Contents of /trunk/installer/doc/manual/scripts/merge_xml

#!/bin/bash

if [ "$1" = "--help" ]; then
    echo "Usage: $0 lang"
    exit 0
fi

language=${1:-en}

basedir="$(pwd)/$(dirname $0)"
TARGET="./integrated"

cd ./$language || exit 1

TEMPDIR=/tmp/merge_xml.$$
LOG=$TEMPDIR/merge_xml.$language.log
[[ -d $TEMPDIR ]] || mkdir $TEMPDIR

#<!ENTITY bookinfo.xml         SYSTEM "en/bookinfo.xml">
OLD_IFS=$IFS
IFS="
"
:>$TEMPDIR/entlist
for ENT in `grep "<!ENTITY" ../build/docstruct.ent` ; do
    echo -n "$(echo $ENT | sed "s/.*ENTITY\ *//" | sed "s/\ *SYSTEM.*$//")" >>$TEMPDIR/entlist
    echo -n ":" >>$TEMPDIR/entlist
    echo    "$(echo $ENT | sed "s/.*SYSTEM\ *\"en\///" | sed "s/\">//")" >>$TEMPDIR/entlist
    
done
IFS=$OLD_IFS

# Make sure that all files are in UTF-8 first
# This is only necessary while we're converting translated files
echo "Converting .xml files to UTF-8..." >>$LOG
for FILE in `find . -name "*.xml"` ; do
    DNAME=$(dirname $FILE); FNAME=$(basename $FILE)
    mkdir -p $TEMPDIR/in/$DNAME
    REGEXP="^<?.*encoding="
    if egrep -q $REGEXP $FILE ; then
        echo "Encoded    : $FILE" >>$LOG
        ENC=$(egrep $REGEXP $FILE | sed "s/.*xml.*encoding=\"//" | sed "s/\"?>//")
        iconv -f $ENC -t utf-8 $FILE | egrep -v $REGEXP >$TEMPDIR/in/$DNAME/$FNAME
    else
        echo "Not encoded: $FILE" >>$LOG
        cp $FILE $TEMPDIR/in/$DNAME/$FNAME
    fi
done
echo "" >>$LOG

# Include lower level xml-files for all the main level xml-files
echo "Merging .xml files per 'chapter'..." >>$LOG
awk -v TARGET="$TEMPDIR" -v LOG=$LOG -v ENTLIST="$TEMPDIR/entlist" \
    -f $basedir/merge_xml.awk ../build/install.${language}.xml

# Copy the results to their proper location
cd ..
TDIR="./integrated/$language"
[[ -d $TDIR ]] && rm $TDIR/*.xml || mkdir -p $TDIR
cp $TEMPDIR/*.xml $TARGET/$language
cp $LOG $TARGET/$language

rm -r $TEMPDIR
exit 0

1	#!/bin/bash
2
3	if [ "$1" = "--help" ]; then
4	echo "Usage: $0 lang"
5	exit 0
6	fi
7
8	language=${1:-en}
9
10	basedir="$(pwd)/$(dirname $0)"
11	TARGET="./integrated"
12
13	cd ./$language \|\| exit 1
14
15	TEMPDIR=/tmp/merge_xml.$$
16	LOG=$TEMPDIR/merge_xml.$language.log
17	[[ -d $TEMPDIR ]] \|\| mkdir $TEMPDIR
18
19	#<!ENTITY bookinfo.xml SYSTEM "en/bookinfo.xml">
20	OLD_IFS=$IFS
21	IFS="
22	"
23	:>$TEMPDIR/entlist
24	for ENT in `grep "<!ENTITY" ../build/docstruct.ent` ; do
25	echo -n "$(echo $ENT \| sed "s/.ENTITY\ //" \| sed "s/\ SYSTEM.$//")" >>$TEMPDIR/entlist
26	echo -n ":" >>$TEMPDIR/entlist
27	echo "$(echo $ENT \| sed "s/.SYSTEM\ \"en\///" \| sed "s/\">//")" >>$TEMPDIR/entlist
28
29	done
30	IFS=$OLD_IFS
31
32	# Make sure that all files are in UTF-8 first
33	# This is only necessary while we're converting translated files
34	echo "Converting .xml files to UTF-8..." >>$LOG
35	for FILE in `find . -name "*.xml"` ; do
36	DNAME=$(dirname $FILE); FNAME=$(basename $FILE)
37	mkdir -p $TEMPDIR/in/$DNAME
38	REGEXP="^<?.*encoding="
39	if egrep -q $REGEXP $FILE ; then
40	echo "Encoded : $FILE" >>$LOG
41	ENC=$(egrep $REGEXP $FILE \| sed "s/.xml.encoding=\"//" \| sed "s/\"?>//")
42	iconv -f $ENC -t utf-8 $FILE \| egrep -v $REGEXP >$TEMPDIR/in/$DNAME/$FNAME
43	else
44	echo "Not encoded: $FILE" >>$LOG
45	cp $FILE $TEMPDIR/in/$DNAME/$FNAME
46	fi
47	done
48	echo "" >>$LOG
49
50	# Include lower level xml-files for all the main level xml-files
51	echo "Merging .xml files per 'chapter'..." >>$LOG
52	awk -v TARGET="$TEMPDIR" -v LOG=$LOG -v ENTLIST="$TEMPDIR/entlist" \
53	-f $basedir/merge_xml.awk ../build/install.${language}.xml
54
55	# Copy the results to their proper location
56	cd ..
57	TDIR="./integrated/$language"
58	[[ -d $TDIR ]] && rm $TDIR/*.xml \|\| mkdir -p $TDIR
59	cp $TEMPDIR/*.xml $TARGET/$language
60	cp $LOG $TARGET/$language
61
62	rm -r $TEMPDIR
63	exit 0
64