teichmann@5844: /* Copyright (C) 2011, 2012, 2013 by Bundesanstalt für Gewässerkunde
teichmann@5844:  * Software engineering by Intevation GmbH
teichmann@5844:  *
teichmann@5992:  * This file is Free Software under the GNU AGPL (>=v3)
teichmann@5844:  * and comes with ABSOLUTELY NO WARRANTY! Check out the
teichmann@5992:  * documentation coming with Dive4Elements River for details.
teichmann@5844:  */
teichmann@5844: 
teichmann@5829: package org.dive4elements.river.importer.parsers;
ingo@2811: 
ingo@2811: import java.io.File;
ingo@2811: 
ingo@2811: import java.math.BigDecimal;
ingo@2811: 
ingo@2811: import java.text.NumberFormat;
ingo@2811: import java.text.ParseException;
ingo@2811: 
ingo@2811: import java.util.ArrayList;
ingo@2811: import java.util.Date;
ingo@2811: import java.util.List;
tom@6273: import java.util.TreeSet;
ingo@2811: import java.util.Locale;
ingo@2811: 
ingo@2811: import java.util.regex.Matcher;
ingo@2811: import java.util.regex.Pattern;
ingo@2811: 
ingo@2811: import java.io.IOException;
ingo@2811: import java.io.LineNumberReader;
ingo@2811: import java.io.FileInputStream;
ingo@2811: import java.io.InputStreamReader;
ingo@2811: 
ingo@2811: import org.apache.log4j.Logger;
ingo@2811: 
tom@8559: import org.dive4elements.river.importer.ImportBedHeight;
tom@8559: import org.dive4elements.river.importer.ImportBedHeightValue;
teichmann@5829: import org.dive4elements.river.importer.ImportBedHeightType;
teichmann@5829: import org.dive4elements.river.importer.ImportElevationModel;
teichmann@5829: import org.dive4elements.river.importer.ImportLocationSystem;
teichmann@5829: import org.dive4elements.river.importer.ImportRange;
teichmann@5829: import org.dive4elements.river.importer.ImportTimeInterval;
teichmann@5829: import org.dive4elements.river.importer.ImportUnit;
teichmann@5829: import org.dive4elements.river.model.BedHeightType;
teichmann@5829: import org.dive4elements.river.importer.ImporterSession;
teichmann@8187: import org.dive4elements.river.backend.utils.EpsilonComparator;
teichmann@8187: import org.dive4elements.river.backend.utils.DateUtil;
ingo@2811: 
tom@8557: public class BedHeightParser {
ingo@2811: 
ingo@2811:     private static final Logger log =
ingo@2811:         Logger.getLogger(BedHeightParser.class);
ingo@2811: 
ingo@2811:     public static final String ENCODING = "ISO-8859-1";
ingo@2811: 
ingo@2811:     public static final Locale DEFAULT_LOCALE = Locale.GERMAN;
ingo@2811: 
ingo@2811:     public static final String START_META_CHAR = "#";
ingo@2811:     public static final String SEPERATOR_CHAR  = ";";
ingo@2811: 
ingo@2811:     public static final Pattern META_YEAR =
felix@3960:         Pattern.compile("^Jahr: [^0-9]*(\\d*).*");
ingo@2811: 
ingo@2811:     public static final Pattern META_TIMEINTERVAL =
ingo@2811:         Pattern.compile("^Zeitraum: Epoche (\\d*)-(\\d*).*");
ingo@2811: 
ingo@2811:     public static final Pattern META_TYPE =
ingo@2811:         Pattern.compile("^Aufnahmeart: (.*).*");
ingo@2811: 
ingo@2811:     public static final Pattern META_LOCATION_SYSTEM =
ingo@2811:         Pattern.compile("^Lagesystem: (.*).*");
ingo@2811: 
ingo@2811:     public static final Pattern META_CUR_ELEVATION_SYSTEM =
ingo@2811:         Pattern.compile("^H.hensystem:\\s(\\w++) (.* )??\\[(.*)\\].*");
ingo@2811: 
ingo@2811:     public static final Pattern META_OLD_ELEVATION_SYSTEM =
ingo@2811:         Pattern.compile("^urspr.ngliches H.hensystem:\\s(\\w++) (.* )??\\[(.*)\\].*");
ingo@2811: 
ingo@2811:     public static final Pattern META_RANGE =
tom@5484:         Pattern.compile("^Strecke:\\D*(\\d++.?\\d*) ?- ?(\\d++.?\\d*).*");
ingo@2811: 
ingo@2811:     public static final Pattern META_EVALUATION_BY =
ingo@2811:         Pattern.compile("^Auswerter: (.*).*");
ingo@2811: 
ingo@2811:     public static final Pattern META_COMMENTS =
ingo@2811:         Pattern.compile("^Weitere Bemerkungen: (.*).*");
ingo@2811: 
ingo@2811: 
ingo@2811:     protected static NumberFormat nf = NumberFormat.getInstance(DEFAULT_LOCALE);
ingo@2811: 
ingo@2811: 
tom@8559:     protected List<ImportBedHeight> bedHeights;
ingo@2811: 
ingo@2811: 
tom@8559:     protected ImportBedHeight newImportBedHeight(String description) {
tom@8559:         return new ImportBedHeight(description);
tom@8557:     }
ingo@2811: 
ingo@2811: 
tom@6273:     protected TreeSet<Double> kmExists;
ingo@2811: 
ingo@2811:     public BedHeightParser() {
tom@8559:         bedHeights = new ArrayList<ImportBedHeight>();
tom@6273:         kmExists = new TreeSet<Double>(EpsilonComparator.CMP);
ingo@2811:     }
ingo@2811: 
ingo@2811: 
tom@8559:     public List<ImportBedHeight> getBedHeights() {
ingo@2811:         return bedHeights;
ingo@2811:     }
ingo@2811: 
ingo@2811: 
ingo@2811:     public void parse(File file) throws IOException {
ingo@2811:         log.info("Parsing bed height single file '" + file + "'");
ingo@2811: 
tom@8559:         ImportBedHeight obj = newImportBedHeight(file.getName().replaceAll("\\.csv", ""));
ingo@2811: 
tom@6273:         kmExists.clear();
teichmann@6285: 
ingo@2811:         LineNumberReader in = null;
ingo@2811:         try {
ingo@2811:             in =
ingo@2811:                 new LineNumberReader(
ingo@2811:                 new InputStreamReader(
ingo@2811:                 new FileInputStream(file), ENCODING));
ingo@2811: 
ingo@2811:             String line = null;
ingo@2811:             while ((line = in.readLine()) != null) {
ingo@2811:                 if ((line = line.trim()).length() == 0) {
ingo@2811:                     continue;
ingo@2811:                 }
ingo@2811: 
ingo@2811:                 if (line.startsWith(START_META_CHAR)) {
ingo@2811:                     handleMetaLine(obj, line);
ingo@2811:                 }
ingo@2811:                 else {
ingo@2811:                     handleDataLine(obj, line);
ingo@2811:                 }
ingo@2811:             }
ingo@2811: 
ingo@2811:             log.info("File contained " + obj.getValueCount() + " values.");
ingo@2811:             bedHeights.add(obj);
ingo@2811:         }
ingo@2811:         finally {
ingo@2811:             if (in != null) {
ingo@2811:                 in.close();
ingo@2811:             }
ingo@2811:         }
ingo@2811:     }
ingo@2811: 
ingo@2811: 
ingo@2811:     protected static String stripMetaLine(String line) {
ingo@2811:         String tmp = line.substring(1, line.length());
ingo@2811: 
ingo@2811:         if (tmp.startsWith(" ")) {
ingo@2811:             return tmp.substring(1, tmp.length());
ingo@2811:         }
ingo@2811:         else {
ingo@2811:             return tmp;
ingo@2811:         }
ingo@2811:     }
ingo@2811: 
ingo@2811: 
tom@8559:     protected void handleMetaLine(ImportBedHeight obj, String line) {
ingo@2811:         String meta = stripMetaLine(line);
ingo@2811: 
ingo@2811:         if (handleMetaYear(obj, meta)) {
ingo@2811:             return;
ingo@2811:         }
ingo@2811:         else if (handleMetaTimeInterval(obj, meta)) {
ingo@2811:             return;
ingo@2811:         }
ingo@2811:         else if (handleMetaComment(obj, meta)) {
ingo@2811:             return;
ingo@2811:         }
ingo@2811:         else if (handleMetaEvaluationBy(obj, meta)) {
ingo@2811:             return;
ingo@2811:         }
ingo@2811:         else if (handleMetaRange(obj, meta)) {
ingo@2811:             return;
ingo@2811:         }
ingo@2811:         else if (handleMetaType(obj, meta)) {
ingo@2811:             return;
ingo@2811:         }
ingo@2811:         else if (handleMetaLocationSystem(obj, meta)) {
ingo@2811:             return;
ingo@2811:         }
ingo@2811:         else if (handleMetaCurElevationModel(obj, meta)) {
ingo@2811:             return;
ingo@2811:         }
ingo@2811:         else if (handleMetaOldElevationModel(obj, meta)) {
ingo@2811:             return;
ingo@2811:         }
ingo@2811:         else {
sascha@3662:             log.warn("BHP: Meta line did not match any known type: " + line);
ingo@2811:         }
ingo@2811:     }
ingo@2811: 
ingo@2811: 
tom@8559:     protected boolean handleMetaYear(ImportBedHeight obj, String line) {
ingo@2811:         Matcher m = META_YEAR.matcher(line);
ingo@2811: 
ingo@2811:         if (m.matches()) {
ingo@2811:             String tmp = m.group(1);
teichmann@5327:             if (tmp.length() > 0) {
teichmann@5327:                 obj.setYear(Integer.parseInt(tmp));
ingo@2811:             }
teichmann@5327:             else {
teichmann@5327:                 log.warn("BHP: No year given.");
ingo@2811:             }
teichmann@5327:             return true;
ingo@2811:         }
ingo@2811: 
ingo@2811:         return false;
ingo@2811:     }
ingo@2811: 
ingo@2811: 
tom@8559:     protected boolean handleMetaTimeInterval(ImportBedHeight obj, String line) {
ingo@2811:         Matcher m = META_TIMEINTERVAL.matcher(line);
ingo@2811: 
ingo@2811:         if (m.matches()) {
ingo@2811:             String lo = m.group(1);
ingo@2811:             String up = m.group(2);
ingo@2811: 
ingo@2811:             log.debug("Found time interval: " + lo + " - " + up);
ingo@2811: 
ingo@2811:             try {
ingo@2811:                 int lower = Integer.valueOf(lo);
ingo@2811:                 int upper = Integer.valueOf(up);
ingo@2811: 
tom@7927:                 Date fromYear = DateUtil.getStartDateFromYear(lower);
tom@7927:                 Date toYear   = DateUtil.getEndDateFromYear(upper);
ingo@2811: 
ingo@2811:                 obj.setTimeInterval(new ImportTimeInterval(fromYear, toYear));
ingo@2811:             }
ingo@2811:             catch (NumberFormatException e) {
tom@5489:                 log.warn("BHP: could not parse timeinterval", e);
ingo@2811:             }
ingo@2811: 
ingo@2811:             return true;
ingo@2811:         }
ingo@2811: 
ingo@2811:         return false;
ingo@2811:     }
ingo@2811: 
ingo@2811: 
tom@8559:     protected boolean handleMetaComment(ImportBedHeight obj, String line) {
ingo@2811:         Matcher m = META_COMMENTS.matcher(line);
ingo@2811: 
ingo@2811:         if (m.matches()) {
ingo@2811:             String tmp = m.group(1);
ingo@2811: 
ingo@2811:             obj.setDescription(tmp);
ingo@2811: 
ingo@2811:             return true;
ingo@2811:         }
ingo@2811: 
ingo@2811:         return false;
ingo@2811:     }
ingo@2811: 
ingo@2811: 
ingo@2811:     protected boolean handleMetaEvaluationBy(
tom@8559:         ImportBedHeight obj,
ingo@2811:         String                line
ingo@2811:     ) {
ingo@2811:         Matcher m = META_EVALUATION_BY.matcher(line);
ingo@2811: 
ingo@2811:         if (m.matches()) {
ingo@2811:             String tmp = m.group(1);
ingo@2811:             tmp = tmp.replace(";", "");
ingo@2811: 
ingo@2811:             obj.setEvaluationBy(tmp);
ingo@2811: 
ingo@2811:             return true;
ingo@2811:         }
ingo@2811: 
ingo@2811:         return false;
ingo@2811:     }
ingo@2811: 
ingo@2811: 
tom@8559:     protected boolean handleMetaRange(ImportBedHeight obj, String line) {
ingo@2811:         Matcher m = META_RANGE.matcher(line);
ingo@2811: 
ingo@2811:         if (m.matches() && m.groupCount() >= 2) {
ingo@2811:             String a = m.group(1).replace(";", "");
ingo@2811:             String b = m.group(2).replace(";", "");
ingo@2811: 
ingo@2811:             try {
ingo@2811:                 BigDecimal lower = new BigDecimal(nf.parse(a).doubleValue());
ingo@2811:                 BigDecimal upper = new BigDecimal(nf.parse(b).doubleValue());
ingo@2811: 
ingo@2811:                 obj.setRange(new ImportRange(lower, upper));
ingo@2811: 
ingo@2811:                 return true;
ingo@2811:             }
ingo@2811:             catch (ParseException e) {
tom@5489:                 log.warn("BHP: could not parse range", e);
ingo@2811:             }
ingo@2811:         }
ingo@2811: 
ingo@2811:         return false;
ingo@2811:     }
ingo@2811: 
ingo@2811: 
tom@8559:     protected boolean handleMetaType(ImportBedHeight obj, String line) {
ingo@2811:         Matcher m = META_TYPE.matcher(line);
ingo@2811: 
ingo@2811:         if (m.matches()) {
tom@6235:             String tmp = m.group(1).replace(";", "").trim();
ingo@2811: 
tom@5278:             BedHeightType bht = BedHeightType.fetchBedHeightTypeForType(
teichmann@7376:                 tmp,
teichmann@7376:                 ImporterSession.getInstance().getDatabaseSession());
ingo@2811: 
teichmann@5273:             if (bht != null) {
teichmann@5273:                 obj.setType(new ImportBedHeightType(bht));
sascha@3798:                 return true;
sascha@3798:             }
teichmann@5273: 
tom@5901:             log.error("Unknown bed height type: '" + tmp + "'. File ignored.");
ingo@2811:         }
ingo@2811: 
ingo@2811:         return false;
ingo@2811:     }
ingo@2811: 
ingo@2811: 
ingo@2811:     protected boolean handleMetaLocationSystem(
tom@8559:         ImportBedHeight obj,
ingo@2811:         String          line
ingo@2811:     ) {
ingo@2811:         Matcher m = META_LOCATION_SYSTEM.matcher(line);
ingo@2811: 
ingo@2811:         if (m.matches()) {
ingo@2811:             String tmp = m.group(1).replace(";", "");
ingo@2811: 
ingo@2811:             obj.setLocationSystem(new ImportLocationSystem(tmp, tmp));
ingo@2811: 
ingo@2811:             return true;
ingo@2811:         }
ingo@2811: 
ingo@2811:         return false;
ingo@2811:     }
ingo@2811: 
ingo@2811: 
ingo@2811:     protected boolean handleMetaCurElevationModel(
tom@8559:         ImportBedHeight obj,
ingo@2811:         String          line
ingo@2811:     ) {
ingo@2811:         Matcher m = META_CUR_ELEVATION_SYSTEM.matcher(line);
ingo@2811: 
ingo@2811:         if (m.matches()) {
ingo@2811:             String name = m.group(1);
ingo@2811:             String num  = m.group(2);
ingo@2811:             String unit = m.group(3);
ingo@2811: 
ingo@2811:             obj.setCurElevationModel(new ImportElevationModel(
ingo@2811:                 name + " " + num,
ingo@2811:                 new ImportUnit(unit)
ingo@2811:             ));
ingo@2811: 
ingo@2811:             return true;
ingo@2811:         }
ingo@2811: 
ingo@2811:         return false;
ingo@2811:     }
ingo@2811: 
ingo@2811: 
ingo@2811:     protected boolean handleMetaOldElevationModel(
tom@8559:         ImportBedHeight obj,
ingo@2811:         String          line
ingo@2811:     ) {
ingo@2811:         Matcher m = META_OLD_ELEVATION_SYSTEM.matcher(line);
ingo@2811: 
ingo@2811:         if (m.matches()) {
ingo@2811:             String name = m.group(1);
ingo@2811:             String num  = m.group(2);
ingo@2811:             String unit = m.group(3);
ingo@2811: 
ingo@2811:             obj.setOldElevationModel(new ImportElevationModel(
ingo@2811:                 name + " " + num,
ingo@2811:                 new ImportUnit(unit)
ingo@2811:             ));
ingo@2811: 
ingo@2811:             return true;
ingo@2811:         }
ingo@2811: 
ingo@2811:         return false;
ingo@2811:     }
tom@8557: 
tom@8559:     protected void handleDataLine(ImportBedHeight obj, String line) {
tom@8557:         String[] values = line.split(SEPERATOR_CHAR, -1);
tom@8557: 
tom@8557:         if (values == null) {
tom@8557:             log.warn("BSP: Error while parsing data line: '" + line + "'");
tom@8557:             return;
tom@8557:         }
tom@8557: 
tom@8557:         Double km;
tom@8557: 
tom@8557:         try {
tom@8557:             km = new Double(nf.parse(values[0]).doubleValue());
tom@8557: 
tom@8557:             Double key = Double.valueOf(km);
tom@8557: 
tom@8557:             if (kmExists.contains(key)) {
tom@8557:                 log.warn("duplicate station '" + km + "': -> ignored");
tom@8557:                 return;
tom@8557:             }
tom@8557: 
tom@8557:             kmExists.add(key);
tom@8557:         }
tom@8557:         catch (ParseException e) {
tom@8557:             // We expect a lot of ";;;;;;" lines.
tom@8557:             return;
tom@8557:         }
tom@8557: 
tom@8557:         // Handle gaps like "10,0;;;;;".
tom@8557:         if (values.length <= 2) {
tom@8557:             // Do not import line without useful data
tom@8557:             if (values.length < 2) {
tom@8557:                 return;
tom@8557:             }
tom@8557:             if (values[1].length() == 0) {
tom@8557:                 return;
tom@8557:             }
tom@8557:         }
tom@8557: 
tom@8557:         Double height = null;
tom@8557:         if (values[1].length() > 0) {
tom@8557:             try {
tom@8557:                 height = new Double(nf.parse(values[1]).doubleValue());
tom@8557:             }
tom@8557:             catch (ParseException e) {
tom@8557:                 log.warn("BSP: unparseable height " + values[1]);
tom@8557:             }
tom@8557:         }
tom@8557: 
tom@8557:         Double uncertainty = null;
tom@8557:         if (values[2].length() > 0) {
tom@8557:             try {
tom@8557:                 uncertainty = new Double(nf.parse(values[2]).doubleValue());
tom@8557:             }
tom@8557:             catch (ParseException e) {
tom@8557:                 log.warn("BSP: unparseable uncertainty value " + values[2]);
tom@8557:             }
tom@8557:         }
tom@8557: 
tom@8557:         Double dataGap = null;
tom@8557:         if (values[3].length() > 0) {
tom@8557:             try {
tom@8557:                 dataGap = new Double(nf.parse(values[3]).doubleValue());
tom@8557:             }
tom@8557:             catch (ParseException e) {
tom@8557:                 log.warn("BSP: unparseable data gap " + values[3]);
tom@8557:             }
tom@8557:         }
tom@8557: 
tom@8557:         Double soundingWidth = null;
tom@8557:         if (values[4].length() > 0) {
tom@8557:             try {
tom@8557:                 soundingWidth = new Double(nf.parse(values[4]).doubleValue());
tom@8557:             }
tom@8557:             catch (ParseException e) {
tom@8557:                 log.warn("BSP: unparseable sounding width " + values[4]);
tom@8557:             }
tom@8557:         }
tom@8557: 
tom@8559:         ImportBedHeightValue value = new ImportBedHeightValue(
tom@8559:             (ImportBedHeight) obj,
tom@8557:             km,
tom@8557:             height,
tom@8557:             uncertainty,
tom@8557:             dataGap,
tom@8557:             soundingWidth);
tom@8557: 
tom@8557:         obj.addValue(value);
tom@8557:     }
ingo@2811: }
ingo@2811: // vim:set ts=4 sw=4 si et sta sts=4 fenc=utf8 :