#!/bin/bash
Version=1.03
Myname="${0##*/}"

:<<'DOC'
= fixfile - convert encoding to UTF-8 and line endings to LF

= Synopsis
fixfile [options] [file file...]	

== Options
-c|--crlf	convert line endings to CRLF; default: LF (unix)
-e|--encoding=X	convert to encoding X
-H|--Help	print full documentation via less and exit
-h|--help	print short help and exit
-v|--verbose	be [not] verbose (quiet is the default)
-V|--version	print version and exit

= Description
fixfile converts the given files in place, as follows:
- the encoding is converted to the encoding specified by the |--encoding|
  option, UTF8 by default.
- line endings are replaced with linefeeds (|\xA|), or,
  with the |--crlf| option, with carriage return + line feed (|\xD\xA|).
- If the file appears to be an HTML file, non-ASCII characters are replaced
  with their HTML equivalents; so for example ö will be replaced with |&#xf6;|.

If no file is given, converts standard input to standard output.
If any of the input files can not be converted for any reason, none of the
files are converted and the script quits with exit value 1.

= Bugs
This script needs the utf82html executable. Sources for it are available
[here](bitbucket.org/wybodekker/utf82html).

= Author
[Wybo Dekker](wybodekker@me.com)

= Copyright
Released under the [GNU General Public License](www.gnu.org/copyleft/gpl.html)
DOC

REd='\e[38;5;9m' Mag='\e[38;5;5m' Nor='\e[0m'
    die() { local i; for i; do echo -e "$Myname: $REd$i"; done 1>&2; exit 1; }
   Warn() { local i; for i; do echo -e "$Myname: $Mag$i$Nor"; done 1>&2; }
   warn() { $verbose && Warn "$@"; }
helpsrt() { sed -n "/^= $Myname/,/^= Desc/{/^= Desc/d;p}" "$0"; exit; }
helpall() { sed -n "/^:<<'DOC'$/,/^DOC/p" "$0"|sed '1d;$d'|
            less -P"$Myname-${Version/./·} (press h for help, q to quit)";exit; }

:<<'DOC' #----------------------------------------------------------------------
= handle_options
synopsis:	 handle_options "$@"
description:	handle the options.
globals used:	 Myname Version
globals  set:	 args
DOC
#-------------------------------------------------------------------------------
handle_options() {
   local options
      options=$(getopt \
      -n "$Myname" \
      -o hHVIve:c \
      -l help,Help,version,verbose,encoding:,crlf -- "$@"
   ) || exit 1
   eval set -- "$options"
   
   outenc=utf8
   linend='\n'
   export verbose=false
   while true; do
      case $1 in
      (-H|--Help)     # print full documentation via less and exit
		      helpall
		      ;;
      (-h|--help)     # print short help and exit
		      helpsrt
		      ;;
      (-V|--version)  # print version and exit
		      echo $Version
		      exit
		      ;;
      (-c|--crlf)     # convert line endings to CRLF; default: LF (unix)
		      linend='\r\n'
		      shift
		      ;;
      (-e|--encoding) # convert to encoding X
		      default: UTF-8
		      outenc="$2"
		      shift 2
		      ;;
      (-v|--verbose)  # be [not] verbose (quiet is the default)
		      verbose=true
		      shift
		      ;;
      (-I)            instscript "$0" ||
			 die 'the -I option is for developers only'
		      exit
		      ;;
      (--)            shift
		      break
		      ;;
      (*)             break
		      ;;
      esac
   done
   args=("$@")
}

handle_options "$@"
set -- "${args[@]}"

# check for valid output encoding:
iconv -l |sed 's=//$==' |grep -iq "^$outenc$" ||
   die "Output encoding ($outenc) is not recognized; try iconv -l"

tmp=$(mktemp -t "$Myname.XXXXXXXXXX") out=
trap 'rm -f $tmp' 0 1 2 15

test $# -eq 0 && set /dev/stdin || out="> \"$i\""

# first, check all input files for potential problems:
for i; do
   name="Input file $i"
   test -z "$out" && name="Input stream"
   test -e "$i"   || die "$name: not found"
   test -r "$i"   || die "$name: is read-protected"
   test -n "$out" -a ! -w "$i" && die "$name: write protected" 2>&1
   cp "$i" "$tmp"
   # check for valid input encoding:
   enc=$(file --mime-encoding --brief "$tmp")
   [[ $enc =~ ^unknown ]] && die "$name: unknown encoding"
   # check for valid input file type (must be text file):
   typ=$(file --mime-type     --brief "$tmp")
   [[ $typ =~ ^text ]] || die "$name: is not a text file"
done

# everthing OK - now do the conversions:
for i; do
   warn "Converting $i"
   cp "$i" "$tmp"
   enc=$(file --mime-encoding -bL "$tmp")
   typ=$(file --mime-type     -bL "$tmp")
   [[ $typ =~ html ]] && filter='utf82html' || filter='cat'
   eval "iconv --from=$enc --to=utf8 \"$tmp\" |
      sed -z '
	s/\([^\r]\)\n/\1\r\n/g
	s/\([^\r\n]\)$/\1\r\n/		# append crlf to lines missing any EOL
	s/\r\r/\r\n\r\n/g		#
	s/\r\([^\n]\|$\)/\r\n\1/g	#
	s/\r\n/$linend/g
	' |$filter $out"
done