Pumbeline - Random Name Generator
This is a very simple Ruby library / script for generating random names with
Markov Chains.
> bin/pumbeline --help
Usage: pumbeline [options] [file or stdin]
-n, --number N Number of words to generate
-N, --newline Use newline as word separator
-f, --filter regex Apply filter to output
-w, --window N Window size (default 3)
-d, --allow_words_from_input Allow words from input
-e, --encoding ENCODING Specify the encoding of the input files
Examples:
> bin/pumbeline words/color.txt -n 20 -w 1
ade alay be berrider bigerin brsichalu bulystemirulapon ffumuruvinawd
fonsnuseria hiturnt jan lis mivo pinglfan prarcratomeo sarint tadeawherudaro
te trimogondowinqumay x
> bin/pumbeline words/color.txt -n 20 -w 2
alacerasierinder bra byzandy copea cory go golf ivendew khalt lap mart
monquine na namokeyden ock pal peat pis rock sin
> bin/pumbeline words/color.txt -n 20 -w 3
amaroonstoneydew denic hunterry iceberry jazzber lavescent napis rass viris
wengeria
> bin/pumbeline words/color.txt -n 20 -w 4
turquoisee
> bin/pumbeline words/color.txt -n 20 -w 4
burn zaffron
> bin/pumbeline words/color.txt -n 20 -w 4
orch torchid ultramaranth
> bin/pumbeline words/color.txt -n 20 -w 4
almon auburnt
> bin/pumbeline words/color.txt -n 20 -w 5
> bin/pumbeline words/color.txt -n 20 -w 5
In the above example, we're generating made-up words from a short list of color
words. We're asking for 20 words, but (by default) words from the input file
are discarded (as are any duplicates).
The window size option (-w) dictates the number of prior characters to look at
when building the probability model from which new words will be generated. A
shorter window will lead to words which are more akin to random gibberish. As
the window size increases, the generated words more closely resemble the input
words, but less and less new results will be produced -- until no results are
produced at all. In my experience, the sweet spot is often 3 (the default).
> bin/pumbeline words/color.txt -n 10000 -f "^p.*a$"
palaya pansa pera peria peridia persia pinereolia pisteria pisterra
In the above example, a large number of words are generated before applying a
filter. In this case, we're looking for new words that begin with 'p' and end
with the letter 'a'.
The name of this project was found by searching for interesting words in this
manner.