#!/bin/bash

# Find standard deviation of packages per category and other fun statistics
# Questions? Ask Donnie Berkholz <spyderous@gentoo.org>

PORTDIR="$(portageq envvar PORTDIR)"
TOTAL="0"
# Multiple of standard deviation to use for category size estimate
SD_OPERATOR="/"
SD_OPERAND="2"

# Color setup, works only for interactive sessions
# Can't quote ${PS1}
if [ -n ${PS1} ]; then
	color="yes"
else
	unset color
fi

bold="${color:+\033[1m}"
unbold="${color:+\033[0m}"
red="${color:+\E[40;31m}"
green="${color:+\E[40;32m}"
yellow="${color:+\E[40;33m}"
blue="${color:+\E[40;34m}"
magenta="${color:+\E[40;35m}"
cyan="${color:+\E[40;36m}"
white="${color:+\E[40;37m}"

# How many categories?
for DIR in $(find ${PORTDIR} -mindepth 1 -maxdepth 1); do
	if [ "${DIR}" != "scripts" -a "${DIR}" != "profiles" -a "${DIR}" != "eclass" ]; then
		CATEGORIES="${CATEGORIES} ${DIR}"
	fi
done
CATEGORY_NUMBER="$(echo ${CATEGORIES} | wc -w)"

# Get total number of packages in tree
for DIR in ${CATEGORIES}; do
	((TOTAL += $(find ${DIR} -mindepth 1 -maxdepth 1 | wc -l)))
done

# Average packages per category
AVERAGE="$((${TOTAL} / ${CATEGORY_NUMBER}))"

# Get sigma( ( x(i) - x(bar) )^^2 )
for DIR in ${CATEGORIES}; do
	PACKAGE_TOTAL="$(find ${DIR} -mindepth 1 -maxdepth 1 | wc -l)"
	SIGMA="$((${SIGMA} + $((${PACKAGE_TOTAL} - ${AVERAGE}))**2))"
done

# Get standard deviation
VARIANCE="$((${SIGMA} / ${CATEGORY_NUMBER}))"
STANDARD_DEVIATION="$(echo "sqrt ( ${VARIANCE} )" | /usr/bin/bc)"

# Suggested category range
MINIMUM="$((${AVERAGE} - ${STANDARD_DEVIATION} ${SD_OPERATOR} ${SD_OPERAND}))"
MAXIMUM="$((${AVERAGE} + ${STANDARD_DEVIATION} ${SD_OPERATOR} ${SD_OPERAND}))"
RANGE="${MINIMUM} to ${MAXIMUM}"

echo
echo -e "${blue}Statistics for ${bold}${white}${PORTDIR}:${unbold}"
echo -e "${yellow}Total packages = ${red}${TOTAL}"
echo -e "${yellow}Total categories = ${red}${CATEGORY_NUMBER}"
echo -e "${yellow}Average packages per category = ${red}${AVERAGE}"
#echo -e "${yellow}Sigma = ${red}${SIGMA}"
#echo -e "${yellow}Variance = ${red}${VARIANCE}"
echo -e "${yellow}Standard deviation of packages per category = ${red}${STANDARD_DEVIATION}"
echo -e "${bold}${green}Suggested category size within (standard deviation ${SD_OPERATOR} ${SD_OPERAND}) of average: ${red}${RANGE}${unbold}"
echo -e "${bold}Split categories with more than ${MAXIMUM} packages, and do not create categories with fewer than ${MINIMUM} packages.${unbold}"
echo