Middleware for operating on models. This is useful for such common middleware tasks as authorizing a user for performing specific operations on a model and for validating changes/updated to a model.

the index.ls file exports the relevant API to external modules (see index_test-ls for examples of usage).

model-mw      = require 'model-mw'
ModelMw       = model-mw.mw
ModelRunner   = model-mw.runner

The runner ModelRunner is registered on Middleware for easy use (by name).

Middleware.register model: ModelRunner

Any runner registered in this fashion can be configured for use by name like this:

middleware = new Middleware('model')

middleware = new Middleware('model', data: users.kris)

The Middleware builder is passed a context object which in this case contains a 'data' entry as expected by the ModelRunner. This data is then passed on to each middleware component registered via use.

middleware = new Middleware('model', data: users.kris).use(validator: validation-mw)

Each middleware component has a connection to the middleware runner which handles running each component in succession. After each component completes the result is collected on the results of the runner.

When the runner is done, the on-success function is normally called. If one or more errors occured, the on-error function is called.

Any mw-component always has access to the runner, and can f.ex use has-errors or similar for decisions in its run method, such as whether it should abort etc. (see specs for more detailed API)

Example:

done-fun = ->
  console.log 'I'm done!', @success, @results

my-error-fun = ->
  throw Error 'Oops! Errors:', @errors

ctx = data: users.kris, on-success: done-fun, on-error: my-error-fun
model-mw = new Middleware('model', ctx)
model-mw.use(authorize-mw).use(validation-mw)
model-mw.run!

You can also set model and/or collection as part of the context.

The runner will try to determine the model and collection based on the data if model or collection are not passed as arguments. Model will always be a singular, underscored entity, such as: project or admin_user. Collection is pluralised by the same rule: admin_users.

If the data is a LiveScript class, the model will be determined by name of the constructor. Otherwise it will try to find a clazz entry on any Object passed as data (see model_container_test` for more).

An Mw-component can issue abort! to abort the current runner from executing any more mw-components. You can also use error(msg) to add an error to the runner and ensure that on-error will be called when the runner completes. By default on-error returns the errors object.

An mw-component can be run "alone" (i.e. without a runner), by calling its run method with the mode 'alone'.

mw-alone.run mode: 'alone', data: users.kris

This can be useful for testing an mw-component stand-alone.

For a more detailed understanding of the internals, please see the specs ;)

Please see:

• middleware
• authorization-mw
• validation-mw

The current model runner is very simple and only supports running every Mw-component in synchronous mode. This is far from optimal! There should be more advanced runners available from the middleware project that support Mw-components that return promises and also one that can run one or more Mw-components asynchronously and wait until they all deliver their results etc.

Please help out with this effort ;) The async runners should support both Q and RSVP, by allowing configuration of defer function (same as LGTM validator approach).

Using mocha

Run all test

$ mocha

Run particular test

$ mocha test/model-mw/mw/model_flow_test.js

Easy :)

Please suggest improvements and help improve this project :)