| 1 |
#!/bin/bash
|
| 2 |
|
| 3 |
# This script is used for translations using .po files.
|
| 4 |
# It merges .xml files per chapter (or appendix).
|
| 5 |
# The reason files are merged is that individual .xml files
|
| 6 |
# are often not well-formed xml.
|
| 7 |
|
| 8 |
if [ "$1" = "--help" ]; then
|
| 9 |
echo "Usage: $0 lang"
|
| 10 |
exit 0
|
| 11 |
fi
|
| 12 |
|
| 13 |
language=${1:-pl}
|
| 14 |
|
| 15 |
SCRIPTDIR="./scripts"
|
| 16 |
BUILDDIR="./build"
|
| 17 |
|
| 18 |
LANGDIR="./$language"
|
| 19 |
[ -d $LANGDIR ] || exit 1
|
| 20 |
|
| 21 |
if [ -z "$PO_USEBUILD" ] ; then
|
| 22 |
WORKDIR="./integrated"
|
| 23 |
else
|
| 24 |
WORKDIR="$BUILDDIR/build.po"
|
| 25 |
fi
|
| 26 |
[ -d $WORKDIR ] || mkdir -p $WORKDIR
|
| 27 |
|
| 28 |
TEMPDIR=/tmp/merge_xml.$$
|
| 29 |
LOG=$TEMPDIR/merge_xml.$language.log
|
| 30 |
[ -d $TEMPDIR ] || mkdir $TEMPDIR
|
| 31 |
|
| 32 |
#<!ENTITY bookinfo.xml SYSTEM "en/bookinfo.xml">
|
| 33 |
OLD_IFS=$IFS
|
| 34 |
IFS="
|
| 35 |
"
|
| 36 |
:>$TEMPDIR/entlist
|
| 37 |
echo "Building list of entities..."
|
| 38 |
for ENT in `grep "<!ENTITY" $BUILDDIR/templates/docstruct.ent` ; do
|
| 39 |
echo -n "$(echo $ENT | sed "s/.*ENTITY\ *//" | sed "s/\ *SYSTEM.*$//")" >>$TEMPDIR/entlist
|
| 40 |
echo -n ":" >>$TEMPDIR/entlist
|
| 41 |
echo "$(echo $ENT | sed "s/.*SYSTEM\ *\"##SRCPATH##\///" | sed "s/\">//")" >>$TEMPDIR/entlist
|
| 42 |
|
| 43 |
done
|
| 44 |
IFS=$OLD_IFS
|
| 45 |
|
| 46 |
# Make sure that all files are in UTF-8 first
|
| 47 |
echo "Converting XML files to UTF-8..."
|
| 48 |
echo "Converting XML files to UTF-8..." >>$LOG
|
| 49 |
for FILE in `find $LANGDIR -name "*.xml"` ; do
|
| 50 |
SUBDIR=$(dirname $FILE | sed "s:$LANGDIR::");
|
| 51 |
XML=$(basename $FILE)
|
| 52 |
mkdir -p $TEMPDIR/in/$SUBDIR
|
| 53 |
REGEXP="^<?.*encoding="
|
| 54 |
if egrep -q $REGEXP $FILE ; then
|
| 55 |
echo "Encoded : $FILE" >>$LOG
|
| 56 |
ENC=$(egrep $REGEXP $FILE | sed "s/.*xml.*encoding=\"//" | sed "s/\"?>//")
|
| 57 |
iconv -f $ENC -t utf-8 $FILE | egrep -v $REGEXP >$TEMPDIR/in/$SUBDIR/$XML
|
| 58 |
else
|
| 59 |
echo "Not encoded: $FILE" >>$LOG
|
| 60 |
cp $FILE $TEMPDIR/in/$SUBDIR/$XML
|
| 61 |
fi
|
| 62 |
done
|
| 63 |
echo "" >>$LOG
|
| 64 |
|
| 65 |
# Include lower level xml-files for all the main level xml-files
|
| 66 |
echo "Merging XML files per 'chapter'..."
|
| 67 |
echo "Merging XML files per 'chapter'..." >>$LOG
|
| 68 |
gawk -v WORKDIR="$TEMPDIR" -v LOG=$LOG -v ENTLIST="$TEMPDIR/entlist" \
|
| 69 |
-f $SCRIPTDIR/merge_xml.awk $BUILDDIR/templates/install.xml.template
|
| 70 |
|
| 71 |
# Copy the results to their proper location
|
| 72 |
TARGET="$WORKDIR/$language"
|
| 73 |
if [ -d $TARGET ]; then
|
| 74 |
rm -r $TARGET
|
| 75 |
fi
|
| 76 |
mkdir -p $TARGET
|
| 77 |
cp -r $TEMPDIR/out/* $TARGET
|
| 78 |
cp $LOG $TARGET
|
| 79 |
|
| 80 |
rm -r $TEMPDIR
|
| 81 |
exit 0
|