teichmann@5829: package org.dive4elements.river.importer.parsers;
sascha@1211: 
sascha@1211: import java.util.HashMap;
sascha@1211: import java.util.TreeSet;
sascha@1211: import java.util.List;
sascha@1211: import java.util.ArrayList;
sascha@1211: 
sascha@1211: import java.io.IOException;
sascha@1211: import java.io.File;
sascha@1211: import java.io.LineNumberReader;
sascha@1211: import java.io.InputStreamReader;
sascha@1211: import java.io.FileInputStream;
sascha@1211: 
sascha@1211: import java.math.BigDecimal;
sascha@1211: 
sascha@1211: import org.apache.log4j.Logger;
sascha@1211: 
teichmann@5829: import org.dive4elements.artifacts.common.utils.FileTools;
sascha@1211: 
teichmann@5829: import org.dive4elements.river.importer.ImportAnnotation;
teichmann@5829: import org.dive4elements.river.importer.ImportRange;
teichmann@5829: import org.dive4elements.river.importer.ImportEdge;
teichmann@5829: import org.dive4elements.river.importer.ImportAnnotationType;
teichmann@5829: import org.dive4elements.river.importer.ImportAttribute;
teichmann@5829: import org.dive4elements.river.importer.ImportPosition;
sascha@1211: 
sascha@1211: public class AnnotationsParser
sascha@1211: {
sascha@1211:     private static Logger log = Logger.getLogger(AnnotationsParser.class);
sascha@1211: 
sascha@1211:     public static final String ENCODING = "ISO-8859-1";
sascha@1211: 
sascha@1211:     public static final String [] TO_SCAN = {
sascha@1211:         "Basisdaten",
ingo@3326:         "Streckendaten",
sascha@3331:         ".." + File.separator +
sascha@3331:             "Morphologie" + File.separator + "Streckendaten"
sascha@1211:     };
sascha@1211: 
sascha@1211:     protected HashMap<String, ImportAttribute> attributes;
sascha@1211:     protected HashMap<String, ImportPosition>  positions;
sascha@1211:     protected TreeSet<ImportAnnotation>        annotations;
sascha@1211:     protected AnnotationClassifier             classifier;
sascha@1211: 
sascha@1211:     public AnnotationsParser() {
sascha@1211:         this(null);
sascha@1211:     }
sascha@1211: 
sascha@1211:     public AnnotationsParser(AnnotationClassifier classifier) {
sascha@1211:         attributes  = new HashMap<String, ImportAttribute>();
sascha@1211:         positions   = new HashMap<String, ImportPosition>();
sascha@1211:         annotations = new TreeSet<ImportAnnotation>();
sascha@1211:         this.classifier = classifier;
sascha@1211:     }
sascha@1211: 
sascha@1211:     public void parseFile(File file) throws IOException {
sascha@1211:         log.info("parsing km file: '" + file + "'");
sascha@1211: 
sascha@1211:         ImportAnnotationType defaultIAT = null;
sascha@1211: 
sascha@1211:         if (classifier != null) {
sascha@1211:             defaultIAT = classifier.classifyFile(
sascha@1211:                 file.getName(),
sascha@1211:                 classifier.getDefaultType());
sascha@1211:         }
sascha@1211: 
sascha@1211:         LineNumberReader in = null;
sascha@1211:         try {
sascha@1211:             in =
sascha@1211:                 new LineNumberReader(
sascha@1211:                 new InputStreamReader(
sascha@1211:                 new FileInputStream(file), ENCODING));
sascha@1211: 
sascha@1211:             String line = null;
sascha@1211:             while ((line = in.readLine()) != null) {
sascha@1211:                 if ((line = line.trim()).length() == 0
sascha@1211:                 || line.startsWith("*")) {
sascha@1211:                     continue;
sascha@1211:                 }
sascha@1211: 
sascha@1211:                 String [] parts = line.split("\\s*;\\s*");
sascha@1211: 
sascha@1211:                 if (parts.length < 3) {
sascha@3662:                     log.warn("ANN: not enough columns in line "
sascha@1211:                         + in.getLineNumber());
sascha@1211:                     continue;
sascha@1211:                 }
sascha@1211: 
sascha@1211:                 ImportPosition position = positions.get(parts[0]);
sascha@1211:                 if (position == null) {
sascha@1211:                     position = new ImportPosition(parts[0]);
sascha@1211:                     positions.put(parts[0], position);
sascha@1211:                 }
sascha@1211: 
sascha@1211:                 ImportAttribute attribute = attributes.get(parts[1]);
sascha@1211:                 if (attribute == null) {
sascha@1211:                     attribute = new ImportAttribute(parts[1]);
sascha@1211:                     attributes.put(parts[1], attribute);
sascha@1211:                 }
sascha@1211: 
sascha@1211:                 String [] r = parts[2].replace(",", ".").split("\\s*#\\s*");
sascha@1211: 
sascha@1211:                 BigDecimal from, to;
sascha@1211: 
sascha@1211:                 try {
sascha@1211:                     from = new BigDecimal(r[0]);
sascha@1211:                     to   = r.length < 2 ? null : new BigDecimal(r[1]);
sascha@1211:                     if (to != null && from.compareTo(to) > 0) {
sascha@1211:                         BigDecimal t = from; from = to; to = t;
sascha@1211:                     }
sascha@1211:                 }
sascha@1211:                 catch (NumberFormatException nfe) {
sascha@3662:                     log.warn("ANN: invalid number in line " + in.getLineNumber());
sascha@1211:                     continue;
sascha@1211:                 }
sascha@1211: 
sascha@1211:                 ImportEdge edge = null;
sascha@1211: 
sascha@1211:                 if (parts.length == 4) { // Only 'Unterkante'
sascha@1211:                     try {
sascha@1211:                         edge = new ImportEdge(
sascha@1211:                             null,
sascha@1211:                             new BigDecimal(parts[3].trim().replace(',', '.')));
sascha@1211:                     }
sascha@1211:                     catch (NumberFormatException nfe) {
sascha@3662:                         log.warn("ANN: cannot parse 'Unterkante' in line " +
sascha@1211:                             in.getLineNumber());
sascha@1211:                     }
sascha@1211:                 }
sascha@1211:                 else if (parts.length > 4) { // 'Unterkante' and 'Oberkante'
sascha@1211:                     String bottom = parts[3].trim().replace(',', '.');
sascha@1211:                     String top    = parts[4].trim().replace(',', '.');
sascha@1211:                     try {
sascha@1211:                         BigDecimal b = bottom.length() == 0
sascha@1211:                             ? null
sascha@1211:                             : new BigDecimal(bottom);
sascha@1211:                         BigDecimal t = top.length() == 0
sascha@1211:                             ? null
sascha@1211:                             : new BigDecimal(top);
sascha@1211:                         edge = new ImportEdge(t, b);
sascha@1211:                     }
sascha@1211:                     catch (NumberFormatException nfe) {
sascha@1211:                         log.warn(
sascha@3662:                             "ANN: cannot parse 'Unterkante' or 'Oberkante' in line "
sascha@1211:                             + in.getLineNumber());
sascha@1211:                     }
sascha@1211:                 }
sascha@1211: 
sascha@1211:                 ImportRange range = new ImportRange(from, to);
sascha@1211: 
sascha@1211:                 ImportAnnotationType type = classifier != null
sascha@1211:                     ? classifier.classifyDescription(line, defaultIAT)
sascha@1211:                     : null;
sascha@1211: 
sascha@1211:                 ImportAnnotation annotation = new ImportAnnotation(
sascha@1211:                     attribute, position, range, edge, type);
sascha@1211: 
sascha@1211:                 if (!annotations.add(annotation)) {
tom@5216:                     log.info("ANN: duplicated annotation '" + parts[0] +
sascha@1211:                         "' in line " + in.getLineNumber());
sascha@1211:                 }
sascha@1211:             }
sascha@1211:         }
sascha@1211:         finally {
sascha@1211:             if (in != null) {
sascha@1211:                 in.close();
sascha@1211:             }
sascha@1211:         }
sascha@1211:     }
sascha@1211: 
sascha@1211:     public void parse(File root) throws IOException {
sascha@1211: 
sascha@1211:         for (String toScan: TO_SCAN) {
sascha@1211:             File directory = FileTools.repair(new File(root, toScan));
sascha@1211:             if (!directory.isDirectory()) {
sascha@3662:                 log.warn("ANN: '" + directory + "' is not a directory.");
sascha@1211:                 continue;
sascha@1211:             }
sascha@1211:             File [] files = directory.listFiles();
sascha@1211:             if (files == null) {
sascha@3662:                 log.warn("ANN: cannot list directory '" + directory + "'");
sascha@1211:                 continue;
sascha@1211:             }
sascha@1211: 
sascha@1211:             for (File file: files) {
sascha@1211:                 if (file.isFile() && file.canRead()
sascha@1211:                 && file.getName().toLowerCase().endsWith(".km")) {
sascha@1211:                     parseFile(file);
sascha@1211:                 }
sascha@1211:             }
sascha@1211:         } // for all directories to scan
sascha@1211:     }
sascha@1211: 
sascha@1211:     public List<ImportAnnotation> getAnnotations() {
sascha@1211:         return new ArrayList<ImportAnnotation>(annotations);
sascha@1211:     }
sascha@1211: }
sascha@1211: // vim:set ts=4 sw=4 si et sta sts=4 fenc=utf8 :