Spaces:
Runtime error
sidebar_position: 1
How to Write a Contract
A smart contract is a class that extends the SmartContract
base class. A simple example is shown below.
import { SmartContract, method, prop, assert } from "scrypt-ts"
class Demo extends SmartContract {
@prop()
readonly x: bigint
constructor(x: bigint) {
super(...arguments)
this.x = x
}
@method()
public unlock(x: bigint) {
assert(this.add(this.x, 1n) == x, 'incorrect sum')
}
@method()
add(x0: bigint, x1:bigint) : bigint {
return x0 + x1
}
}
Class members decorated with @prop
and @method
will end up on the blockchain and thus must be a strict subset of TypeScript. Everywhere decorated with them can be regarded in the on-chain context. Members decorated with neither are regular TypeScript and are kept off chain. The significant benefit of sCrypt
is that both on-chain and off-chain code are written in the same language: TypeScript.
:::note You can use the sCrypt template Repl and play with the code in your browser! :::
Properties
A smart contract can have two kinds of properties:
With
@prop
decorator: these properties are only allowed to have types specified below and they shall only be initialized in the constructor.Without
@prop
decorator: these properties are regular TypeScript properties without any special requirement, meaning they can use any types. Accessing these properties is prohibited in methods decorated with the@method
decorator.
@prop
decorator
Use this decorator to mark any property that intends to be stored on chain.
This decorator takes a boolean
parameter. By default, it is set to false
, meaning the property cannot be changed after the contract is deployed. If the value is true
, the property is a so-called stateful property and its value can be updated in subsequent contract calls.
// good, `a` is stored on chain, and it's readonly after the contract is deployed
@prop()
readonly a: bigint
// valid, but not good enough, `a` cannot be changed after the contract is deployed
@prop()
a: bigint
// good, `b` is stored on chain, and its value can be updated in subsequent contract calls
@prop(true)
b: bigint
// invalid, `b` is a stateful property that cannot be readonly
@prop(true)
readonly b: bigint
// good
@prop()
static c: bigint = 1n
// invalid, static property must be initialized when declared
@prop()
static c: bigint
// invalid, stateful property cannot be static
@prop(true)
static c: bigint = 1n
// good, `UINT_MAX` is a compile-time constant, and no need to typed explicitly
static readonly UINT_MAX = 0xffffffffn
// valid, but not good enough, `@prop()` is not necessary for the CTC
@prop()
static readonly UINT_MAX = 0xffffffffn
// invalid
@prop(true)
static readonly UINT_MAX = 0xffffffffn
Constructor
A smart contract must have an explicit constructor if it has at least one @prop
that is not static
.
The super
method must be called in the constructor and all the arguments of the constructor should be passed to super
in the same order as they are passed into the constructor. For example,
class A extends SmartContract {
readonly p0: bigint
@prop()
readonly p1: bigint
@prop()
readonly p2: boolean
constructor(p0: bigint, p1: bigint, p2: boolean) {
super(...arguments) // same as super(p0, p1, p2)
this.p0 = p0
this.p1 = p1
this.p2 = p2
}
}
arguments
is an array containing the values of the arguments passed to that function. ...
is the spread syntax.
Methods
Like properties, a smart contract can also have two kinds of methods:
With
@method
decorator: these methods can only call methods also decorated by@method
or functions specified below. Also, only the properties decorated by@prop
can be accessed.Without
@method
decorator: these methods are just regular TypeScript class methods.
@method
decorator
- Use this decorator to mark any method that intends to run on chain.
- It takes a sighash flag as a parameter.
Public @method
s
Each contract must have at least one public @method
. It is denoted with the public
modifier and does not return any value. It is visible outside the contract and acts as the main method into the contract (like main
in C and Java).
A public @method
can be called from an external transaction. The call succeeds if it runs to completion without violating any conditions in assert(). An example is shown below.
@method()
public unlock(x: bigint) {
// only succeeds if x is 1
assert(this.add(this.x, 1n) == x, "unequal")
}
:::note
The last function call of a public @method
method must be an assert()
function call, unless it is a console.log()
call.
:::
class PublicMethodDemo extends SmartContract {
@method()
public foo() {
// invalid, the last statement of public method should be an `assert` function call
}
@method()
public bar() {
assert(true);
return 1n; // invalid, because a public method cannot return any value
}
@method()
public foobar() {
console.log();
// valid, `console.log` calling will be ignored when verifying the last `assert` statement
assert(true);
console.log();
console.log();
}
}
Non-public @method
s
Without a public
modifier, a @method
is internal and cannot be directly called from an external transaction.
@method()
add(x0: bigint, x1:bigint) : bigint {
return x0 + x1
}
:::note
Recursion is disallowed. A @method
, public and not, cannot call itself, either directly in its own body or indirectly calls another method that transitively calls itself.
:::
class MethodsDemo extends SmartContract {
@prop()
readonly x: bigint;
@prop()
readonly y: bigint;
constructor(x: bigint, y: bigint) {
super(...arguments);
this.x = x;
this.y = y;
}
// good, non-public static method without access `@prop` properties
@method()
static sum(a: bigint, b: bigint): bigint {
return a + b;
}
// good, non-public method
@method()
xyDiff(): bigint {
return this.x - this.y
}
// good, public method
@method()
public add(z: bigint) {
// good, call `sum` with the class name
assert(z == MethodsDemo.sum(this.x, this.y), 'add check failed');
}
// good, another public method
@method()
public sub(z: bigint) {
// good, call `xyDiff` with the class instance
assert(z == this.xyDiff(), 'sub check failed');
}
// valid but bad, public static method
@method()
public static alwaysPass() {
assert(true)
}
}
Data Types
Types used in @prop
and @method
are restricted to these kinds:
Basic Types
boolean
A simple value true
or false
.
let isDone: boolean = false
bigint
bigint
can represent arbitrarily large integers. A bigint literal is a number with suffix n
:
11n
0x33FEn
const previouslyMaxSafeInteger = 9007199254740991n
const alsoHuge = BigInt(9007199254740991)
// 9007199254740991n
const hugeHex: bigint = BigInt("0x1fffffffffffff")
// 9007199254740991n
ByteString
In a smart contract context (i.e., in @method
s or @prop
s), a ByteString
represents a byte array.
A literal string
can be converted in to a ByteString
using function toByteString(literal: string, isUtf8: boolean = false): ByteString
:
- If not passing
isUtf8
orisUtf8
isfalse
, thenliteral
should be in the format of hex literal, which can be represented by the regular expression:/^([0-9a-fA-F]{2})*$/
- Otherwise,
literal
should be in the format of utf8 literal, e.g.,hello world
.
:::note
toByteString
ONLY accepts string literals for its first argument, and boolean literals for the second.
:::
let a = toByteString('0011') // valid, `0011` is a valid hex literal
// 0011
let b = toByteString('hello world', true) // valid
// 68656c6c6f20776f726c64
toByteString('0011', false) // valid
// 30303131
toByteString(b, true) // invalid, not passing string literal to the 1st parameter
toByteString('001') // invalid, `001` is not a valid hex literal
toByteString('hello', false) // invalid, `hello` is not a valid hex literal
toByteString('hello', 1 === 1) // invalid, not passing boolean literal to the 2nd parameter
let c = true
toByteString('world', c) // invalid, not passing boolean literal to the 2nd parameter
ByteString
has the following operators and methods:
==
/===
: compare+
: concatenate
const str0 = toByteString('01ab23ef68')
const str1 = toByteString('656c6c6f20776f726c64')
// comparison
str0 == str1
str0 === str1
// false
// concatenation
str0 + str1
// '01ab23ef68656c6c6f20776f726c64'
number
Type number
is not allowed in @prop
s and @method
s, except in the following cases. We can use Number()
function to convert bigint
to number
.
- Array index
let arr: FixedArray<bigint, 3> = [1n, 3n, 3n]
let idx: bigint = 2n
let item = arr[Number(idx)]
- Loop variable
for (let i: number = 0 i < 10 i++) {
let j: bigint = BigInt(i) // convert number to bigint
}
It can also be used in defining compile-time constants.
Fixed Size Array
All arrays must be of fixed size and be declared as type of FixedArray<T, SIZE>
, whose SIZE
must be a CTC described later.
The common TypeScript arrays declared as T[]
or Array<T>
are not allowed in @prop
s and @method
s, as they are of dynamic size.
let aaa: FixedArray<bigint, 3> = [1n, 3n, 3n]
// set to all 0s
const N = 20
let aab: FixedArray<bigint, N> = fill(0n, N)
// 2-dimensional array
let abb: FixedArray<FixedArray<bigint, 2>, 3> = [[1n, 3n], [1n, 3n], [1n, 3n]]
:::caution
A FixedArray
behaves differently in an on-chain and off-chain context, when passed as a function argument. It is passed by reference off chain, as a regular TypeScript/JavaScript array, while passed by value on chain. It is thus strongly recommended to NEVER mutate a FixedArray
parameter's element inside a function.
class DemoContract extends SmartContract {
@prop(true)
readonly a: FixedArray<bigint, 3>
constructor(a: FixedArray<bigint, 3>) {
super(...arguments)
this.a = a
}
@method()
onchainChange(a: FixedArray<bigint, 3>) {
a[0] = 0
}
offchainChange(a: FixedArray<bigint, 3>) {
a[0] = 0
}
@method()
public main(a: FixedArray<bigint, 3>) {
this.onchainChange(this.a)
// note: a[0] is not changed on chain
assert(this.a[0] == 1n)
}
}
const arrayA: FixedArray<bigint, 3> = [1n, 2n, 3n]
const instance = new DemoContract(arrayA);
instance.offchainChange(arrayA)
// note: arrayA[0] is changed off chain
assert(arrayA[0] = 0n)
:::
User-defined Types
Users can be define customized types using type
or interface
, made of basic types.[^1]
type ST = {
a: bigint
b: boolean
}
interface ST1 {
x: ST
y: ByteString
}
type Point = {
x: number
y: number
}
function printCoord(pt: Point) {
console.log("The coordinate's x value is " + pt.x)
console.log("The coordinate's y value is " + pt.y)
}
interface Point2 {
x: number
y: number
}
// Exactly the same as the earlier example
function printCoord(pt: Point2) {
console.log("The coordinate's x value is " + pt.x)
console.log("The coordinate's y value is " + pt.y)
}
[^1]: A user-defined type is also passed by value on chain, and by reference off chain, same as a FixedArray
. It is thus strongly recommended to NEVER mutate the field of a parameter, which is of a user-defined type, inside a function.
Domain Types
There are several domain types, specific to the Bitcoin context, used to further improve type safety. They are all subtypes of ByteString
. That is, they can be used where a ByteString
is expected, but not vice versa.
PubKey
- a public keySig
- a signature type in DER format, including sighash flags at the endRipemd160
- a RIPEMD-160 hashPubKeyHash
- an alias forRipemd160
, usually representing a bitcoin address.Sha1
- a SHA-1 hashSha256
- a SHA-256 hashSigHashType
- a sighashSigHashPreimage
- a sighash preimageOpCodeType
- a Script opcode
@method()
public unlock(sig: Sig, pubkey: PubKey) {
// hash160() takes a ByteString as input, but can accept pubkey here, which if of type PubKey
assert(hash160(pubkey) == this.pubKeyHash)
assert(this.checkSig(sig, pubkey), 'signature check failed')
}
Statements
There are some constraints on these following statements within @method
s, except variable declarations.
Variable declarations
Variables can be declared in @method
s by keywords const
/ var
/ let
, like in normal TypeScript.
let a : bigint = 1n
var b: boolean = false
const byte: ByteString = toByteString("ff")
for
Bitcoin does not allow unbounded loops for security reasons, to prevent DoS attacks. All loops must be bounded at compile time. So if you want to loop inside @method
, you must strictly use the following format:
for (let $i = 0; $i < $maxLoopCount; $i++) {
...
}
:::note
- the initial value must be
0
or0n
, the operator<
(no<=
), and increment$i++
(no pre-increment++$i
). $maxLoopCount
must be a CTC.$i
can be arbitrary name, e.g.,i
,j
, ork
. It can be both anumber
or abigint
type.break
andcontinue
are currently not allowed, but can be emulated like :::
// emulate break
let x = 3n
let done = false
for (let i = 0; i < 3; i++) {
if (!done) {
x = x * 2n
if (x >= 8n) {
done = true
}
}
}
return
Due to the lack of native return semantics support in Bitcoin Script, a function currently must end with a return
statement and it is the only valid place for a return
statement. This requirement may be relaxed in the future.
@method() m(x: bigint): bigint {
if (x > 2n) return x // invalid
return x + 1n // valid
}
This is usually not a problem since it can be circumvented as follows:
@method()
abs(a: bigint): bigint {
if (a > 0) {
return a
} else {
return -a
}
}
can be rewritten as
@method()
abs(a: bigint): bigint {
let ret : bigint = 0
if (a > 0) {
ret = a
} else {
ret = -a
}
return ret
}
Compile-time Constant
A compile-time constant, CTC for short, is a special variable whose value can be determined at compile time. A CTC must be defined in one of the following ways.
- A number literal like:
3
- A
const
variable, whose value must be a numeric literal. Expressions cannot be used for now.
const N1 = 3 // valid
const N2: number = 3 // invalid, no explicit type `number` allowed
const N3 = 3 + 3 // invalid, no expression allowed
- A
static
readonly
property:
class X {
static readonly M1 = 3 // valid
static readonly M2: number = 3 // invalid
static readonly M3 = 3 + 3 // invalid
}
A CTC is required in these cases.
- Array size
let arr1: FixedArray<bigint, 3> = [1n, 2n, 3n]
// `typeof` is needed since FixedArray takes a type as the array size, not a value
let arr1: FixedArray<bigint, typeof N1> = [1n, 2n, 3n]
let arr2: FixedArray<bigint, typeof X.M1> = [1n, 2n, 3n]
- Loop count in
for
statement
for(let i=0; i< 3; i++) {}
for(let i=0; i< N1; i++) {}
for(let i=0; i< X.M1; i++) {}
Functions
Built-in Functions
You can refer to Built-ins for a full list of functions and libraries built into scryptTS
.
Whitelisted Functions
Be default, all Javascript/TypeScript built-in functions and global variables are not allowed in @method
s, except the following kinds.
console.log
console.log
can be used for debugging purposes.
@method()
add(x0: bigint, x1:bigint) : bigint {
console.log(x0)
return x0 + x1
}
Operators
sCrypt is a subset of TypeScript. Only the following operators can be used directly.
Operator | Description |
---|---|
+ |
Addition |
- |
Subtraction |
* |
Multiplication |
/ |
Division |
% |
Remainder |
++ |
Increment |
-- |
Decrement |
== |
Equal to |
!= |
Not equal to |
=== |
Same as == |
!== |
Same as != |
> |
Greater than |
>= |
Greater than or equal to |
< |
Less than |
<= |
Less than or equal to |
&& |
Logical AND |
|| |
Logical OR |
! |
Logical NOT |
cond ? expr1 : expr2 |
ternary |
+= |
Add and assign |
-= |
Subtract and assign |
*= |
Multiply and assign |
/= |
Divide and assign |
%= |
Assign remainder |
:::note
**
is not supported currently.
:::