umockdev script format
----------------------
An umockdev script file records the "dialog" between the program under test
("client") and the device or socket it is talking to.  Script files use a
line-based format, one line for an individual operation:

operation value data

operation
---------
single letter for the type of I/O operation. Possible operation codes are:

  r : Client program does a read() or recv() operation; umockdev will send
      data; <data> can be read in several smaller blocks if the first read
      operation does not request the full length of data. During recording, the
      time since the previous operation (in ms) gets recorded into the "value"
      number; during replay, umockdev waits for that time before sending the
      recorded data to mimic the original device or socket.

  w : Client program does a write() or send() operation; it's required that
      this matches the previously recorded data, so that umockdev and the
      client program stay in sync. During recording, multiple successive writes
      which are less than 10 ms apart are merged into one block.
      Correspondingly, the actually written data during replay can be split
      into arbitrary smaller chunks. The time since the previous operation (in
      ms) gets recorded into the "value" number for reference, but is not
      relevant during replay.

   f: Set/change fuzz percentage. This specifies the percentage of allowed
      character mismatches between a recorded 'w'rite block and the actually
      sent data. This defaults to '0' (i. e. exact matching). It is recommended
      to not set this to > 20 to avoid getting out of sync and missing some
      actual misbehaviour; preferrably, only use it for some specific messages
      which are known to have some variance, instead of just setting it at
      the top of a script. The data member is ignored, it's recommended to use
      "-".

   Q : Stop processing the script. This is not required, nor written by
       umockdev-record. You can use this for testing or appending notes at the
       end of the script. value and data are ignored.

   # : Comment; remainder of the line is entirely ignored

data
----
In order to represent the data block as a single line, byte values b <= 32 are
escaped as '^' followed by (b+64). The '^' character itself is represented as
"^`". So the byte value 0 is represented as "^@", 1 as "^A", 10 (line feed) as
"^J", 30 as "^^", 94 as "^`". All values >= 32 (except '^') are verbatim.