Ejemplo n.º 1
0
  /**
   * Opens the NetCDF dataset at the given location, using the dataset cache if {@code location}
   * represents an NcML aggregation. We cannot use the cache for OPeNDAP or single NetCDF files
   * because the underlying data may have changed and the NetcdfDataset cache may cache a dataset
   * forever. In the case of NcML we rely on the fact that server administrators ought to have set a
   * "recheckEvery" parameter for NcML aggregations that may change with time. It is desirable to
   * use the dataset cache for NcML aggregations because they can be time-consuming to assemble and
   * we don't want to do this every time a map is drawn.
   *
   * @param location The location of the data: a local NetCDF file, an NcML aggregation file or an
   *     OPeNDAP location, {@literal i.e.} anything that can be passed to
   *     NetcdfDataset.openDataset(location).
   * @return a {@link NetcdfDataset} object for accessing the data at the given location.
   * @throws IOException if there was an error reading from the data source.
   */
  private NetcdfDataset openAndAggregateDataset(String location) throws IOException, EdalException {
    NetcdfDataset nc;
    if (location.startsWith("dods://") || location.startsWith("http://")) {
      /*
       * We have a remote dataset
       */
      nc = CdmUtils.openDataset(location);
    } else {
      /*
       * We have a local dataset
       */
      List<File> files = null;
      try {
        files = CdmUtils.expandGlobExpression(location);
      } catch (NullPointerException e) {
        System.out.println("NPE processing location: " + location);
        throw e;
      }
      if (files.size() == 0) {
        throw new EdalException(
            "The location " + location + " doesn't refer to any existing files.");
      }
      if (files.size() == 1) {
        location = files.get(0).getAbsolutePath();
        nc = CdmUtils.openDataset(location);
      } else {
        /*
         * We have multiple files in a glob expression. We write some
         * NcML and use the NetCDF aggregation libs to parse this into
         * an aggregated dataset.
         *
         * If we have already generated the ncML on a previous call,
         * just use that.
         */
        if (ncmlString == null) {
          /*
           * Find the name of the time dimension
           */
          NetcdfDataset first = openAndAggregateDataset(files.get(0).getAbsolutePath());
          String timeDimName = null;
          for (Variable var : first.getVariables()) {
            if (var.isCoordinateVariable()) {
              for (Attribute attr : var.getAttributes()) {
                if (attr.getFullName().equalsIgnoreCase("units")
                    && attr.getStringValue().contains(" since ")) {
                  /*
                   * This is the time dimension. Since this is
                   * a co-ordinate variable, there is only 1
                   * dimension
                   */
                  Dimension timeDimension = var.getDimension(0);
                  timeDimName = timeDimension.getFullName();
                }
              }
            }
          }
          first.close();
          if (timeDimName == null) {
            throw new EdalException("Cannot join multiple files without time dimensions");
          }
          /*
           * We can't assume that the glob expression will have
           * returned the files in time order.
           *
           * We could assume that alphabetical == time ordered (and
           * for properly named files it will - but let's not rely on
           * our users having sensible naming conventions...
           *
           * Sort the list using a comparator which opens the file and
           * gets the first value of the time dimension
           */
          final String aggDimName = timeDimName;
          Collections.sort(
              files,
              new Comparator<File>() {
                @Override
                public int compare(File ncFile1, File ncFile2) {
                  NetcdfFile nc1 = null;
                  NetcdfFile nc2 = null;
                  try {
                    nc1 = NetcdfFile.open(ncFile1.getAbsolutePath());
                    nc2 = NetcdfFile.open(ncFile2.getAbsolutePath());
                    Variable timeVar1 = nc1.findVariable(aggDimName);
                    Variable timeVar2 = nc2.findVariable(aggDimName);
                    long time1 = timeVar1.read().getLong(0);
                    long time2 = timeVar2.read().getLong(0);
                    return Long.compare(time1, time2);
                  } catch (Exception e) {
                    /*
                     * There was a problem reading the data. Sort
                     * alphanumerically by filename and hope for the
                     * best...
                     *
                     * This catches all exceptions because however
                     * it fails this is still our best option.
                     *
                     * If the error is a genuine problem, it'll show
                     * up as soon as we try and aggregate.
                     */
                    return ncFile1.getAbsolutePath().compareTo(ncFile2.getAbsolutePath());
                  } finally {
                    if (nc1 != null) {
                      try {
                        nc1.close();
                      } catch (IOException e) {
                        log.error("Problem closing netcdf file", e);
                      }
                    }
                    if (nc2 != null) {
                      try {
                        nc2.close();
                      } catch (IOException e) {
                        log.error("Problem closing netcdf file", e);
                      }
                    }
                  }
                }
              });

          /*
           * Now create the NcML string and use it to create an
           * aggregated dataset
           */
          StringBuffer ncmlStringBuffer = new StringBuffer();
          ncmlStringBuffer.append(
              "<netcdf xmlns=\"http://www.unidata.ucar.edu/namespaces/netcdf/ncml-2.2\">");
          ncmlStringBuffer.append(
              "<aggregation dimName=\"" + timeDimName + "\" type=\"joinExisting\">");
          for (File file : files) {
            ncmlStringBuffer.append("<netcdf location=\"" + file.getAbsolutePath() + "\"/>");
          }
          ncmlStringBuffer.append("</aggregation>");
          ncmlStringBuffer.append("</netcdf>");

          ncmlString = ncmlStringBuffer.toString();
        }
        nc = NcMLReader.readNcML(new StringReader(ncmlString), null);
      }
    }

    return nc;
  }