recoll / Code / [a8e6ce] /src/filters/rcldoc

[a8e6ce]: src / filters / rcldoc History

rcldoc 152 lines (130 with data), 3.7 kB

#!/bin/sh
# @(#$Id: rcldoc,v 1.8 2007-06-08 13:51:08 dockes Exp $  (C) 2004 J.F.Dockes
# Parts taken from Estraier:
#================================================================
# Estraier: a personal full-text search system
# Copyright (C) 2003-2004 Mikio Hirabayashi
#================================================================
#================================================================
# rcldoc
# Extract text from an msword file by executing either antiword 
#  (or wvware maybe if we need it one day)
#
# The default is to use antiword, the code would need modifications to
# work with wvWare
#
#================================================================


# set variables
LANG=C ; export LANG
LC_ALL=C ; export LC_ALL
progname="rcldoc"
filetype=ms-word

decoder="antiword -t -i 1 -m UTF-8"
# Not ready to use this for now (it outputs html, so the code below has to
# be simplified.)
#decoder="wvWare -1 -c UTF-8"



#RECFILTCOMMONCODE
##############################################################################
# !! Leave the previous line unmodified!! Code imported from the
# recfiltcommon file

# Utility code common to all shell filters. This could be sourced at run
# time, but it's slightly more efficient to include the code in the
# filters at build time (with a sed script).

# Describe error in a way that can be interpreted by our caller
senderror()
{
    echo RECFILTERROR $*
    # Also alert on stderr just in case
    echo ":2:$progname::: $*" 1>&2
    exit 1
}

iscmd()
{
    cmd=$1
    case $cmd in
    */*)
	if test -x $cmd -a ! -d $cmd ; then return 0; else return 1; fi ;;
    *)
      oldifs=$IFS; IFS=":"; set -- $PATH; IFS=$oldifs
      for d in $*;do test -x $d/$cmd -a ! -d $d/$cmd && return 0;done
      return 1 ;;
    esac
}

checkcmds()
{
    for cmd in $*;do
      if iscmd $cmd 
      then 
        a=1
      else 
        senderror HELPERNOTFOUND $cmd
      fi
    done
}

# show help message
if test $# -ne 1 -o "$1" = "--help" 
then
  echo "Convert a $filetype file to HTML text for Recoll indexing."
  echo "Usage: $progname [infile]"
  exit 1
fi

infile="$1"

# check the input file existence (may be '-' for stdin)
if test "X$infile" != X- -a ! -f "$infile"
then
  senderror INPUTNOSUCHFILE "$infile"
fi

# protect access to our temp files and directories
umask 77

##############################################################################
# !! Leave the following line unmodified !
#ENDRECFILTCOMMONCODE

checkcmds awk antiword iconv

# output the result
# The strange 'BEGIN' setup is to prevent 'file' from thinking this file
# is an awk program
$decoder "$infile" |
awk 'BEGIN'\
' {
  print "<html><head><title></title>"
  print "<meta http-equiv=\"Content-Type\" content=\"text/html;charset=UTF-8\">"
  print "</head>\n<body>\n<p>"
  cont = ""
  gotdata = 0
}
{
  if (!($0 ~ /^[ 	]*$/))
    gotdata = 1
  $0 = cont $0
  cont = ""

  if ($0 ~ /[-]$/) {
    # Note : soft-hyphen is iso8859 0xad
    # Break at last whitespace
    match($0, "[ \t][^ \t]+$")
    line = substr($0, 0, RSTART)
    cont = substr($0, RSTART, RLENGTH-1)
    $0 = line
  }

  if($0 == "\f") {
    print "</p><hr><p>"
    next
  } 
  gsub(/&/, "\\&amp;", $0)
  gsub(/</, "\\&lt;", $0)
  gsub(/>/, "\\&gt;", $0)

  print $0 "<br>"
}
END {
    if (gotdata == 0)
       exit(1)
    print "</p></body></html>"
}' 

# wvWare is much slower and we don't use it by default, but it handles
# some files that antiword won't, so use it as a last resort. Yes the
# html document will have a second header section, but this doesn't
# seem to be an issue with our brave html input handler...
if test $? -eq 1 ; then
    if iscmd wvWare ; then
	wvWare --nographics --charset=utf-8 $infile
    fi
fi