URL fragment percent-encoding

[kaittodesk]

We are using furl to normalize templated URLs:

normalized_url = furl(rendered_url).tostr("&", False)

But we encounter a bit of an odd behavior when normalizing URLs which have a very specific fragment, and aren't sure if this is an expected or not.

For example normalizing following URL yields an unexpected result (fragment's forward slash gets percent-encoded):

normalized_url = furl("https://example.com/path/#/foo=123").tostr("&", False)
# Actual: https://example.com/path/#%2Ffoo=123
# Expected: https://example.com/path/#/foo=123

However when the URL fragment contains a separator (?) or does not contain an equals sign (=) it works as expected:

normalized_url = furl("https://example.com/path/#/?foo=123").tostr("&", False)
# Actual: https://example.com/path/#/?foo=123
# Expected: https://example.com/path/#/?foo=123

normalized_url = furl("https://example.com/path/#/foo").tostr("&", False)
# Actual: https://example.com/path/#/foo
# Expected: https://example.com/path/#/foo

We are using the latest version of furl==2.1.2.

Looking forward to your feedback.