#!/bin/bash
# a munkakönyvtárban lévő dtd, properties, rdf és html fájlok helyesírás-ellenőrzése

convert_uxxxx()
{
sed "\
s/\\\\u00e1/á/ig;s/\\\\u00c1/Á/ig;\
s/\\\\u00e9/é/ig;s/\\\\u00c9/É/ig;\
s/\\\\u00ed/í/ig;s/\\\\u00cd/Í/ig;\
s/\\\\u00f3/ó/ig;s/\\\\u00d3/Ó/ig;\
s/\\\\u00f6/ö/ig;s/\\\\u00d6/Ö/ig;\
s/\\\\u0151/ő/ig;s/\\\\u0150/Ő/ig;\
s/\\\\u00fa/ú/ig;s/\\\\u00da/Ú/ig;\
s/\\\\u00fc/ü/ig;s/\\\\u00dc/Ü/ig;\
s/\\\\u0171/ű/ig;s/\\\\u0170/Ű/ig;\
s/\\\\u000a/ /g"
}

distill_dtd()
# többsoros kommentekre és azonos sorban levő több kommentre is jól működik
{
sed 's/<!--[^(-->)]*-->//g'|\
sed -e :a -e 's/<!--.*-->//g;/<!--/N;//ba'|\
sed 's/<!ENTITY[ 	]*[^ 	]*[ 	]*\([^>]*\)>/\1/g'
}

distill_properties()
{
sed ':a;/^[ 	]*#.*$/d;ta;s/^[^=]*=[ 	]*\(.*\)$/\1/g;s/\\n/ /g'
}

distill_rdf()
{
grep 'nc:name'|\
perl -p -e 's/^.*nc:name=["](.*?)["].*$/\1/g'
}

distill_html()
{
sed -e :a -e 's/<[^>]*>//g;/</N;//ba'
}

for filename in `find . -noleaf -type f`; do
  echo "$filename"|grep -E "\.dtd$|\.ent$" >/dev/null
  if [ $? -eq 0 ]; then
    cat "$filename"|distill_dtd
  else
    echo "$filename"|grep -E "\.properties$" >/dev/null
    if [ $? -eq 0 ]; then
      cat "$filename"|distill_properties
    else
      echo "$filename"|grep -E "\.rdf$" >/dev/null
      if [ $? -eq 0 ]; then
        cat "$filename"|distill_rdf
      else
        echo "$filename"|grep -E "\.[x]*html$" >/dev/null
        if [ $? -eq 0 ]; then
          cat "$filename"|distill_html
        fi
      fi
    fi
  fi
done | iconv -f utf-8 -t windows-1250 | convert_uxxxx | hunspell -l | sort | uniq >"checkmoz_out_`date +"%Y.%m.%d_%H.%M.%S"`"