Espressif plays websocket

This article is only for ESP32.

ESP32 Use WebSocket There are few scenarios for communication , Most of the application layer protocols used are MQTT and HTTP. But for the WebSocket, Basic understanding is still necessary . I've just had time recently , Just roll it WebSocket.

This article is just a simple description of WebSocket Analyze , If you want to know more about it , Suggestion or reference RFC6455.

WebSocket What is it? ？

WebSocket It is a network communication protocol . Follow MQTT、HTTP agreement , It also belongs to TCP/IP Application layer protocol in the four layer model .

WebSocket The agreement 2008 The year was born ,2011 Became an international standard in .HTML5 Start offering a The network technology of full duplex communication between browser and server . It's based on TCP transmission , And reuse HTTP The handshake .

WebSocket And HTTP The relationship between ？

about HTTP, We all know that it has the following two characteristics ：

1. There is no connection ： Every time HTTP The connection only handles one request . After the server returns the request from the client , This time HTTP Connecting means disconnecting . This feature can greatly improve the execution efficiency of the server .
2. No state ：HTTP Agreement for request / The status information is not saved in the response process . It means that if the subsequent processing requires the previous information , It can only be achieved by retransmission . This feature can reduce the memory burden of the server , Improve the response speed of the server .

These two features greatly improve the execution efficiency and response speed of the server in theory . But it can't be covered up HTTP One drawback of the agreement , That's it Communication can only be initiated by the client . This was no problem before , But now applications have higher and higher requirements for real-time , A typical application scenario is live broadcast , Webpage game , Stock trading, etc . Although it can pass polling The way to solve , But the efficiency of polling is very low , Non stop polling virtually increases the burden on the server , This is related to HTTP The original intention of the design runs counter .

See here , Some readers must have thought of HTTP Long connection of . you 're right , When one HTTP After the server responds to your request ,HTTP Based on TCP The connection won't be closed . If the client wants to send again HTTP Request , It can be done directly at the TCP Send on connection . Compare the way of polling , Long connection is an elegant way .HTTP The long connection of is to add a header field to the response header Connection:keep-alive To achieve . But long connections don't stay connected forever , There is also a holding time .HTTP The long connection of is essentially TCP Long connection of .

• The same thing ：

  1. All belong to TCP/IP Application layer protocol in the four layer model .
  2. All based on TCP.
  3. WebSocket And HTTP Good compatibility , The default port is also 80 and 443.

• Difference ：
  HTTP Is one-way , It can only be initiated by the client Request request ;WebSocket It's two-way .

• contact ：
  WebSocket Just borrow HTTP To complete the handshake . After a successful handshake ,WebSocket It has its own frame for communication .

WebSocket And Socket The relationship between ？

about Socket, We all know that it is just an abstract concept , It will TCP/IP Layer operations are abstracted into several simple interfaces for application programs to call . For users , Just call a few socket Interface can complete the communication between hosts .

Socket It can also be used to represent the two ends of a connection in the network .WebSocket It just borrows this concept to represent the two ends of a connection in the network , These two ends can be regarded as browsers and servers .

WebSocket A simple example

Espressif Already available WebSocket Of demo. But in this demo The server used in is no longer providing services , So you can use a public server or set up a server locally . This paper chooses the latter .

build WebSocket Local server

In this paper Windows Building in the environment is based on nodejs Platform WebSocket Testing tools wscat.

Refer to the official instructions ：wscat

Installation steps ：

1. install nodejs
2. After installation, enter the following command under the command line to install globally wscat

npm install -g wscat

start-up WebSocket Server side

After installation , Enter the following command locally 8080 Port boot WebSocket service .

wscat -l 8080

start-up WebSocket client

stay WebSocket Execute the following commands under the directory to configure （ Set up WebSocket Of URL）：

idf.py menuconfig

Download the firmware to ESP32 And run , You can see in the WebSocket The server continues to receive messages from ESP32 Sent from Text data .

WebSocket agreement

combination Wireshark Grab the bag , You can simply right WebSocket Protocol analysis .

handshake

• establish TCP Connect

  This need not be introduced too much ,TCP Establishing a connection requires 3 The second handshake . package 23545 (SYN), package 23547 (SYN, ACK) And the bag 24983 (ACK) respectively TCP Of 3 Secondary handshake bag .

• handshake

  WebSocket With the help of HTTP To complete the handshake . The client first needs to send a handshake packet , Shake hands with HTTP GET Request The structure of is sent to the server , After the server parses it, use HTTP Response Send the structure of to the client , among Response The status code in is 101 Indicates a successful handshake .

stay WebSocket The protocol document has the following requirements for the handshake request package sent by the client :
1. The handshake request package must be a valid HTTP Request
2. Request Method It has to be for Get,Request Version Must be at least 1.1 edition
3. The request header must contain a Host Key value pair , If not used in the request header port If the key value is right , That's in Host The corresponding value must indicate the port number
4. The request header must contain a Upgrade Key value pair , Value must be websocket
5. The request header must contain a Connection Key value pair , Value must be Upgrade
6. The request header must contain a Sec-WebSocket-Key Key value pair , Values must be randomly generated 16 byte , With base64 code
7. If the client is a browser , Then the request header must contain a Origin Key value pair
8. The request header must contain a Sec-WebSocket-Version Key value pair , Value must be 13
9. The request header can contain a Sec-WebSocket-Protocol Key value pair
10. The request header can contain a Sec-WebSocket-Extensions Key value pair

Let's take a look ESP32 Sent for handshake GET Request package , Basically meet the requirements of the handshake request package in the Protocol .

If the server chooses to receive the connection , It must reply to a valid HTTP Response package .

stay WebSocket Reply to the server in the agreement document HTTP Response There are the following requirements :

1. The status code must be 101, The reason phrase can be Switching Protocols
2. There must be... In the first line Upgrade key , The value is websocket
3. There must be... In the first line Connection key , The value is Upgrade
4. There must be... In the first line Sec-WebSocket-Accept key , The value is based on the client Get Request Medium Sec-WebSocket-Key Key value generation , The process goes through SHA-1 Encrypted into 20 Bytes of data , After taking these data as base64 code .

Let's take a look wscat reply HTTP Response package , Basically meet the requirements in the agreement .

Data interaction

Data communication between client and server , The first thing to understand is WebSocket Data format in .WebSocket The data frame in is relatively simple , The format is as follows ：

1. FIN: 1 bit

   Used to indicate whether this frame is the last frame of the whole message to be sent . If the whole message can be sent in one frame , Then also set .

2. RSV1, RSV2, RSV3: 1 bit

   Retain . It has to be for 0.

3. opcode: 4 bits

   opcode . It's right Payload Data Explanation of domain . If the receiver receives an undefined opcode , Must end WebSocket Connect .

   value	Definition
   %x0	It indicates that this frame data is a continuation of the previous frame data , This frame is a continuous frame
   %x1	It indicates that this frame data is a text frame
   %x2	It indicates that this frame data is a binary data frame
   %x3 - 7	Retain , Applicable to future non control frames
   %x8	Indicates that this frame is a disconnected frame
   %x9	ping frame
   %xA	pong frame
   %xB - F	Retain , Suitable for future control frames

4. MASK: 1 bit

   Mask bit . To specify Payload Data Whether the data in the domain passes XOR( Exclusive or ) operation . If the value is 1, be Masking-key The value in the field is used to restore the process XOR( Exclusive or ) Data from the past . All frames sent from the client to the server must be set 1.

5. Payload len: 7 bits, 7 + 16 bits, 7 + 64 bits

   Load length bit , Must be combined with Extended payload length Fields together indicate Payload Data Length of data in the field , With byte In units of . Be careful Payload len Domain and Extended payload length All domains adopt Network byte order （ Big end model ）.

   1. When Payload len The value in the field is [0, 125] In the middle of the day , The value in this field represents the actual data length , here Extended payload length The domain disappears .Payload len The field is followed by Masking-key Domain .
   2. When Payload len The value in the field is 126 when , The value in this field indicates the subsequent Extended payload length Domain is 16 bits.Payload Data The length of data in the field is determined by Extended payload length The domain specifies .
   3. When Payload len The value in the field is 127 when , The value in this field indicates the subsequent Extended payload length Domain is 64 bits.Payload Data The length of data in the field is determined by Extended payload length The domain specifies .

6. Masking-key: 0 bit or 32 bits

   Mask key bit . MASK Domain is 1 Effective when . All frames sent from the client to the server MASK The domain must be set to 1. let me put it another way Masking-key Domain is specially used to serve the server . The server needs key To decode Payload len Data in domain . The frame settings sent by all servers to clients MASK Domain is 0, In this frame Masking-key Domain ellipsis .

To be continued …