shithub: plumb

Clone

Last commit

"Fix

About

## Overview
Plumb is a drop-in replacement for plan9's [plumb](https://9fans.github.io/plan9port/man/man1/plumb.html) utility.
Notably different from it, are how it handles the `type` attribute of the plumb messages.
They will be set to a proper mimetype, instead of the simple `type is text` that the traditional plumber utilized. 
This is considerably more powerful, as you no longer need to attempt to infer the content based on URIs, extensions, or directory structure.

## Rules
Your plumber rules will have to be updated to reflect this more granular message:

	## In this example, plan9front's plumber is being used; but similar rule changes would apply for plan9 regular.
	
	type is image/png
	# No longer need to match paths
	#data matches '[a-zA-Z¡-￿0-9_\-./@]+'
	#data matches '([a-zA-Z¡-￿0-9_\-./@]+)\.(jpe?g|JPE?G|gif|GIF|tiff?|TIFF?|ppm|bit|png|PNG)'
	arg isfile	$0
	plumb to image
	plumb start 9 page $file
	
	# local html files (Your dev work, for example) can be opened in your editor
	type is text/html
	arg isfile $0
	plumb to edit
	plumb start $editor $0
	
	# remote html files will likely be opened in your browser
	type is text/html
	# With the above case matching local files, we no longer need this convolution
	#data matches '(https?|ftp|file|gopher|mailto|news|nntp|telnet|wais|prospero)://[a-zA-Z0-9_@\-]+([.:][a-zA-Z0-9_@\-]+)*/?[a-zA-Z0-9_?,%#~&/\-+=]+([:.][a-zA-Z0-9_?,%#~&/\-+=]+)*'
	plumb to web
	plumb start web $0
	
	type is application/pdf
	# No longer need to match paths
	#data matches '[a-zA-Z¡-￿0-9_\-./@]+'
	#data matches '([a-zA-Z¡-￿0-9_\-./@]+)\.(ps|PS|eps|EPS|pdf|PDF|dvi|DVI)'
	arg isfile	$0
	plumb to postscript
	plumb start 9 page $file

In practice, setting a rule for most common mimetypes will serve you well. (For example, my last implementation of plumber used a relatively small amount of mimes, after a few years' worth of plumbing in this manner.
See https://github.com/halfwit/Plumber/tree/master/cfg/plumber)

## About application/octet-stream mimetype
If the remote mimetype is `application/octet-stream`, which is a fallback when it cannot infer the mimetype this client will attempt to find a content-type field in any remote URL, finally setting the type to `text`.

- reference for mimetype: https://mimesniff.spec.whatwg.org/
- reasonably robust list of mimetypes: https://www.freeformatter.com/mime-types-list.html