dirgraph

#!/bin/bash

###################################################################
#Script Name	: dirgraph
#
#Description	: Counts the number of subdirectories and files in 
#				specified directory. The output consist of drawn 
#				histogram where all files are sorted according their
#				size. The does not follw symbolic links and hidden
#				files or directories. The histogram can be
#				normalized when parameter -n is provided. Certain
#				filenames or directory can be skipped when parameter
#				-i with valid regular expression is provided. When
#				no specific directory is given, the script begin
#				searching in current ($PWD) directory.	
#
#Args			:[-i FILE_ERE] [-n] [DIR]
#				FILE_ERE -valid regular expression
#				-n -toggles histogram normalization to fit graph into
#					a width the output
#				DIR -specifies the directory to start search in
#
#Author       	: Vladislav Valek
#Email         	: xvalek14@vutbr.cz
###################################################################

#nasteveni konzole
set -u -o pipefail
	# -o pipefail	 navratova hodnoty pipeline je urcena poslednim prikazem
	# -u	povazuje nenastavene vychozi hodnoty promennych za chybu
	# -x	umoznuje sledovani prubehu vykonavani prikazu a jejich tisk na konzoli

export POSIXLY_CORRECT=YES

count_directories () {		#funkce pro pocitani slozek
	num=`find "$1" -type d 2>/dev/null | egrep -v "$2" | wc -l`
	exit_code=$?
	printf "Directories: %d\n" $num
	return $exit_code 
}

count_files () {	#funkce pro pocitani souboru
	num=`find "$1" -type f 2>/dev/null | egrep -v "$2" | wc -l`
	exit_code=$?
	printf "All files: %d\n" $num
	return $exit_code
}

make_fhist () {
	file_list=`find "$1" -type f -print 2>/dev/null | egrep -v "$3"`
	ret_val1=$?     #navratova hodnota posledniho prikazu urcuje navratovy kod dane funkce

	typeset -A cn=([lt100]=0 [lt1Ki]=0 [lt10Ki]=0 [lt100Ki]=0 [lt1Mi]=0 [lt10Mi]=0 [lt100Mi]=0 [lt1Gi]=0 [gt1Gi]=0)

	ret_val2=0
	byte_num=0

	while IFS=$'\n' read -r row
	do
			byte_num=`wc -c "$row" 2>/dev/null | awk '{print $1}'`
			ret_val2=$?

		if [ $ret_val2 -ne 0 ]
		then
			continue
		fi

		if (( $byte_num < 100 ))
		then
			cn[lt100]=`expr ${cn[lt100]} + 1`

		elif (( $byte_num >= 100 &&  $byte_num < 1024 ))
		then
			cn[lt1Ki]=`expr ${cn[lt1Ki]} + 1`

		elif (( $byte_num >= 1024 &&  $byte_num < 10240 ))
		then
			cn[lt10Ki]=`expr ${cn[lt10Ki]} + 1`

		elif (( $byte_num >= 10240 &&  $byte_num < 102400 ))
		then
			cn[lt100Ki]=`expr ${cn[lt100Ki]} + 1`

		elif (( $byte_num >= 102400 &&  $byte_num < 1048576 ))
		then
			cn[lt1Mi]=`expr ${cn[lt1Mi]} + 1`

		elif (( $byte_num >= 1048576 &&  $byte_num < 10485760 ))
		then
			cn[lt10Mi]=`expr ${cn[lt10Mi]} + 1`

		elif (( $byte_num >= 10485760 &&  $byte_num < 104857600 ))
		then
			cn[lt100Mi]=`expr ${cn[lt100Mi]} + 1`

		elif (( $byte_num >= 104857600 &&  $byte_num < (1024*1048576) ))
		then
			cn[lt1Gi]=`expr ${cn[lt1Gi]} + 1`

		else
			cn[gt1Gi]=`expr ${cn[gt1Gi]} + 1`
		fi
	done <<< "$file_list"
	
	cols=1				#pocet sloupcu
	catname_length=11			#delka nazvu kategorie
	norm_on="no"			#rozhodnuti o normalizaci

	if [ "$2" = "-n" ] 	#podminka, zda chci normalizovat graf
	then

		if [ -t 1 ]		#pokud skript bezi v terminalu
		then
			cols=`expr $(tput cols) - 1 - $catname_length`	
		else		#pokud bezi skript nekde jinde
			cols=79
		fi

		max=0
		for idx in ${!cn[@]}		#naleznu maximum
		do
			if [ ${cn[$idx]} -gt $max ]; then max=${cn[$idx]}; fi
		done
		
		cf=0
		if [ $max -gt $cols ]; then    	#pokud je max vetsi jak delka radku, zacnu normalizovat
 			norm_on="yes"
			cf=$(echo "scale=5; $cols / $max" | bc)
		fi

		for idx in "${!cn[@]}"			#normalizuji...
		do
			if [ "$norm_on" = "yes" ]
			then	
				cn[$idx]=$(echo "scale=5; ${cn[$idx]} * $cf" | bc)
				b=${cn[$idx]}			#value before rounding
				cn[$idx]=$(LC_NUMERIC="en_US.UTF-8" printf "%.0f" "${cn[$idx]}")
				a=${cn[$idx]}			#value after rounding
					
				res1=$(echo "if($b!=0) 1" | bc -l)	
				res2=$(echo "if($a==0) 1" | bc -l)

				if [ "$res1" = "1" -a "$res2" = "1" ]  #zaokrouhlovani na jednicku
				then
					cn[$idx]=1
				fi
			else
				break
			fi
		done 
	fi

	echo "File size histogram:"

	printf "  %s%3s " "<100 B" ":"
	idx=0
	while [ $idx -lt ${cn[lt100]} ]; do echo -n "#"; idx=`expr $idx + 1`; done
	echo

	printf "  %s%3s " "<1 KiB" ":"
	idx=0
	while [ $idx -lt ${cn[lt1Ki]} ]; do echo -n "#"; idx=`expr $idx + 1`; done
	echo

	printf "  %s%2s " "<10 KiB" ":"
	idx=0
	while [ $idx -lt ${cn[lt10Ki]} ]; do echo -n "#"; idx=`expr $idx + 1`; done
	echo	

	printf "  %s%s " "<100 KiB" ":"
	idx=0
	while [ $idx -lt ${cn[lt100Ki]} ]; do echo -n "#"; idx=`expr $idx + 1`; done
	echo
	
	printf "  %s%3s " "<1 MiB" ":"
	idx=0
	while [ $idx -lt ${cn[lt1Mi]} ]; do echo -n "#"; idx=`expr $idx + 1`; done
	echo

	printf "  %s%2s " "<10 MiB" ":"
	idx=0
	while [ $idx -lt ${cn[lt10Mi]} ]; do echo -n "#"; idx=`expr $idx + 1`; done
	echo

	printf "  %s%s " "<100 MiB" ":"
	idx=0
	while [ $idx -lt ${cn[lt100Mi]} ]; do echo -n "#"; idx=`expr $idx + 1`; done
	echo

	printf "  %s%3s " "<1 GiB" ":"
	idx=0
	while [ $idx -lt ${cn[lt1Gi]} ]; do echo -n "#"; idx=`expr $idx + 1`; done
	echo

	printf "  %s%2s " ">=1 GiB" ":"
	idx=0
	while [ $idx -lt ${cn[gt1Gi]} ]; do echo -n "#"; idx=`expr $idx + 1`; done
	echo
	
	if [ $ret_val1 -ne 0 -o $ret_val2 -ne 0 ]
	then
		return 1		#navratova hodnota cele funkce
	else
		return 0
	fi
}

if [ $# -eq 0 ]		#pripad, kdy nebyl nalezen zadny parametr, skript bere aktualni adresar
then
	
	echo "Root directory: $PWD"
	count_directories "$PWD" '!.'
	count_files "$PWD" '!.'
	make_fhist "$PWD" x '!.'
	exit $?

elif [ $# -eq 1 ]	#byl zadan jeden pozicni parametr
then

	if [ "$1" = "-n" ]	#pozicni parametr je prepinac normalizace grafu
	then
		echo "Root directory: $PWD"
		count_directories "$PWD" '!.'
		count_files "$PWD" '!.'
		make_fhist "$PWD" $1 '!.'
		exit $?
	
	elif [ -d "$1" ]	#pozicni parametr je nazev adresare
	then
		echo "Root directory: $1"
		count_directories "$1" '!.'
		count_files "$1" '!.'
		make_fhist "$1" x '!.'
		exit $?

	elif [[ "$1" =~ [-*] ]]
	then
		echo "Invalid option!" >/dev/stderr 
		exit 64
	else
		echo "$1:Not a name of a directory!" >/dev/stderr 
		exit 66
	fi

elif [ $# -eq 2 ]	#zadany dva pozicni parametry
then

	if [ "$1" = "-i" ]
	then
		if [[ -n "$2" ]] && [[ ! "$2" =~ [-*] ]]	#test, zda je retezec za prepinacem nenulovy
		then
			if [ "$2" = "/" ]		#nesmi pokryvat nazev korenoveho adresare
			then
				echo "Ignored-files list should not include a name of the root directory!" >/dev/stderr 
				exit 64
			else
				echo "Root directory: $PWD"
				count_directories "$PWD" "$2"
				count_files "$PWD" "$2"
				make_fhist "$PWD" x "$2"
				exit $?

			fi
		else
			echo "Valid regular expression must follow!" >/dev/stderr 
			exit 65
		fi

	elif [ "$1" = "-n" ]
	then
		if [ -d "$2" ]		#testovani, zda druhy pozicni parametr je nazev adresare
		then
			echo "Root directory: $2"
			count_directories "$2" '!.'
			count_files "$2" '!.'
			make_fhist "$2" $1 '!.'
			exit $?

		else
			echo "$2:Not a name of a directory!" >/dev/stderr 
			exit 66		#66- neexistujici jmeno adresare
		fi	
			
	elif [[ "$1" =~ [-*] ]]
	then
		echo "Invalid option!" >/dev/stderr 
		exit 64
	else
		echo "Unrecognized action!" >/dev/stderr 
		exit 67
	fi
elif [ $# -eq 3 ]	#zadany tri pozicni parametry
then

	if [ "$1" = "-i" ]	
	then
		if [[ -n "$2" ]] && [[ ! "$2" =~ [-*] ]]	#test, zda je retezec za prepinacem nenulovy
		then
			if [ "$2" = "/" ]
			then
				echo "Ignored-files list should not include a name of the root directory!" >/dev/stderr 
				exit 64
			else
				if [ "$3" = "-n" ]
				then
					echo "Root directory: $PWD"
					count_directories "$PWD" "$2"
					make_fhist "$PWD" $3 "$2"
					exit $?

				
				elif [ -d "$3" ]
				then
					echo "Root directory: $3"
					count_directories "$3" "$2"
					count_files "$3" "$2"
					make_fhist "$3" x "$2"
					exit $?
				elif [[ "$3" =~ [-*] ]]
				then
					echo "$3:Not a valid parameter!" >/dev/stderr 
					exit 64		# 64 - spatny parametr
				else
					echo "Invalid parameter!"
					exit 64
				fi
			fi
		else
			echo "Valid regular expression must follow!" >/dev/stderr 
			exit 65			# 65- nepresny nebo nulovy regularni vyraz
		fi
	elif [[ "$1" =~ [-*] ]]
	then
		echo "Invalid option!" >/dev/stderr 
		exit 64
	else
		echo "Wrong order of parameters, correct form is: dirgraph [-i FILE_ERE] [-n] [DIR]" >/dev/stderr 
		exit 67
	fi

elif [ $# -eq 4 ]
then

	if [ "$1" = "-i" ]	
	then
		if [[ -n "$2" ]] && [[ ! "$2" =~ [-*] ]]  	#test, zda je retezec za prepinacem nenulovy
		then	
			if [ "$2" = "/" ]		#nesmi pokryvat nazev korenoveho adresare
			then
				echo "Ignored-files list should not include a name of the root directory!" >/dev/stderr 
				exit 64
			else
				if [ "$3" = "-n" ]
				then
					if [ -d "$4" ]
					then
							echo "Root directory: $4"
							count_directories "$4" "$2"
							count_files "$4" "$2"
							make_fhist "$4" $3 "$2"
							exit $?
					else
						echo "$4: Not a name of a directory!" >/dev/stderr 
						exit 66
					fi
				else
					echo "$3: Wrong parameter!"	>/dev/stderr 
					exit 64
				fi
			fi
		else
			echo "Valid regular expression must follow!" >/dev/stderr 
			exit 65
		fi
	elif [[ "$1" =~ [-*] ]]
	then
		echo "Invalid option!"	>/dev/stderr 
		exit 64
	else
		echo "Wrong order of parameters, correct form is: dirgraph [-i FILE_ERE] [-n] [DIR]" >/dev/stderr 
		exit 67
	fi
else
	echo "Too much parameters!" >/dev/stderr
	exit 67
fi
exit 1		#dostat se sem je chyba!