Wheeler

This is a map/reduce proof of concept with a little bit of fun thrown in.

Wheeler is a naive Wheel of Fortune solver. It does this by indexing sampled text from where ever into every possible contiguous combination of words up to a max phrase length.

The mapper in this project is wheeler phrases. Again, this is a naive program. The phrase parser don't use any NLP or grammar logic. It simply splits words by spaces, period, and double quotes. After splitting it outputs joined sets of those words in descending word count.

The reducer is wheeler reduce. It build the phrase index into the .index directory inside this project. Maybe we'll make it an arg in the future. The index structure is: 1st word size/2nd word size/3rd word size/etc.../phrases

So for the phrase: "I love cats". An index entry would be written into .index/1/4/4/phrases. The contents of phrases would be 'I LOVE CATS'

After the index are built, the guess matching is almost trivial.

1. receive a puzzle with underscores and known letters. For example: _ ____ ___s
2. get the word counts of the puzzle. 1, 4, 4.
3. find the phrases file based on the word counts. .index/1/4/4/phrases
4. perform a simple pattern match in that file. grep -e '. .... ....' .index/1/4/4/phrases

The wheeler guess script will do steps 2-4 automatically: wheeler guess '_ ____ ___s'

The more texts indexed, the better chance of solving any given puzzle. This solver is only as good as the index it compiles.

# Map text to phrases
$ wheeler phrases samples/adventure_of_the_speckled_band.txt 5 > tmp/phrases-unsorted.txt

# Sort the mapped phrases
$ sort tmp/phrases-unsorted.txt > tmp/phrases-sorted.txt
 
# Reduce to counts, this writes results to .index
$ wheeler reduce tmp/phrases-sorted.txt

# view phrase indexes
$ find .index -name phrases

# view contents of a phrase
$ less .index/9/8/4/9/3/phrases

$ wheeler guess '____ __ _____ES'
# grep --color=always -e '.... .. .....ES' .index/4/2/7/phrases
# BAND OF GYPSIES

# Map text to phrases
$ wheeler phrases samples/dict.txt 5 > tmp/phrases-unsorted.txt

# Sort the mapped phrases
$ sort tmp/phrases-unsorted.txt > tmp/phrases-sorted.txt
 
# Reduce to counts
$ wheeler reduce tmp/phrases-sorted.txt

# Make the Index
$ curl http://www.gutenberg.org/ebooks/29765 > tmp/dictionary.txt
$ wheeler phrases tmp/dictionary.txt 5 > tmp/phrases-unsorted.txt

# Sort the mapped phrases
$ sort tmp/phrases-unsorted.txt > tmp/phrases-sorted.txt
 
# Reduce to counts
$ wheeler reduce tmp/phrases-sorted.txt

Bug reports and pull requests are welcome on GitHub at https://github.com/ddrscott/wheeler.

The gem is available as open source under the terms of the MIT License.