#!/bin/sh

# HTML spell checker.
# This takes a list of filenames of HTML files and checks
# the spelling of each. It ignores words found in the
# okayword file.

# Note - this program only works on Unix systems, since it
# depends on /bin/sh, spell, and so on. Sorry.

# This also takes an option "-x filename"; the filename following
# -x is eXcluded from the spellcheck.  The -x option must be given before
# the actual filename, or it won't work.
# Currently only one file may be excluded.

# Re-sort okayword, since it must be sorted.
# It's easy to forget to re-sort that file, so let's use some
# time and sort it ourselves.

sort -u okayword > okayword.srt
if cmp -s okayword okayword.srt
then
 # Okayword was already sorted.
 rm okayword.srt
else
 echo "File okayword had to be re-sorted."
 mv okayword.srt okayword
fi

/bin/rm -fr ,errs

if [ "a$1" = 'a-x' ]
then
  exclude=$2
  shift ; shift
else
  exclude=""
fi

for file in $*
do
 if [ "$file" != "$exclude" ]
 then
  # sed -e 's/<[^>]*>//g' -e 's/&[a-z][a-z]*;//' $file | spell +okayword > ,temp
  rmhtml $file | spell +okayword > ,temp
  if [ -s ,temp ]
  then
   # We found some errors, add them to ,errs
   sed -e "s/^/${file}:/" ,temp >> ,errs
  fi
 fi
done

if [ -s ,errs ]
then
  more ,errs
  exit 2 # Return an error code.
else
  exit 0
fi