MDArray Access

#-------------------------------------------------------------------------------------
#
#-------------------------------------------------------------------------------------

should "allow data assignment and retrieval" do

  assert_equal(0, @a[1, 2, 3])
  # assign values to a using [] operator
  @a[0, 0, 0] = 10
  @a[0, 1, 2] = 10
  @a[0, 1, 3] = 10
  assert_equal(10, @a[0, 0, 0])
  assert_equal(10, @a[0, 1, 2])
  assert_equal(10, @a[0, 1, 3])
  assert_equal(0, @a[0, 0, 1])

  # assigning a float value to a byte array type will truncate the data
  @a[0, 0, 0] = 1.5
  assert_equal(1, @a[0, 0, 0])

  assert_equal(0.0, @b[1, 1, 1])

  # assign values to b using set method 
  @b.set([1, 1, 1], 5.25)
  assert_equal(5.25, @b.get([1, 1, 1]))

end

#-------------------------------------------------------------------------------------
#
#-------------------------------------------------------------------------------------

should "generate RuntimeError and RangeError when counter is wrong" do

  # When counter error in [], we get a RuntimeError.
  # RuntimeError... dimension error on b
  assert_raise ( RuntimeError ) { @b[0, 0, 0, 0] = 11.34 }
  # RuntimeError... shape of first dimension too big
  assert_raise ( RuntimeError ) { @b[5, 1, 1] = 7.32 }
  # RangeError... shape of second dimension too big
  assert_raise ( RuntimeError ) { @b[1, 3, 1] = 7.32 }

  # When counter error is in get, we get a RangeError.  Method get is faster than
  # method [] as get does not do any counter validation nor convertion. [] does 
  # counter validation and convertion allowing for the use of negative counter
  # and range.
  assert_raise ( RangeError ) { @b.set([0, 0, 0, 0], 11.34) }
  assert_raise ( RangeError ) { @b.set([5, 1, 1], 7.32) }
  assert_raise ( RangeError ) { @b.set([1, 3, 1], 7.32) }

end

#-------------------------------------------------------------------------------------
# It is possible to use negative indexing with [], but not with methods get and set.
#-------------------------------------------------------------------------------------

should "allow working with negative indices" do

  @b[1, -1, -1] =  4.20
  assert_equal(4.20, @b[1, 2, 3])
  @b[-1, -1, -1] = 10.35

  # using negative indices with set raises an exeption...
  assert_raise ( RangeError ) { @b.set([-1, -1, -1], 11.40) }
  # but works with []
  assert_equal(10.35, @b[-1, -1, -1])
  assert_equal(@b[-1, -1, -1], 10.35)

  # using negative indices with get raises an exeption
  assert_raise ( RangeError ) { @b.get([-1, -1, -1]) }
  assert_equal(10.35, @b[1, 2, 3])

  # RuntimeError... shape of first dimension too big (on the negative side)
  assert_raise ( RangeError ) { @b.set([-5, 1, 1], 7.32) }
  # RuntimeError... shape of second dimension too big (on the negative side)
  assert_raise ( RangeError ) { @b.set([1, -4, 1], 7.32) }

end

#-------------------------------------------------------------------------------------
#
#-------------------------------------------------------------------------------------

should "output data in string format" do
  
  a = MDArray.arange(10)
  assert_equal("0 1 2 3 4 5 6 7 8 9 ", a.to_string)

end

#-------------------------------------------------------------------------------------
#
#-------------------------------------------------------------------------------------

should "allow array traversal in cannonical order" do

  a = MDArray.arange(10)

  # set the value at the current index position. Raise an exception as there
  # is no currently defined position.  Note that a[] refers to the current counter
  # position
  assert_raise ( RuntimeError ) { a[] = 18.25 }

  # gets the value at the current index position.  Raises an exception as 
  # current position not defined
  assert_raise ( RuntimeError) { a[] }

  # reset_traversal initializes the traversal, but leaves the counter in an invalid
  # position
  a.reset_traversal
  # move to the first position
  a.next
  assert_equal(0, a[])
  # sets the value at the current position.  No need to give the counter.
  a[] = 10
  assert_equal(10, a[])
  # move to the next position
  a.next
  assert_equal(1, a[])
  # get_next returns the next position and moves the counter. Method get_next and
  # set_next are "fast" methods in the sense that they do not require any counter
  # correction.
  assert_equal(2, a.get_next)
  assert_equal(2, a[])
  assert_equal(3, a.get_next)
  assert_equal(4, a.get_next)

end

#-------------------------------------------------------------------------------------
#
#-------------------------------------------------------------------------------------

should "allow array traversal using each like methods" do

  a = MDArray.arange(10)

  counter = 0

  a.each do |elmt|
    assert_equal(counter, elmt)
    counter += 1
  end

  a.reset_traversal
  a.next
  assert_equal(0, a[])

  # continues the each from where the counter stoped previously
  counter = 1
  a.each_cont do |elmt|
    assert_equal(counter, elmt)
    counter += 1
  end
  
  # each_with_counter returns the elmt and it's index
  a.reshape!([5, 2])
  counter = 0
  a.each_with_index do |elmt, index|
    assert_equal(counter, elmt)
    counter += 1
  end

  # Collect returns a one dimensional ruby array
  b = a.collect { |val| val * 2}

end

#-------------------------------------------------------------------------------------
#
#-------------------------------------------------------------------------------------

should "have the right precision for the type" do

  b = MDArray.float([2, 3], [0, 1, 2, 3, 4, 5])
  assert_equal(1.0, b[0, 1])

  # float should work with 1 decimal precision
  b[0, 1] = 1.5
  assert_equal(1.5, b[0, 1])

  c = MDArray.build("double", [2, 3], [0, 1, 2, 3, 4, 5])
  c[0, 0] = 1.51
  assert_equal(1.51, c[0, 0])

end