#! /bin/sh

# transform HTML character mnemonic reference web page into mnemonic entries
#	<td class="code"><code>&amp;szlig;</code></td>
#	<td class="code"><code>&amp;#223;</code></td>
# ->
#		{"szlig", 0x00DF /* ß */},

refpage=${1-zeichen.htm}

htmlver=`sed -e '1 s,.*HTML \([0-9.]*\).*,\1,' -e t -e d $refpage`

echo "	/* HTML $htmlver mnemos (named entities) */"

cat $refpage | tr -d '\015' |
sed -e '/class="code"/ b entry' -e d -e ': entry' -e N \
    -e "s@.*&amp;\([^;]*\);.*&amp;#\([^;]*\);.*@echo -n '	{\"\1\", 0x'; printf %04X \2; echo -n ' /* U+'; printf %04X \2; echo ' */},'@" |
sh | ./insutf8 | sed -e 's,U+[^ ]* ,,' |
sed	-e 's,"not","\&not",' \
	-e 's,"oslash","\&oslash",' \
	-e 's,"circ","\&circ",' \
	-e 's,"phi","\&phi",' \
	-e 's,"asymp","\&asymp",'

echo "check uniqueness with ./mnemos.check" >&2
