Incorrect autofix for prefer-math-trunc rule

eslint-plugin-unicorn@23.0.0

Before:

// Very basic hash implementation that mimics Java's hash function, used for generating unique ids for aggregate items
// Found here: https://stackoverflow.com/questions/7616461/generate-a-hash-from-string-in-javascript-jquery
export function generateHash(input) {
  let hash = 0;
  for (let i = 0; i < input.length; i++) {
    const char = input.charCodeAt(i);
    hash = (hash << 5) - hash + char;
    hash |= 0; // Convert to 32bit integer
  }
  return hash;
}

Example result (correct):

generateHash('abc123;')
-1207861333

After autofix by unicorn/prefer-math-trunc:

// Very basic hash implementation that mimics Java's hash function, used for generating unique ids for aggregate items
// Found here: https://stackoverflow.com/questions/7616461/generate-a-hash-from-string-in-javascript-jquery
export function generateHash(input) {
  let hash = 0;
  for (let i = 0; i < input.length; i++) {
    const char = input.charCodeAt(i);
    hash = (hash << 5) - hash + char;
    hash = Math.trunc(hash); // Convert to 32bit integer
  }
  return hash;
}

Example result (after autofix, incorrect):

generateHash('abc123;')
3087105963

Looks related to #862.

[sindresorhus]

Fixed in https://github.com/sindresorhus/eslint-plugin-unicorn/releases/tag/v26.0.1

[noftaly]

I think it’s only for | 0. Are there other cases where the expression is transformed into Math.trunc()? In general, the idea would be that no transformation should return a result different than what it replaces.

Yes, it also does it for ~~x, x >> 0, x << 0 and x ^ 0.

So I’ve done the fix locally (only for | 0), but I’m not sure how to update the test. I’ll open the PR anyway and we’ll figure this out there!