tom@8043: /* Copyright (C) 2011, 2012, 2013 by Bundesanstalt für Gewässerkunde
tom@8043:  * Software engineering by Intevation GmbH
tom@8043:  *
tom@8043:  * This file is Free Software under the GNU AGPL (>=v3)
tom@8043:  * and comes with ABSOLUTELY NO WARRANTY! Check out the
tom@8043:  * documentation coming with Dive4Elements River for details.
tom@8043:  */
tom@8043: 
tom@8043: package org.dive4elements.river.importer.parsers;
tom@8043: 
tom@8043: import java.io.File;
tom@8043: import java.io.IOException;
tom@8043: 
tom@8043: import java.text.NumberFormat;
tom@8043: import java.text.ParseException;
tom@8043: 
tom@8043: import java.util.regex.Matcher;
tom@8043: import java.util.regex.Pattern;
tom@8043: 
tom@8043: import org.apache.log4j.Logger;
tom@8043: 
tom@8043: import org.dive4elements.river.importer.ImporterSession;
tom@8043: import org.dive4elements.river.importer.ImportGrainFraction;
tom@8043: import org.dive4elements.river.importer.ImportTimeInterval;
tom@8043: 
tom@8043: import org.dive4elements.river.model.GrainFraction;
tom@8043: 
tom@8043: import org.dive4elements.river.utils.DateUtil;
tom@8043: import org.dive4elements.river.utils.EpsilonComparator;
tom@8043: 
tom@8043: /** Parses sediment load files. */
tom@8043: public abstract class AbstractSedimentLoadParser extends LineParser {
tom@8043: 
tom@8043:     private static final Logger log =
tom@8043:         Logger.getLogger(AbstractSedimentLoadParser.class);
tom@8043: 
tom@8043: 
tom@8043:     public static final NumberFormat nf = NumberFormat.getInstance(DEFAULT_LOCALE);
tom@8043: 
tom@8043: 
tom@8043:     public static final Pattern TIMEINTERVAL_SINGLE =
tom@8043:         Pattern.compile("\\D*([0-9]+?)\\D*");
tom@8043: 
tom@8043:     public static final Pattern TIMEINTERVAL_EPOCH =
tom@8043:         Pattern.compile("\\D*([0-9]+?)\\s*-\\s*([0-9]+?)\\D*");
tom@8043: 
tom@8043:     public static final Pattern META_FRACTION =
tom@8043:         Pattern.compile("^Fraktion: (.*)");
tom@8043: 
tom@8043:     public static final Pattern META_FRACTION_NAME =
tom@8043:         Pattern.compile("^Fraktionsname: (.*)");
tom@8043: 
tom@8059:     public static final Pattern META_SQ_TIMEINTERVAL =
tom@8059:         Pattern.compile("^S-Q-Beziehung: (.*)");
tom@8059: 
tom@8043:     public static final Pattern META_COLUMN_NAMES =
tom@8043:         Pattern.compile("^Fluss-km.*");
tom@8043: 
tom@8043:     public static final Pattern META_GRAIN_SIZE =
tom@8043:         Pattern.compile("([0-9]*,*[0-9]+)-([0-9]*,*[0-9]+) *mm");
tom@8043: 
tom@8043: 
tom@8043:     protected abstract void handleDataLine(String line);
tom@8043: 
tom@8043:     /** Initialize SedimentLoadLSs from columns, set the kind
tom@8043:      * with respect to file location (offical epoch or not?) */
tom@8043:     protected abstract void initializeSedimentLoads();
tom@8043: 
tom@8056:     protected abstract void handleMetaLine(String line)
tom@8056:         throws LineParserException;
tom@8056: 
tom@8043: 
tom@8043:     protected ImportGrainFraction grainFraction;
tom@8043: 
tom@8059:     protected ImportTimeInterval sqTimeInterval;
tom@8059: 
tom@8043:     protected String description;
tom@8043: 
tom@8043:     protected String[] columnNames;
tom@8043: 
tom@8043:     private String upper;
tom@8043: 
tom@8043:     private String lower;
tom@8043: 
tom@8043: 
tom@8043:     @Override
tom@8043:     public void parse(File file) throws IOException {
tom@8043:         description = file.getName();
tom@8043: 
tom@8043:         super.parse(file);
tom@8043:     }
tom@8043: 
tom@8043: 
tom@8043:     @Override
tom@8043:     protected void handleLine(int lineNum, String line) throws LineParserException {
tom@8043:         if (line.startsWith(START_META_CHAR)) {
tom@8043:             handleMetaLine(stripMetaLine(line));
tom@8043:         }
tom@8043:         else {
tom@8043:             handleDataLine(line);
tom@8043:         }
tom@8043:     }
tom@8043: 
tom@8043: 
tom@8043:     public boolean handleMetaFraction(String line) {
tom@8043:         Matcher m = META_FRACTION.matcher(line);
tom@8043: 
tom@8043:         if (m.matches()) {
tom@8043:             String interval = m.group(1);
tom@8043: 
tom@8043:             Matcher sizes = META_GRAIN_SIZE.matcher(interval);
tom@8043:             if (sizes.matches()) {
tom@8043:                 lower = sizes.group(1);
tom@8043:                 upper = sizes.group(2);
tom@8043: 
tom@8043:                 return true;
tom@8043:             }
tom@8043: 
tom@8043:             log.warn("ASLP: Unrecognized grain-size interval. Ignored.");
tom@8043:             return true;
tom@8043: 
tom@8043:         }
tom@8043: 
tom@8043:         return false;
tom@8043:     }
tom@8043: 
tom@8043: 
tom@8043:     public boolean handleMetaFractionName(String line) throws LineParserException {
tom@8043:         Matcher m = META_FRACTION_NAME.matcher(line);
tom@8043: 
tom@8043:         if (m.matches()) {
tom@8043:             String name = m.group(1);
tom@8043: 
tom@8043: 
tom@8043:             GrainFraction gf = ImporterSession.getInstance().getGrainFraction(name);
tom@8043: 
tom@8043:             if (gf != null) {
tom@8043: 
tom@8043:                 if (lower != null && upper != null) {
tom@8043:                     // Validate grain size interval
tom@8043:                     try {
tom@8043:                         Double lowval = nf.parse(lower).doubleValue();
tom@8043:                         Double upval = nf.parse(upper).doubleValue();
tom@8043: 
tom@8043:                         if (EpsilonComparator.CMP.compare(lowval,
tom@8043:                                 gf.getLower()) != 0 ||
tom@8043:                             EpsilonComparator.CMP.compare(upval,
tom@8043:                                 gf.getUpper()) != 0) {
tom@8043:                             log.warn("ASLP: Invalid grain size for grain fraction '" +
tom@8043:                                      name + "'. Ignored.");
tom@8043:                         }
tom@8043:                     }
tom@8043:                     catch (ParseException pe) {
tom@8043:                         log.warn("ASLP: Could not parse grain-size interval. Ignored.");
tom@8043:                     }
tom@8043:                 }
tom@8043: 
tom@8043:                 grainFraction = new ImportGrainFraction(gf);
tom@8043:                 return true;
tom@8043:             }
tom@8043: 
tom@8043:             throw new LineParserException("ASLP: Unknown grain fraction: '" + 
tom@8043:                 name + "'");
tom@8043:         }
tom@8043: 
tom@8043:         return false;
tom@8043:     }
tom@8043: 
tom@8043: 
tom@8059:     public boolean handleMetaSQTimeInterval(String line) {
tom@8059:         Matcher m = META_SQ_TIMEINTERVAL.matcher(line);
tom@8059: 
tom@8059:         if (m.matches()) {
tom@8059:             String interval = m.group(1);
tom@8059: 
tom@8059:             try {
tom@8059:                 Matcher a = TIMEINTERVAL_EPOCH.matcher(interval);
tom@8059:                 if (a.matches()) {
tom@8059:                     int yearA = nf.parse(a.group(1)).intValue();
tom@8059:                     int yearB = nf.parse(a.group(2)).intValue();
tom@8059: 
tom@8059:                     sqTimeInterval = new ImportTimeInterval(
tom@8059:                         DateUtil.getStartDateFromYear(yearA),
tom@8059:                         DateUtil.getEndDateFromYear(yearB)
tom@8059:                     );
tom@8059:                 }
tom@8059:                 else {
tom@8059:                     log.warn("ASLP: Unknown SQ-time string: '" + interval +
tom@8059:                         "'. Ignored.");
tom@8059:                 }
tom@8059:             }
tom@8059:             catch (ParseException pe) {
tom@8059:                 log.error("ASLP: Could not parse SQ-time string: '" +
tom@8059:                     interval + "'. Ignored.", pe);
tom@8059:             }
tom@8059: 
tom@8059:             return true;
tom@8059: 
tom@8059:         }
tom@8059: 
tom@8059:         return false;
tom@8059:     }
tom@8059: 
tom@8059: 
tom@8043:     public boolean handleColumnNames(String line) throws LineParserException {
tom@8043:         Matcher m = META_COLUMN_NAMES.matcher(line);
tom@8043: 
tom@8043:         if (m.matches()) {
tom@8043:             columnNames = line.split(SEPERATOR_CHAR);
tom@8043: 
tom@8043:             // 'Fluss-km', 'Hinweise' and at least one data column required
tom@8043:             if (columnNames.length < 3) {
tom@8043:                 throw new LineParserException("ASLP: missing columns in '" +
tom@8043:                     line + "'");
tom@8043:             }
tom@8043: 
tom@8043:             initializeSedimentLoads();
tom@8043: 
tom@8043:             return true;
tom@8043:         }
tom@8043: 
tom@8043:         return false;
tom@8043:     }
tom@8043: 
tom@8043: 
tom@8043:     protected ImportTimeInterval getTimeInterval(String column) {
tom@8043:         try {
tom@8043:             Matcher a = TIMEINTERVAL_EPOCH.matcher(column);
tom@8043:             if (a.matches()) {
tom@8043:                 int yearA = nf.parse(a.group(1)).intValue();
tom@8043:                 int yearB = nf.parse(a.group(2)).intValue();
tom@8043: 
tom@8043:                 return new ImportTimeInterval(
tom@8043:                     DateUtil.getStartDateFromYear(yearA),
tom@8043:                     DateUtil.getEndDateFromYear(yearB)
tom@8043:                 );
tom@8043:             }
tom@8043: 
tom@8043:             Matcher b = TIMEINTERVAL_SINGLE.matcher(column);
tom@8043:             if (b.matches()) {
tom@8043:                 int year = nf.parse(b.group(1)).intValue();
tom@8043: 
tom@8043:                 return new ImportTimeInterval(DateUtil.getStartDateFromYear(year));
tom@8043:             }
tom@8043: 
tom@8043:             log.warn("ASLP: Unknown time interval string: '" + column + "'");
tom@8043:         }
tom@8043:         catch (ParseException pe) {
tom@8043:             log.warn("ASLP: Could not parse years: " + column, pe);
tom@8043:         }
tom@8043: 
tom@8043:         return null;
tom@8043:     }
tom@8043: 
tom@8043: }
tom@8043: // vim:set ts=4 sw=4 si et sta sts=4 fenc=utf8 :