types.rst

.. index:: type

.. _types:

Types
#####

Vyper is a statically typed language. The type of each variable (state and local) must be specified or at least known at compile-time. Vyper provides several elementary types which can be combined to form complex types.

In addition, types can interact with each other in expressions containing operators.

.. index:: ! value

Value Types
===========

The following types are also called value types because variables of these
types will always be passed by value, i.e. they are always copied when they
are used as function arguments or in assignments.

.. index:: ! bool, ! true, ! false

Boolean
-------

**Keyword:** ``bool``

A boolean is a type to store a logical/truth value.

Values
******

The only possible values are the constants ``True`` and ``False``.

Operators
*********

====================  ===================
Operator              Description
====================  ===================
``not x``             Logical negation
``x and y``           Logical conjunction
``x or y``            Logical disjunction
``x == y``            Equality
``x != y``            Inequality
====================  ===================

Short-circuiting of boolean operators (``or`` and ``and``) is consistent with
the behavior of Python.

.. index:: ! intN, ! int, ! signed integer

Signed Integer (N bit)
------------------------

**Keyword:** ``intN`` (e.g., ``int128``)

A signed integer which can store positive and negative integers. ``N`` must be a multiple of 8 between 8 and 256 (inclusive).

Values
******

Signed integer values between -2\ :sup:`N-1` and (2\ :sup:`N-1` - 1), inclusive.

Integer literals cannot have a decimal point even if the decimal value is zero. For example, ``2.0`` cannot be interpreted as an integer.

Operators
*********

Comparisons
^^^^^^^^^^^

Comparisons return a boolean value.

==========  ================
Operator    Description
==========  ================
``x < y``   Less than
``x <= y``  Less than or equal to
``x == y``  Equals
``x != y``  Does not equal
``x >= y``  Greater than or equal to
``x > y``   Greater than
==========  ================

``x`` and ``y`` must both be of the same type.

Arithmetic Operators
^^^^^^^^^^^^^^^^^^^^

=============  ======================
Operator       Description
=============  ======================
``x + y``      Addition
``x - y``      Subtraction
``-x``         Unary minus/Negation
``x * y``      Multiplication
``x // y``     Integer division
``x**y``       Exponentiation
``x % y``      Modulo
=============  ======================

``x`` and ``y`` must both be of the same type.

Bitwise Operators
^^^^^^^^^^^^^^^^^

=============  ======================
Operator       Description
=============  ======================
``x & y``      Bitwise and
``x | y``      Bitwise or
``x ^ y``      Bitwise xor
=============  ======================

``x`` and ``y`` must be of the same type.

Shifts
^^^^^^^^^^^^^^^^^

=============  ======================
Operator       Description
=============  ======================
``x << y``     Left shift
``x >> y``     Right shift
=============  ======================

Shifting is only available for 256-bit wide types. That is, ``x`` must be ``int256``, and ``y`` can be any unsigned integer. The right shift for ``int256`` compiles to a signed right shift (EVM ``SAR`` instruction).


.. note::
   While at runtime shifts are unchecked (that is, they can be for any number of bits), to prevent common mistakes, the compiler is stricter at compile-time and will prevent out of bounds shifts. For instance, at runtime, ``1 << 257`` will evaluate to ``0``, while that expression at compile-time will raise an ``OverflowException``.

.. note::
   Integer division has different rounding semantics than Python for negative numbers: Vyper rounds towards zero, while Python rounds towards negative infinity. For example, ``-1 // 2`` will return ``-1`` in Python, but ``0`` in Vyper. This preserves the spirit (but not the text!) of the reasoning for Python's round-towards-negative-infinity behavior, which is that the behavior of ``//`` combined with the behavior of ``%`` preserves the following identity no matter if the quantities are negative or non-negative: ``(x // y) * y + (x % y) == x``.

.. index:: ! uint, ! uintN, ! unsigned integer

Unsigned Integer (N bit)
--------------------------

**Keyword:** ``uintN`` (e.g., ``uint8``)

A unsigned integer which can store positive integers. ``N`` must be a multiple of 8 between 8 and 256 (inclusive).

Values
******

Integer values between 0 and (2\ :sup:`N`-1).

Integer literals cannot have a decimal point even if the decimal value is zero. For example, ``2.0`` cannot be interpreted as an integer.

.. note::
    Integer literals are interpreted as ``int256`` by default. In cases where ``uint8`` is more appropriate, such as assignment, the literal might be interpreted as ``uint8``. Example: ``_variable: uint8 = _literal``. In order to explicitly cast a literal to a ``uint8`` use ``convert(_literal, uint8)``.

Operators
*********

Comparisons
^^^^^^^^^^^

Comparisons return a boolean value.

==========  ================
Operator    Description
==========  ================
``x < y``   Less than
``x <= y``  Less than or equal to
``x == y``  Equals
``x != y``  Does not equal
``x >= y``  Greater than or equal to
``x > y``   Greater than
==========  ================

``x`` and ``y`` must be of the same type.

Arithmetic Operators
^^^^^^^^^^^^^^^^^^^^

===========================  ======================
Operator                     Description
===========================  ======================
``x + y``                    Addition
``x - y``                    Subtraction
``x * y``                    Multiplication
``x // y``                   Integer division
``x**y``                     Exponentiation
``x % y``                    Modulo
===========================  ======================

``x`` and ``y`` must be of the same type.

Bitwise Operators
^^^^^^^^^^^^^^^^^

=============  ======================
Operator       Description
=============  ======================
``x & y``      Bitwise and
``x | y``      Bitwise or
``x ^ y``      Bitwise xor
``~x``         Bitwise not
=============  ======================

``x`` and ``y`` must be of the same type.

.. note::
    The Bitwise ``not`` operator is currently only available for ``uint256`` type.

Shifts
^^^^^^^^^^^^^^^^^

=============  ======================
Operator       Description
=============  ======================
``x << y``     Left shift
``x >> y``     Right shift
=============  ======================

Shifting is only available for 256-bit wide types. That is, ``x`` must be ``uint256``, and ``y`` can be any unsigned integer. The right shift for ``uint256`` compiles to a signed right shift (EVM ``SHR`` instruction).


.. note::
   While at runtime shifts are unchecked (that is, they can be for any number of bits), to prevent common mistakes, the compiler is stricter at compile-time and will prevent out of bounds shifts. For instance, at runtime, ``1 << 257`` will evaluate to ``0``, while that expression at compile-time will raise an ``OverflowException``.



Decimals
--------

**Keyword:** ``decimal``

A decimal is a type to store a decimal fixed point value. As of v0.4.0, decimals must be enabled with the CLI flag ``--enable-decimals``.


Values
******

A value with a precision of 10 decimal places between -18707220957835557353007165858768422651595.9365500928 (-2\ :sup:`167` / 10\ :sup:`10`) and 18707220957835557353007165858768422651595.9365500927 ((2\ :sup:`167` - 1) / 10\ :sup:`10`).

In order for a literal to be interpreted as ``decimal`` it must include a decimal point.

The ABI type (for computing method identifiers) of ``decimal`` is ``int168``.

Operators
*********

Comparisons
^^^^^^^^^^^

Comparisons return a boolean value.

==========  ================
Operator    Description
==========  ================
``x < y``   Less than
``x <= y``  Less or equal
``x == y``  Equals
``x != y``  Does not equal
``x >= y``  Greater or equal
``x > y``   Greater than
==========  ================

``x`` and ``y`` must be of the type ``decimal``.

Arithmetic Operators
^^^^^^^^^^^^^^^^^^^^

=============  ==========================================
Operator       Description
=============  ==========================================
``x + y``      Addition
``x - y``      Subtraction
``-x``         Unary minus/Negation
``x * y``      Multiplication
``x / y``      Decimal division
``x % y``      Modulo
=============  ==========================================

``x`` and ``y`` must be of the type ``decimal``.

.. _address:

Address
-------

**Keyword:** ``address``

The address type holds an Ethereum address.

Values
******

An address type can hold an Ethereum address which equates to 20 bytes or 160 bits. Address literals must be written in hexadecimal notation with a leading ``0x`` and must be `checksummed <https://github.com/ethereum/EIPs/blob/master/EIPS/eip-155.md>`_.

.. _members-of-addresses:

Members
^^^^^^^

=============== =========== ==========================================================================
Member          Type        Description
=============== =========== ==========================================================================
``balance``     ``uint256`` Balance of an address
``codehash``    ``bytes32`` Keccak of code at an address, ``0xc5d2460186f7233c927e7db2dcc703c0e500b653ca82273b7bfad8045d85a470`` if no contract is deployed (see `EIP-1052 <https://eips.ethereum.org/EIPS/eip-1052>`_)
``codesize``    ``uint256`` Size of code deployed at an address, in bytes
``is_contract`` ``bool``    Boolean indicating if a contract is deployed at an address
``code``        ``Bytes``   Contract bytecode
=============== =========== ==========================================================================

Syntax as follows: ``_address.<member>``, where ``_address`` is of the type ``address`` and ``<member>`` is one of the above keywords.

.. note::

    Operations such as ``SELFDESTRUCT`` and ``CREATE2`` allow for the removal and replacement of bytecode at an address. You should never assume that values of address members will not change in the future.

.. note::

    ``_address.code`` requires the usage of :func:`slice <slice>` to explicitly extract a section of contract bytecode. If the extracted section exceeds the bounds of bytecode, this will throw. You can check the size of ``_address.code`` using ``_address.codesize``.

M-byte-wide Fixed Size Byte Array
---------------------------------

**Keyword:** ``bytesM``
This is an M-byte-wide byte array that is otherwise similar to dynamically sized byte arrays. On an ABI level, it is annotated as bytesM (e.g., bytes32).

**Example:**
::

    # Declaration
    hash: bytes32
    # Assignment
    self.hash = _hash

    some_method_id: bytes4 = 0x01abcdef

Operators
*********

====================================  ============================================================
Keyword                               Description
====================================  ============================================================
``keccak256(x)``                      Return the keccak256 hash as bytes32.
``concat(x, ...)``                    Concatenate multiple inputs.
``slice(x, start=_start, len=_len)``  Return a slice of ``_len`` starting at ``_start``.
====================================  ============================================================

Where ``x`` is a byte array and ``_start`` as well as ``_len`` are integer values.

.. index:: !bytes

Byte Arrays
-----------

**Keyword:** ``Bytes``

A byte array with a max size.

The syntax being ``Bytes[maxLen]``, where ``maxLen`` is an integer which denotes the maximum number of bytes.
On the ABI level the Fixed-size bytes array is annotated as ``bytes``.

Bytes literals may be given as bytes strings or as hex strings.

.. code-block:: vyper

    bytes_string: Bytes[100] = b"\x01"
    bytes_string: Bytes[100] = x"01"

.. index:: !string

Strings
-------

**Keyword:** ``String``

Fixed-size strings can hold strings with equal or fewer characters than the maximum length of the string.
On the ABI level the Fixed-size bytes array is annotated as ``string``.

.. code-block:: vyper

    example_str: String[100] = "Test String"

Flags
-----

**Keyword:** ``flag``

Flags are custom defined types. A flag must have at least one member, and can hold up to a maximum of 256 members.
The members are represented by ``uint256`` values in the form of 2\ :sup:`n` where ``n`` is the index of the member in the range ``0 <= n <= 255``.

.. code-block:: vyper

    # Defining a flag with two members
    flag Roles:
        ADMIN
        USER

    # Declaring a flag variable
    role: Roles = Roles.ADMIN

    # Returning a member
    return Roles.ADMIN

Operators
*********

Comparisons
^^^^^^^^^^^

Comparisons return a boolean value.

============== ================
Operator       Description
============== ================
``x == y``     Equals
``x != y``     Does not equal
``x in y``     x is in y
``x not in y`` x is not in y
============== ================

Bitwise Operators
^^^^^^^^^^^^^^^^^

=============  ======================
Operator       Description
=============  ======================
``x & y``      Bitwise and
``x | y``      Bitwise or
``x ^ y``      Bitwise xor
``~x``         Bitwise not
=============  ======================

Flag members can be combined using the above bitwise operators. While flag members have values that are power of two, flag member combinations may not.

The ``in`` and ``not in`` operators can be used in conjunction with flag member combinations to check for membership.

.. code-block:: vyper

    flag Roles:
        MANAGER
        ADMIN
        USER

    # Check for membership
    @external
    def foo(a: Roles) -> bool:
        return a in (Roles.MANAGER | Roles.USER)

    # Check not in
    @external
    def bar(a: Roles) -> bool:
        return a not in (Roles.MANAGER | Roles.USER)

Note that ``in`` is not the same as strict equality (``==``). ``in`` checks that *any* of the flags on two flag objects are simultaneously set, while ``==`` checks that two flag objects are bit-for-bit equal.

The following code uses bitwise operations to add and revoke permissions from a given ``Roles`` object.

.. code-block:: python

    @external
    def add_user(a: Roles) -> Roles:
        ret: Roles = a
        ret |= Roles.USER  # set the USER bit to 1
        return ret

    @external
    def revoke_user(a: Roles) -> Roles:
        ret: Roles = a
        ret &= ~Roles.USER  # set the USER bit to 0
        return ret

    @external
    def flip_user(a: Roles) -> Roles:
        ret: Roles = a
        ret ^= Roles.USER  # flip the user bit between 0 and 1
        return ret

.. index:: !reference

Reference Types
===============

Reference types are those whose components can be assigned to in-place without copying. For instance, array and struct members can be individually assigned to without overwriting the whole data structure.

.. note::

  In terms of the calling convention, reference types are passed by value, not by reference. That means that, a calling function does not need to worry about a callee modifying the data of a passed structure.

.. index:: !arrays

Fixed-size Lists
----------------

Fixed-size lists hold a finite number of elements which belong to a specified type.

Lists can be declared with ``_name: _ValueType[_Integer]``, except ``Bytes[N]``, ``String[N]`` and flags.

.. code-block:: vyper

    # Defining a list
    exampleList: int128[3]

    # Setting values
    exampleList = [10, 11, 12]
    exampleList[2] = 42

    # Returning a value
    return exampleList[0]

Multidimensional lists are also possible. The notation for the declaration is reversed compared to some other languages, but the access notation is not reversed.

A two dimensional list can be declared with ``_name: _ValueType[inner_size][outer_size]``. Elements can be accessed with ``_name[outer_index][inner_index]``.

.. code-block:: vyper

    # Defining a list with 2 rows and 5 columns and set all values to 0
    exampleList2D: int128[5][2] = empty(int128[5][2])

    # Setting a value for row the first row (0) and last column (4)
    exampleList2D[0][4] = 42

    # Setting values
    exampleList2D = [[10, 11, 12, 13, 14], [16, 17, 18, 19, 20]]

    # Returning the value in row 0 column 4 (in this case 14)
    return exampleList2D[0][4]

.. note::
    Defining an array in storage whose size is significantly larger than ``2**64`` can result in security vulnerabilities due to risk of overflow.

.. index:: !dynarrays

Dynamic Arrays
----------------

Dynamic arrays represent bounded arrays whose length can be modified at runtime, up to a bound specified in the type. They can be declared with ``_name: DynArray[_Type, _Integer]``, where ``_Type`` can be of value type or reference type (except mappings).

.. code-block:: vyper

    # Defining a list
    exampleList: DynArray[int128, 3]

    # Setting values
    exampleList = []
    # exampleList.pop()  # would revert!
    exampleList.append(42)  # exampleList now has length 1
    exampleList.append(120)  # exampleList now has length 2
    exampleList.append(356)  # exampleList now has length 3
    # exampleList.append(1)  # would revert!

    myValue: int128 = exampleList.pop()  # myValue == 356, exampleList now has length 2

    # myValue = exampleList[2]  # would revert!

    # Returning a value
    return exampleList[0]


.. note::
    Attempting to access data past the runtime length of an array, ``pop()`` an empty array or ``append()`` to a full array will result in a runtime ``REVERT``. Attempting to pass an array in calldata which is larger than the array bound will result in a runtime ``REVERT``.

.. note::
    To keep code easy to reason about, modifying an array while using it as an iterator is disallowed by the language. For instance, the following usage is not allowed:

    .. code-block:: vyper

        for item: uint256 in self.my_array:
            self.my_array[0] = item

In the ABI, they are represented as ``_Type[]``. For instance, ``DynArray[int128, 3]`` gets represented as ``int128[]``, and ``DynArray[DynArray[int128, 3], 3]`` gets represented as ``int128[][]``.

.. note::
    Defining a dynamic array in storage whose size is significantly larger than ``2**64`` can result in security vulnerabilities due to risk of overflow.


.. _types-struct:

Structs
-------

Structs are custom defined types that can group several variables.

Struct types can be used inside mappings and arrays. Structs can contain arrays and other structs, but not mappings.

Struct members can be accessed via ``struct.argname``.

.. code-block:: vyper

    # Defining a struct
    struct MyStruct:
        value1: int128
        value2: decimal

    # Declaring a struct variable
    exampleStruct: MyStruct = MyStruct(value1=1, value2=2.0)

    # Accessing a value
    exampleStruct.value1 = 1

.. index:: !mapping

Mappings
--------

Mappings are `hash tables <https://en.wikipedia.org/wiki/Hash_table>`_ that are virtually initialized such that every possible key exists and is mapped to a value whose byte-representation is all zeros: a type's :ref:`default value <types-initial>`.

The key data is not stored in a mapping. Instead, its ``keccak256`` hash is used to look up a value. For this reason, mappings do not have a length or a concept of a key or value being "set".

Mapping types are declared as ``HashMap[_KeyType, _ValueType]``.

* ``_KeyType`` can be any base or bytes type. Mappings, arrays or structs are not supported as key types.
* ``_ValueType`` can actually be any type, including mappings.

.. note::
    Mappings are only allowed as state variables.

.. code-block:: vyper

   # Defining a mapping
   exampleMapping: HashMap[int128, decimal]

   # Accessing a value
   exampleMapping[0] = 10.1

.. note::

    Mappings have no concept of length and so cannot be iterated over.

.. index:: !initial

.. _types-initial:

Initial Values
==============

Unlike most programming languages, Vyper does not have a concept of ``null``. Instead, every variable type has a default value. To check if a variable is empty, you must compare it to the default value for its given type.

To reset a variable to its default value, assign to it the built-in ``empty()`` function which constructs a zero value for that type.

.. note::

    Memory variables must be assigned a value at the time they are declared.

Here you can find a list of all types and default values:

=========== ======================================================================
Type        Default Value
=========== ======================================================================
``address`` ``0x0000000000000000000000000000000000000000``
``bool``    ``False``
``bytes32`` ``0x0000000000000000000000000000000000000000000000000000000000000000``
``decimal`` ``0.0``
``uint8``   ``0``
``int128``  ``0``
``int256``  ``0``
``uint256`` ``0``
=========== ======================================================================

.. note::
    In ``Bytes``, the array starts with the bytes all set to ``'\x00'``.

.. note::
    In reference types, all the type's members are set to their initial values.


.. _type_conversions:

Type Conversions
================

All type conversions in Vyper must be made explicitly using the built-in ``convert(a: atype, btype)`` function. Type conversions in Vyper are designed to be safe and intuitive. All type conversions will check that the input is in bounds for the output type. The general principles are:

* Except for conversions involving decimals and bools, the input is bit-for-bit preserved.
* Conversions to bool map all nonzero inputs to 1.
* When converting from decimals to integers, the input is truncated towards zero.
* ``address`` types are treated as ``uint160``, except conversions with signed integers and decimals are not allowed.
* Converting between right-padded (``bytes``, ``Bytes``, ``String``) and left-padded types, results in a rotation to convert the padding. For instance, converting from ``bytes20`` to ``address`` would result in rotating the input by 12 bytes to the right.
* Converting between signed and unsigned integers reverts if the input is negative.
* Narrowing conversions (e.g., ``int256 -> int128``) check that the input is in bounds for the output type.
* Converting between bytes and int types results in sign-extension if the output type is signed. For instance, converting ``0xff`` (``bytes1``) to ``int8`` returns ``-1``.
* Converting between bytes and int types which have different sizes follows the rule of going through the closest integer type, first. For instance, ``bytes1 -> int16`` is like ``bytes1 -> int8 -> int16`` (signextend, then widen). ``uint8 -> bytes20`` is like ``uint8 -> uint160 -> bytes20`` (rotate left 12 bytes).
* Flags can be converted to and from ``uint256`` only.

A small Python reference implementation is maintained as part of Vyper's test suite, it can be found `here <https://github.com/vyperlang/vyper/blob/c4c6afd07801a0cc0038cdd4007cc43860c54193/tests/parser/functions/test_convert.py#L318>`__. The motivation and more detailed discussion of the rules can be found `here <https://github.com/vyperlang/vyper/issues/2507>`__.