CampbellTimeSeriesParser.java

package nl.wldelft.timeseriesparsers;

import nl.wldelft.util.io.LineReader;
import nl.wldelft.util.io.TextParser;
import nl.wldelft.util.timeseries.DefaultTimeSeriesHeader;
import nl.wldelft.util.timeseries.TimeSeriesContentHandler;
import org.apache.log4j.Logger;

import java.io.IOException;

public class CampbellTimeSeriesParser implements TextParser<TimeSeriesContentHandler> {
    private int columnCount = 0;
    private LineReader reader = null;
    private TimeSeriesContentHandler contentHandler = null;
    private boolean skipValues[] = null;
    private static final Logger log = Logger.getLogger(CampbellTimeSeriesParser.class);


    @Override
    public void parse(LineReader reader, String virtualFileName, TimeSeriesContentHandler contentHandler) throws IOException {
        this.contentHandler = contentHandler;
        this.contentHandler.addMissingValue("NAN");

        this.reader = reader;
        parseHeader();

        for (String[] buffer = new String[columnCount]; this.reader.readLine(',', '\"', buffer) != -1; ) {
            this.contentHandler.setTime(this.contentHandler.getDefaultTimeZone(), "yyyy-MM-dd HH:mm:ss", buffer[0]);
            for (int i = 1; i < columnCount; i++) {
                if (!skipValues[i]) {
                    this.contentHandler.setTimeSeriesHeader(i);
                    String valueText = buffer[i];
                    this.contentHandler.setValue('.', valueText);
                    this.contentHandler.applyCurrentFields();
                }
            }
        }
    }

    // The first few lines contain vital information about the file:
    // - Whether the separator character is a , or a ;
    // - The names of the parameters and locations

    private void parseHeader() throws IOException {

        String[] headerLine1 = reader.readLine(',');
        String[] parameterIdsLine = reader.readLine(',');
        String[] unitLine = reader.readLine(',');
        String[] qualifierLine = reader.readLine(',');
        skipValues = new boolean[qualifierLine.length];

        columnCount = parameterIdsLine.length;

        DefaultTimeSeriesHeader header = new DefaultTimeSeriesHeader();
        header.setLocationId(headerLine1[1]);
        if (header.getLocationId() == null) {
            throw new IOException("Location id is missing for first header line ");
        }
        if (!"TIMESTAMP".equals(parameterIdsLine[0])) {
            throw new IOException("First parameter is not TIMESTAMP");
        }
        for (int i = 1; i < columnCount; i++) {
            header.setParameterId(parameterIdsLine[i]);
            header.setUnit(unitLine[i]);
            if ("TMx".equals(qualifierLine[i]) || "TMn".equals(qualifierLine[i])) {
                skipValues[i] = true;
                log.warn("parameter " + parameterIdsLine[i] + " will be skipped because of unknown use");
            } else {
                header.setQualifierIds(qualifierLine[i]);
            }
            if (header.getParameterId() == null) {
                throw new IOException("Parameter id is missing for column " + (i + 1));
            }

            contentHandler.createTimeSeriesHeaderAlias(i, header);

        }
    }
}