A few months ago, I wrote a crossword puzzle library for reading "Across&Down" format .puz files. It's fine, and the features I wanted to add to the library really don't matter too much to using it.
There's a feature I'd really like to add, though, and that's support for "rebus" puzzles. That's when you have something weird in a box, like a bunch of letters. You might have the following:
S?E T E ?RY
Where each ? is actually filled in with "QUAR." The Thursday puzzles in the
New York Times tend to do this. I have a sample file, but I don't think I'm
free to redistribute it openly. Some parts of the format are pretty clear.
There's a chunk that says "the following are the weird values: QUAR," and
there's a chunk that says "weird value 1 appears at these locations," but I'm
mostly not sure how to reliably find the beginning of these sections. I think
I need to build up a small corpus of sample files, then work from there. If I
can't find any, I can always buy the builder and use it to build sample files.
This is largely for the fun of it. Gloria bought me a copy of Black Ink, so I'm in less of a hurry to write my once-considered puz-to-dhtml compiler.
