bin/docx2asn.sh

   1 #!/bin/bash
   2
   3 if [[ $# -eq 0 ]] ; then
   4     echo "Usage: ${0##*/} file [files]"
   5     echo "Extract ASN.1 from one or more 3GPP 38.331 docx files and save then to .asn files"
   6     exit 1
   7 fi
   8
   9 if [[ ! -w . ]] ; then
  10     echo "Cannot write to current directory"
  11     exit 1
  12 fi
  13
  14 platform=$(uname)
  15
  16 case $platform in
  17     Linux) libreoffice="libreoffice" ;;
  18     CYGWIN*) libreoffice="soffice" ;;
  19     *) echo "Unkown platform $platform"
  20        exit 1;;
  21 esac
  22
  23 if [[ ! -x $(which "$libreoffice") ]] ; then
  24     echo "Cannot find libreoffice executable"
  25     echo "Make sure it is installed and the PATH variable includes its path"
  26     exit 1
  27 fi
  28
  29 while [ "$#" -gt 0 ]; do
  30
  31     echo "Processing $1"
  32
  33     if [[ ! -f "$1" ]] || [[ ! -r "$1" ]] ; then
  34         echo "$1 is not a readable file"
  35         shift
  36         continue
  37     fi
  38
  39     if [[ -e "${1%.docx}.txt" ]] ; then
  40         echo "${1%.docx}.txt already exists: skipping conversion"
  41
  42     else
  43         $libreoffice --convert-to txt "$1"
  44
  45         if [[ ! -f "${1%.docx}.txt" ]] || [[ ! -r "${1%.docx}.txt" ]] ; then
  46             echo "libreoffice did not create ${1%.docx}.txt"
  47             shift
  48             continue
  49         fi
  50
  51         # On windows, soffice adds CR characters at line ends and produces EN-DASH in Windows-1252 enconding, i.e. 0x96
  52         # Remove the CR at each line end and convert EN-DASH to normal minus
  53         echo "${1%.docx}.txt"
  54         LC_ALL=C sed -i $'s/\x96/-/g' "${1%.docx}.txt"
  55         LC_ALL=C sed -i $'s/\r$//' "${1%.docx}.txt"
  56
  57     fi
  58
  59     if [[ -e "${1%.docx}.asn" ]] ; then
  60         echo "${1%.docx}.asn already exists: not regenerated"
  61         shift
  62         continue
  63     fi
  64
  65
  66     cat "${1%.docx}.txt" | awk '/^-- ASN1START$/,/^-- ASN1STOP$/' | grep -Evxe '-- ASN1ST(ART|OP)' | grep -v '^--' | awk 'NF || p; { p = NF }' p=0 > "${1%.docx}.asn"
  67
  68     if [[ -e "${1%.docx}.asn" ]] ; then
  69         echo "${1%.docx}.asn was generated"
  70     fi
  71     shift
  72 done
  73