source: CLARIN_KML/exec/generate_CLARIN_KMZ.sh @ 2044

#!/bin/bash

###
###                     Central script to convert CLARIN centers CSV file to Google Earth KMZ file.
###

bold=$(tput bold);
normal=$(tput sgr0);

# As published on Stackoverflow
function robust_directory_of_origin()
{
        SOURCE="${BASH_SOURCE[0]}";
        DIR="$( dirname "$SOURCE" )";
        while [ -h "$SOURCE" ]
        do 
          SOURCE="$(readlink "$SOURCE")";
          [[ $SOURCE != /* ]] && SOURCE="$DIR/$SOURCE"
          DIR="$( cd -P "$( dirname "$SOURCE"  )" && pwd )";
        done
        DIR="$( cd -P "$( dirname "$SOURCE" )" && pwd )";

        echo "${DIR}";
}

CSV_file_URL='http://www.clarin.eu/export_organisations_coordinates';
original_KMZ_file_URL='http://www.mpi.nl/DOBES/dobesmap/language_sites.kmz';

ROOT_directory_path="$(robust_directory_of_origin)/../"; # "$(realpath $(dirname ..) )/"; # X- $0
INPUT_directory_path="${ROOT_directory_path}/.input/";
EXEC_directory_path="${ROOT_directory_path}/exec/";
OUTPUT_directory_path="${ROOT_directory_path}/.output/";
SAXON_directory_path="${EXEC_directory_path}/"; # X- SaxonHE9-4-0-4J/

_SAXON_JAR_file_path="${SAXON_directory_path}/saxon9he.jar";
SAXON_JAR_file_path="$(realpath $_SAXON_JAR_file_path)";

XSLT_1_file_path="${EXEC_directory_path}/XMLdb_to_KML.xsl";
XSLT_2_file_path="${EXEC_directory_path}/KML_original_to_final.xsl";
XSLT_3_file_path="${EXEC_directory_path}/replicate.xsl";
CSV2XML_file_path="${EXEC_directory_path}/csv2xml";

KMZ_original_file_path="${INPUT_directory_path}/language_sites.kmz";
CSV_file_path="${INPUT_directory_path}/export_organisations_coordinates";

KML_original_file_path="${INPUT_directory_path}/doc.kml";
XML_of_CSV_file_path="${OUTPUT_directory_path}/organisations_coordinates.xml";
KML_chunk_file_path="${OUTPUT_directory_path}/KML_chunk.xml";

KML_final_file_path="${OUTPUT_directory_path}/doc.kml";
KMZ_final_file_path="${OUTPUT_directory_path}/language_sites.kmz";
KMZ_publication_directory_path='/var/www/drupal/external/kml/';

timestamp=$( date +%H:%M:%S );

echo -e "\n\n${bold}Central script to convert CLARIN centers CSV file to Google Earth KMZ file. ($timestamp)${normal}";
echo "ROOT: ${ROOT_directory_path}";
echo "Saxon JAR: ${SAXON_JAR_file_path}";

mkdir "${INPUT_directory_path}" 2> '/dev/null';
mkdir "${OUTPUT_directory_path}" 2> '/dev/null';

echo -e "\n\n${bold}Downloading CSV file from "${CSV_file_URL}" to "${CSV_file_path}" ...${normal}";
wget -O - "${CSV_file_URL}" > "${CSV_file_path}";

if [ $? -ne 0 ]
then
        echo "ERROR: the download of the CSV file failed!";
        exit 1;
fi

echo -e "\n\n${bold}Downloading original KMZ file from "${original_KMZ_file_URL}" to "${KMZ_original_file_path}" ...${normal}";
wget -O - "${original_KMZ_file_URL}" > "${KMZ_original_file_path}";

if [ $? -ne 0 ]
then
        echo "ERROR: the download of the original KMZ file failed!";
        exit 1;
fi

if test -r "${OUTPUT_directory_path}" -a -r "${SAXON_JAR_file_path}" -a -x "${CSV2XML_file_path}" -a -r "${XSLT_1_file_path}" -a -r "${XSLT_1_file_path}" -a -r "${XSLT_2_file_path}" -a -r "${XSLT_3_file_path}" -a -r "${KMZ_publication_directory_path}"
then
        if test -r "${KMZ_original_file_path}" -a -r "${CSV_file_path}"
        then
                echo -e "\n\n${bold}Converting CSV file to an XML representation ...${normal}";

                # X- Remove this <table> root element hack.
                echo -e "<table>\n"                                                     > "${XML_of_CSV_file_path}";
                < "${CSV_file_path}" "${CSV2XML_file_path}"     >> "${XML_of_CSV_file_path}";
                echo -e "\n</table>\n"                                                  >> "${XML_of_CSV_file_path}";

                if [ $? -ne 0 ]
                then
                        echo "ERROR: csv2xml command exited with an error!";
                        exit 1;
                fi

                echo -e "\n\n${bold}Transforming XML representation of CSV in ${XML_of_CSV_file_path} to a KML folder XML chunk ${KML_chunk_file_path} ...${normal}";
                java -jar "${SAXON_JAR_file_path}" -a:off -t -xi:off -s:"${XML_of_CSV_file_path}" -xsl:"${XSLT_1_file_path}" -o:"${KML_chunk_file_path}";


                echo -e "\n\n${bold}Unzipping doc.kml from KMZ file in ${KMZ_original_file_path} ...${normal}";
                unzip "${KMZ_original_file_path}" -d "${INPUT_directory_path}";

                echo -e "\n\n${bold}Including generated XML chunk (a KML folder) in the main KML file ...${normal}";
                java -jar "${SAXON_JAR_file_path}" -warnings:fatal -a:off -t -xi:off -s:"${KML_original_file_path}" -xsl:"${XSLT_2_file_path}" -o:"${KML_final_file_path}" KML_chunk_file_path="${KML_chunk_file_path}";
                # Another run, just to resolve the XInclude.
                java -jar "${SAXON_JAR_file_path}" -warnings:fatal -a:off -t -xi:on -xsl:"${XSLT_3_file_path}" -s:"${KML_final_file_path}" -o:"${KML_final_file_path}";
                #<"${KML_final_file_path}" xsltproc --xinclude -o "${KML_final_file_path}"


                if [[ $( grep '<xi:include' -F -l "${KML_final_file_path}" ) = "${KML_final_file_path}" ]]
                then
                        echo "${bold}ERROR: Found one or more string occurrences suggestive of residual unresolved XInclude element in ${KML_final_file_path}!${normal}";
                        exit 3;
                fi

                echo -e "\n\n${bold}Zipping ${KML_final_file_path} to KMZ file ${KMZ_final_file_path} and moving it to "${KMZ_publication_directory_path}" ...${normal}";
                cd $(dirname "${KML_final_file_path}");
                KML_final_file_relpath=$(basename "${KML_final_file_path}");

                zip -9 -D "${KMZ_final_file_path}" "${KML_final_file_relpath}";

                rm "${KML_final_file_relpath}";
                rm "${KML_original_file_path}";

                cp -v --remove-destination "${KMZ_final_file_path}" -t "${KMZ_publication_directory_path}/";
        else
                echo -e "ERROR: The input data file(s) ${KMZ_original_file_path}\nand/or\n${CSV_file_path} is/are not existent as well as readable/executable!";
                exit 2;
        fi
else
        echo -e "ERROR: One or more of the following script files is not existent as well as readable: \n${OUTPUT_directory_path}\n${SAXON_JAR_file_path}\n${CSV2XML_file_path}\n${XSLT_1_file_path}\n${XSLT_1_file_path}\n${XSLT_2_file_path}\n${XSLT_3_file_path}\n${KMZ_publication_directory_path}";
        exit 2;
fi

exit 0;