Textec encoding can be viewed as a three-step transformation:

   <a> compression (from [sourceText] to [compressedText]),
   <b> encryption (from [compressedText] to [cipherext]) and finally
   <c> base64 encoding (from [ciphertext] to [encoded64Text]).

   Decoding reverses the order and the "direction" of the above steps.

   <a> Compression
   ---------------
   The compression level is moderate, ensuring only that (if the
   plaintext consists of natural language words) the volume will be
   reduced more than the later increase caused by the base64 encoding.
   It can also make the known-plaintext attacks considerably more
   difficult. If the volume of text is low or if it has little or
   no redundancy, the compression step may be omitted (see below,
   ~i~ trailer).

   <b> Encryption
   --------------
   Ciphertext is encrypted in 8-byte blocks using Blowfish CBC mode.
   If the last block of text (before the encryption) is "short",
   (i.e., it consists of ~n~ bytes, where 1 <= n <= 7), it will be
   padded (up to 8) with random bytes. If that was the case, a dummy
   "trailer" of ~n~ random bytes is also added to the ciphertext. This
   enables the decryption process to discard the padding: if the size
   of the ciphertext presented to it is not an even multiple of 8
   (i.e., there are n bytes over an even multiple of 8), only the
   first n decrypted bytes of the last full block are included in
   the plaintext.

   <c> Base64 Encoding
   -------------------
   The final base64 encoding transforms 3 bytes of binary data into 4
   characters of ascii text, using only numerals, upper and lower case
   letters and braces. The text is broken into 64-character wide lines.

   Each of the three derived forms of text has some additional
   information added to it (NB.: in the diagram below, relative sizes
   are preserved, but the scale is not):

             [----------------------sourceText------------------------]
   <a>
             [----------compressedText----------]i
   <b>
      ---IV--[------------cypherText------------]-pad8-
   <c>
     -crc-cnt[--------------encoded64Text--------------]--pad64--

   The headers and trailers indicated in the above diagram are:

   i:        single random byte, interpreted as an integer, even if the
             source text was successfully compressed, odd if it was
             "uncompressable" and therefore simply copied from
             [sourceText] to [compressedText].

   IV:       an 8-byte CBC Initialization Vector. Random but not secret.

   pad8:     optional, padding, as described above.

   crc:      4-byte crc32 check value of (IV + cipherText + pad8) before
             base64 encoding.

   cnt:      4-byte integer byte count of (IV + cipherText + pad8) before
             base64 encoding.

   pad64:    up to 63 random bytes, padding (crc + cnt + encoded64Text)
             to a full 64-character (last) line.