Allow access to unsanitized file name

[j0hax]

Hello there! I've come across a situation where I have to save a file with a "double" extension (*.kepub.epub), and the current implementation of (r *Response) FileName() purposefully breaks that by always calling SanitizeFileName().

I propose to provide direct access to an unsanitized file name with the method RawFileName(). This shouldn't affect compatibility and allows for accessing information similar to URL package's RawPath, RawQuery, etc.

[WGH-]

Ehm, perhaps, sanitizing double extension away was a mistake in the first place.

[StantonMatt]

I checked this locally. The change is backwards-compatible for FileName() because it now just sanitizes RawFileName(), and the full Go suite passes on the PR branch.

What I ran:

GOCACHE=.codex-tmp/go-cache GOTMPDIR=.codex-tmp/go-tmp GOMODCACHE=.codex-tmp/go-mod go test ./... -count=1
git diff --check upstream/master...HEAD

Both passed. GitHub reports no checks for this branch.

I also used a local-only test with books/book.kepub.epub and a Content-Disposition filename like ../../unsafe.kepub.epub. That confirmed the tradeoff: RawFileName() preserves the exact header/path-derived name, including path separators and unsafe segments, while FileName() still returns a safe filename and still folds the double extension into names like books_book_kepub.epub.

So this does solve the caller need without changing existing FileName() behavior. The maintainer question above still seems important, though: if the double-extension folding is itself the bug, a smaller change may be to adjust SanitizeFileName()/FileName() so safe double extensions survive. If both APIs are desired, I’d add regression tests for RawFileName() and the existing sanitized behavior so the safety distinction is explicit.