Explaining the Python memoryview

In Python, a "memoryview" is a built-in type that provides a zero-copy, efficient way to access and manipulate binary data stored in objects like "bytes", "bytearray", and other buffer-supporting objects.

It allows direct access to memory buffers without copying the actual data, making it useful for performance optimization when working with large binary datasets.

Creating a "memoryview"

A "memoryview" can be created from any bytes-like object ("bytes", "bytearray", etc.).

data = bytes([0, 1, 2, 3, 4])  # Immutable bytes object
view = memoryview(data)
print(view)  # Output: <memory at 0x...>

Accessing Data in a "memoryview"

A "memoryview" acts like a read-only sequence of bytes when created from an immutable object like "bytes".

print(view[0])  # Output: 0
print(view[1])  # Output: 1

• The "memoryview" does not copy "data", it just provides a view into its memory.

Modifying Data (Using "bytearray")

If created from a mutable object like "bytearray", the "memoryview" allows modifications without copying data.

data = bytearray([0, 1, 2, 3, 4])  # Mutable bytearray
view = memoryview(data)
view[0] = 255  # Modify first byte
print(data)  # Output: bytearray(b'\xff\x01\x02\x03\x04')

• Efficient: No new copy is created; the original "bytearray" is updated.

Slicing a "memoryview"

A "memoryview" supports slicing, which provides a view of part of the memory.

sub_view = view[1:4]
print(sub_view.tolist())  # Output: [1, 2, 3]

• ".tolist()" converts the "memoryview" to a Python list for easy viewing.

Converting a "memoryview" to Bytes

To get back a regular "bytes" object:

bytes_data = bytes(view)
print(bytes_data)  # Output: b'\xff\x01\x02\x03\x04'

• This copies the data into a new "bytes" object.

Checking Memory Layout (".format", ".shape", ".itemsize")

A "memoryview" supports multi-dimensional data structures (like NumPy arrays).

m = memoryview(bytearray([0, 1, 2, 3, 4])).cast("B")  # "B" = unsigned byte
print(m.format)    # Output: 'B' (byte format)
print(m.shape)     # Output: (5,) (one-dimensional)
print(m.itemsize)  # Output: 1 (each item is 1 byte)

When to Use "memoryview"?