . 50
120	     10.7. Sec-WebSocket-Origin . . . . . . . . . . . . . . . . . . . 50
121	     10.8. Sec-WebSocket-Protocol . . . . . . . . . . . . . . . . . . 51
122	     10.9. Sec-WebSocket-Version  . . . . . . . . . . . . . . . . . . 51
123	   11. Using the WebSocket protocol from other specifications . . . . 53
124	   12. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 54
125	   13. References . . . . . . . . . . . . . . . . . . . . . . . . . . 55
126	     13.1. Normative References . . . . . . . . . . . . . . . . . . . 55
127	     13.2. Informative References . . . . . . . . . . . . . . . . . . 56
128	   Author's Address . . . . . . . . . . . . . . . . . . . . . . . . . 57

130	1.  Introduction

132	1.1.  Background

134	   _This section is non-normative._

136	   Historically, creating an instant messenger chat client as a Web
137	   application has required an abuse of HTTP to poll the server for
138	   updates while sending upstream notifications as distinct HTTP
139	   calls.[RFC6202]

141	   This results in a variety of problems:

143	   o  The server is forced to use a number of different underlying TCP
144	      connections for each client: one for sending information to the
145	      client, and a new one for each incoming message.

147	   o  The wire protocol has a high overhead, with each client-to-server
148	      message having an HTTP header.

150	   o  The client-side script is forced to maintain a mapping from the
151	      outgoing connections to the incoming connection to track replies.

153	   A simpler solution would be to use a single TCP connection for
154	   traffic in both directions.  This is what the WebSocket protocol
155	   provides.  Combined with the WebSocket API, it provides an
156	   alternative to HTTP polling for two-way communication from a Web page
157	   to a remote server.  [WSAPI]

159	   The same technique can be used for a variety of Web applications:
160	   games, stock tickers, multiuser applications with simultaneous
161	   editing, user interfaces exposing server-side services in real time,
162	   etc.

164	1.2.  Protocol overview

166	   _This section is non-normative._

168	   The protocol has two parts: a handshake, and then the data transfer.

170	   The handshake from the client looks as follows:

172	        GET /chat HTTP/1.1
173	        Host: server.example.com
174	        Upgrade: websocket
175	        Connection: Upgrade
176	        Sec-WebSocket-Key: dGhlIHNhbXBsZSBub25jZQ==
177	        Sec-WebSocket-Origin: http://example.com
178	        Sec-WebSocket-Protocol: chat, superchat
179	        Sec-WebSocket-Version: 7

181	   The handshake from the server looks as follows:

183	        HTTP/1.1 101 Switching Protocols
184	        Upgrade: websocket
185	        Connection: Upgrade
186	        Sec-WebSocket-Accept: s3pPLMBiTxaQ9kYGzzhZRbK+xOo=
187	        Sec-WebSocket-Protocol: chat

189	   The leading line from the client follows the Request-Line format.
190	   The leading line from the server follows the Status-Line format.  The
191	   Request-Line and Status-Line productions are defined in [RFC2616].

193	   After the leading line in both cases come an unordered set of header
194	   fields.  The meaning of these header fields is specified in Section 5
195	   of this document.  Additional header fields may also be present, such
196	   as cookies [I-D.ietf-httpstate-cookie] required to identify the user.
197	   The format and parsing of headers is as defined in [RFC2616].

199	   Once the client and server have both sent their handshakes, and if
200	   the handshake was successful, then the data transfer part starts.
201	   This is a two-way communication channel where each side can,
202	   independently from the other, send data at will.

204	   Clients and servers, after a successful handshake, transfer data back
205	   and forth in conceptual units referred to in this specification as
206	   "messages".  A message is a complete unit of data at an application
207	   level, with the expectation that many or most applications
208	   implementing this protocol (such as web user agents) provide APIs in
209	   terms of sending and receiving messages.  The websocket message does
210	   not necessarily correspond to a particular network layer framing, as
211	   a fragmented message may be coalesced, or vice versa, e.g. by an
212	   intermediary.

214	   Data is sent on the wire in the form of frames that have an
215	   associated type.  A message is composed of one or more frames, all of
216	   which contain the same type of data.  Broadly speaking, there are
217	   types for textual data, which is interpreted as UTF-8 [RFC3629] text,
218	   binary data (whose interpretation is left up to the application), and
219	   control frames, which are not intended to carry data for the
220	   application, but instead for protocol-level signaling, such as to
221	   signal that the connection should be closed.  This version of the
222	   protocol defines six frame types and leaves ten reserved for future
223	   use.

225	   The WebSocket protocol uses this framing so that specifications that
226	   use the WebSocket protocol can expose such connections using an
227	   event-based mechanism instead of requiring users of those
228	   specifications to implement buffering and piecing together of
229	   messages manually.

231	1.3.  Opening handshake

233	   _This section is non-normative._

235	   The opening handshake is intended to be compatible with HTTP-based
236	   server-side software and intermediaries, so that a single port can be
237	   used by both HTTP clients talking to that server and WebSocket
238	   clients talking to that server.  To this end, the WebSocket client's
239	   handshake is an HTTP Upgrade request:

241	        GET /chat HTTP/1.1
242	        Host: server.example.com
243	        Upgrade: websocket
244	        Connection: Upgrade
245	        Sec-WebSocket-Key: dGhlIHNhbXBsZSBub25jZQ==
246	        Sec-WebSocket-Origin: http://example.com
247	        Sec-WebSocket-Protocol: chat, superchat
248	        Sec-WebSocket-Version: 7

250	   Headers in the handshake are sent by the client in a random order;
251	   the order is not meaningful.

253	   The "Request-URI" of the GET method [RFC2616] is used to identify the
254	   endpoint of the WebSocket connection, both to allow multiple domains
255	   to be served from one IP address and to allow multiple WebSocket
256	   endpoints to be served by a single server.

258	   The client includes the hostname in the Host header of its handshake
259	   as per [RFC2616], so that both the client and the server can verify
260	   that they agree on which host is in use.

262	   Additional headers are used to select options in the WebSocket
263	   protocol.  Options available in this version are the subprotocol
264	   selector, |Sec-WebSocket-Protocol|, and |Cookie|, which can used for
265	   sending cookies to the server (e.g. as an authentication mechanism).

267	   The |Sec-WebSocket-Protocol| request-header field can be used to
268	   indicate what subprotocols (application-level protocols layered over
269	   the WebSocket protocol) are acceptable to the client.  The server
270	   selects one of the acceptable protocols and echoes that value in its
271	   handshake to indicate that it has selected that protocol.
272	        Sec-WebSocket-Protocol: chat

274	   The |Sec-WebSocket-Origin| header is used to protect against
275	   unauthorized cross-origin use of a WebSocket server by scripts using
276	   the |WebSocket| API in a Web browser.  The server is informed of the
277	   script origin generating the WebSocket connection request.  If the
278	   server does not wish to accept connections from this origin, it can
279	   choose to abort the connection.  This header is sent by browser
280	   clients, for non-browser clients this header may be sent if it makes
281	   sense in the context of those clients.

283	   Finally, the server has to prove to the client that it received the
284	   client's WebSocket handshake, so that the server doesn't accept
285	   connections that are not WebSocket connections.  This prevents an
286	   attacker from tricking a WebSocket server by sending it carefully-
287	   crafted packets using |XMLHttpRequest| or a |form| submission.

289	   To prove that the handshake was received, the server has to take two
290	   pieces of information and combine them to form a response.  The first
291	   piece of information comes from the |Sec-WebSocket-Key| header in the
292	   client handshake:

294	        Sec-WebSocket-Key: dGhlIHNhbXBsZSBub25jZQ==

296	   For this header, the server has to take the value (as present in the
297	   header, e.g. the base64-encoded [RFC4648] version minus leading and
298	   trailing whitespace), and concatenate this with the GUID "258EAFA5-
299	   E914-47DA-95CA-C5AB0DC85B11" in string form, which is unlikely to be
300	   used by network endpoints that do not understand the WebSocket
301	   protocol.  A SHA-1 hash, base64-encoded, of this concatenation is
302	   then returned in the server's handshake [FIPS.180-2.2002].

304	   Concretely, if as in the example above, header |Sec-WebSocket-Key|
305	   had the value "dGhlIHNhbXBsZSBub25jZQ==", the server would
306	   concatenate the string "258EAFA5-E914-47DA-95CA-C5AB0DC85B11" to form
307	   the string "dGhlIHNhbXBsZSBub25jZQ==258EAFA5-E914-47DA-95CA-
308	   C5AB0DC85B11".  The server would then take the SHA-1 hash of this,
309	   giving the value 0xb3 0x7a 0x4f 0x2c 0xc0 0x62 0x4f 0x16 0x90 0xf6
310	   0x46 0x06 0xcf 0x38 0x59 0x45 0xb2 0xbe 0xc4 0xea.  This value is
311	   then base64-encoded, to give the value "s3pPLMBiTxaQ9kYGzzhZRbK+
312	   xOo=".  This value would then be echoed in the header |Sec-WebSocket-
313	   Accept|.

315	   The handshake from the server is much simpler than the client
316	   handshake.  The first line is an HTTP Status-Line, with the status
317	   code 101:

319	        HTTP/1.1 101 Switching Protocols

321	   Any status code other than 101 indicates that the WebSocket handshake
322	   has not completed, and that the semantics of HTTP still apply.  The
323	   headers follow the status code.

325	   The |Connection| and |Upgrade| headers complete the HTTP Upgrade.
326	   The |Sec-WebSocket-Accept| header indicates whether the server is
327	   willing to accept the connection.  If present, this header must
328	   include a hash of the client's nonce sent in |Sec-WebSocket-Key|
329	   along with a predefined GUID.  Any other value must not be
330	   interpreted as an acceptance of the connection by the server.

332	        HTTP/1.1 101 Switching Protocols
333	        Upgrade: websocket
334	        Connection: Upgrade
335	        Sec-WebSocket-Accept: s3pPLMBiTxaQ9kYGzzhZRbK+xOo=

337	   These fields are checked by the Web browser when it is acting as a
338	   |WebSocket| client for scripted pages.  If the |Sec-WebSocket-Accept|
339	   value does not match the expected value, or if the header is missing,
340	   or if the HTTP status code is not 101, the connection will not be
341	   established and WebSockets frames will not be sent.

343	   Option fields can also be included.  In this version of the protocol,
344	   the main option field is |Sec-WebSocket-Protocol|, which indicates
345	   the subprotocol that the server has selected.  Web browsers verify
346	   that the server included one of the values as was specified in the
347	   WebSocket client' handshake.  A server that speaks multiple
348	   subprotocols has to make sure it selects one based on the client's
349	   handshake and specifies it in its handshake.

351	        Sec-WebSocket-Protocol: chat

353	   The server can also set cookie-related option fields to _set_
354	   cookies, as in HTTP.

356	1.4.  Closing handshake

358	   _This section is non-normative._

360	   The closing handshake is far simpler than the opening handshake.

362	   Either peer can send a control frame with data containing a specified
363	   control sequence to begin the closing handshake (detailed in
364	   Section 4.5.1).  Upon receiving such a frame, the other peer sends a
365	   close frame in response, if it hasn't already sent one.  Upon
366	   receiving _that_ control frame, the first peer then closes the
367	   connection, safe in the knowledge that no further data is
368	   forthcoming.

370	   After sending a control frame indicating the connection should be
371	   closed, a peer does not send any further data; after receiving a
372	   control frame indicating the connection should be closed, a peer
373	   discards any further data received.

375	   It is safe for both peers to initiate this handshake simultaneously.

377	   The closing handshake is intended to replace the TCP closing
378	   handshake (FIN/ACK), on the basis that the TCP closing handshake is
379	   not always reliable end-to-end, especially in the presence of man-in-
380	   the-middle proxies and other intermediaries.

382	   By sending a close frame and waiting for a close frame in response,
383	   certain cases are avoided where data may be unnecessarily lost.  For
384	   instance, on some platforms, if a socket is closed with data in the
385	   receive queue, a RST packet is sent, which will then cause recv() to
386	   fail for the party that received the RST, even if there was data
387	   waiting to be read.

389	1.5.  Design philosophy

391	   _This section is non-normative._

393	   The WebSocket protocol is designed on the principle that there should
394	   be minimal framing (the only framing that exists is to make the
395	   protocol frame-based instead of stream-based, and to support a
396	   distinction between Unicode text and binary frames).  It is expected
397	   that metadata would be layered on top of WebSocket by the application
398	   layer, in the same way that metadata is layered on top of TCP by the
399	   application layer (HTTP).

401	   Conceptually, WebSocket is really just a layer on top of TCP that
402	   adds a Web "origin"-based security model for browsers; adds an
403	   addressing and protocol naming mechanism to support multiple services
404	   on one port and multiple host names on one IP address; layers a
405	   framing mechanism on top of TCP to get back to the IP packet
406	   mechanism that TCP is built on, but without length limits; and re-
407	   implements the closing handshake in-band.  Other than that, it adds
408	   nothing.  Basically it is intended to be as close to just exposing
409	   raw TCP to script as possible given the constraints of the Web. It's
410	   also designed in such a way that its servers can share a port with
411	   HTTP servers, by having its handshake be a valid HTTP Upgrade request
412	   mechanism also.

414	   The protocol is intended to be extensible; future versions will
415	   likely introduce additional concepts such as multiplexing.

417	1.6.  Security model

419	   _This section is non-normative._

421	   The WebSocket protocol uses the origin model used by Web browsers to
422	   restrict which Web pages can contact a WebSocket server when the
423	   WebSocket protocol is used from a Web page.  Naturally, when the
424	   WebSocket protocol is used by a dedicated client directly (i.e. not
425	   from a Web page through a Web browser), the origin model is not
426	   useful, as the client can provide any arbitrary origin string.

428	   This protocol is intended to fail to establish a connection with
429	   servers of pre-existing protocols like SMTP [RFC5321] and HTTP, while
430	   allowing HTTP servers to opt-in to supporting this protocol if
431	   desired.  This is achieved by having a strict and elaborate
432	   handshake, and by limiting the data that can be inserted into the
433	   connection before the handshake is finished (thus limiting how much
434	   the server can be influenced).

436	   It is similarly intended to fail to establish a connection when data
437	   from other protocols, especially HTTP, is sent to a WebSocket server,
438	   for example as might happen if an HTML |form| were submitted to a
439	   WebSocket server.  This is primarily achieved by requiring that the
440	   server prove that it read the handshake, which it can only do if the
441	   handshake contains the appropriate parts which themselves can only be
442	   sent by a WebSocket handshake.  In particular, at the time of writing
443	   of this specification, fields starting with |Sec-| cannot be set by
444	   an attacker from a Web browser using only HTML and JavaScript APIs
445	   such as |XMLHttpRequest|.

447	1.7.  Relationship to TCP and HTTP

449	   _This section is non-normative._

451	   The WebSocket protocol is an independent TCP-based protocol.  Its
452	   only relationship to HTTP is that its handshake is interpreted by
453	   HTTP servers as an Upgrade request.

455	   By default the WebSocket protocol uses port 80 for regular WebSocket
456	   connections and port 443 for WebSocket connections tunneled over TLS
457	   [RFC2818].

459	1.8.  Establishing a connection

461	   _This section is non-normative._

463	   When a connection is to be made to a port that is shared by an HTTP
464	   server (a situation that is quite likely to occur with traffic to
465	   ports 80 and 443), the connection will appear to the HTTP server to
466	   be a regular GET request with an Upgrade offer.  In relatively simple
467	   setups with just one IP address and a single server for all traffic
468	   to a single hostname, this might allow a practical way for systems
469	   based on the WebSocket protocol to be deployed.  In more elaborate
470	   setups (e.g. with load balancers and multiple servers), a dedicated
471	   set of hosts for WebSocket connections separate from the HTTP servers
472	   is probably easier to manage.  At the time of writing of this
473	   specification, it should be noted that connections on port 80 and 443
474	   have significantly different success rates, with connections on port
475	   443 being significantly more likely to succeed, though this may
476	   change with time.

478	1.9.  Subprotocols using the WebSocket protocol

480	   _This section is non-normative._

482	   The client can request that the server use a specific subprotocol by
483	   including the |Sec-WebSocket-Protocol| field in its handshake.  If it
484	   is specified, the server needs to include the same field and one of
485	   the selected subprotocol values in its response for the connection to
486	   be established.

488	   These subprotocol names do not need to be registered, but if a
489	   subprotocol is intended to be implemented by multiple independent
490	   WebSocket servers, potential clashes with the names of subprotocols
491	   defined independently can be avoided by using names that contain the
492	   domain name of the subprotocol's originator.  For example, if Example
493	   Corporation were to create a Chat subprotocol to be implemented by
494	   many servers around the Web, they could name it "chat.example.com".
495	   If the Example Organization called their competing subprotocol
496	   "example.org's chat protocol", then the two subprotocols could be
497	   implemented by servers simultaneously, with the server dynamically
498	   selecting which subprotocol to use based on the value sent by the
499	   client.

501	   Subprotocols can be versioned in backwards-incompatible ways by
502	   changing the subprotocol name, e.g. going from "bookings.example.net"
503	   to "v2.bookings.example.net".  These subprotocols would be considered
504	   completely separate by WebSocket clients.  Backwards-compatible
505	   versioning can be implemented by reusing the same subprotocol string
506	   but carefully designing the actual subprotocol to support this kind
507	   of extensibility.

509	2.  Conformance requirements

511	   All diagrams, examples, and notes in this specification are non-
512	   normative, as are all sections explicitly marked non-normative.
513	   Everything else in this specification is normative.

515	   The key words "MUST", "MUST NOT", "REQUIRED", "SHOULD", "SHOULD NOT",
516	   "RECOMMENDED", "MAY", and "OPTIONAL" in the normative parts of this
517	   document are to be interpreted as described in RFC2119.  [RFC2119]

519	   Requirements phrased in the imperative as part of algorithms (such as
520	   "strip any leading space characters" or "return false and abort these
521	   steps") are to be interpreted with the meaning of the key word
522	   ("must", "should", "may", etc) used in introducing the algorithm.

524	   Conformance requirements phrased as algorithms or specific steps MAY
525	   be implemented in any manner, so long as the end result is
526	   equivalent.  (In particular, the algorithms defined in this
527	   specification are intended to be easy to follow, and not intended to
528	   be performant.)

530	   Implementations MAY impose implementation-specific limits on
531	   otherwise unconstrained inputs, e.g. to prevent denial of service
532	   attacks, to guard against running out of memory, or to work around
533	   platform-specific limitations.

535	   The conformance classes defined by this specification are user agents
536	   and servers.

538	2.1.  Terminology

540	   *ASCII* shall mean the character-encoding scheme defined in
541	   [ANSI.X3-4.1986].

543	   *Converting a string to ASCII lowercase* means replacing all
544	   characters in the range U+0041 to U+005A (i.e.  LATIN CAPITAL LETTER
545	   A to LATIN CAPITAL LETTER Z) with the corresponding characters in the
546	   range U+0061 to U+007A (i.e.  LATIN SMALL LETTER A to LATIN SMALL
547	   LETTER Z).

549	   Comparing two strings in an *ASCII case-insensitive* manner means
550	   comparing them exactly, code point for code point, except that the
551	   characters in the range U+0041 to U+005A (i.e.  LATIN CAPITAL LETTER
552	   A to LATIN CAPITAL LETTER Z) and the corresponding characters in the
553	   range U+0061 to U+007A (i.e.  LATIN SMALL LETTER A to LATIN SMALL
554	   LETTER Z) are considered to also match.

556	   The term "URI" is used in this document as defined in [RFC3986].

558	   When an implementation is required to _send_ data as part of the
559	   WebSocket protocol, the implementation MAY delay the actual
560	   transmission arbitrarily, e.g. buffering data so as to send fewer IP
561	   packets.

563	3.  WebSocket URIs

565	3.1.  Parsing WebSocket URIs

567	   The steps to *parse a WebSocket URI's components* from a string /uri/
568	   are as follows.  These steps return either a /host/, a /port/, a
569	   /resource name/, and a /secure/ flag, or they fail.

571	   1.   If the /uri/ string is not an absolute URI, then fail this
572	        algorithm.  [RFC3986]

574	   2.   Resolve the /uri/ string using the resolve a Web address
575	        algorithm defined by the Web addresses specification, with the
576	        URI character encoding set to UTF-8.  [RFC3629] [RFC3986]
577	        [RFC3987]

579	        NOTE: It doesn't matter what it is resolved relative to, since
580	        we already know it is an absolute URI at this point.

582	   3.   If /uri/ does not have a <scheme> component whose value, when
583	        converted to ASCII lowercase, is either "ws" or "wss", then fail
584	        this algorithm.

586	   4.   If /uri/ has a <fragment> component, then fail this algorithm.

588	   5.   If the <scheme> component of /uri/ is "ws", set /secure/ to
589	        false; otherwise, if the <scheme> component is "wss", set
590	        /secure/ to true; if neither of the above apply, fail this
591	        algorithm.

593	   6.   Let /host/ be the value of the <host> component of /uri/,
594	        converted to ASCII lowercase.

596	   7.   If /uri/ has a <port> component, then let /port/ be that
597	        component's value; otherwise, there is no explicit /port/.

599	   8.   If there is no explicit /port/, then: if /secure/ is false, let
600	        /port/ be 80, otherwise let /port/ be 443.

602	   9.   Let /resource name/ be the value of the <path> component (which
603	        might be empty) of /uri/.

605	   10.  If /resource name/ is the empty string, set it to a single
606	        character U+002F SOLIDUS (/).

608	   11.  If /uri/ has a <query> component, then append a single U+003F
609	        QUESTION MARK character (?) to /resource name/, followed by the
610	        value of the <query> component.

612	   12.  Return /host/, /port/, /resource name/, and /secure/.

614	3.2.  Constructing WebSocket URIs

616	   The steps to *construct a WebSocket URI* from a /host/, a /port/, a
617	   /resource name/, and a /secure/ flag, are as follows:

619	   1.  Let /uri/ be the empty string.

621	   2.  If the /secure/ flag is false, then append the string "ws://" to
622	       /uri/.  Otherwise, append the string "wss://" to /uri/.

624	   3.  Append /host/ to /uri/.

626	   4.  If the /secure/ flag is false and port is not 80, or if the
627	       /secure/ flag is true and port is not 443, then append the string
628	       ":" followed by /port/ to /uri/.

630	   5.  Append /resource name/ to /uri/.

632	   6.  Return /uri/.

634	3.3.  Valid WebSocket URIs

636	   For a WebSocket URI to be considered valid, the following conditions
637	   MUST hold.

639	   o  The /host/ MUST be ASCII-only (i.e. it MUST have been punycode-
640	      encoded [RFC3492] already if necessary, and MUST NOT contain any
641	      characters above U+007E).

643	   o  The /resource name/ string MUST be a non-empty string of
644	      characters in the range U+0021 to U+007E and MUST start with a
645	      U+002F SOLIDUS character (/).

647	   Any WebSocket URIs not meeting the above criteria are considered
648	   invalid.  A client MUST NOT attempt to make a connection to an
649	   invalid WebSocket URI.  A client SHOULD attempt to parse a URI
650	   obtained from any external source (such as a web site or a user)
651	   using the steps specified in Section 3.1 to obtain a valid WebSocket
652	   URI, but MUST NOT attempt to connect with such an unparsed URI, and
653	   instead only use the parsed version and only if that version is
654	   considered valid by the criteria above.

656	4.  Data Framing

658	4.1.  Overview

660	   In the WebSocket protocol, data is transmitted using a sequence of
661	   frames.  Frames sent from the client to the server are masked to
662	   avoid confusing network intermediaries, such as intercepting proxies.
663	   Frames sent from the server to the client are not masked.

665	   The base framing protocol defines a frame type with an opcode, a
666	   payload length, and designated locations for extension and
667	   application data, which together define the _payload_ data.  Certain
668	   bits and opcodes are reserved for future expansion of the protocol.
669	   As such, in the absence of extensions negotiated during the opening
670	   handshake (Section 5), all reserved bits MUST be 0 and reserved
671	   opcode values MUST NOT be used.

673	   A data frame MAY be transmitted by either the client or the server at
674	   any time after handshake completion and before that endpoint has sent
675	   a close message (Section 4.5.1).

677	4.2.  Base Framing Protocol

679	   This wire format for the data transfer part is described by the ABNF
680	   [RFC5234] given in detail in this section.  A high level overview of
681	   the framing is given in the following figure.

683	      0                   1                   2                   3
684	      0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1
685	     +-+-+-+-+-------+-+-------------+-------------------------------+
686	     |F|R|R|R| opcode|M| Payload len |    Extended payload length    |
687	     |I|S|S|S|  (4)  |A|     (7)     |             (16/63)           |
688	     |N|V|V|V|       |S|             |   (if payload len==126/127)   |
689	     | |1|2|3|       |K|             |                               |
690	     +-+-+-+-+-------+-+-------------+ - - - - - - - - - - - - - - - +
691	     |     Extended payload length continued, if payload len == 127  |
692	     + - - - - - - - - - - - - - - - +-------------------------------+
693	     |                               |Masking-key, if MASK set to 1  |
694	     +-------------------------------+-------------------------------+
695	     | Masking-key (continued)       |          Payload Data         |
696	     +-------------------------------- - - - - - - - - - - - - - - - +
697	     :                     Payload Data continued ...                :
698	     + - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - +
699	     |                     Payload Data continued ...                |
700	     +---------------------------------------------------------------+

702	   FIN:  1 bit

704	      Indicates that this is the final fragment in a message.  The first
705	      fragment MAY also be the final fragment.

707	   RSV1, RSV2, RSV3:  1 bit each

709	      Must be 0 unless an extension is negotiated which defines meanings
710	      for non-zero values

712	   Opcode:  4 bits

714	      Defines the interpretation of the payload data

716	   Mask:  1 bit

718	      Defines whether the payload data is masked.  If set to 1, a
719	      masking key is present in Masking-key, and this is used to unmask
720	      the payload data as per Section 4.3.  All frames sent from client
721	      to server have this bit set to 1.

723	   Payload length:  7 bits, 7+16 bits, or 7+64 bits

725	      The length of the payload: if 0-125, that is the payload length.
726	      If 126, the following 2 bytes interpreted as a 16 bit unsigned
727	      integer are the payload length.  If 127, the following 8 bytes
728	      interpreted as a 64-bit unsigned integer (the most significant bit
729	      MUST be 0) are the payload length.  Multibyte length quantities
730	      are expressed in network byte order.  The payload length is the
731	      length of the Extension data + the length of the Application data.
732	      The length of the Extension data may be zero, in which case the
733	      Payload length is the length of the Application data.  The length
734	      of this field is always at least 7 bits.  If the value of the
735	      first 7 bits is 0-125, the length of this field is 7 bits.  If the
736	      value is 126, there exist 16 additional bits with a 16-bit length.
737	      If the value is 127, there exist 64 additional bits with a 63-bit
738	      length (the most significant bit MUST be 0).

740	   Masking-key:  0 or 4 bytes

742	      All frames sent from the client to the server are masked by a 32-
743	      bit value that is contained within the frame.  This field is
744	      present if the Mask bit is set to 1, and is absent if the Mask bit
745	      is set to 0.  See Section 4.3 for further information on client-
746	      to-server masking.

748	   Payload data:  n bytes

750	      The payload data is defined as Extension Data concatenated with
751	      Application Data.

753	   Extension data:  n bytes

755	      The extension data is 0 bytes unless an extension has been
756	      negotiated.  Any extension MUST specify the length of the
757	      extension data, or how that length may be calculated, and how the
758	      extension use MUST be negotiated during the handshake.  If
759	      present, the extension data is included in the total payload
760	      length.

762	   Application data:  n bytes

764	      Arbitrary application data, taking up the remainder of the frame
765	      after any extension data.  The length of the Application data is
766	      equal to the payload length minus the length of the Extension
767	      data.

769	   The base framing protocol is formally defined by the following ABNF
770	   [RFC5234]:

772	   ws-frame                = frame-fin
773	                             frame-rsv1
774	                             frame-rsv2
775	                             frame-rsv3
776	                             frame-opcode
777	                             frame-masked
778	                             frame-payload-length
779	                             [ frame-masking-key ]
780	                             frame-payload-data

782	   frame-fin               = %x0 ; more frames of this message follow
783	                           / %x1 ; final frame of message

785	   frame-rsv1              = %x0 ; 1 bit, MUST be 0

787	   frame-rsv2              = %x0 ; 1 bit, MUST be 0

789	   frame-rsv3              = %x0 ; 1 bit, MUST be 0

791	   frame-opcode            = %x0 ; continuation frame
792	                           / %x1 ; text frame
793	                           / %x2 ; binary frame
794	                           / %3-7 ; reserved for further non-control frames
795	                           / %x8 ; connection close
796	                           / %x9 ; ping
797	                           / %xA ; pong
798	                           / %xB-F ; reserved for further control frames

800	   frame-masked            = %x0 ; frame is not masked, no frame-masking-key
801	                           = %x1 ; frame is masked, frame-masking-key present

803	   frame-payload-length    = %x00-7D
804	                           / %x7E frame-payload-length-16
805	                           / %x7F frame-payload-length-63

807	   frame-payload-length-16 = %x0000-FFFF

809	   frame-payload-length-63 = %x0000000000000000-7FFFFFFFFFFFFFFF

811	   frame-masking-key       = <4>( %0x00-FF ) ; present only if frame-masked is 1

813	   frame-payload-data      = (frame-masked-extension-data
814	                              frame-masked-application-data)   ; frame-masked 1
815	                           / (frame-unmasked-extension-data
816	                              frame-unmasked-application-data) ; frame-masked 0

818	   frame-masked-extension-data     = *( %x00-FF ) ; to be defined later

820	   frame-masked-application-data   = *( %x00-FF )

822	   frame-unmasked-extension-data   = *( %x00-FF ) ; to be defined later

824	   frame-unmasked-application-data = *( %x00-FF )

826	4.3.  Client-to-Server Masking

828	   The client MUST mask all frames sent to the server.  A server MUST
829	   close the connection upon receiving a frame with the MASK bit set to
830	   0.  In this case, a server MAY send a close frame with a status code
831	   of 1002 (protocol error) as defined in Section 7.4.1.

833	   A masked frame MUST have the field frame-masked set to 1, as defined
834	   in Section 4.2.

836	   The masking key is contained completely within the frame, as defined
837	   in Section 4.2 as frame-masking-key.  It is used to mask the payload
838	   data defined in the same section as frame-payload-data, which
839	   includes extension and application data.

841	   The masking key is a 32-bit value chosen at random by the client.
842	   The masking key MUST be derived from a strong source of entropy, and
843	   the masking key for a given frame MUST NOT make it simple for a
844	   server to predict the masking key for a subsequent frame.

846	   The masking does not affect the length of the payload data.  To
847	   convert masked data into unmasked data, or vice versa, the following
848	   algorithm is applied.  The same algorithm applies regardless of the
849	   direction of the translation - e.g. the same steps are applied to
850	   mask the data as to unmask the data.

852	   Octet i of the transformed data ("transformed-octet-i") is the XOR of
853	   octet i of the original data ("original-octet-i") with octet i modulo
854	   4 of the masking key ("masking-key-octet-j"):

856	     j                   = i MOD 4
857	     transformed-octet-i = original-octet-i XOR masking-key-octet-j

859	   When preparing a masked frame, the client MUST pick a fresh masking
860	   key uniformly at random from the set of allowed 32-bit values.  The
861	   unpredictability of the masking-nonce is essential to prevent the
862	   author of malicious application data from selecting the bytes that
863	   appear on the wire.

865	   The payload length, indicated in the framing as frame-payload-length,
866	   does NOT include the length of the masking key.  It is the length of
867	   the payload data, e.g. the number of bytes following the masking key.

869	4.4.  Fragmentation

871	   The primary purpose of fragmentation is to allow sending a message
872	   that is of unknown size when the message is started without having to
873	   buffer that message.  If messages couldn't be fragmented, then an
874	   endpoint would have to buffer the entire message so its length could
875	   be counted before first byte is sent.  With fragmentation, a server
876	   or intermediary may choose a reasonable size buffer, and when the
877	   buffer is full write a fragment to the network.

879	   A secondary use-case for fragmentation is for multiplexing, where it
880	   is not desirable for a large message on one logical channel to
881	   monopolize the output channel, so the MUX needs to be free to split
882	   the message into smaller fragments to better share the output
883	   channel.

885	   The following rules apply to fragmentation:

887	   o  An unfragmented message consists of a single frame with the FIN
888	      bit set and an opcode other than 0.

890	   o  A fragmented message consists of a single frame with the FIN bit
891	      clear and an opcode other than 0, followed by zero or more frames
892	      with the FIN bit clear and the opcode set to 0, and terminated by
893	      a single frame with the FIN bit set and an opcode of 0.  Its
894	      content is the concatenation of the application data (and any
895	      extension data that may be present) from each of those frames in
896	      order.  As an example, for a text message sent as three fragments,
897	      the first fragment would have an opcode of 0x4 and a FIN bit
898	      clear, the second fragment would have an opcode of 0x0 and a FIN
899	      bit clear, and the third fragment would have an opcode of 0x0 and
900	      a FIN bit that is set.

902	   o  Control frames MAY be injected in the middle of a fragmented
903	      message.  Control frames themselves MUST NOT be fragmented.

905	   o  An endpoint MUST be capable of handling control frames in the
906	      middle of a fragmented message.

908	   o  _Note: if control frames could not be interjected, the latency of
909	      a ping, for example, would be very long if behind a large message.
910	      Hence, the requirement of handling control frames in the middle of
911	      a fragmented message._

913	   o  A sender MAY create fragments of any size for non control
914	      messages.

916	   o  Clients and servers MUST support receiving both fragmented and
917	      unfragmented messages.

919	   o  As control frames cannot be fragmented, an intermediary MUST NOT
920	      attempt to change the fragmentation of a control frame.

922	   o  An intermediary MUST NOT change the fragmentation of a message if
923	      any reserved bit values are used and the meaning of these values
924	      is not known to the intermediary.

926	   o  An intermediary MUST NOT change the fragmentation of any message
927	      in the context of a connection where extensions have been
928	      negotiated and the intermediary is not aware of the semantics of
929	      the negotiated extensions.

931	   o  As a consequence of these rules, all fragments of a message are of
932	      the same type, as set by the first fragment's opcode.  Since
933	      Control frames cannot be fragmented, the type for all fragments in
934	      a message MUST be either text or binary, or one of the reserved
935	      opcodes.

937	4.5.  Control Frames

939	   Control frames are identified by opcodes where the most significant
940	   bit of the opcode is 1.  Currently defined opcodes for control frames
941	   include 0x8 (Close), 0x9 (Ping), and 0xA (Pong).  Opcodes 0xB-0xF are
942	   reserved for further control frames yet to be defined.

944	   Control frames are used to communicate state about the websocket.
945	   Control frames can be interjected in the middle of a fragmented
946	   message.

948	   All control frames MUST have a payload length of 125 bytes or less
949	   and MUST NOT be fragmented.

951	4.5.1.  Close

953	   The Close message contains an opcode of 0x8.

955	   The Close message MAY contain a body (the "application data" portion
956	   of the frame) that indicates a reason for closing, such as an
957	   endpoint shutting down, an endpoint having received a message too
958	   large, or an endpoint having received a message that does not conform
959	   to the format expected by the other endpoint.  If there is a body,
960	   the first two bytes of the body MUST be a 2-byte integer (in network
961	   byte order) representing a status code defined in Section 7.4.
962	   Following the 2-byte integer the body MAY contain UTF-8 encoded data,
963	   the interpretation of which is not defined by this specification.
964	   This data is not necessarily human readable, but may be useful for
965	   debugging or passing information relevant to the script that opened
966	   the connection.

968	   The application MUST NOT send any more data frames after sending a
969	   close message.

971	   If an endpoint receives a Close message and that endpoint did not
972	   previously send a Close message, the endpoint MUST send a Close
973	   message in response.  It SHOULD do so as soon as is practical.

975	   After both sending and receiving a close message, an endpoint
976	   considers the websocket connection closed, and SHOULD close the
977	   underlying TCP connection.

979	   If a client and server both send a Close message at the same time,
980	   both endpoints will have sent and received a Close message and should
981	   consider the websocket connection closed and close the underlying TCP
982	   connection.

984	4.5.2.  Ping

986	   The Ping message contains an opcode of 0x9.

988	   Upon receipt of a Ping message, an endpoint MUST send a Pong message
989	   in response.  It SHOULD do so as soon as is practical.  The message
990	   bodies (i.e. both the Extension data (if any) and the Application
991	   data) of the Ping and Pong MUST be the same.

993	   An endpoint MAY send a Ping message any time after the connection is
994	   established and before the connection is closed.  NOTE: A ping
995	   message may serve either as a keepalive, or to verify that the remote
996	   endpoint is still responsive.

998	4.5.3.  Pong

1000	   The Pong message contains an opcode of 0xA.

1002	   Upon receipt of a Ping message, an endpoint MUST send a Pong message
1003	   in response.  It SHOULD do so as soon as is practical.  The message
1004	   bodies (i.e. both the Extension data (if any) and the Application
1005	   data) of the Ping and Pong MUST be the same.  In the case multiple
1006	   Pings have been received, a Pong MUST be issued only in response to
1007	   the most recent Ping.

1009	   A Pong message MAY be sent unsolicited.  This serves as a
1010	   unidirectional heartbeat.  A response to an unsolicited pong is not
1011	   expected.

1013	4.6.  Data Frames

1015	   Data frames (e.g. non control frames) are identified by opcodes where
1016	   the most significant bit of the opcode is 0.  Currently defined
1017	   opcodes for data frames include 0x1 (Text), 0x2 (Binary).  Opcodes
1018	   0x3-0x7 are reserved for further non-control frames yet to be
1019	   defined.

1021	   Data frames carry application-layer or extension-layer data.  The
1022	   opcode determines the interpretation of the data:

1024	   Text

1026	      The payload data is text data encoded as UTF-8.

1028	   Binary

1030	      The payload data is arbitrary binary data whose interpretation is
1031	      solely up to the application layer.

1033	4.7.  Examples

1035	   _This section is non-normative._

1037	   o  A single-frame unmasked text message

1039	      *  0x81 0x05 0x48 0x65 0x6c 0x6c 0x6f (contains "Hello")

1041	   o  A single-frame masked text message

1043	      *  0x81 0x85 0x37 0xfa 0x21 0x3d 0x7f 0x9f 0x4d 0x51 0x58
1044	         (contains "Hello")

1046	   o  A fragmented unmasked text message

1048	      *  0x01 0x03 0x48 0x65 0x6c (contains "Hel")

1050	      *  0x80 0x02 0x6c 0x6f (contains "lo")

1052	   o  Ping request and response

1054	      *  0x89 0x05 0x48 0x65 0x6c 0x6c 0x6f (contains a body of "Hello",
1055	         but the contents of the body are arbitrary)

1057	      *  0x8a 0x05 0x48 0x65 0x6c 0x6c 0x6f (contains a body of "Hello",
1058	         matching the body of the ping)

1060	   o  256 bytes binary message in a single unmasked frame

1062	      *  0x82 0x7E 0x0100 [256 bytes of binary data]

1064	   o  64KiB binary message in a single unmasked frame

1066	      *  0x82 0x7F 0x0000000000010000 [65536 bytes of binary data]

1068	4.8.  Extensibility

1070	   The protocol is designed to allow for extensions, which will add
1071	   capabilities to the base protocols.  The endpoints of a connection
1072	   MUST negotiate the use of any extensions during the handshake.  This
1073	   specification provides opcodes 0x3 through 0x7 and 0xB through 0xF,
1074	   the extension data field, and the frame-rsv1, frame-rsv2, and frame-
1075	   rsv3 bits of the frame header for use by extensions.  The negotiation
1076	   of extensions is discussed in further detail in Section 8.1.  Below
1077	   are some anticipated uses of extensions.  This list is neither
1078	   complete nor proscriptive.

1080	   o  Extension data may be placed in the payload before the application
1081	      data.

1083	   o  Reserved bits can be allocated for per-frame needs.

1085	   o  Reserved opcode values can be defined.

1087	   o  Reserved bits can be allocated to the opcode field if more opcode
1088	      values are needed.

1090	   o  A reserved bit or an "extension" opcode can be defined which
1091	      allocates additional bits out of the payload area to define larger
1092	      opcodes or more per-frame bits.

1094	5.  Opening Handshake

1096	5.1.  Client Requirements

1098	   User agents running in controlled environments, e.g. browsers on
1099	   mobile handsets tied to specific carriers, may offload the management
1100	   of the connection to another agent on the network.  In such a
1101	   situation, the user agent for the purposes of conformance is
1102	   considered to include both the handset software and any such agents.

1104	   When the user agent is to *establish a WebSocket connection* given
1105	   either a WebSocket URI /uri/ or the constituent components of a URI
1106	   as specified in Section 11, it MUST meet the following requirements.
1107	   In the following text, we will use terms from Section 3 such as
1108	   "/host/" and "/secure/ flag" as defined in that section.

1110	   1.  The WebSocket URI and its components derived by applying the
1111	       steps defined in Section 3.3, or if the following algorithm was
1112	       supplied with the constituent components as defined in Section 11
1113	       then those components provided, MUST be valid according to
1114	       Section 3.3.  If any of the requirements are not met, the client
1115	       MUST fail the WebSocket connection and abort these steps.

1117	   2.  If the user agent already has a WebSocket connection to the
1118	       remote host (IP address) identified by /host/ and port /port/
1119	       pair, even if the remote host is known by another name, the user
1120	       agent MUST wait until that connection has been established or for
1121	       that connection to have failed.  There MUST be no more than one
1122	       connection in a CONNECTING state.  If multiple connections to the
1123	       same IP address are attempted simultaneously, the user agent MUST
1124	       serialize them so that there is no more than one connection at a
1125	       time running through the following steps.

1127	       If the user agent cannot determine the IP address of the remote
1128	       host (for example because all communication is being done through
1129	       a proxy server that performs DNS queries itself), then the user
1130	       agent MUST assume for the purposes of this step that each host
1131	       name refers to a distinct remote host, but should instead limit
1132	       the total number of simultaneous connections that are not
1133	       established to a reasonably low number (e.g., in a Web browser,
1134	       to the number of tabs the user has open).

1136	       NOTE: This makes it harder for a script to perform a denial of
1137	       service attack by just opening a large number of WebSocket
1138	       connections to a remote host.  A server can further reduce the
1139	       load on itself when attacked by making use of this by pausing
1140	       before closing the connection, as that will reduce the rate at
1141	       which the client reconnects.

1143	       NOTE: There is no limit to the number of established WebSocket
1144	       connections a user agent can have with a single remote host.
1145	       Servers can refuse to accept connections from hosts with an
1146	       excessive number of existing connections, or disconnect resource-
1147	       hogging connections when suffering high load.

1149	   3.  _Proxy Usage_: If the user agent is configured to use a proxy
1150	       when using the WebSocket protocol to connect to host /host/
1151	       and/or port /port/, then the user agent SHOULD connect to that
1152	       proxy and ask it to open a TCP connection to the host given by
1153	       /host/ and the port given by /port/.

1155	          EXAMPLE: For example, if the user agent uses an HTTP proxy for
1156	          all traffic, then if it was to try to connect to port 80 on
1157	          server example.com, it might send the following lines to the
1158	          proxy server:

1160	              CONNECT example.com:80 HTTP/1.1
1161	              Host: example.com

1163	          If there was a password, the connection might look like:

1165	              CONNECT example.com:80 HTTP/1.1
1166	              Host: example.com
1167	              Proxy-authorization: Basic ZWRuYW1vZGU6bm9jYXBlcyE=

1169	       If the user agent is not configured to use a proxy, then a direct
1170	       TCP connection SHOULD be opened to the host given by /host/ and
1171	       the port given by /port/.

1173	       NOTE: Implementations that do not expose explicit UI for
1174	       selecting a proxy for WebSocket connections separate from other
1175	       proxies are encouraged to use a SOCKS proxy for WebSocket
1176	       connections, if available, or failing that, to prefer the proxy
1177	       configured for HTTPS connections over the proxy configured for
1178	       HTTP connections.

1180	       For the purpose of proxy autoconfiguration scripts, the URI to
1181	       pass the function MUST be constructed from /host/, /port/,
1182	       /resource name/, and the /secure/ flag using the steps to
1183	       construct a WebSocket URI as given in Section 3.2.

1185	       NOTE: The WebSocket protocol can be identified in proxy
1186	       autoconfiguration scripts from the scheme ("ws:" for unencrypted
1187	       connections and "wss:" for encrypted connections).

1189	   4.  If the connection could not be opened, either because a direct
1190	       connection failed or because any proxy used returned an error,
1191	       then the user agent MUST fail the WebSocket connection and abort
1192	       the connection attempt.

1194	   5.  If /secure/ is true, the user agent MUST perform a TLS handshake
1195	       over the connection [RFC2818].  If this fails (e.g. the server's
1196	       certificate could not be verified), then the user agent MUST fail
1197	       the WebSocket connection and abort the connection.  Otherwise,
1198	       all further communication on this channel MUST run through the
1199	       encrypted tunnel.  [RFC5246]

1201	       User agents MUST use the Server Name Indication extension in the
1202	       TLS handshake.  [RFC6066]

1204	   Once a connection to the server has been established (including a
1205	   connection via a proxy or over a TLS-encrypted tunnel), the client
1206	   MUST send a handshake to the server.  The handshake consists of an
1207	   HTTP upgrade request, along with a list of required and optional
1208	   headers.  The requirements for this handshake are as follows.

1210	   1.   The handshake MUST be a valid HTTP request as specified by
1211	        [RFC2616].

1213	   2.   The Method of the request MUST be GET and the HTTP version MUST
1214	        be at least 1.1.

1216	        For example, if the WebSocket URI is "ws://example.com/chat",
1217	        The first line sent should be "GET /chat HTTP/1.1"

1219	   3.   The request MUST contain a "Request-URI" as part of the GET
1220	        method.  This MUST match the /resource name/ Section 3.

1222	   4.   The request MUST contain a "Host" header whose value is equal to
1223	        /host/

1225	   5.   The request MUST contain an "Upgrade" header whose value is
1226	        equal to "websocket".

1228	   6.   The request MUST contain a "Connection" header whose value MUST
1229	        include the "Upgrade" token.

1231	   7.   The request MUST include a header with the name "Sec-WebSocket-
1232	        Key".  The value of this header MUST be a nonce consisting of a
1233	        randomly selected 16-byte value that has been base64-encoded
1234	        [RFC3548].  The nonce MUST be selected randomly for each
1235	        connection.

1237	        NOTE: As an example, if the randomly selected value was the
1238	        sequence of bytes 0x01 0x02 0x03 0x04 0x05 0x06 0x07 0x08 0x09
1239	        0x0a 0x0b 0x0c 0x0d 0x0e 0x0f 0x10, the value of the header
1240	        would be "AQIDBAUGBwgJCgsMDQ4PEC=="

1242	   8.   The request MUST include a header with the name "Sec-WebSocket-
1243	        Origin" if the request is coming from a browser client.  If the
1244	        connection is from a non-browser client, the request MAY include
1245	        this header if the semantics of that client match the use-case
1246	        described here for browser clients.  The value of this header
1247	        MUST be the ASCII serialization of origin of the context in
1248	        which the code establishing the connection is running, and MUST
1249	        be lower-case.  The value MUST NOT contain letters in the range
1250	        U+0041 to U+005A (i.e.  LATIN CAPITAL LETTER A to LATIN CAPITAL
1251	        LETTER Z) [I-D.ietf-websec-origin].

1253	        As an example, if code is running on www.example.com attempting
1254	        to establish a connection to ww2.example.com, the value of the
1255	        header would be "http://www.example.com".

1257	   9.   The request MUST include a header with the name "Sec-WebSocket-
1258	        Version".  The value of this header MUST be 7.

1260	   10.  The request MAY include a header with the name "Sec-WebSocket-
1261	        Protocol".  If present, this value indicates the subprotocol(s)
1262	        the client wishes to speak.  The elements that comprise this
1263	        value MUST be non-empty strings with characters in the range
1264	        U+0021 to U+007E and MUST all be unique strings.  The ABNF for
1265	        the value of this header is 1#(token | quoted-string), where the
1266	        definitions of constructs and rules are as given in [RFC2616].

1268	   11.  The request MAY include a header with the name "Sec-WebSocket-
1269	        Extensions".  If present, this value indicates the protocol-
1270	        level extension(s) the client wishes to speak.  The
1271	        interpretation and format of this header is described in
1272	        Section 8.1.

1274	   12.  The request MAY include headers associated with sending cookies,
1275	        as defined by the appropriate specifications
1276	        [I-D.ietf-httpstate-cookie].

1278	   Once the client's opening handshake has been sent, the client MUST
1279	   wait for a response from the server before sending any further data.
1280	   The client MUST validate the server's response as follows:

1282	   o  If the status code received from the server is not 101, the client
1283	      handles the response per HTTP procedures.  Otherwise, proceed as
1284	      follows.

1286	   o  If the response lacks an Upgrade header or the Upgrade header
1287	      contains a value that is not an ASCII case-insensitive match for
1288	      the value "websocket", the client MUST fail the WebSocket
1289	      connection.

1291	   o  If the response lacks a Connection header or the Connection header
1292	      contains a value that is not an ASCII case-insensitive match for
1293	      the value "Upgrade", the client MUST fail the WebSocket
1294	      connection.

1296	   o  If the response lacks a Sec-WebSocket-Accept header or the Sec-
1297	      WebSocket-Accept contains a value other than the base64-encoded
1298	      SHA-1 of the concatenation of the Sec-WebSocket-Key (as a string,
1299	      not base64-decoded) with the string "258EAFA5-E914-47DA-95CA-
1300	      C5AB0DC85B11", the client MUST fail the WebSocket connection.

1302	   Where the algorithm above requires that a user agent fail the
1303	   WebSocket connection, the user agent MAY first read an arbitrary
1304	   number of further bytes from the connection (and then discard them)
1305	   before actually *failing the WebSocket connection*.  Similarly, if a
1306	   user agent can show that the bytes read from the connection so far
1307	   are such that there is no subsequent sequence of bytes that the
1308	   server can send that would not result in the user agent being
1309	   required to *fail the WebSocket connection*, the user agent MAY
1310	   immediately *fail the WebSocket connection* without waiting for those
1311	   bytes.

1313	   NOTE: The previous paragraph is intended to make it conforming for
1314	   user agents to implement the algorithm in subtly different ways that
1315	   are equivalent in all ways except that they terminate the connection
1316	   at earlier or later points.  For example, it enables an
1317	   implementation to buffer the entire handshake response before
1318	   checking it, or to verify each field as it is received rather than
1319	   collecting all the fields and then checking them as a block.

1321	5.2.  Server-side requirements

1323	   _This section only applies to servers._

1325	   Servers MAY offload the management of the connection to other agents
1326	   on the network, for example load balancers and reverse proxies.  In
1327	   such a situation, the server for the purposes of conformance is
1328	   considered to include all parts of the server-side infrastructure
1329	   from the first device to terminate the TCP connection all the way to
1330	   the server that processes requests and sends responses.

1332	   EXAMPLE: For example, a data center might have a server that responds
1333	   to WebSocket requests with an appropriate handshake, and then passes
1334	   the connection to another server to actually process the data frames.
1335	   For the purposes of this specification, the "server" is the
1336	   combination of both computers.

1338	5.2.1.  Reading the client's opening handshake

1340	   When a client starts a WebSocket connection, it sends its part of the
1341	   opening handshake.  The server must parse at least part of this
1342	   handshake in order to obtain the necessary information to generate
1343	   the server part of the handshake.

1345	   The client handshake consists of the following parts.  If the server,
1346	   while reading the handshake, finds that the client did not send a
1347	   handshake that matches the description below, the server MUST abort
1348	   the WebSocket connection.

1350	   1.  An HTTP/1.1 or higher GET request, including a "Request-URI"
1351	       [RFC2616] that should be interpreted as a /resource name/
1352	       Section 3.

1354	   2.  A "Host" header containing the server's authority.

1356	   3.  A "Sec-WebSocket-Key" header with a base64-encoded value that,
1357	       when decoded, is 16 bytes in length.

1359	   4.  A "Sec-WebSocket-Version" header, with a value of 7.

1361	   5.  Optionally, a "Sec-WebSocket-Origin" header.  This header is sent
1362	       by all browser clients.  A connection attempt lacking this header
1363	       SHOULD NOT be interpreted as coming from a browser client.

1365	   6.  Optionally, a "Sec-WebSocket-Protocol" header, with a list of
1366	       values indicating which protocols the client would like to speak,
1367	       ordered by preference.

1369	   7.  Optionally, a "Sec-WebSocket-Extensions" header, with a list of
1370	       values indicating which extensions the client would like to
1371	       speak.  The interpretation of this header is discussed in
1372	       Section 8.1.

1374	   8.  Optionally, other headers, such as those used to send cookies to
1375	       a server.  Unknown headers MUST be ignored.

1377	5.2.2.  Sending the server's opening handshake

1379	   When a client establishes a WebSocket connection to a server, the
1380	   server MUST complete the following steps to accept the connection and
1381	   send the server's opening handshake.

1383	   1.  If the server supports encryption, perform a TLS handshake over
1384	       the connection.  If this fails (e.g. the client indicated a host
1385	       name in the extended client hello "server_name" extension that
1386	       the server does not host), then close the connection; otherwise,
1387	       all further communication for the connection (including the
1388	       server handshake) MUST run through the encrypted tunnel.
1389	       [RFC5246]

1391	   2.  Establish the following information:

1393	       /origin/
1394	          The |Sec-WebSocket-Origin| header in the client's handshake
1395	          indicates the origin of the script establishing the
1396	          connection.  The origin is serialized to ASCII and converted
1397	          to lowercase.  The server MAY use this information as part of
1398	          a determination of whether to accept the incoming connection.
1399	          If the server does not validate the origin, it will accept
1400	          connections from anywhere.  For more detail, refer to
1401	          Section 9.

1403	       /key/
1404	          The |Sec-WebSocket-Key| header in the client's handshake
1405	          includes a base64-encoded value that, if decoded, is 16 bytes
1406	          in length.  This (encoded) value is used in the creation of
1407	          the server's handshake to indicate an acceptance of the
1408	          connection.  It is not necessary for the server to base64-
1409	          decode the Sec-WebSocket-Key value.

1411	       /version/
1412	          The |Sec-WebSocket-Version| header in the client's handshake
1413	          includes the version of the WebSocket protocol the client is
1414	          attempting to communicate with.  If this version does not
1415	          match a version understood by the server, the server MUST
1416	          abort the WebSocket connection.  The server MAY send a non-200
1417	          response code with a |Sec-WebSocket-Version| header indicating
1418	          the version(s) the server is capable of understanding.

1420	       /resource name/
1421	          An identifier for the service provided by the server.  If the
1422	          server provides multiple services, then the value should be
1423	          derived from the resource name given in the client's handshake
1424	          from the Request-URI [RFC2616] of the GET method.

1426	       /subprotocol/
1427	          Either a single value or null, representing the subprotocol
1428	          the server is ready to use.  If the server supports multiple
1429	          subprotocols, then the value MUST be derived from the client's
1430	          handshake, specifically by selecting one of the values from
1431	          the "Sec-WebSocket-Protocol" field.  The absence of such a
1432	          field is equivalent to the null value.  The empty string is
1433	          not the same as the null value for these purposes, and is not
1434	          a legal value for this field.  The ABNF for the value of this
1435	          header is (token | quoted-string), where the definitions of
1436	          constructs and rules are as given in [RFC2616].

1438	       /extensions/
1439	          A (possibly empty) list representing the protocol-level
1440	          extensions the server is ready to use.  If the server supports
1441	          multiple extensions, then the value MUST be derived from the
1442	          client's handshake, specifically by selecting one or more of
1443	          the values from the "Sec-WebSocket-Extensions" field.  The
1444	          absence of such a field is equivalent to the null value.  The
1445	          empty string is not the same as the null value for these
1446	          purposes.  Extensions not listed by the client MUST NOT be
1447	          listed.  The method by which these values should be selected
1448	          and interpreted is discussed in Section 8.1.

1450	   3.  If the server chooses to accept the incoming connection, it MUST
1451	       reply with a valid HTTP response indicating the following.

1453	       1.  A 101 response code.  Such a response could look like
1454	           "HTTP/1.1 101 Switching Protocols"

1456	       2.  A "Sec-WebSocket-Accept" header.  The value of this header is
1457	           constructed by concatenating /key/, defined above in
1458	           Paragraph 2 of Section 5.2.2, with the string "258EAFA5-E914-
1459	           47DA-95CA-C5AB0DC85B11", taking the SHA-1 hash of this
1460	           concatenated value to obtain a 20-byte value, and base64-
1461	           encoding this 20-byte hash.

1463	           NOTE: As an example, if the value of the "Sec-WebSocket-Key"
1464	           header in the client's handshake were
1465	           "dGhlIHNhbXBsZSBub25jZQ==", the server would append the
1466	           string "258EAFA5-E914-47DA-95CA-C5AB0DC85B11" to form the
1467	           string "dGhlIHNhbXBsZSBub25jZQ==258EAFA5-E914-47DA-95CA-
1468	           C5AB0DC85B11".  The server would then take the SHA-1 hash of
1469	           this string, giving the value 0xb3 0x7a 0x4f 0x2c 0xc0 0x62
1470	           0x4f 0x16 0x90 0xf6 0x46 0x06 0xcf 0x38 0x59 0x45 0xb2 0xbe
1471	           0xc4 0xea.  This value is then base64-encoded, to give the
1472	           value "s3pPLMBiTxaQ9kYGzzhZRbK+xOo=", which would be returned
1473	           in the "Sec-WebSocket-Accept" header.

1475	       3.  Optionally, a "Sec-WebSocket-Protocol" header, with a value
1476	           /subprotocol/ as defined in Paragraph 2 of Section 5.2.2.

1478	       4.  Optionally, a "Sec-WebSocket-Extensions" header, with a value
1479	           /extensions/ as defined in Paragraph 2 of Section 5.2.2.

1481	   This completes the server's handshake.  If the server finishes these
1482	   steps without aborting the WebSocket connection, and if the client
1483	   does not then fail the WebSocket connection, then the connection is
1484	   established and the server may begin sending and receiving data.

1486	6.  Error Handling

1488	6.1.  Handling errors in UTF-8 from the server

1490	   When a client is to interpret a byte stream as UTF-8 but finds that
1491	   the byte stream is not in fact a valid UTF-8 stream, then any bytes
1492	   or sequences of bytes that are not valid UTF-8 sequences MUST be
1493	   interpreted as a U+FFFD REPLACEMENT CHARACTER.

1495	6.2.  Handling errors in UTF-8 from the client

1497	   When a server is to interpret a byte stream as UTF-8 but finds that
1498	   the byte stream is not in fact a valid UTF-8 stream, behavior is
1499	   undefined.  A server could close the connection, convert invalid byte
1500	   sequences to U+FFFD REPLACEMENT CHARACTERs, store the data verbatim,
1501	   or perform application-specific processing.  Subprotocols layered on
1502	   the WebSocket protocol might define specific behavior for servers.

1504	7.  Closing the connection

1506	7.1.  Definitions

1508	7.1.1.  Close the WebSocket Connection

1510	   To _Close the WebSocket Connection_, an endpoint closes the
1511	   underlying TCP connection.  An endpoint SHOULD use a method that
1512	   cleanly closes the TCP connection, as well as the TLS session, if
1513	   applicable, discarding any trailing bytes that may be received.  An
1514	   endpoint MAY close the connection via any means available when
1515	   necessary, such as when under attack.

1517	   As an example of how to obtain a clean closure in C using Berkeley
1518	   sockets, one would call shutdown() with SHUT_WR on the socket, call
1519	   recv() until obtaining a return value of 0 indicating that the peer
1520	   has also performed an orderly shutdown, and finally calling close()
1521	   on the socket.

1523	7.1.2.  Start the WebSocket Closing Handshake

1525	   To _start the WebSocket closing handshake_, an endpoint MUST send a
1526	   Close control frame, as described in Section 4.5.1.  Once an endpoint
1527	   has both sent and received a Close control frame, that endpoint
1528	   SHOULD _Close the WebSocket Connection_ as defined in Section 7.1.1.

1530	7.1.3.  The WebSocket Connection Is Closed

1532	   When the underlying TCP connection is closed, it is said that _the
1533	   WebSocket connection is closed_.  If the tcp connection was closed
1534	   after the WebSocket closing handshake was completed, the WebSocket
1535	   connection is said to have been closed _cleanly_.

1537	7.1.4.  Fail the WebSocket Connection

1539	   Certain algorithms and specifications require a user agent to _fail
1540	   the WebSocket connection_.  To do so, the user agent MUST _Close the
1541	   WebSocket Connection_, and MAY report the problem to the user (which
1542	   would be especially useful for developers) in an appropriate manner.

1544	   Except as indicated above or as specified by the application layer
1545	   (e.g. a script using the WebSocket API), user agents SHOULD NOT close
1546	   the connection.

1548	7.2.  Abnormal closures
1549	7.2.1.  Client-initiated closure

1551	   Certain algorithms, namely during the initial handshake, require the
1552	   user agent to *fail the WebSocket connection*.  To do so, the user
1553	   agent MUST _Close the WebSocket connection_ as previously defined,
1554	   and MAY report the problem to the user via an appropriate mechanism
1555	   (which would be especially useful for developers).

1557	   Except as indicated above or as specified by the application layer
1558	   (e.g. a script using the WebSocket API), user agents SHOULD NOT close
1559	   the connection.

1561	7.2.2.  Server-initiated closure

1563	   Certain algorithms require or recommend that the server _abort the
1564	   WebSocket connection_ during the opening handshake.  To do so, the
1565	   server MUST simply _close the WebSocket connection_ (Section 7.1.1).

1567	7.3.  Normal closure of connections

1569	   Servers MAY close the WebSocket connection whenever desired.  User
1570	   agents SHOULD NOT close the WebSocket connection arbitrarily.  In
1571	   either case, an endpoint initiates a closure by following the
1572	   procedures to _start the WebSocket closing handshake_
1573	   (Section 7.1.2).

1575	7.4.  Status codes

1577	   When closing an established connection (e.g. when sending a Close
1578	   frame, after the handshake has completed), an endpoint MAY indicate a
1579	   reason for closure.  The interpretation of this reason by an
1580	   endpoint, and the action an endpoint should take given this reason,
1581	   are left undefined by this specification.  This specification defines
1582	   a set of pre-defined status codes, and specifies which ranges may be
1583	   used by extensions, frameworks, and end applications.  The status
1584	   code and any associated textual message are optional components of a
1585	   Close frame.

1587	7.4.1.  Defined Status Codes

1589	   Endpoints MAY use the following pre-defined status codes when sending
1590	   a Close frame.

1592	   1000

1594	      1000 indicates a normal closure, meaning whatever purpose the
1595	      connection was established for has been fulfilled.

1597	   1001

1599	      1001 indicates that an endpoint is "going away", such as a server
1600	      going down, or a browser having navigated away from a page.

1602	   1002

1604	      1002 indicates that an endpoint is terminating the connection due
1605	      to a protocol error.

1607	   1003

1609	      1003 indicates that an endpoint is terminating the connection
1610	      because it has received a type of data it cannot accept (e.g. an
1611	      endpoint that understands only text data MAY send this if it
1612	      receives a binary message).

1614	   1004

1616	      1004 indicates that an endpoint is terminating the connection
1617	      because it has received a message that is too large.

1619	7.4.2.  Reserved status code ranges

1621	   0-999

1623	      Status codes in the range 0-999 are not used.

1625	   1000-1999

1627	      Status codes in the range 1000-1999 are reserved for definition by
1628	      this protocol.

1630	   2000-2999

1632	      Status codes in the range 2000-2999 are reserved for use by
1633	      extensions.

1635	   3000-3999

1637	      Status codes in the range 3000-3999 MAY be used by libraries and
1638	      frameworks.  The interpretation of these codes is undefined by
1639	      this protocol.  End applications MUST NOT use status codes in this
1640	      range.

1642	   4000-4999

1644	      Status codes in the range 4000-4999 MAY be used by application
1645	      code.  The interpretation of these codes is undefined by this
1646	      protocol.

1648	8.  Extensions

1650	   WebSocket clients MAY request extensions to this specification, and
1651	   WebSocket servers MAY accept some or all extensions requested by the
1652	   client.  A server MUST NOT respond with any extension not requested
1653	   by the client.  If extension parameters are included in negotiations
1654	   between the client and the server, those parameters MUST be chosen in
1655	   accordance with the specification of the extension to which the
1656	   parameters apply.

1658	8.1.  Negotiating extensions

1660	   A client requests extensions by including a "Sec-WebSocket-
1661	   Extensions" header, which follows the normal rules for HTTP headers
1662	   (see [RFC2616] section 4.2) and the value of the header is defined by
1663	   the following ABNF.  Note that unlike other section of the document
1664	   this section is using ABNF syntax/rules from [RFC2616].

1666	         extension-list = 1#extension
1667	         extension = extension-token *( ";" extension-param )
1668	         extension-token = registered-token | private-use-token
1669	         registered-token = token
1670	         private-use-token = "x-" token
1671	         extension-param = token [ "=" ( token | quoted-string ) ]

1673	   Note that like other HTTP headers, this header MAY be split or
1674	   combined across multiple lines.  Ergo, the following are equivalent:

1676	         Sec-WebSocket-Extensions: foo
1677	         Sec-WebSocket-Extensions: bar; baz=2

1679	   is exactly equivalent to

1681	         Sec-WebSocket-Extensions: foo, bar; baz=2

1683	   Any extension-token used MUST either be a registered token
1684	   (registration TBD), or have a prefix of "x-" to indicate a private-
1685	   use token.  The parameters supplied with any given extension MUST be
1686	   defined for that extension.  Note that the client is only offering to
1687	   use any advertised extensions, and MUST NOT use them unless the
1688	   server indicates that it wishes to use the extension.

1690	   Note that the order of extensions is significant.  Any interactions
1691	   between multiple extensions MAY be defined in the documents defining
1692	   the extensions.  In the absence of such definition, the
1693	   interpretation is that the headers listed by the client in its
1694	   request represent a preference of the headers it wishes to use, with
1695	   the first options listed being most preferable.  The extensions
1696	   listed by the server in response represent the extensions actually in
1697	   use for the connection.  Should the extensions modify the data and/or
1698	   framing, the order of operations on the data should be assumed to be
1699	   the same as the order in which the extensions are listed in the
1700	   server's response in the opening handshake.

1702	   For example, if there are two extensions "foo" and "bar", if the
1703	   header |Sec-WebSocket-Extensions| sent by the server has the value
1704	   "foo, bar" then operations on the data will be made as
1705	   bar(foo(data)), be those changes to the data itself (such as
1706	   compression) or changes to the framing thay may "stack".

1708	   Non-normative examples of acceptable extension headers:

1710	      Sec-WebSocket-Extensions: deflate-stream
1711	      Sec-WebSocket-Extensions: mux; max-channels=4; flow-control, deflate-stream
1712	      Sec-WebSocket-Extensions: x-private-extension

1714	   A server accepts one or more extensions by including a |Sec-
1715	   WebSocket-Extensions| header containing one or more extensions which
1716	   were requested by the client.  The interpretation of any extension
1717	   parameters, and what constitutes a valid response by a server to a
1718	   requested set of parameters by a client, will be defined by each such
1719	   extension.

1721	8.2.  Known extensions

1723	   Extensions provide a mechanism for implementations to opt-in to
1724	   additional protocol features.  This section defines the meaning of
1725	   well-known extensions but implementations MAY use extensions defined
1726	   separately as well.

1728	8.2.1.  Compression

1730	   The registered extension token for this compression extension is
1731	   "deflate-stream".

1733	   The extension does not have any per message extension data and it
1734	   does not define the use of any WebSocket reserved bits or op codes.

1736	   Senders using this extension MUST apply RFC 1951 encodings to all
1737	   bytes of the data stream following the handshake including both data
1738	   and control messages.  The data stream MAY include multiple blocks of
1739	   both compressed and uncompressed types as defined by [RFC1951].

1741	   Senders MUST NOT delay the transmission of any portion of a WebSocket
1742	   message because the deflate encoding of the message does not end on a
1743	   byte boundary.  The encodings for adjacent messages MAY appear in the
1744	   same byte if no delay in transmission is occurred by doing so.

1746	   Historically there have been some confusion and interoperability
1747	   problems around the specification of compression algorithms.  In this
1748	   specification "deflate-stream" requires a [RFC1951] deflate encoding.
1749	   It MUST NOT be wrapped in any of the header formats often associated
1750	   with RFC 1951 such as "zlib" [RFC1950].  This requirement is given
1751	   special attention with this note because of confusion in this area,
1752	   the presence of some popular open source libraries that create both
1753	   formats under a single API call with confusing naming conventions,
1754	   and the fact that the popular HTTP [RFC2616] specification defines
1755	   "deflate" compression differently than this specification.

1757	9.  Security considerations

1759	   While this protocol is intended to be used by scripts in Web pages,
1760	   it can also be used directly by hosts.  Such hosts are acting on
1761	   their own behalf, and can therefore send fake "Origin" fields,
1762	   misleading the server.  Servers should therefore be careful about
1763	   assuming that they are talking directly to scripts from known
1764	   origins, and must consider that they might be accessed in unexpected
1765	   ways.  In particular, a server should not trust that any input is
1766	   valid.

1768	   EXAMPLE: For example, if the server uses input as part of SQL
1769	   queries, all input text should be escaped before being passed to the
1770	   SQL server, lest the server be susceptible to SQL injection.

1772	   Servers that are not intended to process input from any Web page but
1773	   only for certain sites SHOULD verify the "Origin" field is an origin
1774	   they expect, and should only respond with the corresponding "Sec-
1775	   WebSocket-Origin" if it is an accepted origin.  Servers that only
1776	   accept input from one origin can just send back that value in the
1777	   "Sec-WebSocket-Origin" field, without bothering to check the client's
1778	   value.

1780	   If at any time a server is faced with data that it does not
1781	   understand, or that violates some criteria by which the server
1782	   determines safety of input, or when the server sees a handshake that
1783	   does not correspond to the values the server is expecting (e.g.
1784	   incorrect path or origin), the server SHOULD just disconnect.  It is
1785	   always safe to disconnect.

1787	   The biggest security risk when sending text data using this protocol
1788	   is sending data using the wrong encoding.  If an attacker can trick
1789	   the server into sending data encoded as ISO-8859-1 verbatim (for
1790	   instance), rather than encoded as UTF-8, then the attacker could
1791	   inject arbitrary frames into the data stream.

1793	   In addition to endpoints being the target of attacks via WebSockets,
1794	   other parts of web infrastructure, such as proxies, may be the
1795	   subject of an attack.  In particular, an intermediary may interpret a
1796	   WebSocket message from a client as a request, and a message from the
1797	   server as a response to that request.  For instance, an attacker
1798	   could get a browser to establish a connection to its server, get the
1799	   browser to send a message that looks to an intermediary like a GET
1800	   request for a common piece of JavaScript on another domain, and send
1801	   back a message that is interpreted as a cacheable response to that
1802	   request, thus poisioning the cache for other users.  To prevent this
1803	   attack, messages sent from clients are masked on the wire with a 32-
1804	   bit value, to prevent an attacker from controlling the bits on the
1805	   wire and thus lessen the probability of an attacker being able to
1806	   construct a message that can be misinterpreted by a proxy as a non-
1807	   WebSocket request.

1809	   As mentioned in Section 6.2, servers must be extremely cautious
1810	   interpreting invalid UTF-8 data from the client.  A naive UTF-8
1811	   parsing implementation can result in buffer overflows in the case of
1812	   invalid input data.

1814	10.  IANA considerations

1816	10.1.  Registration of ws: scheme

1818	   A |ws:| URI identifies a WebSocket server and resource name.

1820	   URI scheme name.
1821	      ws

1823	   Status.
1824	      Permanent.

1826	   URI scheme syntax.
1827	      In ABNF terms using the terminals from the URI specifications:
1828	      [RFC5234] [RFC3986]

1830	           "ws" ":" hier-part [ "?" query ]

1832	      The <path> [RFC3986] and <query> components form the resource name
1833	      sent to the server to identify the kind of service desired.  Other
1834	      components have the meanings described in RFC3986.

1836	   URI scheme semantics.
1837	      The only operation for this scheme is to open a connection using
1838	      the WebSocket protocol.

1840	   Encoding considerations.
1841	      Characters in the host component that are excluded by the syntax
1842	      defined above MUST be converted from Unicode to ASCII by applying
1843	      the IDNA ToASCII algorithm to the Unicode host name, with both the
1844	      AllowUnassigned and UseSTD3ASCIIRules flags set, and using the
1845	      result of this algorithm as the host in the URI.  [RFC3490]

1847	      Characters in other components that are excluded by the syntax
1848	      defined above MUST be converted from Unicode to ASCII by first
1849	      encoding the characters as UTF-8 and then replacing the
1850	      corresponding bytes using their percent-encoded form as defined in
1851	      the URI and IRI specifications.  [RFC3986] [RFC3987]

1853	   Applications/protocols that use this URI scheme name.
1854	      WebSocket protocol.

1856	   Interoperability considerations.
1857	      None.

1859	   Security considerations.
1860	      See "Security considerations" section above.

1862	   Contact.
1863	      HYBI WG <hybi@ietf.org>

1865	   Author/Change controller.
1866	      IETF <iesg@ietf.org>

1868	   References.
1869	      RFC XXXX

1871	10.2.  Registration of wss: scheme

1873	   A |wss:| URI identifies a WebSocket server and resource name, and
1874	   indicates that traffic over that connection is to be encrypted.

1876	   URI scheme name.
1877	      wss

1879	   Status.
1880	      Permanent.

1882	   URI scheme syntax.
1883	      In ABNF terms using the terminals from the URI specifications:
1884	      [RFC5234] [RFC3986]

1886	           "wss" ":" hier-part [ "?" query ]

1888	      The <path> and <query> components form the resource name sent to
1889	      the server to identify the kind of service desired.  Other
1890	      components have the meanings described in RFC3986.

1892	   URI scheme semantics.
1893	      The only operation for this scheme is to open a connection using
1894	      the WebSocket protocol, encrypted using TLS.

1896	   Encoding considerations.
1897	      Characters in the host component that are excluded by the syntax
1898	      defined above MUST be converted from Unicode to ASCII by applying
1899	      the IDNA ToASCII algorithm to the Unicode host name, with both the
1900	      AllowUnassigned and UseSTD3ASCIIRules flags set, and using the
1901	      result of this algorithm as the host in the URI.  [RFC3490]

1903	      Characters in other components that are excluded by the syntax
1904	      defined above MUST be converted from Unicode to ASCII by first
1905	      encoding the characters as UTF-8 and then replacing the
1906	      corresponding bytes using their percent-encoded form as defined in
1907	      the URI and IRI specification.  [RFC3986] [RFC3987]

1909	   Applications/protocols that use this URI scheme name.
1910	      WebSocket protocol over TLS.

1912	   Interoperability considerations.
1913	      None.

1915	   Security considerations.
1916	      See "Security considerations" section above.

1918	   Contact.
1919	      HYBI WG <hybi@ietf.org>

1921	   Author/Change controller.
1922	      IETF <iesg@ietf.org>

1924	   References.
1925	      RFC XXXX

1927	10.3.  Registration of the "WebSocket" HTTP Upgrade keyword

1929	   Name of token.
1930	      WebSocket

1932	   Author/Change controller.
1933	      IETF <iesg@ietf.org>

1935	   Contact.
1936	      HYBI <hybi@ietf.org>

1938	   References.
1939	      RFC XXXX

1941	10.4.  Sec-WebSocket-Key

1943	   This section describes a header field for registration in the
1944	   Permanent Message Header Field Registry.  [RFC3864]

1946	   Header field name
1947	      Sec-WebSocket-Key

1949	   Applicable protocol
1950	      http

1952	   Status
1953	      standard

1955	   Author/Change controller
1956	      IETF

1958	   Specification document(s)
1959	      RFC XXXX

1961	   Related information
1962	      This header field is only used for WebSocket handshake.

1964	   The |Sec-WebSocket-Key| header is used in the WebSocket handshake.
1965	   It is sent from the client to the server to provide part of the
1966	   information used by the server to prove that it received a valid
1967	   WebSocket handshake.  This helps ensure that the server does not
1968	   accept connections from non-WebSocket clients (e.g.  HTTP clients)
1969	   that are being abused to send data to unsuspecting WebSocket servers.

1971	10.5.  Sec-WebSocket-Extensions

1973	   This section describes a header field for registration in the
1974	   Permanent Message Header Field Registry.  [RFC3864]

1976	   Header field name
1977	      Sec-WebSocket-Extensions

1979	   Applicable protocol
1980	      http

1982	   Status
1983	      standard

1985	   Author/Change controller
1986	      IETF

1988	   Specification document(s)
1989	      RFC XXXX

1991	   Related information
1992	      This header field is only used for WebSocket handshake.

1994	   The |Sec-WebSocket-Extensions| header is used in the WebSocket
1995	   handshake.  It is initially sent from the client to the server, and
1996	   then subsequently sent from the server to the client, to agree on a
1997	   set of protocol-level extensions to use for the duration of the
1998	   connection.

2000	10.6.  Sec-WebSocket-Accept

2002	   This section describes a header field for registration in the
2003	   Permanent Message Header Field Registry.  [RFC3864]

2005	   Header field name
2006	      Sec-WebSocket-Accept

2008	   Applicable protocol
2009	      http

2011	   Status
2012	      standard

2014	   Author/Change controller
2015	      IETF

2017	   Specification document(s)
2018	      RFC XXXX

2020	   Related information
2021	      This header field is only used for WebSocket handshake.

2023	   The |Sec-WebSocket-Accept| header is used in the WebSocket handshake.
2024	   It is sent from the server to the client to confirm that the server
2025	   is willing to initiate the connection.

2027	10.7.  Sec-WebSocket-Origin

2029	   This section describes a header field for registration in the
2030	   Permanent Message Header Field Registry.  [RFC3864]

2032	   Header field name
2033	      Sec-WebSocket-Origin

2035	   Applicable protocol
2036	      http

2038	   Status
2039	      standard

2041	   Author/Change controller
2042	      IETF

2044	   Specification document(s)
2045	      RFC XXXX

2047	   Related information
2048	      This header field is only used for WebSocket handshake.

2050	   The |Sec-WebSocket-Origin| header is used in the WebSocket handshake.
2051	   It is sent from the server to the client to confirm the origin of the
2052	   script that opened the connection.  This enables user agents to
2053	   verify that the server is willing to serve the script that opened the
2054	   connection.

2056	10.8.  Sec-WebSocket-Protocol

2058	   This section describes a header field for registration in the
2059	   Permanent Message Header Field Registry.  [RFC3864]

2061	   Header field name
2062	      Sec-WebSocket-Protocol

2064	   Applicable protocol
2065	      http

2067	   Status
2068	      standard

2070	   Author/Change controller
2071	      IETF

2073	   Specification document(s)
2074	      RFC XXXX

2076	   Related information
2077	      This header field is only used for WebSocket handshake.

2079	   The |Sec-WebSocket-Protocol| header is used in the WebSocket
2080	   handshake.  It is sent from the client to the server and back from
2081	   the server to the client to confirm the subprotocol of the
2082	   connection.  This enables scripts to both select a subprotocol and be
2083	   sure that the server agreed to serve that subprotocol.

2085	10.9.  Sec-WebSocket-Version

2087	   This section describes a header field for registration in the
2088	   Permanent Message Header Field Registry.  [RFC3864]

2090	   Header field name
2091	      Sec-WebSocket-Version

2093	   Applicable protocol
2094	      http

2096	   Status
2097	      standard

2099	   Author/Change controller
2100	      IETF

2102	   Specification document(s)
2103	      RFC XXXX

2105	   Related information
2106	      This header field is only used for WebSocket handshake.

2108	   The |Sec-WebSocket-Version| header is used in the WebSocket
2109	   handshake.  It is sent from the client to the server to indicate the
2110	   protocol version of the connection.  This enables servers to
2111	   correctly interpret the handshake and subsequent data being sent from
2112	   the data, and close the connection if the server cannot interpret
2113	   that data in a safe manner.

2115	11.  Using the WebSocket protocol from other specifications

2117	   The WebSocket protocol is intended to be used by another
2118	   specification to provide a generic mechanism for dynamic author-
2119	   defined content, e.g. in a specification defining a scripted API.

2121	   Such a specification first needs to "establish a WebSocket
2122	   connection", providing that algorithm with:

2124	   o  The destination, consisting of a /host/ and a /port/.

2126	   o  A /resource name/, which allows for multiple services to be
2127	      identified at one host and port.

2129	   o  A /secure/ flag, which is true if the connection is to be
2130	      encrypted, and false otherwise.

2132	   o  An ASCII serialization of an origin that is being made responsible
2133	      for the connection.  [I-D.ietf-websec-origin]

2135	   o  Optionally a string identifying a protocol that is to be layered
2136	      over the WebSocket connection.

2138	   The /host/, /port/, /resource name/, and /secure/ flag are usually
2139	   obtained from a URI using the steps to parse a WebSocket URI's
2140	   components.  These steps fail if the URI does not specify a
2141	   WebSocket.

2143	   If a connection can be established, then it is said that the
2144	   "WebSocket connection is established".

2146	   If at any time the connection is to be closed, then the specification
2147	   needs to use the "close the WebSocket connection" algorithm.

2149	   When the connection is closed, for any reason including failure to
2150	   establish the connection in the first place, it is said that the
2151	   "WebSocket connection is closed".

2153	   While a connection is open, the specification will need to handle the
2154	   cases when "a WebSocket message has been received" with text /data/.

2156	   To send some text /data/ to an open connection, the specification
2157	   needs to "send /data/ using the WebSocket".

2159	12.  Acknowledgements

2161	   Special thanks are due to Ian Hickson, who was the original author
2162	   and editor of this protocol.  The initial design of this
2163	   specification benefitted from the participation of many people in the
2164	   WHATWG and WHATWG mailing list.  Contributions to that specification
2165	   are not tracked by section, but a list of all who contributed to that
2166	   specification is given in the WHATWG HTML specification at
2167	   http://whatwg.org/html5.

2169	   Special thanks also to John Tamplin for providing a significant
2170	   amount of text for the Data Framing section of this specification.

2172	   Special thanks also to Adam Barth for providing a significant amount
2173	   of text and background research for the Data Masking section of this
2174	   specification.

2176	13.  References

2178	13.1.  Normative References

2180	   [ANSI.X3-4.1986]
2181	              American National Standards Institute, "Coded Character
2182	              Set - 7-bit American Standard Code for Information
2183	              Interchange", ANSI X3.4, 1986.

2185	   [FIPS.180-2.2002]
2186	              National Institute of Standards and Technology, "Secure
2187	              Hash Standard", FIPS PUB 180-2, August 2002, <http://
2188	              csrc.nist.gov/publications/fips/fips180-2/fips180-2.pdf>.

2190	   [RFC1951]  Deutsch, P., "DEFLATE Compressed Data Format Specification
2191	              version 1.3", RFC 1951, May 1996.

2193	   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
2194	              Requirement Levels", BCP 14, RFC 2119, March 1997.

2196	   [RFC2616]  Fielding, R., Gettys, J., Mogul, J., Frystyk, H.,
2197	              Masinter, L., Leach, P., and T. Berners-Lee, "Hypertext
2198	              Transfer Protocol -- HTTP/1.1", RFC 2616, June 1999.

2200	   [RFC2818]  Rescorla, E., "HTTP Over TLS", RFC 2818, May 2000.

2202	   [RFC3490]  Faltstrom, P., Hoffman, P., and A. Costello,
2203	              "Internationalizing Domain Names in Applications (IDNA)",
2204	              RFC 3490, March 2003.

2206	   [RFC3492]  Costello, A., "Punycode: A Bootstring encoding of Unicode
2207	              for Internationalized Domain Names in Applications
2208	              (IDNA)", RFC 3492, March 2003.

2210	   [RFC3548]  Josefsson, S., "The Base16, Base32, and Base64 Data
2211	              Encodings", RFC 3548, July 2003.

2213	   [RFC3629]  Yergeau, F., "UTF-8, a transformation format of ISO
2214	              10646", STD 63, RFC 3629, November 2003.

2216	   [RFC3864]  Klyne, G., Nottingham, M., and J. Mogul, "Registration
2217	              Procedures for Message Header Fields", BCP 90, RFC 3864,
2218	              September 2004.

2220	   [RFC3986]  Berners-Lee, T., Fielding, R., and L. Masinter, "Uniform
2221	              Resource Identifier (URI): Generic Syntax", STD 66,
2222	              RFC 3986, January 2005.

2224	   [RFC3987]  Duerst, M. and M. Suignard, "Internationalized Resource
2225	              Identifiers (IRIs)", RFC 3987, January 2005.

2227	   [RFC5246]  Dierks, T. and E. Rescorla, "The Transport Layer Security
2228	              (TLS) Protocol Version 1.2", RFC 5246, August 2008.

2230	   [RFC6066]  Eastlake, D., "Transport Layer Security (TLS) Extensions:
2231	              Extension Definitions", RFC 6066, January 2011.

2233	   [RFC4648]  Josefsson, S., "The Base16, Base32, and Base64 Data
2234	              Encodings", RFC 4648, October 2006.

2236	   [RFC5234]  Crocker, D. and P. Overell, "Augmented BNF for Syntax
2237	              Specifications: ABNF", STD 68, RFC 5234, January 2008.

2239	13.2.  Informative References

2241	   [WSAPI]    Hickson, I., "The Web Sockets API", August 2010,
2242	              <http://dev.w3.org/html5/websockets/>.

2244	   [I-D.ietf-httpstate-cookie]
2245	              Barth, A., "HTTP State Management Mechanism",
2246	              draft-ietf-httpstate-cookie-20 (work in progress),
2247	              December 2010.

2249	   [I-D.ietf-websec-origin]
2250	              Barth, A., "The Web Origin Concept",
2251	              draft-ietf-websec-origin-00 (work in progress),
2252	              December 2010.

2254	   [RFC1950]  Deutsch, L. and J-L. Gailly, "ZLIB Compressed Data Format
2255	              Specification version 3.3", RFC 1950, May 1996.

2257	   [RFC5321]  Klensin, J., "Simple Mail Transfer Protocol", RFC 5321,
2258	              October 2008.

2260	   [RFC6202]  Loreto, S., Saint-Andre, P., Salsano, S., and G. Wilkins,
2261	              "Known Issues and Best Practices for the Use of Long
2262	              Polling and Streaming in Bidirectional HTTP", RFC 6202,
2263	              April 2011.

2265	Author's Address

2267	   Ian Fette
2268	   Google, Inc.

2270	   Email: ifette+ietf@google.com
2271	   URI:   http://www.ianfette.com/