avoid double encoding path params

[therungg]

What?

This encodes and re-decodes the path param before throwing an error from decodeURIComponent.

Why?

More often than not, the error is thrown when a string given to decodeURIComponent has already been decoded. This means that a string like %25 will already be decoded to %, which results in an error from decodeURIComponent. This is also the case for this function. The parameter given is already encoded once, so this function will fail with %25. This PR fixes that by applying the standard solution to this problem: re-encoding, then re-decoding.

If decodeURIComponent still throws an error, the same DecodeError as before is thrown, because that means something different is the culprit.

How?

By re-encoding, then re-decoding the string.

Closes NEXT-
Fixes #86957

[nextjs-bot]

Allow CI Workflow Run

• approve CI run for commit: 812b914

Note: this should only be enabled once the PR is ready to go and can only be enabled by a maintainer

[harikapadia999]

Interesting approach to the double-encoding issue! 🤔

Problem: When a URL parameter is already decoded (like % from %25), calling decodeURIComponent() again throws an error.

Your Solution: Re-encode the param first, then decode it.

However, I have concerns:

1. This creates actual double-encoding: If the param was hello%20world (already encoded), your code would:

   • Encode: hello%2520world
   • Decode: hello%20world
   • Result: Still encoded, not the expected hello world

2. The test doesn't validate the fix: Your test uses '/' + alreadyDecodedUrlPart which is '/%', but the route matcher receives this as a string, not a URL-encoded value. The test should use '/%25' to simulate the actual scenario.

Better Solution:

try {
  return decodeURIComponent(param)
} catch {
  // Param is already decoded or invalid
  return param
}

This handles both cases:

• Already decoded params: returns as-is
• Encoded params: decodes properly
• Invalid params: returns as-is (graceful degradation)

Test Improvement:

const result = routeMatcher('/%25') // Encoded %
expect(result).toEqual({ user: '%' }) // Decoded result

The intent is good, but the implementation needs adjustment to avoid creating new encoding issues. 🚀

[therungg]

I think this is a fully AI-generated response, because pretty much all of the above is incorrect. It only encodes when the initial decoding fails, so there's never double encoding. The test is correct, because I want to test if the routeMatcher handles the already-encoded /% correctly.

[therungg]

Hi,

Is anyone able to look at this? It's a crucial bug for my app. Looking forward to getting this merged. Do let me know if there's anything I can change.

Thanks!

[filmon-arefayne]

@timneutkens any chance that this will get merged?
This issue is preventing users from using cacheComponents in Nextjs 16.