#!/bin/bash
# stats - calculate statistics from a whitespace-separated list of integers
#
# Computes count, total, average, max, min, range (max-min), and outliers
# (1.5 * IQR method) from stdin input
#
# Usage:
#   command | stats [-n]
#   stats -h

_stats() (
    local SCRIPT_NAME; SCRIPT_NAME="$(basename "${BASH_SOURCE[0]}")"
    case "${BASH_SOURCE[0]}" in /dev/*|/proc/*) SCRIPT_NAME="" ;; esac
    case "$SCRIPT_NAME" in ""|bash|sh|zsh|dash) SCRIPT_NAME="stats" ;; esac

    _error() { echo "[ERR][$SCRIPT_NAME] $*" >&2; }

    _show_help() {
        local s; [ -t 1 ] && s=$'\033[4m'
        local r; [ -t 1 ] && r=$'\033[24m'
        echo "NAME"
        echo "  $SCRIPT_NAME - calculate statistics from a list of integers"
        echo "SYNOPSIS"
        echo "  ${s}command${r} | $SCRIPT_NAME [-n]"
        echo "  $SCRIPT_NAME -h"
        echo "DESCRIPTION"
        echo "  Reads whitespace-separated integers from stdin and prints count, total,"
        echo "  average, range (min to max), and outliers (1.5 * IQR method). Input may"
        echo "  be newline-, space-, or tab-delimited (or any mix)."
        echo "OPTIONS"
        echo "  -n             Print the dataset before statistics"
        echo "  -h, --help     Show this help message"
        echo "EXIT STATUS"
        echo "  0  Success"
        echo "  2  Usage error (non-integer input)"
        echo "  3  Dependency error (bc missing)"
        echo "DEPENDENCIES"
        echo "  bc, sort"
    }

    case "${1-}" in
        -h|--help) _show_help; return 0 ;;
    esac

    if ! command -v bc >/dev/null 2>&1; then
        _error "bc is required"
        return 3
    fi

    # Read a list of whitespace-separated integers from stdin (newlines, spaces, tabs all OK)
    local -a numbers
    local line
    while IFS= read -r line || [ -n "$line" ]; do
        [ -n "$line" ] && numbers+=("$line")
    done < <(tr -s '[:space:]' '\n')

    # Check for any non-digits (which would most likely be a decimal point) - allow spaces as the numbers are concatenated with spaces
    grep -q '[^0-9 ]' <<<"${numbers[*]}" && { _error "Integers only"; return 2; }
    # If -n is provided, list the dataset first
    [ "$1" = "-n" ] && printf '%s, ' "${numbers[@]}" | sed 's/, $/\n/'

    # Sort the numbers in ascending order (for outlier calculation, but it gets us min and max too)
    local -a sorted
    while IFS= read -r line || [ -n "$line" ]; do sorted+=("$line"); done < <(printf '%s\n' "${numbers[@]}" | sort -n)
    #mapfile -t sorted <<< "$(printf '%s\n' "${numbers[@]}" | sort -n)"
    #sorted=($(printf '%s\n' "${numbers[@]}" | sort -n))

    # Find outliers via Interquartile Range (IQR)
    # Calculate quartiles
    local -i n=${#sorted[@]}
    local -i Q1="${sorted[$((n / 4))]}"
    local -i Q3="${sorted[$((n * 3 / 4))]}"
    # Calculate the IQR
    local -i IQR=$((Q3 - Q1))
    # Define lower and upper bounds for outliers
    local -i lower_bound; lower_bound="$(printf '%.0f' "$(bc <<< "$Q1 - 1.5 * $IQR")")"
    local -i upper_bound; upper_bound="$(printf '%.0f' "$(bc <<< "$Q3 + 1.5 * $IQR")")"
    # Find the position of the first and last data points within the bounds
    local -i lower_index=0
    local -i upper_index=$((n - 1))
    while ((sorted[lower_index] < lower_bound)); do
        ((lower_index++))
    done
    while ((sorted[upper_index] > upper_bound)); do
        ((upper_index--))
    done
    # Collect lower and upper outliers
    local -a outliers=("${sorted[@]:0:$lower_index}" "${sorted[@]:$((upper_index + 1))}")
    local percent_outliers; percent_outliers="$(printf '%.1f%%' "$(bc <<< "scale=4; ${#outliers[@]} / ${#numbers[@]} * 100")")"

    # Get total, min, and max
    local -i total; total=$(printf '%s+' "${sorted[@]}" | sed 's/+$//' | bc)
    local -i min="${sorted[0]}"
    local -i max="${sorted[$((${#sorted[@]} - 1))]}"
    local -i range=$((max - min))

    # Print results
    echo "Count: ${#numbers[@]}"
    echo "Total: $total"
    echo "Average: $(bc <<< "$total / ${#numbers[@]}")"
    echo "Range: $range ($min to $max)"
    echo "Outliers: ${outliers[*]}"
    echo "          ${#outliers[@]} of ${#numbers[@]} ($percent_outliers)"
)

_stats "$@"
__stats_rc=$?
unset -f _stats
if [ -n "${BASH_SOURCE[0]}" ] && [ "${BASH_SOURCE[0]}" != "$0" ]; then
    eval "unset __stats_rc; return $__stats_rc"
fi
eval "unset __stats_rc; exit $__stats_rc"