Typographer

#!/bin/bash

# The script replaces some keyboard characters and their combinations
# with typographically correct ones, for example hyphens padded with
# spaces are converted to em- or en-dashes. Works with both English
# and Russian text. The text for conversion should be placed in the
# clipboard; the converted text is returned by the script to the
# clipboard too. espeak, xclip and yad need to be installed for the
# script to work, otherwise they should be replaced with alternatives;
# espeak is only used for verbal notification so it can be removed
# from the code if such notification is not needed.

function quitting()
{
    espeak -v english "$1"
    exit "$2"
}

function cyrillic()
{
    echo -n "$input" | sed -r 's/(^| )([^ A-Za-zА-Яа-я0-9]*)"([^ A-Za-zА-Яа-я0-9]*)"([^ A-Za-zА-Яа-я0-9]*)([A-Za-zА-Яа-я0-9])/\1\2«\3«\4\5/g; s/([A-Za-zА-Яа-я0-9])([^ A-Za-zА-Яа-я0-9]*)"([^ A-Za-zА-Яа-я0-9]*)"([^ A-Za-zА-Яа-я0-9]*)( |$)/\1\2»\3»\4\5/g; s/(^| )([^ A-Za-zА-Яа-я0-9]*)"([^ A-Za-zА-Яа-я0-9]*)([A-Za-zА-Яа-я0-9])/\1\2«\3\4/g; s/([A-Za-zА-Яа-я0-9])([^ A-Za-zА-Яа-я0-9]*)"([^ A-Za-zА-Яа-я0-9]*)( |$)/\1\2»\3\4/g; s/^([^«»]*)«([^«»]*)«/\1„\2«/g; s/»([^«»]*)»([^«»]*)$/»\1“\2/g; s/^-+ /— /g; s/ -+ / — /g; s/\.\.\./…/g; s/([0-9]) *-+ *([0-9])/\1—\2/g' | xclip -i -selection clipboard
}

function latin()
{
    echo -n "$input" | sed -r 's/(^| )([^ A-Za-zА-Яа-я0-9]*)"([^ A-Za-zА-Яа-я0-9]*)"([^ A-Za-zА-Яа-я0-9]*)([A-Za-zА-Яа-я0-9])/\1\2“\3“\4\5/g; s/([A-Za-zА-Яа-я0-9])([^ A-Za-zА-Яа-я0-9]*)"([^ A-Za-zА-Яа-я0-9]*)"([^ A-Za-zА-Яа-я0-9]*)( |$)/\1\2”\3”\4\5/g; s/(^| )([^ A-Za-zА-Яа-я0-9]*)"([^ A-Za-zА-Яа-я0-9]*)([A-Za-zА-Яа-я0-9])/\1\2“\3\4/g; s/([A-Za-zА-Яа-я0-9])([^ A-Za-zА-Яа-я0-9]*)"([^ A-Za-zА-Яа-я0-9]*)( |$)/\1\2”\3\4/g; s/^([^“”]*)“([^“”]*)“/\1‘\2“/g; s/”([^“”]*)”([^“”]*)$/”\1’\2/g; s/^-+ /– /g; s/ -+ / – /g; s/\.\.\./…/g; s/([0-9]) *-+ *([0-9])/\1–\2/g' | xclip -i -selection clipboard
}

input=$(xclip -o -selection clipboard)

if [[ $input =~ ^.*[A-Za-z].*[А-Яа-я].*$ ]] || [[ $input =~ ^.*[А-Яа-я].*[A-Za-z].*$ ]]; then

    yad --title="Typograph" --text="Select a character set: " --button='Cyrillic:0' --button='Latin:1' --button="Abort:2" --borders=8
    case "$?" in
        0)
            cyrillic
            ;;
        1)
            latin
            ;;
        2)
            quitting "Aborted!" 1
            ;;
        *)
            true
            ;;
    esac
else
    if [[ $input =~ ^.*[А-Яа-я].*$ ]]; then
        cyrillic
    else
        latin
    fi
fi

quitting "Done!" 0