tris/recoll
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

try to use wvWare if present and antiword fails

Browse Source
This commit is contained in:
dockes 2009-01-21 11:11:03 +00:00
parent e8d2885728
commit 876747da2a
1 changed files with 15 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
src/filters/rcldoc
View File

@ -28,8 +28,6 @@ decoder="antiword -t -i 1 -m UTF-8"
#decoder="wvWare -1 -c UTF-8" #decoder="wvWare -1 -c UTF-8"
#RECFILTCOMMONCODE #RECFILTCOMMONCODE
############################################################################## ##############################################################################
# !! Leave the previous line unmodified!! Code imported from the # !! Leave the previous line unmodified!! Code imported from the
@ -108,8 +106,11 @@ awk 'BEGIN'\
print "<meta http-equiv=\"Content-Type\" content=\"text/html;charset=UTF-8\">" print "<meta http-equiv=\"Content-Type\" content=\"text/html;charset=UTF-8\">"
print "</head>\n<body>\n<p>" print "</head>\n<body>\n<p>"
cont = "" cont = ""
gotdata = 0
} }
{ {
if (!($0 ~ /^[ ]*$/))
gotdata = 1
$0 = cont $0 $0 = cont $0
cont = "" cont = ""
@ -133,8 +134,17 @@ awk 'BEGIN'\
print $0 "<br>" print $0 "<br>"
} }
END { END {
if (gotdata == 0)
exit(1)
print "</p></body></html>" print "</p></body></html>"
}' | iconv -f UTF-8 -t UTF-8 -c -s }'
# exit normally # wvWare is much slower and we don't use it by default, but it handles
exit 0 # 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