PREEN - Preen current path by file type

preen

#! /bin/sh

# PREEN - Preen current/working directory by file type
# Copyright (C) 2013,2018,2020 Matous J. Fialka, <https://mjf.cz/>
# Released under the terms of The MIT License

PREENDIR="${PREENDIR:-.preen}"
PREENRC="${PREENRC:-.preenrc}"

for util in which find file test mkdir ln
do
	if ! which "$util" 1>/dev/null 2>/dev/null
	then
		echo "Fatal error: which(1) at \`$util'" 1>&2
		exit 1
	fi
done

if test -e "$PWD/$PREENDIR"
then
	echo "Fatal error: path \`$PWD/$PREENDIR' must not exist" 1>&2
	exit 1
fi

if test -e "$PWD/$PREENRC"
then
	if ! source "$PWD/$PREENRC"
	then
		echo "Fatal error: source \`$PWD/$PREENRC'" 1>&2
		exit 1
	fi
fi

skipped=0
processed=0
failed=0
suspicious=0
maybe_dangerous=0
maybe_url=0
assorted=0
excluded=0
included=0

SIGNALS=(HUP USR1 USR2)

sighandler()
{
	cat <<- EOT
	STATISTICS
	----------
	Processed  : $processed
	Excluded   : $excluded
	Included   : $included
	Skipped    : $skipped
	Failed     : $failed
	Suspicious : $suspicious (dangerous: $maybe_dangerous, URL: $maybe_url)
	Assorted   : $assorted
	TOTAL      : $((skipped + failed + processed + suspicious + assorted))

	EOT

	sighandled=0
	for sig in ${SIGNALS[@]}; do
		if [ "$sig" = "$1" ]; then
			sighandled=1
		fi
	done

	if [ $sighandled -eq 0 ]; then
		for sig in ${SIGNALS[@]}; do
			trap - $sig
		done
		kill -$1 $$
		exit $2
	fi
}

siginit() {
	for sig in ${SIGNALS[@]}; do
		trap 'sighandler '$sig' $?' $sig
	done
}

siginit

while read -r name
do
	if ! type=`file -b --mime-type -e cdf "$name"` 2>/dev/null
	then
		echo "Fatal error: file(1) at \`$name'" 1>&2
		failed=$((failed + 1))
		continue
	fi

	if test "${type///}" = "${type}"
	then
		type="assorted/$type"
		assorted=$((assorted + 1))
	fi

	case "$name" in
	*[\*]*)
		type="suspicious/maybe-dangerous/$type"
		suspicious=$((suspicious + 1))
		maybe_dangerous=$((maybe_dangerous + 1))
		;;
	*[%\?\&]*)
		type="suspicious/maybe-url/$type"
		suspicious=$((suspicious + 1))
	        maybe_url=$((maybe_url + 1))
		;;
	esac

	if grep -q -E -i "$PREEN_EXCLUDE_REGEX" <<< "$type"
	then
		if grep -q -E -i -v "$PREEN_INCLUDE_REGEX" <<< "$type"
		then
			excluded=$((excluded + 1))
			skipped=$((skipped + 1))
			processed=$((processed + 1))
			continue
		else
			included=$((included + 1))
		fi
	fi

	if ! test -d "$PWD/$PREENDIR/$type"
	then
		if ! mkdir -p "$PWD/$PREENDIR/$type"
		then
			echo "Fatal error: mkdir(1) at \`$PWD/$PREENDIR/$type'" 1>&2
			exit 1
		fi
	fi

	if ! ln --backup=numbered -s "$name" "$PWD/$PREENDIR/$type/${name##*/}" 2>/dev/null
	then
		echo "Non-fatal error: skipped \`$name'" 1>&2
		skipped=$((skipped + 1))
	else
		processed=$((processed + 1))
	fi
done <<< "$(find "$PWD" -mount -name '.?*' -prune -o -follow -type f -print 2>/dev/null)"

sighandler 15 $?

# vi:ft=sh

USAGE

Use as simply as follows:

cd /some/path
preen
ls -R .preen

or, alternatively, as follows (see TODO)

cd /some/path
preen
cp -R -L .preen /backup/path

or, even, as follows (see TODO):

cd /some/path
preen
tar chf backup.tar .preen

or (see TODO):

cd /some/path
preen
rsync -a .preen /backup/path

STATISTICS

You can see runtime statistics by signaling the PID of the preen(1) process by one of HUP, USR1 or USR2. Run preen(1) in one terminal:

preen
STATISTICS
----------
Processed  : 123456
Excluded   : 0
Included   : 123400
Skipped    : 56
Failed     : 0
Suspicious : 5 (dangerous: 1, URL: 4)
Assorted   : 0
TOTAL      : 123456

and signal it from another terminal:

while true; do
  pkill -USR1 preen
  sleep 5
done

or, simply:

watch -n5 pkill -USR1 preen

Enjoy!

TODO

• Handle dupes so that it makes sense with backuping