any2md

#!/bin/bash
usage()
{
cat << EOF
Usage: `basename $0` [options] file(s)
        cat file | `basename $0` [options]
 
OPTIONS:
    -h              Show this message
    -t <title>      Specify title of document
    -a <author>     Specify author of document
    -d <date>       Specify creation date of document
NOTE:
    This script is sloppy when it comes to adding metadata.
    In particular, if you pass it a document that already has
    pandoc-style metadata, this will not check for that, and will
    simply prepend metadata to the original.
EOF
}
 
defined()
{
[ "${!1-X}" == "${!1-Y}" ]
}
 
convertFile() 
{
	orig="$@"
	ext=${orig##*.}
	# odt, doc, docx, rtf, rtfd, wordml, webarchive
	if [[ "$ext" = "md" || "$ext" = "markdown" || "$ext" = "mdown" ||  "$ext" = "txt" ]]; then
		cat "$orig"
	elif [[ "$ext" = "html" || "$ext" = "htm" || "$ext" = "shtml" ]]; then
		html2markdown -s --no-wrap --reference-links "$orig"
	elif [[ "$ext" = "tex" || "$ext" = "latex" || "$ext" = "rst" ]]; then
		pandoc -s --no-wrap --reference-links -t markdown "$orig"
	elif [[ "$ext" = "pdf" || "$ext" = "ps" ]]; then
		pdftohtml -noframes -stdout "$orig" | html2markdown -s --no-wrap --reference-links
	elif [[ "$ext" = "odt" || "$ext" = "doc" || "$ext" = "$docx" || "$ext" = "rtf" || "$ext" = "rtfd" || "$ext" = "wordml" || "$ext" = "webarchive" ]]; then 
		textutil -convert html -stdout "$orig" | html2markdown -s --no-wrap --reference-links
	else
			echo "unsupported file type" >&1
	fi
}
 
# parse options
 
while getopts "ht:a:d:" option
do
     case $option in
         h)
             usage
             exit 1
             ;;
         t)
             title=$OPTARG
             shift $((OPTIND-1)); OPTIND=1
             ;;
         d)
             date=$OPTARG
             shift $((OPTIND-1)); OPTIND=1
             ;;
         a)
             author=$OPTARG
             shift $((OPTIND-1)); OPTIND=1  
             ;;
         ?)
             usage
             exit
             ;;
     esac
done
 
# If no arguments, process STDIN
 
if [ $# -eq 0 ]; then 
    files=`mktemp -t any2md`
    cat > $files
else
    files="$@"
fi
 
for file in "$files"; do
	if [ -f "$file" -a -r "$file" ]; then
	    if defined "title" -o defined "author" -o defined "date"; then
	     echo "% $title"
	     echo "% $author"
	     echo "% $date"
	     echo
	    fi   
		convertFile "$file"
	else 
		echo "$file is not a file or is unreadable; skipping..."
	fi
done