#26 Server-generated data

February 16, 2023
See the code for this post on the server-generated-data branch.

Let's generate location updates on the backend and make them available to the client via an API.

For start, let's create a file dbconfig.json that will store our database configuration.

"user": "postgres",
"password": "mysecretpassword",
"host": "db",
"port": 5432,
"dbname": "postgres"

We will import dbconfig.json into the two components accessing our database: the web server and the simulation. To make sure the file is available in both containers, we need to update their respective Dockerfiles in the server and the simulation folders:

COPY dbconfig.json .

Let's now update db.go to load the configuration from dbconfig.json rather than hardcoding it. Here we open the file, decode it into a config variable using the Config struct and then create the connection string using the config values.

type Config struct {
User string `json:"user"`
Password string `json:"password"`
Host string `json:"host"`
Port int `json:"port"`
DBname string `json:"dbname"`
// initDB creates a new instance of DB
func InitDB() {
configFile, err := os.Open("../dbconfig.json")
if err != nil {
log.Fatal("Error opening dbconfig.json:", err)
defer configFile.Close()
var config Config
jsonParser := json.NewDecoder(configFile)
if err = jsonParser.Decode(&config); err != nil {
log.Fatal("Error decoding dbconfig.json", err)
connStr := fmt.Sprintf(
"user=%s password=%s host=%s port=%d dbname=%s sslmode=disable",
config.User, config.Password, config.Host, config.Port, config.DBname,
# ...

Let's now edit simulation/main.js. Here, we will generate simple stream of data that will allow us to check the correctness of our animation code in the frontend. Similarly as in server/db.go, let's initiate the database connector using dbconfig.json. At the start, we need to wait a few seconds to give the database container some time to initiate.

const fs = require('fs');
const { Client } = require('pg');
const dbConfig = JSON.parse(fs.readFileSync('../dbconfig.json', 'utf8'));
const main = async () => {
await wait (5000);
const { host, port, user, password, dbname } = dbConfig;
const client = new Client({ host, port, user, password, database: dbname });
client.connect((err) => {
if (err) console.error('connection error', err.stack);
else console.log('connected');
// ...

Now, how can we actually generate a stream of data? For now, we can use a series of coordinates that will form a cycle on the map. For that, we need two paths. A car will start along the first path. Once it completes it, it will switch to the second path. Afterwards, it will start traversing the first path again. This will go on indefinitely.

const paths = {
first: [
// ...
second: [
// ...

Now, we can set up our loop for pushing location updates. Every 200 ms, we will update the row in the rides table with the new location. We also update the path, in case the path changes. Notice that the query is an upsert operation. If the row doesn't exist, it will create it, otherwise it will update the existing one. We do this so that we don't have to initialize this row manually.

const main = async () => {
// ...
let path = 'first';
let i = 0;
while (true) {
const [x, y] = paths[path][i];
const res = await client.query(
INSERT INTO rides (car_id, location, path)
VALUES ('car1', '${x}:${y}', '${JSON.stringify(paths[path])}')
ON CONFLICT (car_id)
DO UPDATE SET location = EXCLUDED.location, path = EXCLUDED.path;
if (res.rowCount) console.log(`${x}:${y}`);
if (i === paths[path].length - 1) {
path = path === 'first' ? 'second' : 'first';
i = 0;
await wait(3000);
} else {
await wait(200);

You may be asking - can't we just merge these two paths into a single path, loop over it, and once we reach the end, start the loop from the beginning again? Recall that we are passing the path to the move() method in the frontend. This method looks for the current position on the path and assumes that no coordinate appears more than once on the path. This is a reasonable assumption - if you are traversing a path, you don't want to visit any point twice. So for creating our cycle, we need to set up our iteration as shown above.

Before we run the simulation container, we need to set up the rides table in our database. On both the local machine and the production server, let's open a terminal session in the database container with sudo docker exec -it app-db-1 bash and initiate the table:

location VARCHAR(255) NOT NULL,
path TEXT

Finally, in server/main.go, let's add a list API for retrieving all records from the rides table and add the /rides endpoint.

type Ride struct {
Id string `json:"id"`
CarId string `json:"car_id"`
Location string `json:"location"`
Path string `json:"path"`
func getRides(w http.ResponseWriter, req *http.Request) {
rows, err := db.Connection.Query("SELECT * FROM rides")
if err != nil {
http.Error(w, "Failed to get rides: "+err.Error(), http.StatusInternalServerError)
defer rows.Close()
var rides []Ride
for rows.Next() {
var ride Ride
rows.Scan(&ride.Id, &ride.CarId, &ride.Location, &ride.Path)
rides = append(rides, ride)
ridesBytes, _ := json.MarshalIndent(rides, "", "\t")
w.Header().Set("Content-Type", "application/json")
func main() {
# ...
http.HandleFunc("/rides", getRides)
# ...

Previously we used the /drivers endpoint to test the database connection. We can remove it now, along with its associated handler.

Now we are ready to test our changes. Start the stack, head to Docker Desktop, and peek into the simulation container. Since we are logging every coordinate after a database update, you should see the stream of the generated data. Then head to the browser, and visit the /rides endpoint. Refresh it a few times and check that you are polling the updated location.

See the code for this post on the server-generated-data branch.