tc: Trie Count Utility

A small command-line utility that takes a list of strings as input and produces a trie (prefix tree) as output. The trie associates a count with each prefix in the input lines (i.e., shows how common a particular prefix is). It's useful for quick data exploration. One can iteratively explore the most common prefixes, in particular when viewing the output in an editor with indentation-based folding (e.g., VS Code).

Usage Examples

Quickly determine that there are 4 strings in the input file that start with a, 3 strings that start with ab, etc.:

$ cat tests/inputs/simple.txt
abc
abd
abe
a

$ tc --quote tests/inputs/simple.txt
4 'a'
  3 'b'
    1 'c'
    1 'd'
    1 'e'
  1 ''

A trie can be useful for exploring how many files are in which part of a directory tree:

$ find -type f
./Cargo.toml
./Cargo.lock
./src/main.rs
./src/options.rs

$ find -type f | tc --split-delimiter='/' --sort=count
4 ./
  2 src/
    1 main.rs
    1 options.rs
  1 Cargo.toml
  1 Cargo.lock

In combination with other UNIX tools one can, e.g., determine which parts of a directory tree contains the most words:

$ find -type f | xargs wc -w | head -n-1
42 src/options.rs
1337 src/main.rs

$ find -type f | xargs wc -w | head -n-1 | tc --counted-input -d'/' -sc
1379 src/
  1337 main.rs
  42 options.rs

For more options (e.g., filtering out subtries with a count below a threshold) see tc --help.

Installation / Building from Source

Get the source code:

git clone https://github.com/danleh/trie-count.git
cd trie-count

Get the Rust installer from https://rustup.rs/ and install a nightly toolchain. (Currently nightly is required because of the unstable pattern feature of the regex crate. Feel free to contribute a stable replacement.)

rustup toolchain install nightly

Build with optimizations and install locally:

cargo +nightly install --path .

Should now work:

tc --help

License

MIT. See LICENSE.