177 lines
4.4 KiB
Bash
Executable File
177 lines
4.4 KiB
Bash
Executable File
#!/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
|
||
#================================================================
|
||
#================================================================
|
||
# Extract text from an msword file by executing either antiword
|
||
# or 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"
|
||
|
||
|
||
|
||
#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
|
||
|
||
# We need to do some strange stuff to retrieve the status from antiword. Things
|
||
# would be simpler if we relied on using bash.
|
||
# Explanations:
|
||
#http://stackoverflow.com/questions/1221833/bash-pipe-output-and-capture-exit-status
|
||
|
||
stdintoexitstatus() {
|
||
read exitstatus
|
||
return $exitstatus
|
||
}
|
||
|
||
# The strange 'BEGIN' setup is to prevent 'file' from thinking this file
|
||
# is an awk program
|
||
(((($decoder "$infile"; echo $? >&3) |
|
||
awk 'BEGIN'\
|
||
' {
|
||
cont = ""
|
||
gotdata = 0
|
||
}
|
||
{
|
||
if (!($0 ~ /^[ ]*$/) && gotdata == 0) {
|
||
print "<html><head><title></title>"
|
||
print "<meta http-equiv=\"Content-Type\" content=\"text/html;charset=UTF-8\">"
|
||
print "</head>\n<body>\n<p>"
|
||
gotdata = 1
|
||
}
|
||
$0 = cont $0
|
||
cont = ""
|
||
|
||
if ($0 ~ /[<5B>-]$/) {
|
||
# 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
|
||
}
|
||
|
||
if (gotdata == 1) {
|
||
gsub(/&/, "\\&", $0)
|
||
gsub(/</, "\\<", $0)
|
||
gsub(/>/, "\\>", $0)
|
||
|
||
print $0 "<br>"
|
||
}
|
||
}
|
||
END {
|
||
if (gotdata == 1)
|
||
print "</p></body></html>"
|
||
}' >&4) 3>&1) | stdintoexitstatus) 4>&1
|
||
|
||
|
||
# Antiword rarely fails, we try to catch the most common reasons:
|
||
if test $? -eq 1 ; then
|
||
# Check actual document type
|
||
mtype=`file -b -i "$infile" | awk '{sub(";", "", $1);print $1}'`
|
||
|
||
if test X"$mtype" = Xtext/rtf; then
|
||
# RTF document disguising as msword either because it has a .doc
|
||
# extension or because it's an attachment with a wrong mime.
|
||
exec `dirname $0`/rclrtf "$infile"
|
||
fi
|
||
|
||
if test X"$mtype" = Xtext/plain; then
|
||
# Someone gave a .doc ext to their texts. Happens...
|
||
exec `dirname $0`/rcltext "$infile"
|
||
fi
|
||
|
||
if test X"$mtype" = Xapplication/msword; then
|
||
# Actually application/msword: try wvWare, which is much
|
||
# slower and we don't use it by default, but it handles some
|
||
# files that antiword won't, so use it as a last resort.
|
||
if iscmd wvWare ; then
|
||
exec wvWare --nographics --charset=utf-8 "$infile"
|
||
fi
|
||
fi
|
||
|
||
# else let the error be...
|
||
exit 1
|
||
fi
|