richmond
v0.4.12
Published
Map REST calls to MongoDB
Downloads
10
Maintainers
Readme
richmond.js
Map REST calls to MongoDB
Installation
$ npm init
$ npm install richmond --save
$ npm install richmond-web-controller --save
Important Note: be sure to be using the latest richmond-web-controller. There was a database id validation bug in earlier versions (fixed in 0.1.3).
Terminology
- demo controller - refers to the richmond-web-controller that you install separately.
Usage
Step 1: Visit MongoLab
Use your own local install of MongoDB or visit https://mlab.com and create a free test database, writing down the credentials.
Step 2: Edit ~/.bash_profile
Using your favorite plain text editor add the lines below to ~/.bash_profile (if on a Mac or Linux).
Replace SUBDOMAIN, DBPORT, DATABASE, USER and PASSWORD with your values.
You can also choose other values for APP_SECRET and TEST_PORT.
# Application
export APP_SECRET=testsecret
# MONGO LABS DB
export TEST_MONGO_DB=mongodb://SUBDOMAIN.mongolab.com:DBPORT/DATABASE
export TEST_MONGO_USER=USER
export TEST_MONGO_PASS=PASSWORD
export TEST_PORT=3030
When you are done with that, execute the following at the command line:
$ source ~/.bash_profile
Step 3: Create a config.js file in your projects root folder:
This file can also be found in the online source under examples/demo
Note that this has changed since version 0.1.x.
var Controller = require('richmond-web-controller');
module.exports = {
controller: new Controller(),
service: {
logFile: "my-test.log",
port: process.env.TEST_PORT || null,
secret: process.env.APP_SECRET || null,
prefix: "/API",
database: {
uri: process.env.TEST_MONGO_DB || 'mongodb://localhost/mytest',
options: {
user: process.env.TEST_MONGO_USER || null,
pass: process.env.TEST_MONGO_PASS || null
}
}
}
};
Step 4: Create index.js in your projects root folder:
This file can also be found in the online source under examples/demo
Important Note: if you omit a line in the controller setup for an HTTP method (POST, PUT, etc) trying to make an HTTP request using that method will result in a returned status of 404 (Not Found).
That may actually be desired. For example you may want to build a Web service where people can only read (GET) records from your database. So you would only include the setup lines for getOne (get one record) and getMany (get collection) in the demo controller.
As of 0.3.0 you can pass all of the service options to the constructor instead of .setup.
As of 0.4.0 it is no longer necessary to call .connect if the database configs were already setup. .addModel will call .connect directly if no connection can be found.
var Richmond = require('richmond'),
config = require('./config'),
controller = config.controller,
service = config.service,
micro = new Richmond(service),
modelName = "MyTest";
micro.controller(
controller.setup({
del: [{ model: modelName, rights: "PUBLIC" }],
getOne: [{ model: modelName, rights: "PUBLIC" }],
getMany: [{ model: modelName, rights: "PUBLIC" }],
post: [{ model: modelName, rights: "PUBLIC" }],
put: [{ model: modelName, rights: "PUBLIC" }],
patch: [{ model: modelName, rights: "PUBLIC" }],
})
);
micro.addModel(modelName, {
email: { type: String, required: true },
status: { type: String, required: true },
password: { type: String, select: false },
});
micro.listen();
console.log("Listening on port:", service.port);
Also note that the lines above in the controller.setup code only apply to the demo controller. Future and third party controllers may implement their own strategy and options for defining how the controller works.
Step 5: Install and run the app
From your projects root folder, execute the following at the command line:
$ node index.js
Step 6: Test the app using curl commands
Note that these commands refer to how the demo controller works. Future and third party controllers may operate differently.
POST
Create a new record at the command line using curl (assumes port 3030):
$ curl -i -X POST -H "Content-Type: application/json"
-d '{"email":"[email protected]","password":"foo","status":"OK"}'
http://localhost:3030/api/mytest
Repeat that step a few times. Change the values to create a small set of records to experiment with.
GET
Now get all the records (by default non-selected fields, like password, will not be returned):
$ curl -X GET -H "Accept: applications/json"
http://localhost:3030/api/mytest
In some browsers, like Chrome, you can also see the raw JSON returned by browsing to: http://localhost:3030/api/mytest
The demo controller lets you get an individual document using the record id like this (substitute for a record from your database):
$ curl -X GET -H "Accept: applications/json"
http://localhost:3030/api/mytest/54ce6eca470103ca057b0097
You can append a filter to a GET request like this (%7B = '{' and %7D = '}'):
$ curl -X GET -H "Accept: applications/json"
'http://localhost:3030/api/mytest?filter=%7B"email":"[email protected]"%7D'
You can also select what fields to show (%20 is a space), even non-selected fields (like password):
$ curl -X GET -H "Accept: applications/json"
'http://localhost:3030/api/mytest?fields=email%20status%20password'
Note that if a field was never set in the record, you will not see it listed in the returned record.
PUT
PUT is similar to POST where you have to pass in data. But since you are updating a record that already exists, you need to also append a record id to the URL.
$ curl -i -X PUT -H "Content-Type: application/json"
-d '{"email":"[email protected]","password":"foo","status":"UPDATED"}'
http://localhost:3030/api/mytest/54ce6eca470103ca057b0097
The general philosophy with PUT is that you should use it to replace the entire record, and you should use PATCH to do partial updates. If you only pass in an incomplete set of fields, the demo controller does a merge. Technically that isn't very RESTful. There is no guarantee that future controllers may be more strict. If you want to be more strict, simply make sure to pass in all fields.
This is what the demo PUT controller currently does behind the scenes via mongoose:
collection.update( { _id : req.params.id }, { $set : body }, ... )
PATCH
PATCH is simular to PUT. You still need to include the id of the record in the URL. But the data you pass in is not a set of fields. Instead it's a set of instructions for how to patch the record.
$ curl -i -X PATCH -H "Content-Type: application/json"
-d '[{"op":"replace","path":"/status","value":"PATCH THE STATUS"}]'
http://localhost:3030/api/mytest/54ce741e470103ca057b0098
Behind the scenes the demo controller currently uses fast-json-patch. Search for that on npm for more info how to apply patches.
DELETE
Deleting a record through curl is similar to getting a record - you simply append the id to the URL.
$ curl -i -X DELETE -H "Content-Type: application/json"
http://localhost:3021/api/mytest/54ce5b2109f8166e04258d31
Middleware
Under the hood this module is using express.js and wraps the app.use call.
To inject middleware, like CORS, you can do the following (assumes you installed cors and required it):
micro.use( cors() );
micro.listen();
SSL
The Web controller in the demo supports SSL.
To use SSL with the demo controller you must add an SSL key with a value of 404 or 302 in the setup.
For example, here is how you would do it for getOne:
getOne: [{ model: modelName, rights: "PUBLIC", ssl: 404 }],
A value of 404 means that if a user attempts to browse to the Non-SSL version of the URL a 404 (Not Found) status will be returned.
A value of 302 (Moved) will result in the user being redirected to the SSL equivalent of the request.
Wrappers
You can add before and after wrappers to the demo controller like this:
post: [{ model: modelName, rights: "PUBLIC",
before: beforePost, after: afterPost }],
This example uses a before method to hash a password before saving it.
The after method demonstrates how to remove the hashed password before the doc is returned.
The example also includes showing how to pass through extra data to the after method.:
var testExtraMessage = 'Testing 123';
var beforePost =
function (prop, next) {
var extras = { message: testExtraMessage };
var body = prop.req.body;
if (body.password) {
bcrypt.hash(body.password, 10, function (err, hash) {
if (err) {
throw err;
}
body.password = hash;
next(body, extras);
});
} else {
next(body, extras);
}
};
var afterPost =
function (prop, next) {
var doc = JSON.parse(JSON.stringify(prop.result));
thepatch = [ { "op": "remove", "path": "/password" } ];
jsonpatch.apply( doc, thepatch );
var extras = prop.extras;
if (extras.message != testExtraMessage) {
throw new Error("Test extra message not what expected.");
}
next(doc);
};
Multiple Models
This module supports multiple models. The setup could look something like this:
modelName = ["AlphaTest","BetaTest"];
micro
.logFile("two-models-test.log")
.controller(
controller.setup({
del: [ { model: modelName[0], rights: "PUBLIC" },
{ model: modelName[1], rights: "PUBLIC" } ],
getOne: [ { model: modelName[0], rights: "PUBLIC" },
{ model: modelName[1], rights: "PUBLIC" } ],
getMany: [ { model: modelName[0], rights: "PUBLIC" },
{ model: modelName[1], rights: "PUBLIC" } ],
post: [ { model: modelName[0], rights: "PUBLIC" },
{ model: modelName[1], rights: "PUBLIC" } ],
put: [ { model: modelName[0], rights: "PUBLIC" },
{ model: modelName[1], rights: "PUBLIC" } ]
}));
// Model[0]
micro.addModel(modelName[0], {
email: { type: String, required: true },
status: { type: String, required: true },
});
// Model[1]
micro.addModel(modelName[1], {
user: { type: String, required: true },
level: { type: String, required: true },
});
micro.listen();
API
Constructor
You can still call the constructor with no arguments. As of 0.3.0 you can also call the constructor passing in the options for setup.
The constructor will still initialize internal values, but then call setup with the options as a final step. If you pass in the options, there would no need to call setup later. That will save you a step and simplify your code.
Richmond = require('richmond');
micro = new Richmond(options);
Or
Richmond = require('richmond');
micro = new Richmond();
micro.setup(options)
See the setup method for more info on what options are available.
.setup(options)
A method introduced in 0.2.0 that lets you setup several parameters simultaneously. This allows you to move some functionality to a config file, etc.
You do not need to include every option in the usage example below.
You can still change a value (example: prefix) after calling setup. Just call the related method afterwards (i.e.: micro.prefix("/v1"));
As of 0.3.0 an alternative is to pass the options to the constructor which will call setup internally.
Usage
micro = new Richmond();
var options = {
logFile: "my-test.log",
port: process.env.TEST_PORT || null,
secret: process.env.APP_SECRET || null,
prefix: "/API",
database: {
uri: process.env.TEST_MONGO_DB || 'mongodb://localhost/mytest',
options: {
user: process.env.TEST_MONGO_USER || null,
pass: process.env.TEST_MONGO_PASS || null
}
}
};
micro.setup(options);
Or:
var options = {
logFile: "my-test.log",
port: process.env.TEST_PORT || null,
secret: process.env.APP_SECRET || null,
prefix: "/API",
database: {
uri: process.env.TEST_MONGO_DB || 'mongodb://localhost/mytest',
options: {
user: process.env.TEST_MONGO_USER || null,
pass: process.env.TEST_MONGO_PASS || null
}
}
};
micro = new Richmond(options);
.logFile(filename)
- If no path is specified, will just write to the apps current directory.
- If writing to a folder, the folder must already exist.
- If this is not called all log output will go through the console.
- You can also skip this call by setting the value through setup or the constructor.
Usage
micro.logFile("my-test.log");
Or via setup:
var options = {
logFile: "my-test.log"
};
micro.setup(options);
Or via the constructor:
var options = {
logFile: "my-test.log"
};
micro = new Richmond(options);
.prefix()
Will return the prefix string.
Usage
var s = micro.prefix();
.prefix(string)
Allows defining a prefix for all routes.
Will validate the string and convert it internally to lowercase.
micro.prefix("/API")
Routes will contain "/api" like this: http://localhost:3030/api/mytest
micro.prefix("/v1")
Routes will contain "/v1" like this: http://localhost:3030/v1/mytest
Prefix validation rules:
- prefix can't be null
- prefix must begin with a slash
- prefix must not end with a slash
- prefix must not contain whitepace
- You can also skip this call by setting the value through setup or the constructor.
Usage
micro.prefix("/v1");
Or via setup:
var options = {
prefix: "/v1"
};
micro.setup(options);
Or via the constructor:
var options = {
prefix: "/v1"
};
micro = new Richmond(options);
.addModel(name,model)
- If .connect was not called .addModel will call it with no parameters. It assumes that all database connection info was set via setup or the constructor.
- Assigns a name to a Mongoose model and saves it.
- The model name will be used in routes, like this: http://localhost:3030/api/mytest
- The name will be validated internally with a call to normalizeModelName.
Usage:
micro.connect();
MyTests = micro.addModel("mytest", {
email: { type: String, required: true },
status: { type: String, required: true },
password: { type: String, select: false },
});
// PURGE all records (don't do in production!)
MyTests.remove({"email": /@/ }, function (err) {
if (err) {
console.error(err);
}
}
.model(name)
Returns the Mongoose model that was stored via addModel. You can then use that to create a record and use other Mongoose methods, like .save.
Usage
This example would most likely be used in a POST controller.
function (req, res, next) {
var Collection = micro.model(name),
record = new Collection(req.body);
record.save(function (err, doc) {
if (err) { throw err; }
res
.location("/" + name + "/" + doc._id)
.status(201) // Created
.json(doc);
}
}
}
.normalizeModelName(name)
Makes sure the name is not null, has no spaces and returns the result in all lowercase. Called internally by addModel and model to make sure that internal keys are lowercase and not incompatible.
Usage
var modelName = micro.normalizeModelName(name);
connect()
connect(uri,options)
The uri would be in a form like this: mongodb://HOST:PORT/DATABASE
This is a wrapper for Mongoose.createConnection. See their documentation for more option parameters.
Usage
var options = {
user: dbConfig.user,
pass: dbConfig.pass
};
micro.connect(dbConfig.uri, options);
Alternatively you can set the parameters via setup and then call connect with no parameters:
var options = {
database: {
uri: process.env.TEST_MONGO_DB || 'mongodb://localhost/mytest',
options: {
user: process.env.TEST_MONGO_USER || null,
pass: process.env.TEST_MONGO_PASS || null
}
}
};
micro.setup(options);
micro.connect();
Or via the constructor:
var options = {
database: {
uri: process.env.TEST_MONGO_DB || 'mongodb://localhost/mytest',
options: {
user: process.env.TEST_MONGO_USER || null,
pass: process.env.TEST_MONGO_PASS || null
}
}
};
micro = new Richmond(options);
micro.connect();
.closeConnecton()
If there is an existing Mongoose connection this will close it.
.controller(controller)
Used to assign a controller (like richmond-web-controller) to process HTTP requests.
Usage
var controller = config.controller;
micro.controller(
controller.setup({
getOne: [ { model: modelName, rights: "PUBLIC" } ],
getMany: [ { model: modelName, rights: "PUBLIC" } ],
})
)
.use(middleware)
Wrapper for internal express.js app.
Usage
micro.use( cors() );
.listen()
.listen(port)
When everything is good to go, make this call last to start listening for requests on a particular port.
Usage
micro.listen(port);
Alternatively you can define the port via the setup method or the constructor. Then call listen with no parameters.
var options = {
port: process.env.TEST_PORT || null,
};
micro.setup(options);
...
micro.listen();
Or via the constructor:
var options = {
port: process.env.TEST_PORT || null,
};
micro = new Richmond(options);
...
micro.listen();
.closeService()
Closes service and any open connections.
Usage
micro.closeService();
.secret(secret)
Used internally and externally combined with jwt-simple to encrypt and decrypt strings of data.
Do NOT hardcode the value. Always get it from the environment.
Internally requests are intercepted by middleware and the headers are scanned for 'x-auth'. If x-auth is found it is decoded using jwt-simple's decode method and the string set by .secret(). The result is assigned via the middleware to req.token and carried on to the next method in the chain. Controllers then have the option to look for the decoded token and use it as they see fit.
Review the rights tests for more info.
Usage
micro.secret(appSecret);
request(testHost)
.post(testUrl)
.send(testObject)
.set('x-auth', jwt.encode({ username: "Mitch", role: "admin" }, appSecret))
.set('Content-Type', 'application/json')
.expect(201)
.end(function (err, res) {
should.not.exist(err);
done();
});
});
Alterntively you can set the secret through the setup method.
var options = {
secret: process.env.APP_SECRET || null
};
micro.setup(options);
Or via the constructor:
var options = {
secret: process.env.APP_SECRET || null
};
micro = new Richmond(options);
Then there is no need to call the .secret method directly But you can if you want to override the value set through setup
Tests
In order to run the tests, you need to add two more variables to your environment: TEST_HOST and TEST_SSL
export TEST_HOST=http://localhost:8081
export TEST_SSL=https://localhost:8081
The SSL version must be the https equivelant of the host value.
Internally the test cases use ngrok for SSL testing.
Source the changes:
$ source ~/.bash_profile
Tests assume that mocha has been installed globally. If not execute the following:
$ npm install -g mocha
Run the tests from the projects root folder in one of the following ways:
$ mocha --recursive --timeout 20000
Or
$ npm test
Or if you feel like kickin' it old skool:
make test
Testing by Version
To run the tests for recent versions:
$ mocha --timeout 5000 --recursive test/v0004/*test.js
The tests generate log files in a logs/ folder under the projects root folder.
Repos
Contributing
In lieu of a formal style guide, take care to maintain the existing coding style. Add unit tests for any new or changed functionality. Lint and test your code.
Version History
Version 0.4.12 release notes
- upgraded test dependency to richmond-web-controller 0.1.9
Version 0.4.11 release notes
- upgraded test dependency to richmond-web-controller 0.1.8
Version 0.4.10 release notes
- mongoose now uses global.Promise
Version 0.4.9 release notes
- refactored some test cases
Version 0.4.8 release notes
- removed obsolete Makefile
Version 0.4.7 release notes
- brought code coverage up to 100%
Version 0.4.6 release notes
- changed description
- added keyworks to package.json
Version 0.4.5 release notes
- brought code statement coverage up to 97%
Version 0.4.4 release notes
- integrated travis-ci and codecov.io support
- integrated ngrok module into SSL testing
- updated dependencies
- added grunt build support
- brought test cases up to date
- removed test cases for old, outdated versions
Version 0.4.3 release notes
- Updated test cases use logs/ folder
Version 0.4.2 release notes
- Added new github repo to package.json
Version 0.4.1 release notes
- Created an examples folder and updated the README.
Version 0.4.0 release notes
- .addModel will now automatically call .connect (with no arguments) if no connection has been made.
- .addModel assumes all connection parameters have been set via .setup or the constructor.
Version 0.3.1 release notes
- Refined some of the README examples.
Version 0.3.0 release notes
- You can now pass options to the constructor which will call .setup for you.
- Cloned tests to a new version 0.3.x suite and updated for new functionality.
- Added version info to all test suite output and test log file names.
Version 0.2.0 release notes
- Updated test cases to use richmond-web-controller 1.3.0 due to id validation issue
- Added .setup method
- Refactored tests by version to support backward compatibility
- Added a PATCH test
Version 0.1.3 release notes
- Removed redundant code, updated documentation.
Version 0.1.2 release notes
- Updated test cases to use demo controller 0.1.2
Version 0.1.1 release notes
- Updated repo url and test controller dependency.
Version 0.1.0 release notes
- Initial release