Tools for solving wordplay puzzles. Heavily inspired by tools.qhex.org, Nutrimatic, and Qat.

Noodle has the following key features that differentiate it from basic regex/anagram tools:

• Evaluate complex anagram constraints
  • "(anagram of pasta, except one letter) followed by led": stapled.
• Generate phrases from the input wordlist which match the query
  • a..mong.u.s.ntencewithmul.iplewords: a humongous sentence with multiple words
• "Fuzzy" search, to find phrases which almost match the constraints
  • "phrases within edit distance 2 of breadfast": breakfast(s), broadcast, bead east...
• Sorted results where matches with common/long words are prioritized
• Responsive, showing matches as they are found, even on long queries
• A simple GET API, allowing it to be easily integrated into Google Sheets or bots

$ cargo install --path noodle-cli

This installs a noodle binary to your path.

> noodle --help
noodle 0.1.0

USAGE:
    noodle [OPTIONS] <query>

FLAGS:
    -h, --help       Prints help information
    -V, --version    Prints version information

OPTIONS:
    -n, --count <count>                    Number of results to return
    -i, --input <input>                    Input wordlist file [default: /usr/share/dict/words]
    -m, --phrase-length <phrase-length>    Maximum number of words to combine to make a matching phrase [default: 10]

ARGS:
    <query>    Noodle query string

$ cargo run --release --bin noodle-webapp

This launches the Noodle server bound to http:https://localhost:8082

$ flyctl launch

This will create & launch a fly.io application (named noodle).

(Although fly.io is a bit overkill as a hosting provider, they have a generous & easy-to-use free tier.)

See Noodle Help.

• Document theory of operation (See architecture.md)
  • NX basics, tradeoffs (NFA, small alphabet)
  • Fuzzy matches
  • Multi-word matches
  • Multi-NX matches
  • Sugar (anagrams)
• Select wordlist
• Mix & match wordlists ("one word from this list of 100 words, plus any other 3 words")
• Maybe use fst crate to represent wordlists?
  • Representation is memory-optimized, unlike the current struct Word
  • Can re-use prefix structure in matcher.rs?
  • Likely overkill (but it's a relatively mature crate for this sort of data structure?)
• Pre/post filters (regex)
• "Extract"/re-write rules for matching "inner" words, etc. ("cross-filtering" on qhex)
• "Inverse" NX expressions? ("does not match") -- (this is hard with NFAs)
• Fuzzy matching + anagrams are weak; add post filter
• Python library
• Other ways of sorting the output in the UI (e.g. by length, alphabetical, etc.)
• Add support for (...:-n)/(...:+n) syntax
• More powerful macro/preprocessing language?
  • "Length of macro" would help with certain repetitive lookups
• Heuristically re-sort constraints from most-to-least constraining (for speed)

Released under the MIT License.

Copyright (c) 2020-2022 Zach Banks