teichmann@5844: /* Copyright (C) 2011, 2012, 2013 by Bundesanstalt für Gewässerkunde
teichmann@5844:  * Software engineering by Intevation GmbH
teichmann@5844:  *
teichmann@5844:  * This file is Free Software under the GNU AGPL (>=v3) 
teichmann@5844:  * and comes with ABSOLUTELY NO WARRANTY! Check out the
teichmann@5844:  * documentation coming with Dive4Elements River for details. 
teichmann@5844:  */
teichmann@5844: 
teichmann@5829: package org.dive4elements.river.importer.parsers;
felix@4681: 
teichmann@5829: import org.dive4elements.artifacts.common.utils.FileTools;
felix@4681: 
teichmann@5829: import org.dive4elements.river.importer.XY;
teichmann@4735: 
teichmann@5829: import org.dive4elements.river.utils.EpsilonComparator;
felix@4681: 
felix@4681: import java.io.File;
felix@4681: import java.io.IOException;
felix@4681: 
teichmann@4735: import java.util.ArrayList;
teichmann@4735: import java.util.HashMap;
teichmann@4735: import java.util.List;
teichmann@4735: import java.util.Map;
teichmann@4735: import java.util.TreeMap;
felix@4681: 
teichmann@4735: import java.util.regex.Matcher;
teichmann@4735: import java.util.regex.Pattern;
felix@4681: 
teichmann@4735: import org.apache.log4j.Logger;
felix@4681: 
felix@4681: 
felix@4681: /**
felix@4681:  * To create cross-sections, generate: Map<double,list<xy>> from files
felix@4681:  * in da66 format.
felix@4681:  */
felix@4715: public class DA66Parser extends LineParser implements CrossSectionParser
felix@4681: {
felix@4681:     /** Private logger. */
felix@4681:     private static Logger logger = Logger.getLogger(DA66Parser.class);
felix@4681: 
felix@4712:     private static String HEAD_HEAD = "00";
felix@4712:     private static String HEAD_GEOM = "66"; // "Values"
felix@4712:     private static String HEAD_ENDG = "88"; // Probably never used.
felix@4712: 
felix@4717:     /** Regex to match lines of files in da66 format. */
felix@4712:     private static final Pattern LINE_PATTERN =
felix@4712:         Pattern.compile("^([0-9 -]{2})" + // Type (00|66|88)
felix@4681:                         "([0-9 -]{5})" + // unset
felix@4681:                         "([0-9 -]{2})" + // id
felix@4681:                         "([0-9 -]{9})" + // station
felix@4681:                         "([0-9 -]{2})" + // running number
felix@4712:                         "([0-9 -]{1})?" + // point id
felix@4681:                         /*
felix@4681:                         Would be great if we could express the pattern as this:
felix@4681:                         ([0-9 -]{1})([0-9 -JKMLMNOPQR]{7})([0-9 -]{7})+
felix@4681:                         */
felix@4712:                         "([0-9 -JKMLMNOPQR]{7})?" + // y
felix@4712:                         "([0-9 -]{7})?" + // z
felix@4712:                         "([0-9 -]{1})?" + // point id
felix@4712:                         "([0-9 -JKMLMNOPQR]{7})?" + // y
felix@4712:                         "([0-9 -]{7})?" + // z
felix@4712:                         "([0-9 -]{1})?" + // point id
felix@4712:                         "([0-9 -JKMLMNOPQR]{7})?" + // y
felix@4712:                         "([0-9 -]{7})?" + // z
felix@4712:                         "([0-9 -]{1})?" + // point id
felix@4712:                         "([0-9 -JKMLMNOPQR]{7})?" + // y
felix@4712:                         "([0-9 -]{7})?" // z
felix@4681:                         );
felix@4681: 
felix@4723: 
felix@4712:     /** Indices to match group of main regex. */
felix@4712:     private static enum FIELD {
felix@4712:         HEAD      ( 1),
felix@4712:         UNSET     ( 2),
felix@4712:         ID        ( 3),
felix@4712:         STATION   ( 4),
felix@4712:         RUNNR     ( 5),
felix@4712:         POINT_1_ID( 6),
felix@4712:         POINT_1_Y ( 7),
felix@4712:         POINT_1_Z ( 8),
felix@4712:         POINT_2_ID( 9),
felix@4712:         POINT_2_Y (10),
felix@4712:         POINT_2_Z (11),
felix@4712:         POINT_3_ID(12),
felix@4712:         POINT_3_Y (13),
felix@4712:         POINT_3_Z (14),
felix@4712:         POINT_4_ID(15),
felix@4712:         POINT_4_Y (16),
felix@4712:         POINT_4_Z (17);
teichmann@4735: 
felix@4712:         private int idx;
felix@4712:         FIELD(int idx) {
felix@4712:             this.idx = idx;
felix@4712:         }
felix@4712:         int getIdx() {
felix@4712:             return idx;
felix@4712:         }
felix@4712:     }
felix@4712: 
felix@4723: 
felix@4712:     /** Header lines of da66 can define a type. */
felix@4681:     private static enum Type {
felix@4681:         DATE                     ( 0),
felix@4681:         HEKTOSTONE_LEFT          ( 1), //grm. "Standlinie"
felix@4681:         HEKTOSTONE_RIGHT         ( 2),
felix@4681:         CHANNEL_LEFT             ( 3), //grm. "Fahrrinne"
felix@4681:         CHANNEL_RIGHT            ( 4),
felix@4681:         CHANNEL_2_LEFT           ( 5),
felix@4681:         CHANNEL_2_RIGHT          ( 6),
felix@4681:         GIW_1972                 ( 7),
felix@4681:         GROIN_DIST_LEFT          ( 8), //grm. "Buhnenkopfabstand links"
felix@4681:         GROIN_HEIGHT_LEFT        ( 9),
felix@4681:         GROIN_SLOPE_LEFT         (10),
felix@4681:         GROIN_DIST_RIGHT         (11),
felix@4681:         GROIN_HEIGHT_RIGHT       (12),
felix@4681:         GROIN_SLOPE_RIGHT        (13),
felix@4681:         STRIKE_LEFT              (14), //grm. "Streichlinie links"
felix@4681:         AXIS                     (15),
felix@4681:         STRIKE_RIGHT             (16),
felix@4681:         GROIN_BACK_SLOPE_LEFT    (17), //grm. "Buhnenrueckenneigung"
felix@4681:         GROIN_BACK_SLOPE_RIGHT   (18),
felix@4681:         GIW_1932                 (19),
felix@4681:         GIW_1982                 (20),
felix@4681:         STAND_ISLAND_1           (21),
felix@4681:         STAND_ISLAND_2           (22),
felix@4681:         STAND_ISLAND_3           (23),
felix@4681:         STAND_ISLAND_4           (24),
felix@4681:         UNSPECIFIED_1            (25),
felix@4681:         UNSPECIFIED_2            (26),
felix@4681:         HHW                      (27),
felix@4681:         OLD_PROFILE_NULL         (28),
felix@4681:         AW_1978                  (29),
felix@4681:         SIGN_LEFT                (30),
felix@4681:         SIGN_RIGHT               (31),
felix@4681:         DIST_SIGNAL_CHANNEL_LEFT (32),
felix@4681:         DIST_SIGNAL_CHANNEL_RIGHT(33),
felix@4681:         UNSPECIFIED_3            (34),
felix@4681:         UNSPECIFIED_4            (35),
felix@4681:         UNSPECIFIED_5            (36),
felix@4681:         UNSPECIFIED_6            (37),
felix@4681:         SHORE_LEFT               (38),
felix@4681:         SHORE_RIGHT              (39),
felix@4681:         UNSPECIFIED_7            (40);
felix@4681: 
felix@4681:         private final int id;
felix@4681:         Type(int id) {
felix@4681:             this.id = id;
felix@4681:         }
felix@4681:         public int getId() {
felix@4681:             return id;
felix@4681:         }
felix@4681:     }
felix@4681: 
felix@4723: 
felix@4681:     /** Available types. */
felix@4681:     private static HashMap<Integer, Type> typeMap;
felix@4681: 
felix@4723: 
felix@4681:     /** Types we can deal with. */
felix@4681:     private static List<Type> implementedTypes;
felix@4681: 
felix@4723: 
felix@4681:     static {
felix@4681:         typeMap = new HashMap<Integer, Type>();
felix@4681:         for (Type t: Type.values()) {
felix@4681:             typeMap.put(new Integer(t.getId()), t);
felix@4681:         }
felix@4712:         // TODO populate and respect header type.
felix@4681:         implementedTypes = new ArrayList<Type>();
felix@4681:         //implementedTypes.add(..);
felix@4681:     }
felix@4681: 
felix@4712: 
felix@4712:     /** The current line to which add points. */
felix@4712:     private List<XY> currentLine;
felix@4712: 
felix@4712: 
felix@4712:     /** Data collected so far, last element will be currentLine. */
felix@4681:     protected Map<Double, List<XY>> data;
felix@4681: 
felix@4681: 
felix@4717:     /** Trivial constructor. */
felix@4681:     public DA66Parser() {
teichmann@4735:         data = new TreeMap<Double, List<XY>>(EpsilonComparator.CMP);
felix@4681:     }
felix@4681: 
felix@4717: 
felix@4717:     /** Get the description of the cross section parsed. */
felix@4715:     @Override
felix@4715:     public String getDescription() {
felix@4729:         return FileTools.removeExtension(getFileName());
felix@4715:     }
felix@4715: 
felix@4717: 
felix@4717:     /** Get the year of this cross sections measurement. */
felix@4715:     @Override
felix@4715:     public Integer getYear() {
felix@4726:         return null;
felix@4715:     }
felix@4715: 
felix@4717: 
felix@4717:     /**
felix@4717:      * Return the data parsed.
felix@4717:      * @return map of stations (km) to list of points.
felix@4717:      */
felix@4715:     @Override
felix@4681:     public Map<Double, List<XY>> getData() {
felix@4681:         return data;
felix@4681:     }
felix@4681: 
felix@4681: 
felix@4791:     /**
felix@4791:      * Walk a directory tree, parse its *.da66 files and store the
felix@4791:      * data found.
felix@4791:      */
felix@4681:     public void parseDA66s(File root, final Callback callback) {
felix@4681: 
felix@4681:         FileTools.walkTree(root, new FileTools.FileVisitor() {
felix@4681:             @Override
felix@4681:             public boolean visit(File file) {
felix@4681:                 if (file.isFile() && file.canRead()
felix@4681:                 && file.getName().toLowerCase().endsWith(".d66")
felix@4715:                 && (callback == null || callback.accept(file))) {
felix@4681:                     reset();
felix@4681:                     try {
felix@4681:                         parse(file);
felix@4681:                         logger.info("parsing done");
felix@4681:                         if (callback != null) {
felix@4715:                             callback.parsed(DA66Parser.this);
felix@4681:                         }
felix@4681:                     }
felix@4681:                     catch (IOException ioe) {
felix@4681:                         logger.error("IOException while parsing file");
felix@4681:                         return false;
felix@4681:                     }
felix@4681:                 }
felix@4681:                 return true;
felix@4681:             }
felix@4681:         });
felix@4681:     }
felix@4681: 
felix@4681: 
felix@4712:     /**
felix@4712:      * Get the Index of the last cross-section lines point.
felix@4712:      * @return last points index, -1 if not available.
felix@4712:      */
felix@4712:     private int lastPointIdx() {
felix@4712:         if (currentLine == null || currentLine.isEmpty()) {
felix@4712:             return -1;
felix@4712:         }
felix@4712:         XY lastPoint = this.currentLine.get(currentLine.size()-1);
felix@4712:         return lastPoint.getIndex();
felix@4712:     }
felix@4712: 
felix@4712: 
felix@4720:     /** Returns station, deciding if it could in cm, in which case convert. */
felix@4720:     private double stationInKm(double station) {
felix@4720:         if (station > 10000) {
felix@4720:             return station/100000d;
felix@4720:         }
felix@4720:         else {
felix@4720:             return station;
felix@4720:         }
felix@4720:     }
felix@4720: 
felix@4791: 
felix@4746:     /** Apply the convention how to deal with numbers < -99.999 .*/
felix@4746:     private String applyLetterConvention(String orig) {
felix@4746:         if (orig.endsWith("-")) {
felix@4746:             return "-" + orig.replace("-","");
felix@4746:         }
felix@4746:         else if (orig.endsWith("J")) {
felix@4746:             return "-" + orig.replace("J","1");
felix@4746:         }
felix@4746:         else if (orig.endsWith("K")) {
felix@4746:             return "-" + orig.replace("K","2");
felix@4746:         }
felix@4746:         else if (orig.endsWith("L")) {
felix@4746:             return "-" + orig.replace("L","3");
felix@4746:         }
felix@4746:         else if (orig.endsWith("M")) {
felix@4746:             return "-" + orig.replace("M","4");
felix@4746:         }
felix@4746:         else if (orig.endsWith("N")) {
felix@4746:             return "-" + orig.replace("N","5");
felix@4746:         }
felix@4746:         else if (orig.endsWith("O")) {
felix@4746:             return "-" + orig.replace("O","6");
felix@4746:         }
felix@4746:         else if (orig.endsWith("P")) {
felix@4746:             return "-" + orig.replace("P","7");
felix@4746:         }
felix@4746:         else if (orig.endsWith("Q")) {
felix@4746:             return "-" + orig.replace("Q","8");
felix@4746:         }
felix@4746:         else if (orig.endsWith("R")) {
felix@4746:             return "-" + orig.replace("R","9");
felix@4746:         }
felix@4746:         else {
felix@4746:             return orig;
felix@4746:         }
felix@4746:     }
felix@4720: 
felix@4712:     /**
felix@4712:      * Add a Point (YZ,Index) to the current cross section line.
felix@4712:      * @param y The y coordinate of new point.
felix@4712:      * @param z The z coordinate of new point.
felix@4712:      * @param idx Ignored, the parameter of new point.
felix@4712:      * @return true if point could been added, false otherwise (e.g. not
felix@4712:      *         parsable y or z values.
felix@4712:      */
felix@4712:     private boolean addPoint(String y, String z, String idx) {
felix@4712:         if (z == null || y == null || idx == null) {
felix@4712:             logger.error("Incomplete point definition");
felix@4712:             return false;
felix@4712:         }
felix@4712: 
teichmann@4735:         double iy;
teichmann@4735:         double iz;
felix@4746:         // Handle letter convention.
felix@4746:         y = applyLetterConvention(y);
felix@4712:         try {
felix@4727:             iy = Double.parseDouble(y) / 1000d;
felix@4727:             iz = Double.parseDouble(z) / 1000d;
felix@4712:         }
felix@4712:         catch(java.lang.NumberFormatException nfe) {
felix@4712:             logger.error("Could not parse Number: " + nfe.getMessage());
felix@4712:             return false;
felix@4712:         }
felix@4712: 
felix@4712:         // We ignore idx, and increment instead.
teichmann@4735:         int index;
felix@4712:         int lastPointIdx = lastPointIdx();
felix@4712:         if (lastPointIdx <= 0) {
felix@4712:             index = 1;
felix@4712:         } else {
felix@4712:             index = lastPointIdx + 1;
felix@4712:         }
felix@4712: 
felix@4712:         currentLine.add(new XY(iy, iz, index));
felix@4712:         return true;
felix@4712:     }
felix@4712: 
felix@4712: 
felix@4712:     /** Called before consuming first line of file. */
felix@4712:     public void reset() {
felix@4712:         data.clear();
felix@4712:         currentLine = new ArrayList<XY>();
felix@4712:     }
felix@4712: 
felix@4712: 
felix@4712:     /**
felix@4712:      * Called for each line. Try to extract info from a da66 line.
felix@4712:      */
felix@4681:     @Override
felix@4681:     protected void handleLine(int lineNum, String line) {
felix@4712:         String head = line.substring(0,2);
felix@4712:         if (HEAD_HEAD.equals(head)) {
felix@4715:                 //logger.debug("New station");
felix@4712:                 Matcher m = LINE_PATTERN.matcher(line);
felix@4712:                 if (m.find()) {
felix@4712:                     // Actually matches!
felix@4792:                     // TODO 'move' last line to match river axis
felix@4792:                     // TODO find river axis intersection
felix@4712:                     currentLine = new ArrayList<XY>();
felix@4720:                     double station = stationInKm(Double.parseDouble(m.group(FIELD.STATION.getIdx())));
felix@4720:                     data.put(station, currentLine);
felix@4712:                 }
felix@4712:                 else {
felix@4712:                     logger.error("HEAD line bad.");
felix@4712:                 }
felix@4681:         }
felix@4712:         else if (HEAD_GEOM.equals(head)) {
felix@4681:             Matcher m = LINE_PATTERN.matcher(line);
felix@4712:             if (m.find()) {
felix@4712:                 //logger.info("Station: " + m.group(FIELD.STATION.getIdx()));
felix@4712:                 // TODO if last station differs, error and abort
felix@4712:                 if (m.group(FIELD.POINT_1_ID.getIdx()) != null) {
felix@4712:                     // Point 1
felix@4712:                     if(addPoint(
felix@4712:                         m.group(FIELD.POINT_1_Y.getIdx()),
felix@4712:                         m.group(FIELD.POINT_1_Z.getIdx()),
felix@4712:                         m.group(FIELD.POINT_1_ID.getIdx()))) {
felix@4712:                         // Point added.
felix@4712:                     }
felix@4712:                     else {
felix@4712:                         // Problematic point.
felix@4712:                         logger.error("A point could not be added");
felix@4712:                     }
felix@4712:                 }
felix@4712:                 if (m.group(FIELD.POINT_2_ID.getIdx()) != null) {
felix@4712:                     // Point 2
felix@4712:                     if(addPoint(
felix@4712:                         m.group(FIELD.POINT_2_Y.getIdx()),
felix@4712:                         m.group(FIELD.POINT_2_Z.getIdx()),
felix@4712:                         m.group(FIELD.POINT_2_ID.getIdx()))) {
felix@4712:                         // Point added.
felix@4712:                     }
felix@4712:                     else {
felix@4712:                         // Problematic point.
felix@4712:                         logger.error("A point could not be added");
felix@4712:                     }
felix@4712:                 }
felix@4712:                 if (m.group(FIELD.POINT_3_ID.getIdx()) != null) {
felix@4712:                     // Point 3
felix@4712:                     if(addPoint(
felix@4712:                         m.group(FIELD.POINT_3_Y.getIdx()),
felix@4712:                         m.group(FIELD.POINT_3_Z.getIdx()),
felix@4712:                         m.group(FIELD.POINT_3_ID.getIdx()))) {
felix@4712:                         // Point added.
felix@4712:                     }
felix@4712:                     else {
felix@4712:                         // Problematic point.
felix@4712:                         logger.error("A point could not be added");
felix@4712:                     }
felix@4712:                 }
felix@4712:                 if (m.group(FIELD.POINT_4_ID.getIdx()) != null) {
felix@4712:                     // Point 4
felix@4712:                     if(addPoint(
felix@4712:                         m.group(FIELD.POINT_4_Y.getIdx()),
felix@4712:                         m.group(FIELD.POINT_4_Z.getIdx()),
felix@4712:                         m.group(FIELD.POINT_4_ID.getIdx()))) {
felix@4712:                         // Point added.
felix@4712:                     }
felix@4712:                     else {
felix@4712:                         // Problematic point.
felix@4712:                         logger.error("A point could not be added");
felix@4712:                     }
felix@4712:                 }
felix@4712:             }
felix@4712:             else {
felix@4712:                 logger.warn("Line could not be parsed: ");
felix@4712:                 logger.warn(line);
felix@4712:             }
felix@4681:         }
felix@4712:         else if (HEAD_GEOM.equals(head)) {
felix@4712:             logger.debug("Hit a 88");
felix@4712:         }
felix@4712:         else {
felix@4712:             logger.error("Do not know how to treat da66 line:");
felix@4712:             logger.error(line);
felix@4712:         }
felix@4712:     }
felix@4712: 
felix@4712: 
felix@4712:     /** Called when file is fully consumed. */
felix@4712:     @Override
felix@4712:     protected void finish() {
felix@4792:         // TODO 'move' last line to match river axis
felix@4712:         logger.info("Parsed " + data.size() + " lines");
felix@4681:     }
felix@4681: 
felix@4681: 
felix@4681:     /** Parses files given as arguments. */
felix@4681:     public static void main(String [] args) {
felix@4681: 
felix@4681:         DA66Parser parser = new DA66Parser();
felix@4681: 
felix@4719:         logger.warn("Start parsing files.");
felix@4681:         for (String arg: args) {
felix@4681:             parser.parseDA66s(new File(arg), null);
felix@4681:             logger.warn("Parsing a file.");
felix@4681:         }
felix@4719:         logger.error("Finished parsing files.");
felix@4681:     }
felix@4681: }
felix@4681: // vim:set ts=4 sw=4 si et sta sts=4 fenc=utf8 :