pkgstats

#!/usr/bin/env bash
#cito M:755 O:0 G:0 T:/usr/local/bin/pkgstats
#------------------------------------------------------------------------------
# Project Name      - ShellProjects/source/pkgstats
# Started On        - Sat 17 Jun 09:08:12 BST 2023
# Last Change       - Mon 19 Jun 23:34:57 BST 2023
# Author E-Mail     - terminalforlife@yahoo.com
# Author GitHub     - https://github.com/terminalforlife
#------------------------------------------------------------------------------
# Gather, store, and maintain list of available packages and their file sizes.
#
# Example:
#
#   $ pkgstats -s packages_bytes_bionic.txt
#    Packages: 82795
#    TTL Size: 252.2G
#   Avg. Size: 3.1M
#     >= 500M: 6 (3.9G, 0.01%)
#      <= 10M: 74287 (66G, 89.72%)
#
#   Largest:
#
#     - 'flightgear-data-base' (1004.6M)
#     - 'redeclipse-data' (822.3M)
#     - '0ad-data' (610.5M)
#
#   Smallest:
#
#     - 'libc6-armel-armhf-cross' (776)
#     - 'libc6-armhf-armel-cross' (776)
#     - 'libc6-dev-armel-armhf-cross' (780)
#
# NOTE: This can take a very long time, depending on the repositories you have
#       available. Thanks to the ability for PKGStats to pick up where it left
#       off, you needn't do it all in one session.
#
#       If you run this, please be mindful of the servers. While sending a
#       spider to grab the header is surely harmless, it's still constant
#       connections, so please don't run this excessively!
#
# Features:
#
# N/A
#
# Bugs:
#
# N/A
#
# Dependencies:
#
#   apt (>= 1.6.17)
#   bash (>= 4.0)
#   gawk (>= 4.1.4)
#   wget (>= 1.19.4)
#------------------------------------------------------------------------------

CurVer='2023-06-19'
Progrm='pkgstats'
ProgrmFancy='PKGStats'

Usage() {
	read -d '' <<-EOF
		Usage: $Progrm [OPTS] FILE

		  -h, --help               - Display this help information.
		  -v, --version            - Output the version datestamp.
		  -s, --summary-only       - Just display summary of existing FILE.

		  The use of '--' to ignore proceeding options is supported.
	EOF

	printf '%s' "$REPLY"
}

Err() {
	printf 'Err: %s\n' "$2" 1>&2
	(( $1 > 0 )) && exit $1
}

SummaryOnly='False'

Skip=
Args=()
while [[ -n $1 ]]; do
	case $1 in
		--)
			Args+=(--)
			Skip='True' ;;
		-[^-]*)
			if [[ $Skip == True ]]; then
				Args+=("$1")
			else
				Str=${1#-}
				Len=${#Str}
				for (( Index = 0; Index < Len; Index++ )); {
					Args+=(-"${Str:Index:1}")
				}
			fi ;;
		*)
			Args+=("$1") ;;
	esac
	shift
done

set -- "${Args[@]}"

while [[ -n $1 ]]; do
	case $1 in
		--)
			break ;;
		--help|-h)
			Usage; exit 0 ;;
		--version|-v)
			printf '%s\n' "$CurVer"; exit 0 ;;
		--summary-only|-s)
			SummaryOnly='True' ;;
		-*)
			Err 1 'Incorrect option(s) specified.' ;;
		*)
			break ;;
	esac
	shift
done

Args=$#
if (( Args < 1 )); then
	Err 0 'Argument required.'

	read -d '' <<-EOF

		     $ProgrmFancy requires a file into which to save the package
		     information, consisting of a 'BYTES PACKAGE' format.

		Nothing to do -- quitting.
	EOF

	printf '%s' "$REPLY"
	exit 1
elif (( Args > 1 )); then
	Err 1 'Invalid argument(s).'
fi

DepCount=0
for Dep in apt-get apt-cache wget awk; do
	if ! type -P "$Dep" &> /dev/null; then
		Err 0 "Dependency '$Dep' not met."
		(( DepCount++ ))
	fi
done

(( DepCount > 0 )) && exit 1

Interactive=
[[ -t 1 ]] && Interactive='True'

if [[ -d $1 ]]; then
	Err 1 'Argument is a directory.'
elif [[ -f $1 ]]; then
	FileExists='True'

	if [[ $SummaryOnly == True ]]; then
		[[ -s $1 ]] || Err 1 'Unable to summarize empty FILE.'
	else
		printf "NOTE: File '%s' already exists.\n" "$1"

		while :; do
			read -p '      Continue? [Y/N] '
			case ${REPLY,,} in
				y|yes)
					break ;;
				n|no)
					exit 1 ;;
				'')
					Err 0 'Empty response -- try again.' ;;
				*)
					Err 0 'Unrecognised response -- try again.' ;;
			esac
		done
	fi
else
	if [[ $SummaryOnly == True ]]; then
		Err 1 'Unable to summarize non-existent FILE.'
	fi
fi

#-----------------------------------Gather Data from and Validate Existing File

if [[ $FileExists == True ]]; then
	LineNr=0
	LineErrs=0
	declare -A FilePKGs=()
	while read Bytes FilePKG; do
		(( LineNr++ ))

		[[ ${Bytes:0:1} == '#' ]] && continue

		if [[ $Bytes != +([[:digit:]]) ]]; then
			Err 0 "Invalid bytes on line $LineNr."

			(( LineErrs++ ))
		fi

		if [[ $FilePKG == *[[:space:]]* ]]; then
			Err 0 "Invalid package on line $LineNr."

			(( LineErrs++ ))
		fi

		(( LineErrs > 0 )) || FilePKGs["$FilePKG"]=$Bytes
	done < "$1"

	(( LineErrs > 0 )) && exit 1
fi

#---------------------------------------Begin Gathering Package Names and Sizes

if [[ $SummaryOnly != True ]]; then
	SigHandler() {
		printf '\n'
		exit 130
	}

	trap SigHandler INT

	readarray -t Packages <<< "$(apt-cache pkgnames 2>&-)" || exit 1

	[[ $FileExists == True ]] || > "$1"

	Exists=
	Count=0
	Len=${#Packages[@]}
	for Package in "${Packages[@]}"; {
		if [[ $FileExists == True ]]; then
			Exists=
			if [[ -n ${FilePKGs["$Package"]} ]]; then
				Bytes=${FilePKGs["$Package"]}
				Exists='True'
			fi
		fi

		if [[ $Exists != True ]]; then
			while read Line; do
				if [[ ${Line:0:1} == "'" ]]; then
					Link=${Line%\'*}
					Link=${Link#\'}

					break
				fi
			done <<< "$(apt-get download --print-uris "$Package" 2>&-)"

			while read F1 Bytes _; do
				if [[ $F1 == Length: ]]; then
					printf '%s %s\n' "$Bytes" "$Package" >> "$1"

					break
				fi
			done <<< "$(wget --spider "$Link" 2>&1)"
		fi

		if [[ $Interactive == True ]]; then
			if [[ $Exists == True ]]; then
				printf '\r[???.????] \e[92m%s\e[0m is \e[93m%d\e[0m bytes\e[K'\
					"$Package" $Bytes

				continue
			fi

			awk -v C=$(( ++Count )) -v L=$Len -v PKG="$Package" -v B=$Bytes '
				BEGIN {
					Format = "\r[%08.4f%%] \033[92m%s\033[0m "
					Format = Format "is \033[93m%d\033[0m bytes\033[K"
					printf(Format, C / L * 100, PKG, B)
				}
			' 2>&-
		fi
	}
fi

#---------------------------Display Summary (AWK for Floating Point Arithmetic)

Interactive=True
if [[ $Interactive == True ]]; then
	[[ $SummaryOnly == True ]] || printf '\rDone! Summary below.\e[K\n\n'

	awk '
		BEGIN {
			GE500M = 0
			LE10M = 0
		}

		# Underscores because AWK sucks at localizing variables.
		function Human(_Size) {
			_UnitStr = "b K M G T P E Z Y"
			_Len = split(_UnitStr, _Units)
			for (_Index = 1; _Index <= _Len; _Index++) {
				if (_Size < 1024) {
					_Result = sprintf("%.1f", _Size)
					_Len = length(_Result)

					if (substr(_Result, _Len - 1) == ".0") {
						_Result = substr(_Result, 0, _Len - 2)
					}

					if (_Units[_Index] == "b") {
						return(substr(_Result, 0, _Len - 1))
					} else {
						return(_Result _Units[_Index])
					}
				}

				_Size = _Size / 1024
			}
		}

		function Top10(__Sorted_Array, __Len) {
			for (__Index = 1; __Index <= __Len; __Index++) {
				split(__Sorted_Array[__Index], __Fields)

				printf("  - '"'%s'"' (%s)\n", __Fields[2], Human(__Fields[1]))

				if (__Index == 3) break
			}
		}

		{
			Lines[$0]

			Total += $1
		}

		$1 >= 524288000 {
			GE500M++
			GE500M_Size += $1
		}

		$1 <= 10485760 {
			LE10M++
			LE10M_Size += $1
		}

		END {
			printf(" Packages: %s\n", NR)
			printf(" TTL Size: %s\n", Human(Total))
			printf("Avg. Size: %s\n", Human(Total / NR))

			Perc = GE500M / NR * 100
			Size = Human(GE500M_Size)
			printf("  >= 500M: %s (%s, %0.2f%%)\n", GE500M, Size, Perc)

			Perc = LE10M / NR * 100
			Size = Human(LE10M_Size)
			printf("   <= 10M: %s (%s, %0.2f%%)\n\n", LE10M, Size, Perc)

			print("Largest:\n")
			Len = asorti(Lines, Sorted, "@ind_num_desc")
			Top10(Sorted, Len)

			print("\nSmallest:\n")
			Len = asorti(Lines, Sorted, "@ind_num_asc")
			Top10(Sorted, Len)
		}
	' "$1"
fi