Batch Mojang API requests using dataloader pattern

[Copilot]

Mojang's API provides bulk endpoints that accept up to 10 requests. Individual calls were inefficient and hitting rate limits unnecessarily.

Implementation

Dataloader pattern with goroutines:

• Two background goroutines collect requests from buffered channels
• Batch triggers on 10 requests or 50ms timeout, whichever comes first
• POST /profiles/minecraft bulk endpoint for usernames (up to 10)
• UUID requests batched but use individual calls (no bulk endpoint exists)

API modifications:

// Before: direct HTTP call per request
func (m *Mojang) GetAccountByUsername(ctx context.Context, username string) (domain.Account, error) {
    return m.getProfile(ctx, fmt.Sprintf("https://api.mojang.com/users/profiles/minecraft/%s", username))
}

// After: enqueue to batch processor
func (m *Mojang) GetAccountByUsername(ctx context.Context, username string) (domain.Account, error) {
    responseChan := make(chan usernameResponse, 1)
    req := usernameRequest{username: username, response: responseChan}
    
    select {
    case m.usernameRequestChan <- req:
    case <-ctx.Done():
        return domain.Account{}, ctx.Err()
    }
    
    select {
    case resp := <-responseChan:
        return resp.account, resp.err
    case <-ctx.Done():
        return domain.Account{}, ctx.Err()
    }
}

Fallback handling:

• Bulk request failures automatically retry as individual calls
• Case-insensitive username matching (Minecraft usernames are case-insensitive)

Lifecycle management:

• Shutdown() method for graceful cleanup
• sync.WaitGroup ensures goroutines complete before shutdown

Original prompt

Chunk requests to the mojang account provider by collecting them and making a request towards their bulk endpoints (one for username and one for uuid). Use a dataloader approach where the get single methods put a request onto a queue (one for uuid and one for name). These channels are listened to by two goroutines that should fire a bulk request once they hit 10 requests or a timeout is reached. The single calls should then be notified of the response (or error) and handle it accordingly

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.