prism

The goal of prism came is to take output from a process and split it up into buckets based on a regular expression, and then letting you view the contents of each bucket individually using a nice TUI.

prism was created primarily to manage the output from turborepo, but it will split output based on a custom regex -- defaulting to one that supports turborepo output -- so it can be used for anything; it can even display its own log output as it is running.

This may all sound a bit vague, so it's probably better explained with some demos:

Running it with a dummy turborepo project

Running it on its own log output:

Features

• Custom regular expression
• Color support

Usage

NOTE: This is alpha level software. Use at your own risk.

$ prism -p <prefix_regex> <command>

where prefix_regex is a regex with at least two capture groups. The first capture group will be the prefix, and the second the message.

In the TUI, use j/k to navigate prefixes, and tab to cycle between messages, stderr and unparsable messages.

Examples:

Run yarn dev with the default regex

$ prism yarn dev

Run cat file with a regex that parses lines like DEBUG This is a message

$ prism -p '^([A-Z]*?) (.*)' cat file

Run a command with command line flags:

$ prism "tail -f file"

Use it to split its own log output(NB: if you want to try this, the log files grows very quickly due to the "recursive" nature of doing this)

$ RUST_LOG=debug prism -p '\[.* ([A-Z]+ .*?)\] (.*)' "tail -f log" 2>log

Known issues

• When used with turborepo, child processes are not terminated reliably
• The terminal is not always left in a great state when exiting and requires a reset
• The default regular expression is probably not very good

TODO

• Scrolling messages
• tests
• Configurable scrollback limit
• Show an indicator for when the process has exited