#!/bin/sh
#
# tidy-us

for f in *.html *.htm
do

  case "$f" in
  '*.html'|'*.htm')
    continue
    ;;
  esac

  # only process real files
  if [ -d "$f" -o -h "$f" ]
  then
    continue
  fi

  # also skip MS Word HTML
  if grep -i 'generator.*Microsoft Word' "$f" >/dev/null
  then
    continue
  fi

 #tidy --word-2000 true $f > $f.tidy 2> $f.err
  tidy $f > $f.tidy 2> $f.err

  if cmp $f $f.tidy >/dev/null
  then
    # keep the original, with its file attributes and possibly hardlinks
    rm $f.tidy
    echo "$f: no changes from tidy"
  else
    mv -f $f $f.untidy &&
    mv -f $f.tidy $f &&
    echo "$f:  tidied" ||
    echo "$f: untidy, replacing with tidied version failed"
  fi

  if fgrep 'No warnings or errors were found.' $f.err >/dev/null
  then
    echo "$f: no errors or warnings from tidy"
    rm $f.err
  else
    < $f.err perl -ne 's/line (\d+) column (\d+) - Warning:/$1.$2 / and printf "%s:%s", '"'$f'"', $_'
  fi

done