ritual

Database and API for enhancing history and clipboard processing

Usage no npm install needed!

<script type="module">
  import ritual from 'https://cdn.skypack.dev/ritual';
</script>

README

ritual

npm package version

Database and API for enhancing command line history, common paths and files, and clipboard handling.

Installation

The server is configured via environment variables.

  • RITUAL_PORT: set this to the TCP port you want to listen on.
    • Defaults to 7483 ("RITE" on a telephone's keypad).
  • RITUAL_VERBOSE: set this to '1' (or anything except the empty string) to increase the logging threshold to 'debug'.
    • Defaults to 'info'.
  • RITUAL_HOST: set this to the IPv4 address to listen on.
    • Defaults to '127.0.0.1'.

Load bash integration by putting the following in your ~/.bashrc or somewhere:

source ~/github/ritual/install/ritual.bash

Run at load:

cp install/github.chbrown.ritual.plist ~/Library/LaunchAgents/github.chbrown.ritual.plist
launchctl load ~/Library/LaunchAgents/github.chbrown.ritual.plist

When running under launchctl, there are two useful commands:

To kill the ritual server process until the next system boot:

launchctl remove github.chbrown.ritual

To restart it:

launchctl stop github.chbrown.ritual

Calling stop kills the current process, but launchd will immediately restart it since the launchd config file, github.chbrown.ritual.plist, has KeepAlive set to true.

IPC

I've chosen TCP for inter-process communication. Linux supports IPC message queues at some level, but there doesn't appear to be a trivial way to add messages to the queue via the command line.

Meanwhile, bash makes TCP requests easy with its echo 'hello' > /dev/tcp/<host>/<port> syntax.

The Ritual TCP server is obligated to return one (new)line for every line that it receives. If it does not, the client may hang, waiting for a response.

Ideas / TODO

  • Maybe load top 10 paths into $CDPATH?
  • Smarter path selection (right now it only selects the most recent exact match)
    • Exclude current path from candidates
    • Weight directories that have been recently used higher
    • Weight directories that have been frequently used higher
  • Add weight for directories successfully fetched with j or, more generally, get_directory.
  • Add clipboard watching
  • Add shell history watching
  • Prune directories that don't exist, or maybe just skip them in case they are mounted directories that may exist again.
  • There are a lot of good ideas on the iTerm2 shell integration page which seem to me like overkill for a terminal GUI, and much more suited to a third party service (like ritual).
  • Maybe use a coproc in the shell script instead of kicking off a new subshell whenever the bash user enters a ritual command?

Examples

Want to grab the matching directories for several queries at once?

echo 

q\ntmp\nmovies\nnode' |\
  sv --json |\
  jq -c '{action: "get_directory", q: .q}' |\
  ritual_tcp

Replace a substring with another substring, e.g., after moving a common directory:

ritual_replace "/github/business-" "/business/"

Autojump import

cat <(echo 

score\tpath') <(sort -g ~/.local/share/autojump/autojump.txt) |\
  sv --json |\
  jq -c '{action: "add_scored_directory", score: .score | tonumber, path: .path}' |\
  ritual_tcp

License

Copyright 2015-2017 Christopher Brown. MIT Licensed.