source: other-projects/maori-lang-detection/src/org/greenstone/atea/CountryCodeCountsMapData.java@ 33805

package org.greenstone.atea;

import java.io.BufferedReader;
import java.io.BufferedWriter;
import java.io.File;
import java.io.FileReader;
import java.io.FileWriter;
import java.io.Writer;

import java.util.HashMap;
import java.util.LinkedList;
import java.util.List;
import java.util.Map;

import org.apache.commons.csv.*;
import org.apache.log4j.Logger;

// Google's gson imports for parsing any kind of json
import com.google.gson.JsonArray;
import com.google.gson.JsonElement;
import com.google.gson.JsonObject;
import com.google.gson.JsonParser;

// For working with GeoJSON's Simple Features in Java
import mil.nga.sf.geojson.Feature;
import mil.nga.sf.geojson.FeatureCollection;
import mil.nga.sf.geojson.FeatureConverter;
import mil.nga.sf.geojson.Geometry;
import mil.nga.sf.geojson.MultiPoint;
import mil.nga.sf.geojson.Polygon;
import mil.nga.sf.geojson.Position;


/**
 * Run a mongodb query that produces counts per countrycode like in the following 2 examples:
 *
 * 1. count of country codes for all sites
 * db.Websites.aggregate([
 *
 *   { $unwind: "$geoLocationCountryCode" },
 *   {
 *       $group: {
 *           _id: "$geoLocationCountryCode",
 *           count: { $sum: 1 }
 *       }
 *   },
 *   { $sort : { count : -1} }
 * ]);
 * 
 * Then store the mongodb query result's JSON format output in a file called "counts.json".
 * Then run this program with counts.json as parameter
 * Copy the geojson output into http://geojson.tools/
 *
 * 2. count of country codes for sites that have at least one page detected as MRI 
 *
 * db.Websites.aggregate([
 *   {
 *       $match: {
 *           numPagesInMRI: {$gt: 0}
 *       }
 *   },
 *   { $unwind: "$geoLocationCountryCode" },
 *   {
 *       $group: {
 *           _id: {$toLower: '$geoLocationCountryCode'},
 *           count: { $sum: 1 }
 *       }
 *   },
 *   { $sort : { count : -1} }
 * ]);
 *
 * Store the mongodb query result's JSON format output in a file called "counts_sitesWithPagesInMRI.json".
 * Then run this program with counts_sitesWithPagesInMRI.json as parameter.
 * Copy the geojson output into http://geojson.tools/
 * 
 * ##################
 * TO COMPILE:
 *    maori-lang-detection/src$
 *       javac -cp ".:../conf:../lib/*" org/greenstone/atea/CountryCodeCountsMapData.java
 *
 * TO RUN:
 *    maori-lang-detection/src$
 *       java -cp ".:../conf:../lib/*" org/greenstone/atea/CountryCodeCountsMapData ../mongodb-data/countrycodes.json ../mongodb-data/counts.json
 *###################
 *
 * This class needs the gson library, and now the sf-geojson(-2.02).jar and 
 * helper jars sf(-2.02).jar and 3 jackson jars too,
 * to create and store Simple Features geo json objects with Java.
 * I copied the gson jar file from GS3.
 *
 * Simple Features GeoJSON Java
 * https://ngageoint.github.io/simple-features-geojson-java/ - liks to API and more
 *
 * https://mvnrepository.com/artifact/mil.nga.sf/sf-geojson (https://github.com/ngageoint/simple-features-geojson-java/)
 *
 * Also need the basic data types used by the Geometry objects above: 
 * https://mvnrepository.com/artifact/mil.nga/sf (https://github.com/ngageoint/simple-features-java)
 *
 * Further helper jars needed (because of encountering the exception documented at
 * stackoverflow.com/questions/36278293/java-lang-classnotfoundexception-com-fasterxml-jackson-core-jsonprocessingexcep/36279872)
 *   https://mvnrepository.com/artifact/com.fasterxml.jackson.core/jackson-core/2.10.0
 *   https://mvnrepository.com/artifact/com.fasterxml.jackson.core/jackson-databind
 *   https://mvnrepository.com/artifact/com.fasterxml.jackson.core/jackson-annotations/2.10.0
 */
public class CountryCodeCountsMapData {
    static Logger logger = Logger.getLogger(org.greenstone.atea.CountryCodeCountsMapData.class.getName());

    //Map<String, JsonObject> countryToJsonMap;
    JsonArray countryCodesJsonArray;
    JsonArray countryCountsJsonArray;

    // North-central Antarctica coords
    private final double ANTARCTICA_LNG = 57.0d;
    private final double ANTARCTICA_LAT = -70.0d;
    // For EU coords, spot in Atlantic Ocean close to western European coast.
    private final double EU_LNG = -20.0d;
    private final double EU_LAT = 50.0d;

    private final String geoJsonFilenameWithSuffix;
    private final File outputFolder;
    
    public CountryCodeCountsMapData(String countryCountsJSONFilename) throws Exception {

    // work out the unique filename we're going to save the geojson files under
    // and the folder we're going to save them into
    File countryCountsJSONFile = new File(countryCountsJSONFilename);
    String tailname = countryCountsJSONFile.getName();
    this.geoJsonFilenameWithSuffix = (tailname.startsWith("counts_")) ? tailname.substring("counts_".length()) : tailname;  
    this.outputFolder = countryCountsJSONFile.getParentFile().getCanonicalFile(); // canonical resolves any .. and . in path    
    
    // locate the countrycodes.json file
    File countryCoordsJSONFile = new File(this.getClass().getClassLoader().getResource("countrycodes.json").getFile());
    
    // Create a map of ALL country code names to ALL the country code json objects
    // that contain the location (lat, lng) info for each country code
    Map<String, JsonObject> countryToJsonMap = new HashMap<String, JsonObject>();

    // Parse json file of country codes and put into a JsonArray.
    // then put into map of each country code to its JsonObject.
    countryCodesJsonArray = parseJSONFile(countryCoordsJSONFile);
    for(JsonElement obj : countryCodesJsonArray) {
        JsonObject countryCodeJson = obj.getAsJsonObject();
        countryToJsonMap.put(countryCodeJson.get("country").getAsString(), countryCodeJson);
    }

    // Parse json file of country code counts
    // Then for each JsonObject in this file,
    // find a match on its country code in the map created above to get a country code JsonObject
    // Get the longitude and latitude of the JsonObject that matched that country code.
    // Add this lng,lat location information to the current JsonObject from the counts file.
    countryCountsJsonArray = parseJSONFile(countryCountsJSONFile);  

    for(JsonElement obj : countryCountsJsonArray) {
        JsonObject json = obj.getAsJsonObject();
        String countryCode = json.get("_id").getAsString().toUpperCase();
        // set the property back as uppercase and with property name "countrycode" instead of "_id"
        json.remove("_id");
        json.addProperty("countrycode", countryCode);
        
        int count = (int)json.get("count").getAsDouble();
        
        //logger.info("Got country code: " + countryCode);
        //logger.info("   count: " + count);

        // locate in countryCode map
        JsonObject countryCodeJson = countryToJsonMap.get(countryCode);

        if(countryCodeJson != null) {
        //logger.info("Found in map: " + countryCodeJson.toString());
        
        // for geojson, want longitude then latitude
        Double lng = countryCodeJson.get("longitude").getAsDouble();
        Double lat = countryCodeJson.get("latitude").getAsDouble();
        //logger.info("long: " + Double.toString(lng) + ", lat: " + Double.toString(lat));
        String countryName = countryCodeJson.get("name").getAsString();
        
        // let's add lat and lng fields to countryCounts object
        json.addProperty("lng", lng); // adds Number: https://javadoc.io/static/com.google.code.gson/gson/2.8.5/com/google/gson/JsonObject.html
        json.addProperty("lat", lat);
        json.addProperty("region", countryName);
        
        } else {
        logger.info("No geolocation info found for country code " + countryCode);
        if(countryCode.equals("EU")) {
            //logger.info("Unlisted country code: EU");
            // add lat and lng for Europe
            json.addProperty("lng", EU_LNG); 
            json.addProperty("lat", EU_LAT);
            json.addProperty("region", "Europe");
        }
        else if(countryCode.equals("UNKNOWN")) {
            //logger.info("Unlisted country code: UNKNOWN");
            // add lat and lng for Antarctica
            json.addProperty("lng", ANTARCTICA_LNG); 
            json.addProperty("lat", ANTARCTICA_LAT);
            json.addProperty("region", "UNKNOWN");
        } else {
            logger.error("ERROR: entirely unknown country code: " + countryCode);
        }
        }
    }

    }

    /** Convert mongodb tabular output of json records stored in the given file
     * into a JsonArray.
     */
    public JsonArray parseJSONFile(File file) throws Exception {
    JsonArray jsonArray = null;
    // read into string
    try (
         BufferedReader reader = new BufferedReader(new FileReader(file));
         ) {
        StringBuilder str = //new StringBuilder();
        new StringBuilder("[");
        String line;
        while((line = reader.readLine()) != null) {
        line = line.replaceAll("/\\* [^\\/]* \\*/", ""); // get rid of any multiline comments symbols on a single line
        str.append(line);
        if(line.endsWith("}")) {
            str.append(",\n");
        }
        }
        // replace last comma with closing bracket
        String fileContents = str.substring(0, str.length()-2) + "]";
        
        //System.err.println("Got file:\n" + fileContents);
        
        // https://stackoverflow.com/questions/2591098/how-to-parse-json-in-java
        jsonArray = new JsonParser().parse(fileContents).getAsJsonArray();    
        
    } catch(Exception e) {
        throw e;
    }
    

    return jsonArray;
    }

    /**
     * Reading
     * https://www.here.xyz/api/concepts/geojsonbasics/
     * https://ngageoint.github.io/simple-features-geojson-java/docs/api/
     *
     * https://stackoverflow.com/questions/55621480/cant-access-coordinates-member-of-geojson-feature-collection
     *
     * Downloaded geojson simple features' jar file from maven, but it didn't work:
     * a more private version of MultiPoint.java is not included in the jar file (there's only
     * mil.nga.sf.geojson.MultiPoint  , whereas
     * mil.nga.sf.MultiPoint          is missing
     * 
     * This seems to have gone wrong at
     * https://github.com/ngageoint/simple-features-geojson-java/tree/master/src/main/java/mil/nga/sf
     * but the one at
     * https://github.com/ngageoint/simple-features-java/tree/master/src/main/java/mil/nga/sf
     * has it. So I've been trying to build that, but don't have the correct version of maven.
     */
    public Geometry toMultiPointGeoJson() {
    //System.err.println("toGeoJSON() is not yet implemented.");

    List<Position> points = new LinkedList<Position>();
    
    for(JsonElement obj : this.countryCountsJsonArray) {
        JsonObject json = obj.getAsJsonObject();
        Double lng = json.get("lng").getAsDouble();
        Double lat = json.get("lat").getAsDouble();

        Position point = new Position(lng, lat);
        points.add(point);
    }
    
    Geometry multiPoint = new MultiPoint(points);
    
    return multiPoint;
    }

    // https://javadoc.io/static/com.google.code.gson/gson/2.8.5/index.html
    public FeatureCollection toFeatureCollection() {
    final int HISTOGRAM_WIDTH = 4;
    
    FeatureCollection featureCollection = new FeatureCollection();
    
    for(JsonElement obj : this.countryCountsJsonArray) {
        JsonObject json = obj.getAsJsonObject();
        
        String countryCode = json.get("countrycode").getAsString();
        String region = json.get("region").getAsString();
        int count = json.get("count").getAsInt();

        // make a histogram for each country
        Geometry rectangle = this.toPolygon(json, count, HISTOGRAM_WIDTH);

        Feature countryFeature = new Feature(rectangle);
        Map<String, Object> featureProperties = new HashMap<String, Object>();
        featureProperties.put("count", new Integer(count));
        featureProperties.put("code", countryCode);
        featureProperties.put("region", region);
        countryFeature.setProperties(featureProperties);

        featureCollection.addFeature(countryFeature);
    }

    return featureCollection;
    }

    // create rectangular "histogram" for each country code
    private Geometry toPolygon(JsonObject json, int count, int HISTOGRAM_WIDTH) {
    int half_width = HISTOGRAM_WIDTH/2;
    double vertical_factor = 1.0;
    
    Double lng = json.get("lng").getAsDouble();
    Double lat = json.get("lat").getAsDouble();

    String countryCode = json.get("countrycode").getAsString();


    //create the 4 corners of the rectangle
    // West is negative, east is positive, south is negative, north is positive
    // See http://www.learnz.org.nz/sites/learnz.org.nz/files/lat-long-geo-data-01_0.jpg
    // But since the histograms grow vertically/northwards and we can't go past a latitude of 90,
    // to compensate, we increase the width of the histograms by the same factor as our inability
    // to grow northwards.
    Double north = lat + (vertical_factor * count);
    while (north > 90) {
        // recalculate north after decreasing histogram's vertical growth
        // by the same factor as we increase its width
        vertical_factor = vertical_factor/2.0; 
        half_width = 2 * half_width;
        north = lat + (vertical_factor * count);
    }
    Double east = lng + half_width;
    Double west = lng - half_width;
    Double south = lat;
    
    List<List<Position>> outerList = new LinkedList<List<Position>>();
    List<Position> points = new LinkedList<Position>(); 
    outerList.add(points);

    
    points.add(new Position(west, south)); // Position(lng, lat) not Position(lat, lng)
    points.add(new Position(west, north));
    points.add(new Position(east, north));
    points.add(new Position(east, south));


    Geometry rectangle = new Polygon(outerList);
    
    // Coords: a List of List of Positions, see https://ngageoint.github.io/simple-features-geojson-java/docs/api/
    // https://www.here.xyz/api/concepts/geojsonbasics/#polygon

    return rectangle;
    }
    
    public String writeMultiPointGeoJsonToFile() {
    final String filename = "multipoint_" + this.geoJsonFilenameWithSuffix;
    File outFile = new File(this.outputFolder, filename);

    Geometry geometry = this.toMultiPointGeoJson();
    String multiPointGeojsonString = FeatureConverter.toStringValue(geometry);
    System.err.println("\nMap data as MultiPoint geometry:\n" + multiPointGeojsonString + "\n");
    try (
         Writer writer = new BufferedWriter(new FileWriter(outFile));
         ) {

        // Some basic re-formatting for some immediate legibility
        // But pasting the contents of the file (or the System.err output above)
        // directly into http://geojson.tools/ will instantly reformat the json perfectly anyway.
        multiPointGeojsonString = multiPointGeojsonString.replace("[[", "\n[\n\t[");
        multiPointGeojsonString = multiPointGeojsonString.replace("],[", "],\n\t[");
        multiPointGeojsonString = multiPointGeojsonString.replace("]]", "]\n]");
        
        writer.write(multiPointGeojsonString + "\n");
    } catch(Exception e) {
        logger.error("Unable to write multipoint geojson:\n**********************");
        logger.error(multiPointGeojsonString);
        logger.error("**********************\ninto file " + outFile.getAbsolutePath());
        logger.error(e.getMessage(), e);
    }

    return outFile.getAbsolutePath();
    
    }
    
    public String writeFeaturesGeoJsonToFile() {
    final String filename = "geojson-features_" + this.geoJsonFilenameWithSuffix;
    File outFile = new File(this.outputFolder, filename);

    FeatureCollection featureColl = this.toFeatureCollection();
    String featuresGeojsonString = FeatureConverter.toStringValue(featureColl);
    System.err.println("\nMap data as featurecollection:\n" + featuresGeojsonString + "\n");
    try (
         Writer writer = new BufferedWriter(new FileWriter(outFile));
         ) {
        
        writer.write(featuresGeojsonString + "\n");
    } catch(Exception e) {
        logger.error("Unable to write multipoint geojson:\n**********************");
        logger.error(featuresGeojsonString);
        logger.error("**********************\ninto file " + outFile.getAbsolutePath());
        logger.error(e.getMessage(), e);
    }

    return outFile.getAbsolutePath();
    
    }

    
    public int getTotalCount() {
    int total = 0;
    for(JsonElement obj : this.countryCountsJsonArray) {
        JsonObject json = obj.getAsJsonObject();
        int count = json.get("count").getAsInt();
        total += count;     
    }
    return total;
    }

    
    // Unfinished and unused
    public void parseCSVFile(String filename) throws Exception {
    File csvData = new File(filename);
    CSVParser parser = CSVParser.parse(csvData, java.nio.charset.Charset.forName("US-ASCII"), CSVFormat.RFC4180);
    for (CSVRecord csvRecord : parser) {
        logger.info("Got record: " + csvRecord.toString());
    }
    }
    
    public static void printUsage() {
    System.err.println("CountryCodeCountsMapData <counts-by-countrycode-file>.json");
    }
    
    public static void main(String args[]) {
    if(args.length != 1) {
        printUsage();
        System.exit(-1);
    }
    
    try {
        File countsFile = new File(args[0]);
        
        CountryCodeCountsMapData mapData = new CountryCodeCountsMapData(args[0]);

        String multipointOutFileName = mapData.writeMultiPointGeoJsonToFile();      
        String featuresOutFileName = mapData.writeFeaturesGeoJsonToFile();
        
        System.err.println("***********\nWrote mapdata to files " + multipointOutFileName
                   + " and " + featuresOutFileName);
        System.err.println("You can paste the geojson contents of either of these files into the "
                   + "editor at http://geojson.tools/ to see the data arranged on a world map");

        System.err.println("Total count for query: " + mapData.getTotalCount());
        
    } catch(Exception e) {
        logger.error(e.getMessage(), e);
    }
    }
}