TAO (aspect-oriented) modeling for MongoDB


Keywords
mongodb, aspect, tao, orm, db, schema, casting, stream, mongoose, aspect-oriented-framework, aspect-oriented-programming, middleware, nodejs, observable, odm, rxjs, streams
License
Apache-2.0
Install
npm install monogram@0.2.5

Documentation

monogram

Action-based anti-ODM for MongoDB and Node.js

Read the intro blog post here.

Usage

const { connect } = require('monogram');
const db = await connect('mongodb://localhost:27017/test');

Usage

Actions

From an end developer perspective, monogram behaves just like the MongoDB Node.js driver. The key difference is that monogram converts collection functions into actions under the hood. Actions are an object representation of a function call.

    const Test = db.collection('Test');

    let called = 0;
    Test.pre(action => {
      ++called;
      // An _action_ is an object representation of a function call.
      // It has an `_id` property to uniquely identify it, and
      // some other properties:
      assert.deepEqual(_.omit(action, ['_id']), {
        collection: 'Test', // The name of the collection
        name: 'insertOne', // The name of the function called
        params: [{
          hello: 'world'
        }], // The parameters passed to the function
        chained: [] // Function calls chained onto this one
      });
    });

    await Test.insertOne({ hello: 'world' });

    assert.equal(called, 1);
  

Motivation: Logging

Monogram isn't an ODM/ORM like its uncle mongoose, It's a new abstraction entirely. You can call it an AOM, "action-object mapping". Why is this abstraction better? Consider the problem of logging all database operations to the console in an ODM. In mongoose, this is hard, because there's a lot of different types of middleware. In monogram, this is trivial, because all database operations are represented in a common form, actions, and all actions go through one pipeline.

    const Test = db.collection('Test');

    let called = 0;

    Test.action$.subscribe(action => {
      ++called;
      const params = action.params.
        map(p => util.inspect(p, { depth: 5 })).
        join(', ');
      const msg = `${action.collection}.${action.name}(${params})`

      assert.equal(msg,
        `Test.updateOne({ _id: 1 }, { '$set': { hello: 'world' } })`);
    });

    await Test.updateOne({ _id: 1 }, {
      $set: { hello: 'world' }
    });

    assert.equal(called, 1);
  

Enforcing Internal Best Practices

The purpose of monogram is to allow you to enforce best practices, not to prescribe best practices. Beginners are best served using a tool like mongoose, which has a lot of baked-in best practices to prevent you from shooting yourself in the foot. Monogram is more for advanced users who have established best practices they want to enforce. For example, here's how you would prevent users from calling updateOne() or updateMany() without any update operators, which would overwrite the document.

    const Test = db.collection('Test');

    let called = 0;

    // Will catch `updateOne()`, `updateMany()`, and `findOneAndUpdate()`
    // actions
    Test.pre(/update/i, action => {
      const update = action.params[1] || {};
      const keys = Object.keys(update);
      if (keys.length > 0 && !keys[0].startsWith('$')) {
        throw new Error('Not allowed to overwrite document ' +
          'using `updateOne()`, use `replaceOne() instead`');
      }
    });

    let threw = false;
    try {
      // Normally this would delete all properties on the document
      // other than `_id` and `overwrite`. This is expected behavior,
      // but you might want to disallow it. Monogram gives you a
      // framework to do so.
      await Test.updateOne({ _id: 1 }, { overwrite: 'woops!' });
    } catch (error) {
      threw = true;
      assert.equal(error.message, 'Not allowed to overwrite document ' +
        'using `updateOne()`, use `replaceOne() instead`');
    }

    assert.ok(threw);