带有数据库的Express的Node.js中的路由器实例

Question

As far as I can gather from the Express documentation , when you declare an express.Router() , it creates a single instance of a router that you can then assign a routing path and execute logic with. The documentation says to think of a router like a mini-app for a specific route or routes, which makes sense.

I'm trying to strategize what to wrap my database connection around (using mongodb, let's say via mongoose, but it shouldn't matter). I of course don't want to open a new database connection on every route call, and I assume that if I wrap it around a router only one Router() instance will only be created. In other words, if I went to /routes/index.js , defined a Router() and then opened a database connection, then within it did router.get (or router.post , etc.), I would be opening one database connection when launching the app, not one per hit on that route.

Yet there might be other routes beyond index.js where I want access to the database. So alternatively, could I wrap the database connection around the app.use route handlers and other middleware within the main app.js file, then require('../app') in /routes files and add the database connection to module.exports in app.js , and finally define route logic in other files like /routes/index.js ?

I'm a little confused on how to structure everything. Any guidance would be much appreciated.

Answer 1

If you are using mongoose, you can just connect once with some code like this:

mongoose.connect("mongodb://127.0.0.1:27017/test");
mongoose.connection.on('error', console.error.bind(console, 'Db connection error:'));
// Start the server once we have connected to the database.
mongoose.connection.once('open', () => {
    console.log('Db connection open.');
    app.listen(3000, function () {
        console.log('Listening on port 3000');
    });
});

And then, if you have a mongoose model named Foo set up like

const Foo = mongoose.model('Foo', fooSchema); // fooSchema is a  mongoose.Schema

Then in your route you can use it like so:

const router = express.Router();
const Foo = require('./models/foo');
router
    .route('/foos/:foo_id')
    .get((req, res)=> {
        Foo.findById(req.params.foo_id, (err, foo) => {
        if (err) return res.sendStatus(500);
        if (!foo) return res.status(404).send('Foo not found.');
        res.status(200).json(foo);
    });
});

This kind of setup lets mongoose handle connection pooling.