Types & Operators
CashScript is a statically typed language, which means that the type of each variable needs to be specified. Types can also be implicitly or explicitly cast to other types. For a quick reference of the various casting possibilities, see Type Casting.
Boolean
bool
: The possible values are constants true
and false
.
Operators:
!
(logical negation)&&
(logical conjunction, “and”)||
(logical disjunction, “or”)==
(equality)!=
(inequality)
The operators ||
and &&
don't apply common short-circuiting rules. This means that in the expression f(x) || g(y)
, g(y)
will still be executed even if f(x)
evaluates to true.
Integer
int
: Signed integer of arbitrary size (BigInt).
Operators:
- Comparisons:
<=
,<
,==
,!=
,>=
,>
(all evaluate tobool
) - Arithmetic operators:
+
,-
, unary-
,*
,/
,%
(modulo).
Note the lack of the **
(exponentiation) operator as well as any bitwise operators.
Number Formatting
Underscores can be used to separate the digits of a numeric literal to aid readability, e.g. 1_000_000
. Numbers can also be formatted in scientific notation, e.g. 1e6
or 1E6
. These can also be combined, e.g. 1_000e6
.
Division by Zero
The script will fail when the right hand side of division or modulo operations is zero.
Contract authors should always consider whether /
and %
operations have division-by-zero edge cases and how this would impact contract security.
Date Parsing
Dates and times are always represented as integers. To get the UTC timestamp of a date use the built-in parser to avoid any potential errors. This will take a date in the format date("YYYY-MM-DDThh:mm:ss")
and convert it to an integer timestamp.
Example
int timestamp = date("2021-02-17T01:30:00");
require(timestamp == 1613554200);
String
string
: UTF8-encoded byte sequence.
Operators:
+
(concatenation)==
(equality)!=
(inequality)
Members:
length
: Number of characters in the string.split(int)
: Splits the string at the specified index and returns a tuple with the two resulting strings.reverse()
: Reverses the string.
The script will fail if split()
is called with an index that is out of bounds.
Bytes
bytes
: Byte sequence. Prefixed with 0x
to indicate hexadecimal sequence. Can optionally be bound to a byte length by specifying e.g. bytes4
, bytes32
, bytes64
. It is also possible to use byte
as an alias for bytes1
.
Operators:
+
(concatenation)==
(equality)!=
(inequality)&
(bitwise AND)|
(bitwise OR)^
(bitwise XOR)
Members:
length
: Number of bytes in the sequence.split(int)
: Splits the byte sequence at the specified index and returns a tuple with the two resulting byte sequences.reverse()
: Reverses the byte sequence.
The script will fail if split()
is called with an index that is out of bounds.
Example
bytes mintingCapability = 0x02;
bytes noCapability = 0x;
Bytes types with semantic meaning
Some byte sequences hold specific meanings inside Bitcoin Cash contracts. These have been granted their own types, separate from the regular bytes
type.
Public Key
pubkey
: Byte sequence representing a public key. Generally 33 bytes long.
Operators:
==
(equality)!=
(inequality)
Transaction Signature
sig
: Byte sequence representing a transaction signature. Generally 65 bytes long.
Operators:
==
(equality)!=
(inequality)
Data Signature
datasig
: Byte sequence representing a data signature. Generally 64 bytes long.
Operators:
==
(equality)!=
(inequality)
Array
Arrays are not assignable and can only be used with the checkMultisig
function using the following syntax:
checkMultisig([sig1, sig2], [pk1, pk2, pk3]);
Tuple
Tuples are the type that is returned when calling the split
member function on a string
or bytes
type. Their first or second element can be accessed through an indexing syntax similar to other languages:
string question = "What is Bitcoin Cash?";
string answer = question.split(15)[0].split(8)[1];
It is also possible to assign both sides of the tuple at once with a destructuring syntax:
string bitcoin, string cash = "BitcoinCash".split(7);
require(bitcoin == cash);
Type Casting
Type casting can be done both explicitly and implicitly as illustrated below. pubkey
, sig
and datasig
can be implicitly cast to bytes
, meaning they can be used anywhere where you would normally use a bytes
type. Explicit type casting can be done with a broader range of types, but is still limited. The syntax of this explicit type casting is illustrated below. Note that you can also cast to bounded bytes
types.
See the following table for information on which types can be cast to other which other types.
Type | Implicitly castable to | Explicitly castable to |
---|---|---|
int | bytes, bool | |
bool | int | |
string | bytes | |
bytes | sig, pubkey, int | |
pubkey | bytes | bytes |
sig | bytes | bytes |
datasig | bytes | bytes |
Example
pubkey pk = pubkey(0x0000);
bytes editedPk = bytes(pk) + 0x1234;
bytes4 integer = bytes4(25);
Int to Byte Casting
When casting integer types to bytes of a certain size, the integer value is padded with zeros, e.g. bytes4(0) == 0x00000000
. It is also possible to pad with a variable number of zeros by passing in a size
parameter, which indicates the size of the output, e.g. bytes(0, 4 - 2) == 0x0000
. The size casting can be a very important feature when keeping local state in an nftCommitment or in the simulated state.
VM numbers follow Script Number format (A.K.A. CSCriptNum), to convert VM number to bytes or the reverse, it's recommended to use helper functions for these conversions from libraries like Libauth.
Semantic Byte Casting
When casting unbounded bytes
types to bounded bytes
types (such as bytes20
or bytes32
), this is a purely semantic cast. The bytes are not padded with zeros, and no checks are performed to ensure the cast bytes are of the correct length. This can be helpful in certain cases, such as LockingBytecode
, which expects a specific length input.
Example
bytes pkh = nftCommitment.split(20)[0]; // (type = bytes, content = 20 bytes)
bytes20 bytes20Pkh = bytes20(pkh); // (type = bytes20, content = 20 bytes)
bytes25 lockingBytecode = new LockingBytecodeP2PKH(bytes20Pkh);
If you do need to pad bytes to a specific length, you can convert the bytes to int
first, and then cast to the bounded bytes
type. This will pad the bytes with zeros to the specified length, like specified in the Int to Byte Casting section above.
Example
bytes data = nftCommitment.split(10)[0]; // (type = bytes, content = 10 bytes)
bytes20 paddedData = bytes20(int(data)); // (type = bytes20, content = 20 bytes)
require(storedContractState == paddedData);
Operators
An overview of all supported operators and their precedence is included below. Notable is a lack of exponentiation, since these operations are not supported by the underlying Bitcoin Script.
Precedence | Description | Operator |
---|---|---|
1 | Parentheses | (<expression>) |
2 | Type cast | <type>(<expression>) |
3 | Object instantiation | new <class>(<args...>) |
4 | Function call | <function>(<args...>) |
5 | Tuple index | <tuple>[<index>] |
6 | Member access | <object>.<member> |
7 | Unary minus | - |
7 | Logical NOT | ! |
8 | Multiplication, division and modulo | * , / , % |
9 | Addition and subtraction | + , - |
9 | String / bytes concatenation | + |
10 | Numeric comparison | < , > , <= , >= |
11 | Equality and inequality | == , != |
12 | Bitwise AND | & |
13 | Bitwise XOR | ^ |
14 | Bitwise OR | | |
15 | Logical AND | && |
16 | Logical OR | || |
17 | Assignment | = |