package org.djunits.vecmat.storage;
   
   import java.util.Arrays;
   import java.util.Collection;
   import java.util.Objects;
   
   import org.djunits.quantity.def.Quantity;
   import org.djutils.exceptions.Throw;
   
  /**
   * SparseFloatData implements a sparse data grid for N x M matrices or N x 1 or 1 x N vectors with float values. The sparse grid
   * is implemented with an index array that indicates the position of the data values in the dense array. Any index that is
   * missing indicates a data value of 0.
   * <p>
   * Copyright (c) 2025-2026 Delft University of Technology, Jaffalaan 5, 2628 BX Delft, the Netherlands. All rights reserved. See
   * for project information <a href="https://djunits.org" target="_blank">https://djunits.org</a>. The DJUNITS project is
   * distributed under a <a href="https://djunits.org/docs/license.html" target="_blank">three-clause BSD-style license</a>.
   * @author Alexander Verbraeck
   */
  public class SparseFloatDataSi implements DataGridSi<SparseFloatDataSi>
  {
      /** The sparse data stored in row-major format, where only non-zero values are stored. */
      private float[] sparseData;
  
      /** The index array, giving the position in the dense data array of the values in data. */
      private int[] indexes;
  
      /** The number of rows. */
      private final int rows;
  
      /** the number of columns. */
      private final int cols;
  
      /**
       * Instantiate a data object with one array in row-major format. Note that NO safe copy of the data is stored. Also note
       * that get(r, c) methods will be 0-based in this underlying class, whereas the Matrix.get(r, c) method is 1-based.
       * @param sparseData the sparse data values
       * @param indexes the indexes with the data coordinates, where index = row * cols() + col (0-based)
       * @param rows the number of rows (1-based)
       * @param cols the number of columns (1-based)
       * @throws IllegalArgumentException when the size of the data object is not equal to rows*cols, or when the number of rows
       *             or columns is not positive, or when indexes is not in strictly increasing order
       * @throws IndexOutOfBoundsException when one of the entries in indexes is out of bounds
       */
      protected SparseFloatDataSi(final float[] sparseData, final int[] indexes, final int rows, final int cols)
      {
          Throw.whenNull(sparseData, "sparseData");
          Throw.whenNull(indexes, "indexes");
          Throw.when(rows <= 0, IllegalArgumentException.class, "Number of rows <= 0");
          Throw.when(cols <= 0, IllegalArgumentException.class, "Number of columns <= 0");
          Throw.when(sparseData.length != indexes.length, IllegalArgumentException.class,
                  "sparseData array (%d) has different length from indexes array (%d)", sparseData.length, indexes.length);
          this.rows = rows;
          this.cols = cols;
          checkIndexes(indexes);
          this.sparseData = sparseData;
          this.indexes = indexes;
      }
  
      /**
       * Instantiate a data object with one array in row-major format. Note that NO safe copy of the data is stored.
       * @param denseData the dense data as double values in row-major format
       * @param rows the number of rows
       * @param cols the number of columns
       * @throws IllegalArgumentException when the size of the data object is not equal to rows*cols, or when the number of rows
       *             or columns is not positive
       */
      public SparseFloatDataSi(final double[] denseData, final int rows, final int cols)
      {
          Throw.whenNull(denseData, "denseData");
          Throw.when(rows <= 0, IllegalArgumentException.class, "Number of rows <= 0");
          Throw.when(cols <= 0, IllegalArgumentException.class, "Number of columns <= 0");
          Throw.when(denseData.length != rows * cols, IllegalArgumentException.class,
                  "denseData.length (%d) != rows x cols (%d x %d)", denseData.length, rows, cols);
          this.rows = rows;
          this.cols = cols;
          storeSparse(denseData);
      }
  
      /**
       * Instantiate a data object with one array in row-major format. Note that NO safe copy of the data is stored.
       * @param denseData the dense data as float values in row-major format
       * @param rows the number of rows
       * @param cols the number of columns
       * @throws IllegalArgumentException when the size of the data object is not equal to rows*cols, or when the number of rows
       *             or columns is not positive
       */
      public SparseFloatDataSi(final float[] denseData, final int rows, final int cols)
      {
          Throw.whenNull(denseData, "denseData");
          Throw.when(rows <= 0, IllegalArgumentException.class, "Number of rows <= 0");
          Throw.when(cols <= 0, IllegalArgumentException.class, "Number of columns <= 0");
          Throw.when(denseData.length != rows * cols, IllegalArgumentException.class,
                  "denseData.length (%d) != rows x cols (%d x %d)", denseData.length, rows, cols);
          this.rows = rows;
          this.cols = cols;
          storeSparse(denseData);
      }
  
     /**
      * Instantiate a data object with a dense double[rows][cols]. A sparse, safe copy of the data is stored.
      * @param denseData the data in row-major format as a double[][]
      * @throws IllegalArgumentException when the matrix is ragged
      */
     @SuppressWarnings("checkstyle:needbraces")
     public SparseFloatDataSi(final double[][] denseData)
     {
         Throw.whenNull(denseData, "denseData");
         Throw.when(denseData.length == 0, IllegalArgumentException.class, "Number of rows in the data matrix = 0");
         this.rows = denseData.length;
         this.cols = denseData[0].length;
         for (int r = 1; r < this.rows; r++)
             Throw.when(denseData[r].length != this.cols, IllegalArgumentException.class,
                     "Number of columns in row %d (%d) is not equal to number of columns in row 0 (%d)", r, denseData[r].length,
                     this.cols);
         storeSparse(denseData);
     }
 
     /**
      * Instantiate a data object with a dense double[rows][cols]. A sparse, safe copy of the data is stored.
      * @param denseData the data in row-major format as a float[][]
      * @throws IllegalArgumentException when the matrix is ragged
      */
     @SuppressWarnings("checkstyle:needbraces")
     public SparseFloatDataSi(final float[][] denseData)
     {
         Throw.whenNull(denseData, "denseData");
         Throw.when(denseData.length == 0, IllegalArgumentException.class, "Number of rows in the data matrix = 0");
         this.rows = denseData.length;
         this.cols = denseData[0].length;
         for (int r = 1; r < this.rows; r++)
             Throw.when(denseData[r].length != this.cols, IllegalArgumentException.class,
                     "Number of columns in row %d (%d) is not equal to number of columns in row 0 (%d)", r, denseData[r].length,
                     this.cols);
         storeSparse(denseData);
     }
 
     /**
      * Instantiate a data object with one array in row-major format. Note that a safe copy of the data is stored.
      * @param sparseData the sparse data in row-major format
      * @param indexes the indexes with the data coordinates, where index = row * cols() + col (0-based)
      * @param rows the number of rows
      * @param cols the number of columns
      * @throws IllegalArgumentException when the size of the data object is not equal to rows*cols, or when the number of rows
      *             or columns is not positive, or when indexes is not in strictly increasing order
      * @throws IndexOutOfBoundsException when one of the entries in indexes is out of bounds
      * @param <Q> the quantity type
      */
     @SuppressWarnings("checkstyle:needbraces")
     public <Q extends Quantity<Q>> SparseFloatDataSi(final Q[] sparseData, final int[] indexes, final int rows, final int cols)
     {
         Throw.whenNull(sparseData, "sparseData");
         Throw.whenNull(indexes, "indexes");
         Throw.when(rows <= 0, IllegalArgumentException.class, "Number of rows <= 0");
         Throw.when(cols <= 0, IllegalArgumentException.class, "Number of columns <= 0");
         Throw.when(sparseData.length != indexes.length, IllegalArgumentException.class,
                 "sparseData array (%d) has different length from indexes array (%d)", sparseData.length, indexes.length);
         this.rows = rows;
         this.cols = cols;
         checkIndexes(indexes);
         this.sparseData = new float[sparseData.length];
         for (int i = 0; i < sparseData.length; i++)
             this.sparseData[i] = sparseData[i].floatValue();
         this.indexes = indexes.clone();
     }
 
     /**
      * Instantiate a data object with one array in row-major format. Note that a safe copy of the data is stored.
      * @param denseData the dense data in row-major format
      * @param rows the number of rows
      * @param cols the number of columns
      * @throws IllegalArgumentException when the size of the data object is not equal to rows*cols, or when the number of rows
      *             or columns is not positive
      * @param <Q> the quantity type
      */
     public <Q extends Quantity<Q>> SparseFloatDataSi(final Q[] denseData, final int rows, final int cols)
     {
         Throw.whenNull(denseData, "denseData");
         Throw.when(rows <= 0, IllegalArgumentException.class, "Number of rows <= 0");
         Throw.when(cols <= 0, IllegalArgumentException.class, "Number of columns <= 0");
         Throw.when(denseData.length != rows * cols, IllegalArgumentException.class,
                 "denseData.length (%d) != rows x cols (%d x %d)", denseData.length, rows, cols);
         this.rows = rows;
         this.cols = cols;
         storeSparse(denseData);
     }
 
     /**
      * Instantiate a data object with a dense double[rows][cols]. A sparse, safe copy of the data is stored.
      * @param denseData the data in row-major format as a double[][]
      * @throws IllegalArgumentException when the data matrix is ragged
      * @param <Q> the quantity type
      */
     @SuppressWarnings("checkstyle:needbraces")
     public <Q extends Quantity<Q>> SparseFloatDataSi(final Q[][] denseData)
     {
         Throw.whenNull(denseData, "denseData");
         Throw.when(denseData.length == 0, IllegalArgumentException.class, "Number of rows in the data matrix = 0");
         this.rows = denseData.length;
         this.cols = denseData[0].length;
         for (int r = 1; r < this.rows; r++)
             Throw.when(denseData[r].length != this.cols, IllegalArgumentException.class,
                     "Number of columns in row %d (%d) is not equal to number of columns in row 0 (%d)", r, denseData[r].length,
                     this.cols);
         storeSparse(denseData);
     }
 
     /**
      * Instantiate a data object with an indexed collection of values. Note that a safe copy of the data is stored.
      * @param indexedData the sparse data in an indexed format
      * @param rows the number of rows
      * @param cols the number of columns
      * @throws IndexOutOfBoundsException when a row or column index of an element is out of bounds
      * @param <Q> the quantity type
      */
     @SuppressWarnings("checkstyle:needbraces")
     public <Q extends Quantity<Q>> SparseFloatDataSi(final Collection<FloatSparseValue<Q>> indexedData, final int rows,
             final int cols)
     {
         Throw.whenNull(indexedData, "indexedData");
         Throw.when(rows <= 0, IllegalArgumentException.class, "Number of rows <= 0");
         Throw.when(cols <= 0, IllegalArgumentException.class, "Number of columns <= 0");
         this.rows = rows;
         this.cols = cols;
         this.sparseData = new float[indexedData.size()];
         this.indexes = new int[indexedData.size()];
         int index = 0;
         for (var value : indexedData)
         {
             Throw.when(value.getRow() >= rows, IndexOutOfBoundsException.class, "Row index for indexed value %s out of bounds",
                     value.toString());
             Throw.when(value.getColumn() >= cols, IndexOutOfBoundsException.class,
                     "Column index for indexed value %s out of bounds", value.toString());
             this.sparseData[index] = (float) value.si();
             this.indexes[index++] = value.getRow() * this.cols + value.getColumn();
         }
     }
 
     /**
      * Check the correctness of the indexes array.
      * @param indexArray the indexes with the data coordinates, where index = row * cols() + col (0-based)
      * @throws IllegalArgumentException when indexes is not in strictly increasing order
      * @throws IndexOutOfBoundsException when one of the entries in indexes is out of bounds
      */
     protected void checkIndexes(final int[] indexArray)
     {
         for (int i = 0; i < indexArray.length; i++)
         {
             if (indexArray[i] < 0 || indexArray[i] >= this.rows * this.cols)
             {
                 throw new IndexOutOfBoundsException(
                         String.format("indexes[%d] out of bounds, %d rows x %d cols; value should be 0..%d", i, this.rows,
                                 this.cols, this.rows * this.cols - 1));
             }
         }
         for (int i = 1; i < indexArray.length; i++)
         {
             if (indexArray[i] <= indexArray[i - 1])
             {
                 throw new IllegalArgumentException(
                         "indexes[] must be strictly increasing, found " + indexArray[i - 1] + " then " + indexArray[i]);
             }
         }
     }
 
     /**
      * Store sparse data[] and indexes[].
      * @param denseData the dense data in row-major format
      */
     @SuppressWarnings("checkstyle:needbraces")
     public void storeSparse(final double[] denseData)
     {
         int nonzero = 0;
         for (int i = 0; i < denseData.length; i++)
             if (denseData[i] != 0.0)
                 nonzero++;
         this.sparseData = new float[nonzero];
         this.indexes = new int[nonzero];
         int index = 0;
         for (int i = 0; i < denseData.length; i++)
             if (denseData[i] != 0.0)
             {
                 this.sparseData[index] = (float) denseData[i];
                 this.indexes[index] = i;
                 index++;
             }
     }
 
     /**
      * Store sparse data[] and indexes[].
      * @param denseData the dense data in row-major format
      */
     @SuppressWarnings("checkstyle:needbraces")
     public void storeSparse(final float[] denseData)
     {
         int nonzero = 0;
         for (int i = 0; i < denseData.length; i++)
             if (denseData[i] != 0.0f)
                 nonzero++;
         this.sparseData = new float[nonzero];
         this.indexes = new int[nonzero];
         int index = 0;
         for (int i = 0; i < denseData.length; i++)
             if (denseData[i] != 0.0f)
             {
                 this.sparseData[index] = denseData[i];
                 this.indexes[index] = i;
                 index++;
             }
     }
 
     /**
      * Store sparse data[] and indexes[].
      * @param denseData the dense data in row-major format
      */
     @SuppressWarnings("checkstyle:needbraces")
     public void storeSparse(final double[][] denseData)
     {
         int nonzero = 0;
         for (int i = 0; i < denseData.length; i++)
             for (int j = 0; j < denseData[i].length; j++)
                 if (denseData[i][j] != 0.0)
                     nonzero++;
         this.sparseData = new float[nonzero];
         this.indexes = new int[nonzero];
         int index = 0;
         for (int i = 0; i < denseData.length; i++)
             for (int j = 0; j < denseData[i].length; j++)
                 if (denseData[i][j] != 0.0)
                 {
                     this.sparseData[index] = (float) denseData[i][j];
                     this.indexes[index] = i * this.cols + j;
                     index++;
                 }
     }
 
     /**
      * Store sparse data[] and indexes[].
      * @param denseData the dense data in row-major format
      */
     @SuppressWarnings("checkstyle:needbraces")
     public void storeSparse(final float[][] denseData)
     {
         int nonzero = 0;
         for (int i = 0; i < denseData.length; i++)
             for (int j = 0; j < denseData[i].length; j++)
                 if (denseData[i][j] != 0.0f)
                     nonzero++;
         this.sparseData = new float[nonzero];
         this.indexes = new int[nonzero];
         int index = 0;
         for (int i = 0; i < denseData.length; i++)
             for (int j = 0; j < denseData[i].length; j++)
                 if (denseData[i][j] != 0.0f)
                 {
                     this.sparseData[index] = (float) denseData[i][j];
                     this.indexes[index] = i * this.cols + j;
                     index++;
                 }
     }
 
     /**
      * Store sparse data[] and indexes[].
      * @param denseData the dense data in row-major format
      * @param <Q> the quantity type
      */
     @SuppressWarnings("checkstyle:needbraces")
     public <Q extends Quantity<Q>> void storeSparse(final Q[] denseData)
     {
         int nonzero = 0;
         for (int i = 0; i < denseData.length; i++)
             if (denseData[i].ne0())
                 nonzero++;
         this.sparseData = new float[nonzero];
         this.indexes = new int[nonzero];
         int index = 0;
         for (int i = 0; i < denseData.length; i++)
             if (denseData[i].ne0())
             {
                 this.sparseData[index] = denseData[i].floatValue();
                 this.indexes[index] = i;
                 index++;
             }
     }
 
     /**
      * Store sparse data[] and indexes[].
      * @param denseData the dense data in row-major format
      * @param <Q> the quantity type
      */
     @SuppressWarnings("checkstyle:needbraces")
     public <Q extends Quantity<Q>> void storeSparse(final Q[][] denseData)
     {
         int nonzero = 0;
         for (int i = 0; i < denseData.length; i++)
             for (int j = 0; j < denseData[i].length; j++)
                 if (denseData[i][j].ne0())
                     nonzero++;
         this.sparseData = new float[nonzero];
         this.indexes = new int[nonzero];
         int index = 0;
         for (int i = 0; i < denseData.length; i++)
             for (int j = 0; j < denseData[i].length; j++)
                 if (denseData[i][j].ne0())
                 {
                     this.sparseData[index] = denseData[i][j].floatValue();
                     this.indexes[index] = i * this.cols + j;
                     index++;
                 }
     }
 
     @Override
     public int rows()
     {
         return this.rows;
     }
 
     @Override
     public int cols()
     {
         return this.cols;
     }
 
     @Override
     public boolean isDense()
     {
         return false;
     }
 
     @Override
     public boolean isDouble()
     {
         return false;
     }
 
     /**
      * Check whether the row and column are within bounds.
      * @param row the row number
      * @param col the column number
      * @throws IndexOutOfBoundsException when row &gt; rows() or col &gt; cols() or row &lt; 0 or col &lt; 0
      */
     private void checkRowCol(final int row, final int col) throws IndexOutOfBoundsException
     {
         Throw.when(row < 0 || row >= this.rows, IndexOutOfBoundsException.class, "row %d not in range 0..%d", row,
                 this.rows - 1);
         Throw.when(col < 0 || col >= this.cols, IndexOutOfBoundsException.class, "column %d not in range 0..%d", col,
                 this.cols - 1);
     }
 
     @Override
     public double get(final int row, final int col)
     {
         checkRowCol(row, col);
         int index = row * this.cols + col; // zero-based
         final int pos = Arrays.binarySearch(this.indexes, index);
         return (pos >= 0) ? this.sparseData[pos] : 0.0;
     }
 
     @SuppressWarnings("checkstyle:needbraces")
     @Override
     public double[] getSiArray()
     {
         double[] denseData = new double[rows() * cols()];
         for (int i = 0; i < this.sparseData.length; i++)
             denseData[this.indexes[i]] = this.sparseData[i];
         return denseData;
     }
 
     @Override
     public double[] unsafeSiArray()
     {
         return getSiArray();
     }
 
     @Override
     public SparseFloatDataSi copy()
     {
         return new SparseFloatDataSi(this.sparseData.clone(), this.indexes.clone(), rows(), cols());
     }
 
     @SuppressWarnings("checkstyle:needbraces")
     @Override
     public int nonZeroCount()
     {
         int result = 0;
         for (int i = 0; i < this.sparseData.length; i++)
             result += this.sparseData[i] == 0.0 ? 0 : 1;
         return result;
     }
 
     @Override
     public SparseFloatDataSi instantiateNew(final double[] denseData)
     {
         Throw.when(denseData.length != rows() * cols(), IllegalArgumentException.class,
                 "Data object length != rows * cols, %d != %d * %d", denseData.length, rows(), cols());
         return new SparseFloatDataSi(denseData, rows(), cols());
     }
 
     @Override
     public SparseFloatDataSi instantiateNew(final double[] denseData, final int newRows, final int newCols)
     {
         Throw.when(denseData.length != newRows * newCols, IllegalArgumentException.class,
                 "Data object length != rows * cols, %d != %d * %d", denseData.length, newRows, newCols);
         return new SparseFloatDataSi(denseData, newRows, newCols);
     }
 
     @Override
     public int hashCode()
     {
         final int prime = 31;
         int result = 1;
         result = prime * result + Arrays.hashCode(unsafeSiArray());
         result = prime * result + Objects.hash(this.cols, this.rows);
         return result;
     }
 
     @SuppressWarnings("checkstyle:needbraces")
     @Override
     public boolean equals(final Object obj)
     {
         if (this == obj)
             return true;
         if (obj == null)
             return false;
         if (getClass() != obj.getClass())
         {
             if (obj instanceof DataGridSi dg)
                 return this.cols == dg.cols() && this.rows == dg.rows() && Arrays.equals(unsafeSiArray(), dg.unsafeSiArray());
             return false;
         }
         SparseFloatDataSi other = (SparseFloatDataSi) obj;
         return this.cols == other.cols && this.rows == other.rows && Arrays.equals(this.sparseData, other.sparseData)
                 && Arrays.equals(this.indexes, other.indexes);
     }
 
 }