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         case $platform in
  52             # Replace EN-DASH (UTF-8 encoding) at line start with -
  53             Linux) LC_ALL=C sed -i $'s/^\u2013/-/g' "${1%.docx}.txt" ;;
  54
  55             # Replace EN-DASH (Windows-1252 encoding) at line start with -
  56             # Remove CR (0x0d) put by libreoffice in windows before each LF (0x0a)
  57             CYGWIN*) LC_ALL=C sed -i $'s/^\x96/-/g; s/\r$//' "${1%.docx}.txt" ;;
  58             *) echo "Unkown platform $platform"
  59                exit 1;;
  60         esac
  61
  62         # Changes things that should not be so in the specification
  63         # Change "-- Cond N3C MP" to "-- Cond N3C-MP" (to respect condition naming rules)
  64         LC_ALL=C sed -i 's/-- Cond N3C MP/-- Cond N3C-MP/' "${1%.docx}.txt"
  65         LC_ALL=C sed -i 's/^N3C MP/N3C-MP/' "${1%.docx}.txt"
  66         LC_ALL=C sed -i 's/-- Cond HOAndServCellAdd,$/-- Cond HOAndServCellAdd/' "${1%.docx}.txt"
  67     fi
  68
  69     if [[ -e "${1%.docx}.asn" ]] ; then
  70         echo "${1%.docx}.asn already exists: not regenerated"
  71         shift
  72         continue
  73     fi
  74
  75
  76     cat "${1%.docx}.txt" | awk '/^-- ASN1START$/,/^-- ASN1STOP$/' | grep -Evxe '-- ASN1ST(ART|OP)' | grep -v '^--' | awk 'NF || p; { p = NF }' p=0 > "${1%.docx}.asn"
  77
  78     if [[ -e "${1%.docx}.asn" ]] ; then
  79         echo "${1%.docx}.asn was generated"
  80     fi
  81     shift
  82 done
  83