The general algorithm is:

Encrypt:

• Transform InputText to InputBytes via encoding TE (text encoding).
• Encrypt InputBytes to OutputBytes
• Transform OutputBytes to OutputText via BE (binary encoding).

Decrypt (reverse BE and TE from Encrypt):

• Transform InputText to InputBytes via encoding BE.
• Decrypt InputBytes to OutputBytes
• Transform OutputBytes to OutputText via TE.

The most common mistake is to choose a "text encoding" instead of a "binary encoding" for BE, which is a problem if any encrypted byte (or any IV byte) is outside the range 0x20-0x7E (for UTF-8 or ASCII). Since the "safe range" is less than half of the byte space the chances of a text encoding being successful are vanishingly small.

• If post-encryption string contains a 0x00 then C/C++ programs will likely misinterpret that as the end of the string.
• If a console-based program sees 0x08 it may erase the previous character (and the control code), making the InputText value to Decrypt have the wrong value (and the wrong length).