extents-expr

#!/bin/bash

# SPDX-FileCopyrightText: 2016-2020 Graham R. Cobb
# SPDX-License-Identifier: GPL-3.0-or-later
#
# Copyright (C) 2016 Graham R. Cobb
# Released under GPL V3.0 -- see LICENSE
#
# extents-expr [-s] <directory>|@<file>... [<operator> <directory>|@<file>...]...
#
# Extents lists are generated from each <directory> and the operators are applied.  The default
# operator used between adjacent directories is "union".  Specifying @ causes the extents to be
# read from an existing file.
#
# Operators are:
#   + union
#   ^ intersection
#   - difference
#
# There is no operator precedence, no parentheses and all evaluation is strictly left-to-right.
# This means that using wildcards or concatenation between directories in the righthand-side
# of an operator will almost certainly not achieve the desired effect (see example below).
#
# If the -s option is specified then the resulting list is automatically sent to extents-size.
#
# Example: to find out how much space is being wasted by keeping historical btrbk snapshots
# of a subvolume called 'cobb' you could use:
#
# 	extents-expr -s cobb.* - cobb.20160815T000602+0100
#
# Or, to find out how much data is shared between two specific snapshots use:
#
#	extents-expr -s Media.20160801T030601+0100 ^ Media.20160815T000602+0100
#
# However, the following expression DOES NOT give the result that is probably intended:
#
#	extents-expr -s cobb.20160815T000602+0100 - cobb.*
#
# The desired calculation was probably to determine how much of the data in the latest snapshot is
# unique to that snapshot (i.e. new).  However the effect is to subtract the first of the cobb.*
# directories and then add in all the rest.  The desired effect would be something like:
#
# 	extents-expr -s cobb.20160815T000602+0100 - ( cobb.* )
#
# but parentheses (and precedence) are not available.
#
# THIS MIGHT CHANGE at some time in the future.
#

# Make sure errors are passed back
set -o pipefail

if [ "$1" = "-s" ]
then
    shift
    $0 "$@" | extents-size
    exit $?
fi

extents-file-or-list() {
    # Either generate an extent list or use an existing file
    if [ -z "${1##@*}" ]
    then
	# @file
	cat "${1#@}"
    else
	extents-list "$1"
    fi
}

outfile=""
while [ $# -gt 0 ]
do
    if [ "$1" = "+" ]
    then
	: Skip, as + is the default operator
    elif [ "$1" = "^" ]
    then
	# Intersection
	shift
	
	# If existing outfile is null then we do nothing as intersection with null is still null
	if [ -n "$outfile" ]
	then
	    tmpfile="$(mktemp)"
	    status=$?
	    if [ $status != 0 ] ; then rm $outfile ; exit $status ; fi

	    extents-file-or-list "$1" >$tmpfile
	    status=$?
	    if [ $status != 0 ] ; then rm $tmpfile ; rm $outfile ; exit $status ; fi

	    newoutfile="$(mktemp)"
	    extents-intersection $outfile $tmpfile >$newoutfile
	    status=$?
	    rm $outfile
	    rm $tmpfile
	    if [ $status != 0 ] ; then rm $newoutfile ; exit $status ; fi
	    outfile=$newoutfile
	fi
    elif [ "$1" = "-" ]
    then
	# Difference
	shift
	
	# If existing outfile is null then we do nothing as difference from null is still null
	if [ -n "$outfile" ]
	then
	    tmpfile="$(mktemp)"
	    status=$?
	    if [ $status != 0 ] ; then rm $outfile ; exit $status ; fi

	    extents-file-or-list "$1" >$tmpfile
	    status=$?
	    if [ $status != 0 ] ; then rm $tmpfile ; rm $outfile ; exit $status ; fi

	    newoutfile="$(mktemp)"
	    status=$?
	    if [ $status != 0 ] ; then rm $tmpfile ; rm $outfile ; exit $status ; fi

	    extents-difference $outfile $tmpfile >$newoutfile
	    status=$?
	    rm $outfile
	    rm $tmpfile
	    if [ $status != 0 ] ; then rm $newoutfile ; exit $status ; fi
	    outfile=$newoutfile
	fi
    else
	# Default operation (union)
	tmpfile="$(mktemp)"
	status=$?
	if [ $status != 0 ] ; then [ $outfile ] && rm $outfile ; exit $status ; fi

	extents-file-or-list "$1" >$tmpfile
	status=$?
	if [ $status != 0 ] ; then rm $tmpfile ; [ $outfile ] && rm $outfile ; exit $status ; fi

	if [ -n "$outfile" ]
	then
	    newoutfile="$(mktemp)"
	    status=$?
	    if [ $status != 0 ] ; then rm $tmpfile ; rm $outfile ; exit $status ; fi

	    extents-union $outfile $tmpfile >$newoutfile
	    status=$?
	    rm $outfile
	    rm $tmpfile
	    if [ $status != 0 ] ; then rm $newoutfile ; exit $status ; fi
	    outfile=$newoutfile
	else
	    outfile=$tmpfile
	fi
    fi
    
    shift
done
if [ -n "$outfile" ]
then
    cat $outfile
    rm $outfile
fi