Select Encode and RFC 3986 component, then enter café & tea exactly. ASCII letters stay literal. The UTF-8 bytes C3 A9 for é become %C3%A9, each space byte becomes %20, and the ampersand becomes %26 because it is data inside this one component rather than a query-pair separator.
Expected byte evidence
The visible rows show c as 63, a as 61, f as 66, é as C3 A9, space as 20, ampersand as 26, and the remaining letters. The final output is caf%C3%A9%20%26%20tea. Decode it once under RFC 3986 and the exact original text returns; the canonical source status is then matched.
Reproduce the round trip
Select the stated profile, run the plain-text input in Encode mode, and record the exact uppercase percent escapes. Switch to Decode without changing the profile and paste that output. The original Unicode text must return exactly, including spaces and literal plus signs. A lower-case escape may decode, but URLCodec marks it noncanonical and shows the unique uppercase spelling used by its round trip.