Braille encode

The letters are, the rest is not. The 26 letter cells are the standard English Braille Grade 1 dot patterns, written as Unicode Braille characters from the U+2800 block. Digits and punctuation are not the authentic notation: real Braille prefixes digits with a number sign and reuses the a to j cells, and punctuation is context-sensitive. Here every character gets its own dedicated cell so the conversion round-trips.

The 26 letters a to z, the ten digits, the space, and six punctuation marks: full stop, comma, question mark, exclamation mark, apostrophe and hyphen. Anything else, including other punctuation and any non-Latin script, is passed through unchanged rather than dropped, so you can see in the output exactly what was not converted.

They are lower-cased before conversion, and the capitalisation is lost. There is no capital sign in this mapping, so an upper-case and a lower-case letter produce the same cell and decoding gives you lower case back. If case matters for your text, keep the original alongside the Braille.

Not directly, and not reliably. The output is Unicode Braille pattern characters, meant for display and for round-tripping through this tool. It is not Grade 2 contracted Braille and it does not follow a transcription standard, so an embosser fed this text would produce something a Braille reader would find wrong in the digits and punctuation. Use a proper transcriber for anything a person will read.

Run the output through Braille decode, which uses the exact same table in reverse. Every mapped character round-trips, apart from capitalisation, which was already discarded. Characters that passed through unconverted on the way in pass through again on the way back.