The puzzle file format consists of a grid size (as a single integer -
the grid is assumed to be square) then a square list of piece
characters, with . indicating an empty square.  For example, puzzle 1
(which can be found as puzzles/01) is represented as

6
jj...m
p..n.m
paan.m
p..n..
e...ll
e.ooo.

which corresponds to what I might represent in ASCII (well, 8859-1)
graphics thusly (* marks the goal piece which must exit via the gap):

+---------------------------+
| +-------+           +---+ |
| |       | ·   ·   · |   | |
| +---+---+   +---+   |   | |
| |   | ·   · |   | · |   | |
| |   +-------+   |   |   |
| |   | *   * |   | · |   |
| |   +-------+   |   +---+
| |   | ·   · |   | ·   ·   |
| +---+       +---+-------+ |
| |   | ·   ·   · |       | |
| |   |   +-------+---+---+ |
| |   | · |           | ·   |
| +---+   +-----------+     |
+---------------------------+

All pieces must be either 1x2 or 1x3 rectangles.  The assignment of
letters to peices is arbitrary, except that `a' must be the character
for the goal piece, which must be a two- or three-cell horizontal
piece.  The exit gate is considered to be at the right edge of the
diagram in line with this piece.  (Different pieces must, of course,
use different letters.)