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

use a temporary symbolic link to avoid djvu issues with binary path names

Browse Source
This commit is contained in:
Jean-Francois Dockes 2010-04-16 15:55:29 +02:00
parent b703332324
commit 6f13c8c5f7
1 changed files with 27 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

34
src/filters/rcldjvu
View File

@ -25,17 +25,18 @@
# Of course this only means anything if the djvu document actually has # Of course this only means anything if the djvu document actually has
# a text layer ! # a text layer !
# #
# djvu utilities (04-2010) have a bug in which they try to interpret
# and convert file paths as character data, and fail miserably if the
# locale is not consistent with the actual encoding of the path (which
# could be arbitrary binary for all they know). We use a temporary
# symbolic link to get around this.
#
#================================================================ #================================================================
LANG=C ; export LANG
LC_ALL=C ; export LC_ALL
progname="rcldjvu" progname="rcldjvu"
filetype=dejavu filetype=dejavu
#RECFILTCOMMONCODE #RECFILTCOMMONCODE
############################################################################## ##############################################################################
# !! Leave the previous line unmodified!! Code imported from the # !! Leave the previous line unmodified!! Code imported from the
@ -104,11 +105,30 @@ umask 77
checkcmds djvutxt djvused awk checkcmds djvutxt djvused awk
# We need a temporary symlink to avoid path encoding issues
if test z"$RECOLL_TMPDIR" != z; then
ttdir=$RECOLL_TMPDIR
elif test z"$TMPDIR" != z ; then
ttdir=$TMPDIR
else
ttdir=/tmp
fi
tmplink=$ttdir/rcldjvu_tmp$$.djvu
rm -f $tmplink
ln -s $infile $tmplink || exit 1
cleanup()
{
rm -f $tmplink
}
trap cleanup EXIT HUP QUIT INT TERM
# Title: we try to extract it from the annotations. djvused outputs string # Title: we try to extract it from the annotations. djvused outputs string
# in C/awk \-escaped notation. Awk can only process this in string # in C/awk \-escaped notation. Awk can only process this in string
# constants, so we have a first awk pass to create an awk program to parse # constants, so we have a first awk pass to create an awk program to parse
# the string as a constant (...). This is not exactly robust or nice # the string as a constant (...). This is not exactly robust or nice
title=`djvused "$infile" -e 'select 1;output-ant' | \ title=`djvused "$tmplink" -e 'select 1;output-ant' | \
grep ' (title ' | sed -e 's/^.* (title //' -e 's/)$//' |\ grep ' (title ' | sed -e 's/^.* (title //' -e 's/)$//' |\
awk ' awk '
{ {
@ -128,7 +148,7 @@ EOF
# The strange 'BEGIN' setup is to prevent 'file' from thinking this file # The strange 'BEGIN' setup is to prevent 'file' from thinking this file
# is an awk program # is an awk program
djvutxt "$infile" | sed -e 's/[ ][ ]*$//' | \ djvutxt "$tmplink" | sed -e 's/[ ][ ]*$//' | \
awk 'BEGIN'\ awk 'BEGIN'\
' { ' {
cont = "" cont = ""