Requesting /favicon.ico or /robots.txt invokes function, although it shouldn't according to doc

[alsoba13]

Expected behaviour

According to Functions framework documentation, requests to /favicon.ico and /robots.txt should return a 404 status without calling the function:

Note that the framework must listen to all inbound paths (.*) and route these requests to the function, with the following exceptions:

• /robots.txt - must respond with 404 without invoking function
• /favicon.ico - must response with 404 without invoking function

Actual behaviour

When requesting a /favicon.ico or /robots.txt, a 404 status is returned, but the function is actually invoked, with all its possible side effects.

Reproduction

1. Take the Hello World example from this repository, but write directly to stdout:

package com.example;

import com.google.cloud.functions.HttpFunction;
import com.google.cloud.functions.HttpRequest;
import com.google.cloud.functions.HttpResponse;

public class HelloWorld implements HttpFunction {
  @Override
  public void service(HttpRequest request, HttpResponse response)
      throws Exception {
    System.out.println("Hello World! Path: " + request.getPath());
  }
}

2. Run function:

./gradlew runFunction -Prun.functionTarget=com.example.HelloWorld

3. Curl to /favicon.ico:

curl localhost:8080/favicon.ico

4. See the result in the console to check that the function is executed with /favicon.ico:

Hello World! Path: /favicon.ico

Possible fix

I believe that the bug is at the NotFoundHandler class, that executes the super after writing the 404 error. I am not familiar with Jetty's handlers API, but we may see that other jetty handlers (InetAccessHandler.java) do return right after generating the sendError.

PR for this: #226

Related

• Most of the other Functions frameworks have a special handler for this (except for golang), and even nodejs framework has deal with a conceptually similar issue: Requesting /favicon.ico URL invokes function, ignores its response, logs error to console functions-framework-nodejs#191
• For production environment, there exist some workaround but it needs to be very precisely executed, otherwise some odd errors (502 codes) will occur. I can give more detail if someone is interested.

[HKWinterhalter]

Thanks for finding and reporting on this! I have approved PR #226