#! /bin/bash

trace () {
#	echo $* >&2
	true
}

wgetfirst=false
#wgetfirst=true

case "$1" in
-sh)	postproc=cat
	shift;;
*)	postproc=bash;;
esac

case `basename "$PWD"` in
mintty)
	echo "You seem to be in a mintty config directory:" >&2
	echo "	$PWD" >&2
	echo "For direct emoji deployment, run this script from its subdirectory 'emojis'." >&2
	exit;;
esac

emojisurl0=https://www.unicode.org/emoji/charts/full-emoji-list.html
emojisurl1=https://www.unicode.org/emoji/charts/full-emoji-modifiers.html

download () {
	if $wgetfirst
	then	wget -N -t 1 "$1"
	elif type curl
	then	# prefer curl because it works better behind proxies
		if [ -f `basename "$1"` ]
		then	curl -RO -z `basename "$1"` "$1"
		else	# avoid "Warning: Illegal date format for -z"
			curl -RO "$1"
		fi
	else	wget -N -t 1 "$1"
	fi ||
	(echo "FILE: $1" >&2
	 echo "ERROR: download failed, file incomplete" >&2
	 if [ -f `basename "$1"` ]
	 then	# enforce fresh download next time
		touch -d 1970-01-01 `basename "$1"`
	 fi
	 return 9
	)
}

case "$1" in
""|-h|--help)
	echo "Usage: `basename $0` [-d | DIR | .../full-emoji-list.html] [EMOJI_STYLE]..." >&2
	echo >&2
	echo "This script extracts emojis graphics (.png format) from a downloaded copy of" >&2
	echo "	$emojisurl0" >&2
	echo "	$emojisurl1" >&2
	echo "for the selected emoji style sets, or (if none selected) all of them:" >&2
	echo "	apple google facebook windows twitter emojione samsung" >&2
	echo "and always extracts common emoji graphics." >&2
	echo >&2
	echo "Options:" >&2
	echo "	-d	Download the chart files" >&2
	echo "	DIR	Expect the chart files in given directory" >&2
	echo >&2
	echo "Warning: with all styles selected, this may take a while." >&2
	echo >&2
	echo "Note: for direct deployment, first go into subdirectory 'emojis' of one of the" >&2
	echo "mintty config directories:" >&2
	echo '  ~/.mintty' >&2
	echo '  ~/.config/mintty' >&2
	echo '  $APPDATA/mintty' >&2
	echo '  /usr/share/mintty' >&2
	exit;;
-d|--download)
	if download $emojisurl0 && download $emojisurl1
	then	emojis0=full-emoji-list.html
		emojis1=full-emoji-modifiers.html
	else	echo Download failed >&2
		exit
	fi
	shift;;
*.html)	emojis1=`dirname "$1"`/full-emoji-modifiers.html
	if [ -r "$1" -a -r "$emojis1" ]
	then	emojis0="$1"
	else	echo Not readable: "$1" "$emojis1" >&2
		exit
	fi
	shift;;
*)	if [ -d "$1" ]
	then	emojis0="$1"/full-emoji-list.html
		emojis1="$1"/full-emoji-modifiers.html
		if [ -r "$emojis0" -a -r "$emojis1" ]
		then	true
		else	echo Not readable: "$emojis0" "$emojis1" >&2
			exit
		fi
		shift
	else	echo Missing file name of full emoji list >&2
		exit
	fi;;
esac

echo -n "Using " >&2
cat "$emojis0" | sed -e "s,<title>\(.*\)</title>,\1," -e t -e d >&2

#cat "$emojis0" |
#sed -e "/^<\/tr/ q" -e "s/.*<th.*#col-vendor.>\([^.<]*\).*/\1/" -e t -e d |
#pr -t -n | sed -e "s,^,vendor ," -e 7q
# 11.0:
#	vendor     1    Appl
#	vendor     2    Goog
#	vendor     3    Twtr
#	vendor     4    One
#	vendor     5    FB
#	vendor     6    Sams
#	vendor     7    Wind
# 12.0:
#	vendor     1    Appl
#	vendor     2    Goog
#	vendor     3    FB
#	vendor     4    Wind
#	vendor     5    Twtr
#	vendor     6    Joy
#	vendor     7    Sams

seli=0
styles=(common)
for vendor in `cat "$emojis0" |
  sed -e "/^<\/tr/ q" -e "s/.*<th.*#col-vendor.>\([^.<]*\).*/\1/" -e t -e d |
  sed -e 7q`
do	seli=`expr $seli + 1`
	case "$vendor" in
	Appl)	st=apple; apple=$seli;;
	Goog)	st=google; google=$seli;;
	FB)	st=facebook; facebook=$seli;;
	Wind)	st=windows; windows=$seli;;
	Twtr)	st=twitter; twitter=$seli;;
	One|Joy) st=emojione; emojione=$seli;;
	Sams)	st=samsung; samsung=$seli;;
	esac
	styles[$seli]=$st
done

case "$1" in
"")	set - apple google facebook windows twitter emojione samsung;;
esac
sel=
while	case "$1" in
	apple)		seli="$apple";;
	google)		seli="$google";;
	facebook)	seli="$facebook";;
	windows)	seli="$windows";;
	twitter)	seli="$twitter";;
	joy|emojione)	seli="$emojione";;
	samsung)	seli="$samsung";;
	"")	false;;
	*)	echo emoji set "$1" not known; exit;;
	esac
do	sel="$sel$seli"
	mkdir -p "$1"
	shift
done
mkdir -p common

export sel

echo "Warning: this may take a while on Cygwin" >&2

LC_ALL=C
export LC_ALL

total=`grep -e "name='\([^']*\)'.*U+" "$emojis0" "$emojis1" | wc -l`
export total

(
echo "Extracting $total emojis " >&2

echo LC_ALL=C
echo export LC_ALL
echo total=$total
echo sel=$sel
echo "styles=(${styles[*]})"

cat <<\/EOS
n=0

name () {
  ename=$1
  style=0
  n=$(( $n + 1 ))
  p=$(( ${n}00 / $total ))
  echo "emoji $ename (${p}%)" >&2
}

img0 () {
  echo " common $ename.png" >&2
  echo "$1" | base64 -d > common/$ename.png
}

img () {
  style=$(( $style + 1 ))
  case $sel in
  *$style*)	echo "$1" | base64 -d > ${styles[$style]}/$ename.png;;
  esac
}

imgskip () {
  style=$(( $style + 1 ))
  case $sel in
  *$style*)	echo " skip ${styles[$style]}/$ename.png" >&2;;
  esac
}

/EOS

cat "$emojis0" "$emojis1" |
sed -e "s/^.*name='\([^']*\)'.*U+.*/name \1/" -e "t name" \
    -e "s/.*—.*/imgskip/" -e t \
    -e "s@^.*….*src='data:image/png;base64,\([^']*\)'.*@img0 \1@" -e t \
    -e "s@^.*src='data:image/png;base64,\([^']*\)'.*@img \1@" -e t \
    -e d \
    -e ": name" \
    -e "s,_,-,g"
) | $postproc