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

powerpoint: decide to use unoconv based on the number of lines in catppt output

Browse Source
This commit is contained in:
Jean-Francois Dockes 2013-11-12 10:40:07 +01:00
parent a9358d2f03
commit 134153e412
1 changed files with 63 additions and 51 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

114
src/filters/rclppt
View File

@ -30,8 +30,6 @@ LC_ALL=C ; export LC_ALL
progname="rclppt"
filetype=powerpoint
RCLPPT_CATPPT=${RCLPPT_CATPPT:=yes}
#RECFILTCOMMONCODE
##############################################################################
# !! Leave the previous line unmodified!! Code imported from the
@ -98,54 +96,68 @@ umask 77
# !! Leave the following line unmodified !
#ENDRECFILTCOMMONCODE
if test X$RCLPPT_CATPPT = Xyes ; then
checkcmds catppt
havecappt=no
iscmd cappt && havecappt=yes
haveunoconv=no
iscmd unoconv && haveunoconv=yes
iscmd pdftotext || haveunoconv=no
# output the result
echo '<html><head>'
#echo '<title>' "$title" '</title>'
echo '<meta http-equiv="Content-Type" content="text/html;charset=UTF-8">'
echo '</head><body>'
echo '<pre>'
catppt -d utf-8 "$infile" | \
sed -e 's/</&lt;/g' -e 's/&/&amp;/g'
echo '</pre>'
echo '</body></html>'
# exit normally
exit 0
else
# Using unoconv
checkcmds unoconv pdftotext
# This needs a temp dir because we first output pdf (outputting html
# would produce one file per page), and pdftotext can't read from
# stdin
if test z"$RECOLL_TMPDIR" != z; then
ttdir=$RECOLL_TMPDIR
elif test z"$TMPDIR" != z ; then
ttdir=$TMPDIR
else
ttdir=/tmp
fi
tmpdir=$ttdir/rclppt_tmp$$
mkdir $tmpdir || exit 1
mkdir $tmpdir/rclppttmp || exit 1
unopdf=$tmpdir/rclppttmp/output.pdf
cleanup()
{
# Note that we're using a constant part (rclkwdtmp), that hopefully
# guarantees that we can't do big mistakes here.
rm -rf $tmpdir/rclppttmp
rmdir $tmpdir
}
trap cleanup EXIT HUP QUIT INT TERM
unoconv -f pdf -o $unopdf "$infile"
`dirname $0`/rclpdf $unopdf
if test X$havecatppt = Xno -a X$haveunoconv = Xno ; then
# checkcmds will exit with the appropriate salutations
checkcmds catppt unoconv pdftotext
fi
# This needs a temp dir because we first output pdf (outputting html
# would produce one file per page), and pdftotext can't read from
# stdin
if test z"$RECOLL_TMPDIR" != z; then
ttdir=$RECOLL_TMPDIR
elif test z"$TMPDIR" != z ; then
ttdir=$TMPDIR
else
ttdir=/tmp
fi
tmpdir=$ttdir/rclppt_tmp$$
mkdir $tmpdir || exit 1
mkdir $tmpdir/rclppttmp || exit 1
unopdf=$tmpdir/rclppttmp/output.pdf
cattxt=$tmpdir/rclppttmp/output.txt
cleanup()
{
# Note that we're using a constant part (rclkwdtmp), that hopefully
# guarantees that we can't do big mistakes here.
rm -rf $tmpdir/rclppttmp
rmdir $tmpdir
}
trap cleanup EXIT HUP QUIT INT TERM
# Try catppt. If the output looks too small and unoconv is available, use this
# instead. unoconv is very slow but it handles newer files that catppt will
# not convert.
#
# I'm not sure of the right test for detecting catppt failure. On the
# sample I have, it outputs Azure\n1_Azure\n\n. I don't know if Azure
# is a good marker of failure. Anyway, it seems unlikely that a real
# ppt would have fewer than 5 lines
catppt -d utf-8 "$infile" > $cattxt
lines=`wc -l < $cattxt`
if test $lines -lt 5 -a X$haveunoconv = Xyes; then
unoconv -f pdf -o $unopdf "$infile"
`dirname $0`/rclpdf $unopdf
else
# output the catppt result
echo '<html><head>'
#echo '<title>' "$title" '</title>'
echo '<meta http-equiv="Content-Type" content="text/html;charset=UTF-8">'
echo '</head><body>'
echo '<pre>'
catppt -d utf-8 "$infile" | \
sed -e 's/</&lt;/g' -e 's/&/&amp;/g' < $cattxt
echo '</pre>'
echo '</body></html>'
fi