mathml2tex

# Script to replace MathML with LaTeX
saxonpath="/var/www/html/george01/question/format/mapleta/SaxonHE"
mml2texpath="/var/www/html/george01/question/format/mapleta/mml2tex"

# Create copy of the original file with _tex appended to the file name
file="$1"
#if [[ $file =~ '/.*\.zip/' ]]; then zipped=true; fi
#echo $zipped

if [ ${file: -4} == ".zip" ]; then
  echo "Unzipping...";
  newfile=$(sed 's/\.zip/\.xml/g' <<< $file)
  unzip $file -d tmp
  file=$newfile
  mv "tmp/manifest.xml" $newfile
fi

prefile=$(sed 's/\.xml/_tidied\.xml/g' <<< $file) 
outputfile=$(sed 's/\.xml/_tex\.xml/g' <<< $file) 
echo "Output will be saved in $outputfile"
cp /dev/null $outputfile

#
# Process the original Maple TA XML file to prepare it for mml2tex
#

# Add entity declarations:
cat xmlentities.txt $file >> $outputfile

# Remove CDATA tags
sed -i "s/<text><\!\[CDATA\[/<text>/g" $outputfile
sed -i "s/\]\]><\/text>/<\/text>/g" $outputfile
sed -i "s/<comment><\!\[CDATA\[/<comment>/g" $outputfile
sed -i "s/\]\]><\/comment>/<\/comment>/g" $outputfile
sed -i "s/<choice><\!\[CDATA\[/<choice>/g" $outputfile
sed -i "s/\]\]><\/choice>/<\/choice>/g" $outputfile

# Rewrite <N> as <optN></optN>, N=1..10 (these are fill-in-the-blanks placeholders)
for i in `seq 1 10`;
do
sedarg="s/<$i>/<opt$i><\/opt$i>/g"
#echo $sedarg
	sed -i $sedarg $outputfile
done  

# Tidy up <br> tags to make them XML compliant
#first make any properly formatted <br/> tags into bad ones, and get rid of any </br>'s so we don't double-fix them
sed -i "s/<br\(.*\)\/>/<br \1 >/g" $outputfile
sed -i "s/<\/br>//g" $outputfile
#now fix all <br*> tags by making them <br*/>
sed -i "s/<br\([^<]*\)>/<br \1 \/>/g" $outputfile

# Also fix <hr> and <img> tags
sed -i "s/<hr\([^<]*\)>/<hr\/>/g" $outputfile
sed -i "s/<img\([^<]*\)>/<img \1 \/>/g" $outputfile

# remove bad MathML
sed -i "s/<mspace class=\"nbsp\"><\/mspace>//g" $outputfile

#remove <pre> and <script> tags which break some questions
sed -i "s/<pre>//g" $outputfile
sed -i "s/<\/pre>//g" $outputfile
sed -i "s/<script>//g" $outputfile
sed -i "s/<\/script>//g" $outputfile


cp $outputfile $prefile
mv $prefile tmp

#
# Use saxon to run mml2tex on the new file
#
echo "Replacing MathML with LaTeX..."
args="$saxonpath/saxon9he.jar -s:$outputfile -xsl:$mml2texpath/xsl/invoke-mml2tex.xsl -o:$outputfile"
java -jar $args

#
# Tidy up the mml2tex output
#

# Restore CDATA tags
sed -i "s/<text> /<text><\!\[CDATA\[ /g" $outputfile
sed -i "s/ <\/text>/ \]\]><\/text>/g" $outputfile
sed -i "s/<comment> /<comment><\!\[CDATA\[ /g" $outputfile
sed -i "s/ <\/comment>/ \]\]><\/comment>/g" $outputfile
sed -i "s/<choice> /<choice><\!\[CDATA\[ /g" $outputfile
sed -i "s/ <\/choice>/ \]\]><\/choice>/g" $outputfile
sed -i "s/<algorithm> /<algorithm><\!\[CDATA\[ /g" $outputfile
sed -i "s/ <\/algorithm>/ \]\]><\/algorithm>/g" $outputfile

cp $outputfile testing/geoseqser_prefinal.xml

# Restore <N> in place of <optN/>
for i in `seq 1 10`;
do
sedarg="s/<opt$i\/>/<$i>/g"
	sed -i $sedarg $outputfile
done 
echo "Complete."
echo ""