teichmann@8025: /* Copyright (C) 2011, 2012, 2013 by Bundesanstalt für Gewässerkunde
teichmann@8025:  * Software engineering by Intevation GmbH
teichmann@8025:  *
teichmann@8025:  * This file is Free Software under the GNU AGPL (>=v3)
teichmann@8025:  * and comes with ABSOLUTELY NO WARRANTY! Check out the
teichmann@8025:  * documentation coming with Dive4Elements River for details.
teichmann@8025:  */
teichmann@8025: 
teichmann@8025: package org.dive4elements.river.importer.parsers;
teichmann@8025: 
teichmann@8025: import java.io.File;
teichmann@8025: import java.io.IOException;
teichmann@8025: 
teichmann@8025: import java.text.NumberFormat;
teichmann@8025: import java.text.ParseException;
teichmann@8025: 
teichmann@8025: import java.util.ArrayList;
teichmann@8025: import java.util.List;
teichmann@8025: import java.util.regex.Matcher;
teichmann@8025: import java.util.regex.Pattern;
teichmann@8025: 
teichmann@8025: import org.apache.log4j.Logger;
teichmann@8025: 
tom@8032: import org.dive4elements.river.importer.ImporterSession;
teichmann@8025: import org.dive4elements.river.importer.ImportGrainFraction;
teichmann@8031: import org.dive4elements.river.importer.ImportSedimentLoadLS;
teichmann@8025: import org.dive4elements.river.importer.ImportSedimentLoadLSValue;
teichmann@8025: import org.dive4elements.river.importer.ImportTimeInterval;
teichmann@8025: import org.dive4elements.river.importer.ImportUnit;
tom@8032: 
teichmann@8025: import org.dive4elements.river.model.GrainFraction;
tom@8032: 
teichmann@8025: import org.dive4elements.river.utils.DateUtil;
tom@8032: import org.dive4elements.river.utils.EpsilonComparator;
teichmann@8025: 
tom@8032: /** Parses sediment load longitudinal section files. */
teichmann@8025: public class SedimentLoadLSParser extends LineParser {
teichmann@8025: 
teichmann@8025:     private static final Logger log =
teichmann@8025:         Logger.getLogger(SedimentLoadLSParser.class);
teichmann@8025: 
teichmann@8025: 
teichmann@8025:     public static final NumberFormat nf = NumberFormat.getInstance(DEFAULT_LOCALE);
teichmann@8025: 
teichmann@8025: 
teichmann@8025:     public static final Pattern TIMEINTERVAL_SINGLE =
teichmann@8025:         Pattern.compile("\\D*([0-9]+?)\\D*");
teichmann@8025: 
teichmann@8025:     public static final Pattern TIMEINTERVAL_EPOCH =
teichmann@8025:         Pattern.compile("\\D*([0-9]+?)\\s*-\\s*([0-9]+?)\\D*");
teichmann@8025: 
teichmann@8025:     public static final Pattern META_FRACTION =
teichmann@8025:         Pattern.compile("^Fraktion: (.*)");
teichmann@8025: 
tom@8032:     public static final Pattern META_FRACTION_NAME =
tom@8032:         Pattern.compile("^Fraktionsname: (.*)");
tom@8032: 
teichmann@8025:     public static final Pattern META_UNIT =
teichmann@8025:         Pattern.compile("^Einheit: \\[(.*)\\].*");
teichmann@8025: 
teichmann@8025:     public static final Pattern META_COLUMN_NAMES =
teichmann@8025:         Pattern.compile("^Fluss-km.*");
teichmann@8025: 
tom@8032:     public static final Pattern META_GRAIN_SIZE =
tom@8032:         Pattern.compile("([0-9]*,*[0-9]+)-([0-9]*,*[0-9]+) *mm");
teichmann@8025: 
teichmann@8025: 
tom@8032:     protected List<ImportSedimentLoadLS> sedimentLoadLSs;
teichmann@8025: 
teichmann@8031:     protected ImportSedimentLoadLS[] current;
teichmann@8025: 
teichmann@8025:     protected ImportGrainFraction grainFraction;
teichmann@8025: 
teichmann@8025:     protected ImportUnit unit;
teichmann@8025: 
teichmann@8025:     protected String description;
teichmann@8025: 
teichmann@8025:     protected String[] columnNames;
teichmann@8025: 
tom@8032:     private String upper;
tom@8032: 
tom@8032:     private String lower;
tom@8032: 
teichmann@8025: 
teichmann@8025:     public SedimentLoadLSParser() {
tom@8032:         sedimentLoadLSs = new ArrayList<ImportSedimentLoadLS>();
teichmann@8025:     }
teichmann@8025: 
teichmann@8025: 
teichmann@8025:     @Override
teichmann@8025:     public void parse(File file) throws IOException {
teichmann@8025:         description = file.getName();
teichmann@8025: 
teichmann@8025:         super.parse(file);
teichmann@8025:     }
teichmann@8025: 
teichmann@8025: 
teichmann@8025:     @Override
teichmann@8025:     protected void reset() {
teichmann@8025:         current       = null;
teichmann@8025:         grainFraction = null;
teichmann@8025:         unit          = null;
teichmann@8025:     }
teichmann@8025: 
teichmann@8025: 
teichmann@8025:     @Override
teichmann@8025:     protected void finish() {
teichmann@8025:         if (current != null) {
teichmann@8031:             for (ImportSedimentLoadLS isy: current) {
tom@8032:                 sedimentLoadLSs.add(isy);
teichmann@8025:             }
teichmann@8025:         }
teichmann@8025: 
teichmann@8025:         description = null;
teichmann@8025:     }
teichmann@8025: 
teichmann@8025: 
teichmann@8025:     @Override
teichmann@8042:     protected void handleLine(int lineNum, String line) throws LineParserException {
teichmann@8025:         if (line.startsWith(START_META_CHAR)) {
teichmann@8025:             handleMetaLine(stripMetaLine(line));
teichmann@8025:         }
teichmann@8025:         else {
teichmann@8025:             handleDataLine(line);
teichmann@8025:         }
teichmann@8025:     }
teichmann@8025: 
teichmann@8025: 
teichmann@8042:     protected void handleMetaLine(String line) throws LineParserException {
teichmann@8025:         if (handleMetaUnit(line)) {
teichmann@8025:             return;
teichmann@8025:         }
teichmann@8042:         if (handleMetaFraction(line)) {
tom@8032:             return;
tom@8032:         }
teichmann@8042:         if (handleMetaFractionName(line)) {
teichmann@8025:             return;
teichmann@8025:         }
teichmann@8042:         if (handleColumnNames(line)) {
teichmann@8042:             return;
teichmann@8025:         }
teichmann@8042:         log.warn("SLLSP: Unknown meta line: '" + line + "'");
teichmann@8025:     }
teichmann@8025: 
teichmann@8025: 
teichmann@8025:     protected boolean handleMetaUnit(String line) {
teichmann@8025:         Matcher m = META_UNIT.matcher(line);
teichmann@8025: 
teichmann@8025:         if (m.matches()) {
teichmann@8025:             unit = new ImportUnit(m.group(1));
teichmann@8025:             return true;
teichmann@8025:         }
teichmann@8025: 
teichmann@8025:         return false;
teichmann@8025:     }
teichmann@8025: 
teichmann@8025: 
teichmann@8025:     public boolean handleMetaFraction(String line) {
teichmann@8025:         Matcher m = META_FRACTION.matcher(line);
teichmann@8025: 
teichmann@8025:         if (m.matches()) {
tom@8032:             String interval = m.group(1);
teichmann@8025: 
tom@8032:             Matcher sizes = META_GRAIN_SIZE.matcher(interval);
tom@8032:             if (sizes.matches()) {
tom@8032:                 lower = sizes.group(1);
tom@8032:                 upper = sizes.group(2);
teichmann@8025: 
teichmann@8025:                 return true;
teichmann@8025:             }
tom@8032: 
tom@8032:             log.warn("SLLSP: Unrecognized grain-size interval. Ignored.");
tom@8032:             return true;
tom@8032: 
tom@8032:         }
tom@8032: 
tom@8032:         return false;
tom@8032:     }
tom@8032: 
tom@8032: 
tom@8032:     public boolean handleMetaFractionName(String line) {
tom@8032:         Matcher m = META_FRACTION_NAME.matcher(line);
tom@8032: 
tom@8032:         if (m.matches()) {
tom@8032:             String name = m.group(1);
tom@8032: 
tom@8032: 
tom@8032:             GrainFraction gf = ImporterSession.getInstance().getGrainFraction(name);
tom@8032: 
tom@8032:             if (gf != null) {
tom@8032: 
tom@8032:                 if (lower != null && upper != null) {
tom@8032:                     // Validate grain size interval
tom@8032:                     try {
tom@8032:                         Double lowval = nf.parse(lower).doubleValue();
tom@8032:                         Double upval = nf.parse(upper).doubleValue();
tom@8032: 
tom@8032:                         if (EpsilonComparator.CMP.compare(lowval,
tom@8032:                                 gf.getLower()) != 0 ||
tom@8032:                             EpsilonComparator.CMP.compare(upval,
tom@8032:                                 gf.getUpper()) != 0) {
tom@8032:                             log.warn("SLLSP: Invalid grain size for grain fraction '" +
tom@8032:                                      name + "'. Ignored.");
tom@8032:                         }
tom@8032:                     }
tom@8032:                     catch (ParseException pe) {
tom@8032:                         log.warn("SLLSP: Could not parse grain-size interval. Ignored.");
tom@8032:                     }
tom@8032:                 }
tom@8032: 
tom@8032:                 grainFraction = new ImportGrainFraction(gf);
tom@8032:                 return true;
tom@8032:             }
tom@8032: 
tom@8032:             log.error("SLLSP: Unknown grain fraction: '" + name + "'");
teichmann@8025:         }
teichmann@8025: 
teichmann@8025:         return false;
teichmann@8025:     }
teichmann@8025: 
teichmann@8025: 
teichmann@8042:     public boolean handleColumnNames(String line) throws LineParserException {
teichmann@8025:         Matcher m = META_COLUMN_NAMES.matcher(line);
teichmann@8025: 
teichmann@8025:         if (m.matches()) {
teichmann@8025:             columnNames = line.split(SEPERATOR_CHAR);
teichmann@8025: 
tom@8032:             // 'Fluss-km', 'Hinweise' and at least one data column required
tom@8032:             if (columnNames.length < 3) {
teichmann@8042:                 throw new LineParserException("SLLSP: missing columns.");
tom@8032:             }
tom@8032: 
tom@8032:             initializeSedimentLoadLSs();
teichmann@8025: 
teichmann@8025:             return true;
teichmann@8025:         }
teichmann@8025: 
teichmann@8025:         return false;
teichmann@8025:     }
teichmann@8025: 
teichmann@8025: 
teichmann@8025:     protected void handleDataLine(String line) {
teichmann@8025:         String[] vals = line.split(SEPERATOR_CHAR);
teichmann@8025: 
teichmann@8025:         if (vals == null || vals.length < columnNames.length-1) {
tom@8032:             log.warn("SLLSP: skip invalid data line: '" + line + "'");
teichmann@8025:             return;
teichmann@8025:         }
teichmann@8025: 
teichmann@8025:         try {
teichmann@8025:             Double km = nf.parse(vals[0]).doubleValue();
teichmann@8025: 
teichmann@8025:             for (int i = 1, n = columnNames.length-1; i < n; i++) {
teichmann@8025:                 String curVal = vals[i];
teichmann@8025: 
teichmann@8025:                 if (curVal != null && curVal.length() > 0) {
teichmann@8025:                     current[i-1].addValue(new ImportSedimentLoadLSValue(
teichmann@8025:                         km, nf.parse(vals[i]).doubleValue()
teichmann@8025:                     ));
teichmann@8025:                 }
teichmann@8025:             }
teichmann@8025:         }
teichmann@8025:         catch (ParseException pe) {
tom@8032:             log.warn("SLLSP: unparseable number in data row '" + line + "':", pe);
teichmann@8025:         }
teichmann@8025:     }
teichmann@8025: 
teichmann@8025: 
tom@8032:     /** Initialize SedimentLoadLSs from columns, set the kind
teichmann@8025:      * with respect to file location (offical epoch or not?) */
tom@8032:     private void initializeSedimentLoadLSs() {
teichmann@8025:         // skip first column (Fluss-km) and last column (Hinweise)
teichmann@8031:         current = new ImportSedimentLoadLS[columnNames.length-2];
teichmann@8025: 
teichmann@8025:         Integer kind;
teichmann@8025: 
teichmann@8025:         if (inputFile.getAbsolutePath().contains("amtliche Epochen")) {
teichmann@8025:             kind = new Integer(1);
teichmann@8025:         }
teichmann@8025:         else {
teichmann@8025:             kind = new Integer(0);
teichmann@8025:         }
teichmann@8025: 
teichmann@8025:         for (int i = 0, n = columnNames.length; i < n-2; i++) {
teichmann@8031:             current[i] = new ImportSedimentLoadLS(this.description);
teichmann@8025:             current[i].setTimeInterval(getTimeInterval(columnNames[i+1]));
teichmann@8025:             current[i].setUnit(unit);
teichmann@8025:             current[i].setGrainFraction(grainFraction);
teichmann@8025:             current[i].setKind(kind);
teichmann@8025:         }
teichmann@8025:     }
teichmann@8025: 
teichmann@8025: 
teichmann@8025:     private ImportTimeInterval getTimeInterval(String column) {
teichmann@8025:         try {
teichmann@8025:             Matcher a = TIMEINTERVAL_EPOCH.matcher(column);
teichmann@8025:             if (a.matches()) {
teichmann@8025:                 int yearA = nf.parse(a.group(1)).intValue();
teichmann@8025:                 int yearB = nf.parse(a.group(2)).intValue();
teichmann@8025: 
teichmann@8025:                 return new ImportTimeInterval(
teichmann@8025:                     DateUtil.getStartDateFromYear(yearA),
teichmann@8025:                     DateUtil.getEndDateFromYear(yearB)
teichmann@8025:                 );
teichmann@8025:             }
teichmann@8025: 
teichmann@8025:             Matcher b = TIMEINTERVAL_SINGLE.matcher(column);
teichmann@8025:             if (b.matches()) {
teichmann@8025:                 int year = nf.parse(b.group(1)).intValue();
teichmann@8025: 
teichmann@8025:                 return new ImportTimeInterval(DateUtil.getStartDateFromYear(year));
teichmann@8025:             }
teichmann@8025: 
tom@8032:             log.warn("SLLSP: Unknown time interval string: '" + column + "'");
teichmann@8025:         }
teichmann@8025:         catch (ParseException pe) {
tom@8032:             log.warn("SLLSP: Could not parse years: " + column, pe);
teichmann@8025:         }
teichmann@8025: 
teichmann@8025:         return null;
teichmann@8025:     }
teichmann@8025: 
teichmann@8025: 
tom@8032:     public List<ImportSedimentLoadLS> getSedimentLoadLSs() {
tom@8032:         return sedimentLoadLSs;
teichmann@8025:     }
teichmann@8025: }
teichmann@8025: // vim:set ts=4 sw=4 si et sta sts=4 fenc=utf8 :