The Base45 Data Encoding

A QR-code is used to encode text as a graphical image. Depending on the characters used in the text various encoding options for a QR-code exist, e.g. Numeric, Alphanumeric and Byte mode. Even in Byte mode a typical QR-code reader tries to interpret a byte sequence as a UTF-8 or ISO/IEC 8859-1 encoded text. Thus, QR-codes cannot be used to encode arbitrary binary data directly. Such data has to be converted into an appropriate text before that text could be encoded as a QR-code. Compared to already established Base64, Base32 and Base16 encoding schemes, that are described in RFC 4648, the Base45 scheme described in this document offer a more compact QR-code encoding. One important difference from those others and Base45 is the key table and that the padding with '=' is not required.

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in BCP 14 when, and only when, they appear in all capitals, as shown here.

Encoded data is to be interpreted as described in RFC 4648 with the exception that a different alphabet is selected.

QR codes have a limited ability to store binary data. In practice binary data have to be encoded in characters according to one of the modes already defined in the standard for QR codes. The easiest mode to use in called Alphanumeric mode (see section 7.3.4 and Table 2 of ISO/IEC 18004:2015). Unfortunately Alphanumeric mode uses 45 different characters which implies neither Base32 nor Base64 are very effective encodings. A 45-character subset of US-ASCII is used; the 45 characters usable in a QR code in Alphanumeric mode (see section 7.3.4 and Table 2 of ISO/IEC 18004:2015). Base45 encodes 2 bytes in 3 characters, compared to Base64, which encodes 3 bytes in 4 characters. For encoding, two bytes [a, b] MUST be interpreted as a number n in base 256, i.e. as an unsigned integer over 16 bits so that the number n = (a*256) + b. This number n is converted to base 45 [c, d, e] so that n = c + (d*45) + (e*45*45). Note the order of c, d and e which are chosen so that the left-most [c] is the least significant. The values c, d and e are then looked up in Table 1 to produce a three character string. The process is reversed when decoding. For encoding a single byte [a], it MUST be interpreted as a base 256 number, i.e. as an unsigned integer over 8 bits. That integer MUST be converted to base 45 [c d] so that a = c + (45*d). The values c and d are then looked up in Table 1 to produce a two character string. A byte string [a b c d ... x y z] with arbitrary content and arbitrary length MUST be encoded as follows: From left to right pairs of bytes MUST be encoded as described above. If the number of bytes is even, then the encoded form is a string with a length which is evenly divisible by 3. If the number of bytes is odd, then the last (rightmost) byte MUST be encoded on two characters as described above. For decoding a Base45 encoded string the inverse operations are performed.

If binary data is to be stored in a QR-Code, the suggested mechanism is to use the Alphanumeric mode that uses 11 bits for 2 characters as defined in section 7.3.4 in ISO/IEC 18004:2015. The ECI mode indicator for this encoding is 0010. On the other hand if the data is to be sent via some other transport, a transport encoding suitable for that transport should be used instead of Base45. For example, it is not recommended to first encode data in Base45 and then encode the resulting string in Base64 if the data is to be sent via email. Instead, the Base45 encoding should be removed, and the data itself should be encoded in Base64.

The Alphanumeric mode is defined to use 45 characters as specified in this alphabet.

Table 1: The Base45 Alphabet Value Encoding Value Encoding Value Encoding Value Encoding 00 0 12 C 24 O 36 Space 01 1 13 D 25 P 37 $ 02 2 14 E 26 Q 38 % 03 3 15 F 27 R 39 * 04 4 16 G 28 S 40 + 05 5 17 H 29 T 41 - 06 6 18 I 30 U 42 . 07 7 19 J 31 V 43 / 08 8 20 K 32 W 44 : 09 9 21 L 33 X 10 A 22 M 34 Y 11 B 23 N 35 Z

It should be noted that although the examples are all text, Base45 is an encoding for binary data where each octet can have any value 0-255. Encoding example 1: The string "AB" is the byte sequence [65 66]. The 16 bit value is 65 * 256 + 66 = 16706. 16706 equals 11 + 45 * 11 + 45 * 45 * 8, so the sequence in base 45 is [11 11 8]. By looking up these values in the Table 1 we get the encoded string "BB8". Encoding example 2: The string "Hello!!" as ASCII is the byte sequence [72 101 108 108 111 33 33]. If we look at each 16 bit value, it is [18533 27756 28449 33]. Note the 33 for the last byte. When looking at the values in base 45, we get [[38 6 9] [36 31 13] [9 2 14] [33 0]] where the last byte is represented by two. The resulting string "%69 VD92EX0" is created by looking up these values in Table 1. It should be noted it includes a space. Encoding example 3: The string "base-45" as ASCII is the byte sequence [98 97 115 101 45 52 53]. If we look at each 16 bit value, it is [25185 29541 11572 53]. Note the 53 for the last byte. When looking at the values in base 45, we get [[30 19 12] [21 26 14] [7 32 5] [8 1]] where the last byte is represented by two. By looking up these values in the Table 1 we get the encoded string "UJCLQE7W581".

Decoding example 1: The string "QED8WEX0" represents, when looked up in Table 1, the values [26 14 13 8 32 14 33 0]. We arrange the numbers in chunks of three, except for the last one which can be two, and get [[26 14 13] [8 32 14] [33 0]]. In base 45 we get [26981 29798 33] where the bytes are [[105 101] [116 102] [33]]. If we look at the ASCII values we get the string "ietf!".

There are no considerations for IANA in this document.

When implementing encoding and decoding it is important to be very careful so that buffer overflow or similar does not occur. This of course includes the calculations in base 45 and lookup in the table of characters (Table 1). A decoder must also be robust regarding input, including proper handling of any octet value 0-255, including the NUL character (ASCII 0). It should be noted that Base64 and some other encodings pad the string so that the encoding starts with an aligned number of characters while Base45 specifically avoids padding. Because of this, special care has to be taken when odd number of octets are to be encoded. Similarly, care must be taken if the number of characters to decode are not evenly divisible by 3. Base encodings use a specific, reduced alphabet to encode binary data. Non-alphabet characters could exist within base-encoded data, caused by data corruption or by design. Non-alphabet characters may be exploited as a "covert channel", where non-protocol data can be sent for nefarious purposes. Non-alphabet characters might also be sent in order to exploit implementation errors leading to, e.g., buffer overflow attacks. Implementations MUST reject any input that is not a valid encoding. For example, it MUST reject the input (encoded data) if it contains characters outside the base alphabet (in Table 1) when interpreting base-encoded data. Even though a Base45 encoded string contains only characters from the alphabet in Table 1, cases like the following has to be considered: The string "FGW" represents 65535 (FFFF in base 16), which is a valid encoding of 16 bits. A slightly different encoded string of the same length, "GGW", would represent 65536 (10000 in base 16), which is represented by more than 16 bits. Implementations MUST also reject the encoded data if it contains a triplet of characters which, when decoded, results in an unsigned integer which is greater than 65535 (ffff in base 16). It should be noted that the resulting string after encoding to Base45 might include non-URL-safe characters so if the URL including the Base45 encoded data has to be URL safe, one has to use %-encoding.

The authors thank Mark Adler, Anders Ahl, Alan Barrett, Sam Spens Clason, Alfred Fiedler, Tomas Harreveld, Erik Hellman, Joakim Jardenberg, Michael Joost, Erik Kline, Christian Landgren, Anders Lowinger, Mans Nilsson, Jakob Schlyter, Peter Teufl and Gaby Whitehead for the feedback. Also, everyone that have been working with Base64 over a long period of years and have proven the implementations are stable.