package nom.tam.fits;
   
   /*-
    * #%L
    * nom.tam FITS library
    * %%
    * Copyright (C) 1996 - 2024 nom-tam-fits
    * %%
    * This is free and unencumbered software released into the public domain.
   * 
   * Anyone is free to copy, modify, publish, use, compile, sell, or
   * distribute this software, either in source code form or as a compiled
   * binary, for any purpose, commercial or non-commercial, and by any
   * means.
   * 
   * In jurisdictions that recognize copyright laws, the author or authors
   * of this software dedicate any and all copyright interest in the
   * software to the public domain. We make this dedication for the benefit
   * of the public at large and to the detriment of our heirs and
   * successors. We intend this dedication to be an overt act of
   * relinquishment in perpetuity of all present and future rights to this
   * software under copyright law.
   * 
   * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
   * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
   * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
   * IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR
   * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
   * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
   * OTHER DEALINGS IN THE SOFTWARE.
   * #L%
   */
  
  import java.io.IOException;
  import java.nio.Buffer;
  import java.util.Arrays;
  
  import nom.tam.fits.header.Bitpix;
  import nom.tam.fits.header.NonStandard;
  import nom.tam.fits.header.Standard;
  import nom.tam.image.ImageTiler;
  import nom.tam.image.StandardImageTiler;
  import nom.tam.util.ArrayDataInput;
  import nom.tam.util.ArrayDataOutput;
  import nom.tam.util.ArrayFuncs;
  import nom.tam.util.ComplexValue;
  import nom.tam.util.Cursor;
  import nom.tam.util.FitsEncoder;
  import nom.tam.util.Quantizer;
  import nom.tam.util.RandomAccess;
  import nom.tam.util.array.MultiArrayIterator;
  import nom.tam.util.type.ElementType;
  
  /**
   * <p>
   * Image data. Essentially these data are a primitive multi-dimensional array, such as a <code>double[]</code>,
   * <code>float[][]</code>, or <code>short[][][]</code>. Or, as of version 1.20, they may also be {@link ComplexValue}
   * types also.
   * </p>
   * <p>
   * Starting in version 0.9 of the FITS library, this class allows users to defer the reading of images if the FITS data
   * is being read from a file. An {@link ImageTiler} object is supplied which can return an arbitrary subset of the image
   * as a one dimensional array -- suitable for manipulation by standard Java libraries. The image data may not be read
   * from the input until the user calls a method that requires the actual data (e.g. the {@link #getData()} /
   * {@link #getKernel()}, {@link #convertTo(Class)} or {@link #write(ArrayDataOutput)} methods).
   * </p>
   * 
   * @see ImageHDU
   */
  public class ImageData extends Data {
  
      private static final String COMPLEX_TYPE = "COMPLEX";
  
      /**
       * This class describes an array
       */
      protected static class ArrayDesc {
  
          private final Class<?> type;
          private int[] dims;
  
          private Quantizer quant;
  
          private int complexAxis = -1;
  
          ArrayDesc(int[] dims, Class<?> type) {
              this.dims = dims;
              this.type = type;
  
              if (ComplexValue.class.isAssignableFrom(type)) {
                  complexAxis = dims.length;
              }
          }
      }
  
      /**
       * This inner class allows the ImageTiler to see if the user has read in the data.
       */
      protected class ImageDataTiler extends StandardImageTiler {
 
         ImageDataTiler(RandomAccess o, long offset, ArrayDesc d) {
             super(o, offset, d.dims, d.type);
         }
 
         @Override
         protected Object getMemoryImage() {
             return dataArray;
         }
     }
 
     // private static final Logger LOG = getLogger(ImageData.class);
 
     /** The size of the data */
     private long byteSize;
 
     /**
      * The actual array of data. This is normally a multi-dimensional primitive array. It may be null until the
      * getData() routine is invoked, or it may be filled by during the read call when a non-random access device is
      * used.
      */
     private Object dataArray;
 
     /** A description of what the data should look like */
     private ArrayDesc dataDescription;
 
     /** The image tiler associated with this image. */
     private StandardImageTiler tiler;
 
     /**
      * Create the equivalent of a null data element.
      */
     public ImageData() {
         this(new byte[0]);
     }
 
     /**
      * (<i>for internal use</i>) Create an array from a header description. This is typically how data will be created
      * when reading FITS data from a file where the header is read first. This creates an empty array.
      *
      * @param  h             header to be used as a template.
      *
      * @throws FitsException if there was a problem with the header description.
      */
     public ImageData(Header h) throws FitsException {
         dataDescription = parseHeader(h);
     }
 
     /**
      * Create an ImageData object using the specified object to initialize the data array.
      *
      * @param  x                        The initial data array. This should be a primitive array but this is not checked
      *                                      currently.
      * 
      * @throws IllegalArgumentException if x is not a suitable primitive array
      */
     public ImageData(Object x) throws IllegalArgumentException {
         try {
             checkCompatible(x);
         } catch (FitsException e) {
             throw new IllegalArgumentException(e.getMessage(), e);
         }
 
         dataDescription = new ArrayDesc(ArrayFuncs.getDimensions(x), ArrayFuncs.getBaseClass(x));
         dataArray = x;
         byteSize = FitsEncoder.computeSize(x);
     }
 
     @Override
     protected void loadData(ArrayDataInput in) throws IOException, FitsException {
         if (tiler != null) {
             dataArray = tiler.getCompleteImage();
         } else {
             dataArray = ArrayFuncs.newInstance(getType(), getDimensions());
             in.readImage(dataArray);
         }
     }
 
     @Override
     public void read(ArrayDataInput in) throws FitsException {
         tiler = (in instanceof RandomAccess) ?
                 new ImageDataTiler((RandomAccess) in, ((RandomAccess) in).getFilePointer(), dataDescription) :
                 null;
         super.read(in);
     }
 
     @Override
     protected Object getCurrentData() {
         return dataArray;
     }
 
     /**
      * Returns the an image tiler instance that can be used to divide this image into tiles that may be processed
      * separately (and in parallel). A default tiler, which returns the image in memory, is returned for images not
      * associated with a random-accessible input.
      * 
      * @return image tiler for this image instance.
      */
     public StandardImageTiler getTiler() {
         return tiler != null ? tiler : new ImageDataTiler(null, 0, dataDescription);
     }
 
     /**
      * Sets the buffer that may hold a serialized version of the data for this image.
      * 
      * @param data the buffer that may hold this image's data in serialized form.
      */
     public void setBuffer(Buffer data) {
         ElementType<Buffer> elementType = ElementType.forClass(getType());
         dataArray = ArrayFuncs.newInstance(getType(), getDimensions());
         MultiArrayIterator<?> iterator = new MultiArrayIterator<>(dataArray);
         Object array = iterator.next();
         while (array != null) {
             elementType.getArray(data, array);
             array = iterator.next();
         }
         tiler = new ImageDataTiler(null, 0, dataDescription);
     }
 
     @SuppressWarnings({"resource", "deprecation"})
     @Override
     public void write(ArrayDataOutput o) throws FitsException {
         // Don't need to write null data (noted by Jens Knudstrup)
         if (byteSize == 0) {
             return;
         }
 
         if (o != getRandomAccessInput()) {
             ensureData();
         }
 
         try {
             o.writeArray(dataArray);
         } catch (IOException e) {
             throw new FitsException("IO Error on image write" + e);
         }
 
         FitsUtil.pad(o, getTrueSize());
     }
 
     @SuppressWarnings("deprecation")
     @Override
     protected void fillHeader(Header head) throws FitsException {
 
         if (dataArray == null) {
             head.nullImage();
             return;
         }
 
         Standard.context(ImageData.class);
 
         // We'll assume it's a primary image, until we know better...
         // Just in case, we don't want an XTENSION key lingering around...
         head.deleteKey(Standard.XTENSION);
 
         Cursor<String, HeaderCard> c = head.iterator();
         c.add(HeaderCard.create(Standard.SIMPLE, true));
 
         Class<?> base = getType();
         int[] dims = getDimensions();
 
         if (ComplexValue.class.isAssignableFrom(base)) {
             dims = Arrays.copyOf(dims, dims.length + 1);
             dims[dims.length - 1] = 2;
             base = ComplexValue.Float.class.isAssignableFrom(base) ? float.class : double.class;
         }
 
         c.add(HeaderCard.create(Standard.BITPIX, Bitpix.forPrimitiveType(base).getHeaderValue()));
 
         c.add(HeaderCard.create(Standard.NAXIS, dims.length));
         for (int i = 1; i <= dims.length; i++) {
             c.add(HeaderCard.create(Standard.NAXISn.n(i), dims[dims.length - i]));
         }
 
         // Just in case!
         c.add(HeaderCard.create(Standard.PCOUNT, 0));
         c.add(HeaderCard.create(Standard.GCOUNT, 1));
         c.add(HeaderCard.create(Standard.EXTEND, true));
 
         if (isComplexValued()) {
             c.add(HeaderCard.create(Standard.CTYPEn.n(dims.length - dataDescription.complexAxis), COMPLEX_TYPE));
         }
 
         if (dataDescription.quant != null) {
             dataDescription.quant.editImageHeader(head);
         }
 
         Standard.context(null);
     }
 
     @Override
     protected long getTrueSize() {
         return byteSize;
     }
 
     /**
      * Returns the image specification based on its description in a FITS header.
      * 
      * @param  h             the FITS header that describes this image with the standard keywords for an image HDU.
      * 
      * @return               an object that captures the description contained in the header for internal use.
      * 
      * @throws FitsException If there was a problem accessing or interpreting the required header values.
      */
     protected ArrayDesc parseHeader(Header h) throws FitsException {
         String ext = h.getStringValue(Standard.XTENSION, Standard.XTENSION_IMAGE);
 
         if (!ext.equalsIgnoreCase(Standard.XTENSION_IMAGE) && !ext.equalsIgnoreCase(NonStandard.XTENSION_IUEIMAGE)) {
             throw new FitsException("Not an image header (XTENSION = " + h.getStringValue(Standard.XTENSION) + ")");
         }
 
         int gCount = h.getIntValue(Standard.GCOUNT, 1);
         int pCount = h.getIntValue(Standard.PCOUNT, 0);
         if (gCount > 1 || pCount != 0) {
             throw new FitsException("Group data treated as images");
         }
 
         Bitpix bitpix = Bitpix.fromHeader(h);
         Class<?> baseClass = bitpix.getPrimitiveType();
         int ndim = h.getIntValue(Standard.NAXIS, 0);
         int[] dims = new int[ndim];
         // Note that we have to invert the order of the axes
         // for the FITS file to get the order in the array we
         // are generating.
 
         byteSize = ndim > 0 ? 1 : 0;
         for (int i = 1; i <= ndim; i++) {
             int cdim = h.getIntValue(Standard.NAXISn.n(i), 0);
             if (cdim < 0) {
                 throw new FitsException("Invalid array dimension:" + cdim);
             }
             byteSize *= cdim;
             dims[ndim - i] = cdim;
         }
         byteSize *= bitpix.byteSize();
 
         ArrayDesc desc = new ArrayDesc(dims, baseClass);
 
         if (COMPLEX_TYPE.equals(h.getStringValue(Standard.CTYPEn.n(1))) && dims[ndim - 1] == 2) {
             desc.complexAxis = ndim - 1;
         } else if (COMPLEX_TYPE.equals(h.getStringValue(Standard.CTYPEn.n(ndim))) && dims[0] == 2) {
             desc.complexAxis = 0;
         }
 
         desc.quant = Quantizer.fromImageHeader(h);
         if (desc.quant.isDefault()) {
             desc.quant = null;
         }
 
         return desc;
     }
 
     void setTiler(StandardImageTiler tiler) {
         this.tiler = tiler;
     }
 
     /**
      * (<i>for expert users</i>) Overrides the image size description in the header to the specified Java array
      * dimensions. Typically users should not call this method, unless they want to define the image dimensions in the
      * absence of the actual complete image data. For example, to describe the dimensions when using low-level writes of
      * an image row-by-row, without ever storing the entire image in memory.
      * 
      * @param  header                   A FITS image header
      * @param  sizes                    The array dimensions in Java order (fastest varying index last)
      * 
      * @throws FitsException            if the size has negative values, or the header is not that for an image
      * @throws IllegalArgumentException should not actually happen
      * 
      * @since                           1.18
      * 
      * @see                             #fillHeader(Header)
      */
     public static void overrideHeaderAxes(Header header, int... sizes) throws FitsException, IllegalArgumentException {
         String extType = header.getStringValue(Standard.XTENSION, Standard.XTENSION_IMAGE);
         if (!extType.equals(Standard.XTENSION_IMAGE) && !extType.equals(NonStandard.XTENSION_IUEIMAGE)) {
             throw new FitsException("Not an image header (XTENSION = " + extType + ")");
         }
 
         // Remove prior NAXISn values
         int n = header.getIntValue(Standard.NAXIS);
         for (int i = 1; i <= n; i++) {
             header.deleteKey(Standard.NAXISn.n(i));
         }
 
         Cursor<String, HeaderCard> c = header.iterator();
         c.setKey(Standard.NAXIS.key());
 
         c.add(HeaderCard.create(Standard.NAXIS, sizes.length));
 
         for (int i = 1; i <= sizes.length; i++) {
             int l = sizes[sizes.length - i];
             if (l < 0) {
                 throw new FitsException("Invalid size[ " + i + "] = " + l);
             }
             c.add(HeaderCard.create(Standard.NAXISn.n(i), l));
         }
     }
 
     /**
      * Creates a new FITS image using the specified primitive numerical Java array containing data.
      * 
      * @param  data                     A regulatly shaped primitive numerical Java array, which can be
      *                                      multi-dimensional.
      * 
      * @return                          A new FITS image that encapsulates the specified array data.
      * 
      * @throws IllegalArgumentException if the argument is not a primitive numerical Java array.
      * 
      * @since                           1.19
      */
     public static ImageData from(Object data) throws IllegalArgumentException {
         return new ImageData(data);
     }
 
     /**
      * Checks if a given data object may constitute the kernel for an image. To conform, the data must be a regularly
      * shaped primitive numerical array of ant dimensions, or <code>null</code>.
      * 
      * @param  data                     A regularly shaped primitive numerical array of ny dimension, or
      *                                      <code>null</code>
      * 
      * @throws IllegalArgumentException If the array is not regularly shaped.
      * @throws FitsException            If the argument is not a primitive numerical array type
      * 
      * @since                           1.19
      */
     static void checkCompatible(Object data) throws IllegalArgumentException, FitsException {
         if (data != null) {
             Class<?> base = ArrayFuncs.getBaseClass(data);
             if (ComplexValue.Float.class.isAssignableFrom(base)) {
                 base = float.class;
             } else if (ComplexValue.class.isAssignableFrom(base)) {
                 base = double.class;
             }
             Bitpix.forPrimitiveType(base);
             ArrayFuncs.checkRegularArray(data, false);
         }
     }
 
     @Override
     @SuppressWarnings("deprecation")
     public ImageHDU toHDU() throws FitsException {
         Header h = new Header();
         fillHeader(h);
         return new ImageHDU(h, this);
     }
 
     /**
      * Sets the conversion between decimal and integer data representations. The quantizer for the image is set
      * automatically if the image was read from a FITS input, and if any of the associated BSCALE, BZERO, or BLANK
      * keywords were defined in the HDU's header. User may use this methods to set a different quantization or to use no
      * quantization at all when converting between floating-point and integer representations.
      * 
      * @param quant the quantizer that converts between floating-point and integer data representations, or <code>
      *          null</code> to not use quantization and instead rely on simple rounding for decimal-ineger conversions..
      * 
      * @see         #getQuantizer()
      * @see         #convertTo(Class)
      * 
      * @since       1.20
      */
     public void setQuantizer(Quantizer quant) {
         dataDescription.quant = quant;
     }
 
     /**
      * Returns the conversion between decimal and integer data representations.
      * 
      * @return the quantizer that converts between floating-point and integer data representations, which may be
      *             <code>null</code>
      * 
      * @see    #setQuantizer(Quantizer)
      * @see    #convertTo(Class)
      * 
      * @since  1.20
      */
     public final Quantizer getQuantizer() {
         return dataDescription.quant;
     }
 
     /**
      * Returns the element type of this image in its current representation.
      * 
      * @return The element type of this image, such as <code>int.class</code>, <code>double.class</code> or
      *             {@link ComplexValue}<code>.class</code>.
      * 
      * @see    #getDimensions()
      * @see    #isComplexValued()
      * @see    #convertTo(Class)
      * 
      * @since  1.20
      */
     public final Class<?> getType() {
         return dataDescription.type;
     }
 
     /**
      * Returns the dimensions of this image.
      * 
      * @return An array containing the sizes along each data dimension, in Java indexing order. The returned array is
      *             not used internally, and therefore modifying it will not damage the integrity of the image data.
      * 
      * @see    #getType()
      * 
      * @since  1.20
      */
     public final int[] getDimensions() {
         return Arrays.copyOf(dataDescription.dims, dataDescription.dims.length);
     }
 
     /**
      * Checks if the image data is explicitly designated as a complex-valued image. An image may be designated as
      * complex-valued either because it was created with {@link ComplexValue} type data, or because it was read from a
      * FITS file in which one image axis of dimension 2 was designated as an axis containing complex-valued components
      * with the corresponding CTYPEn header keyword set to 'COMPLEX'. The complex-valued deignation checked by this
      * method is not the same as {@link #getType()}, as it does not necesarily mean that the data itself is currently in
      * {@link ComplexValue} type representation. Rather it simply means that this data can be represented as
      * {@link ComplexValue} type, possibly after an appropriate conversion to a {@link ComplexValue} type.
      * 
      * @return <code>true</code> if the data is complex valued or has been explicitly designated as complex valued.
      *             Otherwise <code>false</code>.
      * 
      * @see    #convertTo(Class)
      * @see    #getType()
      * 
      * @since  1.20
      */
     public final boolean isComplexValued() {
         return dataDescription.complexAxis >= 0;
     }
 
     /**
      * Converts this image HDU to another image HDU of a different type, possibly using a qunatizer for the
      * integer-decimal conversion of the data elements. In all other respects, the returned image is identical to the
      * the original. If th conversion is th indetity, it will return itself and the data may remain in deferred mode.
      * 
      * @param  type          The primitive numerical type (e.g. <code>int.class</code> or <code>double.class</code>), or
      *                           else a {@link ComplexValue} type in which data should be represented. Complex
      *                           representations are normally available for data whose first or last CTYPEn axis was
      *                           described as 'COMPLEX' by the FITS header with a dimensionality is 2 corresponfing to a
      *                           pair of real and imaginary data elements. Even without the CTYPEn designation, it is
      *                           always possible to convert to complex all arrays that have a trailing Java dimension
      *                           (NAXIS1 in FITS) equal to 2.
      * 
      * @return               An image HDU containing the same data in the chosen representation by another type. (It may
      *                           be the same as this HDU if the type is unchanged from the original).
      * 
      * @throws FitsException if the data cannot be read from the input.
      * 
      * @see                  #isComplexValued()
      * @see                  ArrayFuncs#convertArray(Object, Class, Quantizer)
      * 
      * @since                1.20
      */
     public ImageData convertTo(Class<?> type) throws FitsException {
         if (type.isAssignableFrom(getType())) {
             return this;
         }
 
         ensureData();
 
         ImageData typed = null;
 
         boolean toComplex = ComplexValue.class.isAssignableFrom(type) && !ComplexValue.class.isAssignableFrom(getType());
 
         if (toComplex && dataDescription.complexAxis == 0) {
             // Special case of converting separate re/im arrays to complex...
 
             // 1. Convert to intermediate floating-point class as necessary (with quantization if any)
             Class<?> numType = ComplexValue.Float.class.isAssignableFrom(type) ? float.class : double.class;
             Object[] t = (Object[]) ArrayFuncs.convertArray(dataArray, numType, getQuantizer());
             ImageData f = new ImageData(ArrayFuncs.decimalsToComplex(t[0], t[1]));
             f.dataDescription.quant = getQuantizer();
 
             // 2. Assemble complex from separate re/im components.
             return f.convertTo(type);
         }
 
         typed = new ImageData(ArrayFuncs.convertArray(dataArray, type, getQuantizer()));
         typed.dataDescription.quant = getQuantizer();
         return typed;
     }
 }