grpc-devtool

WIP - beta expected soon

gRPC-devtool

gRPC-devtool is a cli program to monitor, record and playback gRPC traffic.

Features

• Monitor gRPC traffic
• Record gRPC traffic
• Playback recorded traffic
• Define matcher rules to match and respond to an endpoint
• Use templates to generate responses dynamically
• Support for sessions

Setup

A sample demonstration of creating a project from scratch

• Run as node module

  # Install with npm
  npm install -g grpc-devtool
  
  # Create a project
  grpc create --protos path/to/protos
  
  # Recrod traffic
  grpc record 
  
  # Serve mapped responses 
  grpc start
  

• Run as docker container

  TODO

• Use API in nodejs project

  TODO

Sample strucuture of config folder

my-stub/
  - config/
  - grpc.yaml
  - greet/
    - default.yaml
    - rohit.yaml
    - virat.yaml
  - prices/
    - defualt.yaml
  - common/
  - message.yaml

Configurations

Configuration file shoud be placed in the config dir and named as grpc.yaml

Example of a cofiguration file :

protos : '../Protos'
host : localhost
port : 3000
trimmedStreamSize : 10

All configuration values can be configured at runtime (e.g. in your ci build).

e.g.

grpc start --port 50055 

Configuration options

Name	default
host	localhost	Address of server ran by grpc-devtool (use it in you app)
port	3009	Port of server exposed by grpc-devtool (use it in you app)
protos		Relative or absolute path to the directory containing proto files
expressionSymbols	["{{", "}}"]	e.g {{request.body.name}}
sessionEnabled	true	Enable session support
keywordSuffix	@	ends with @
trimmedStreamSize	10	number for responses for a streaming server to keep in mappings file. Will repeast the responses unless configured otherwise per request basis.

Defining Mappings

To define a stubbed grpc endpoint, first update the config/mappings.yaml in config directory

# Name of endpoint and response rules in order
helloworld.greet.Greeter.SayHello : [
  "greet/rohit.js.yaml",
  "greet/virat.js.yaml",
  "greet/default.js.yaml",
]

prices.streaming.Pricing.Subscribe : [
  "prices/211-Stock.js.yaml",
]

Remember that the responses are applied in the order declared in mappings file. So if "greet/rohit.js.yaml" matches the request, the files next to it won't be matched with the request.

Define responses

A simple unary request :

request@ : {
  name: "rohit"
}

response@ : {
  message : "Hello Rohit"
}

For a streaming response :

# prices/211-Stock.js.yaml
request@ : {
  uic: 211,
  assetType: "Stock"
}

# Response for a streaming request
response@ : {
  stream@  : [
    {quote: "quote:one"}, 
    {quote: "quote:two"}, 
    {quote: "quote:three"}
  ],
  doNotRepeat@ : true,   # if not defined, will stream infinitely
  streamInterval@ : 500  # 500 ms of delay between consecutive messages
}

Using matchers and templates

For some endpoint, it is must to have part of response based on request body. In such case we can use the template :

# Responds to any request
request@ : {
  name: "any@"
}

# Using expressions in response
response@ : {
  message : "Glad to meet you {{request.body.name}}"
}

WIP

Specs below this are not implemented or guranteed to work for now.

Including templates

If your response is extremly complex and you need to parameterize certaion parts of it, you can use partial templates. E.g.

request@: {
  uic: 211,
  assetType: "Stock"
}

reply@: {
  stream@: [
    { message: "this is your first message"},
    { message: "this is your second message"},
  
    {
    	# path is relative to config directory
      include@: "shared/message-template.js.yaml",
      param@: {username: "{{request.body.name}}"}
    },
    
  ],
  repeat@: false
}

Scripting in templates

You can add custom script to handle complex scnarios :

request@ : {
	name    : "any@"    
  lastName: "Singh"  
  # applied only if rest of body matches
  # template ignored if it return false
  js@: `
  	return request.matches && request.body.age > 18 
  `
}

You can even write your own code to create responses dynamically using javascript :

request@: {
  stream@: [
  {name: "first-user"},
  {name: "second-user"},
  ]
}

@reply: {
  stream@: {
    js@: "
      endpoint.calls = endpoint.calls  || 0;
      endpoint.calls++;
      
      scope.calls = scope.calls  || 0;
      scope.calls++;
      
      var message = `
        hello : ${request.body.name},
        total calls to this endpoint : ${endpoint.calls},
        total replies by this rule : ${scope.calls},
        sequence in stream : ${stream.$index}`;
      
      return {message};
    "
    }
  }

Extensions

You can write your own javascript code to add custom logic to templates. Create a directory config/ext and simply put you javascript file there.

Create custom matchers

To create custom matcher, create a file ``config/ext /asset-types.js` as :

const matchers = require('miraje/matchers');

const validAssetTypes = ['Stock', 'CfdOnFutures', 'FxSpot'];

module.exports = {
	appliesTo : (str) => str === 'assetTypes@',
  matches   : (value) => validAssetTypes.includes(value)
}

Sessions

In case you wan't to build a stateful stub (not recommened), you can use sessions. To enable sessions make sure thatsessionEnabled: true is set inconfig/grpc.yaml

Then you can use sessions in your matchers or templates :

request@: {
  name: 'rohit'
}

@reply: {
  stream@: {
    response: {
      message: 'You called me {{session[request.name]}} times.'
    }
    js@: '
      session[request.name] = session[request.name] || 0;
      session[request.name] += session[request.name]; 
    '
    }
  }

Roadmap

• Test with prices service at saxo
• Add Oauth
• Add MTLS