bin/docx2txt.sh

   1 #!/bin/bash
   2
   3 if [[ $# -eq 0 ]] ; then
   4     echo "Usage: ${0##*/} file [files]"
   5     echo "Extract ASN.1 from one or more 3GPP 38.331 docx files and save them to .txt files"
   6     exit 1
   7 fi
   8
   9 if [[ ! -w . ]] ; then
  10     echo "Cannot write to current directory"
  11     exit 1
  12 fi
  13
  14 platform=$(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     $libreoffice --convert-to txt "$1"
  40
  41     if [[ ! -f "${1%.docx}.txt" ]] || [[ ! -r "${1%.docx}.txt" ]] ; then
  42         echo "libreoffice did not create ${1%.docx}.txt"
  43         shift
  44         continue
  45     fi
  46
  47     case $platform in
  48         # Replace EN-DASH (UTF-8 encoding) at line start with -
  49         Linux) LC_ALL=C sed -i $'s/^\u2013/-/g' "${1%.docx}.txt" ;;
  50
  51         # Replace EN-DASH (Windows-1252 encoding) at line start with -
  52         # Remove CR (0x0d) put by libreoffice in windows before each LF (0x0a)
  53         CYGWIN*) LC_ALL=C sed -i $'s/^\x96/-/g; s/\r$//' "${1%.docx}.txt" ;;
  54         *) echo "Unkown platform $platform"
  55            exit 1;;
  56     esac
  57
  58     # Changes things that should not be so in the specification
  59     # Change "-- Cond N3C MP" to "-- Cond N3C-MP" (to respect condition naming rules)
  60     LC_ALL=C sed -i 's/-- Cond N3C MP/-- Cond N3C-MP/' "${1%.docx}.txt"
  61     LC_ALL=C sed -i 's/^N3C MP/N3C-MP/' "${1%.docx}.txt"
  62     LC_ALL=C sed -i 's/-- Cond HOAndServCellAdd,$/-- Cond HOAndServCellAdd/' "${1%.docx}.txt"
  63     shift
  64 done
  65