This repository has a number of Solidity contracts in it. Some of these contracts will potentially make it into an official Solidity standard library.
Contract imports uses the github path, so can be imported into the online compiler, or the repo can be cloned and contract compiled locally using solc's path re-mapping features.
Short descriptions of contracts and methods can be found in this README, as well as some examples. More detailed documentation can be found in the contracts themselves.
Warning: All the code in here is still in development, and should be considered unreliable. Do not deploy any of this code onto a public node and use it in production.
This is also a personal side project with no deadlines or dev plan.
There is a system of grades for contracts. Grades are found along with the version in contract descriptions. They are temporary and will be changed when the library is stable (new contracts will be developed on separate branches then).
Generally speaking, contracts with a grade less then 3
should not be used, and those less then 5
should only be used when testing.
-
(
Incomplete
) Contract is incomplete; code is still being added. -
(
Draft
) First draft is complete, and contract is being tested. -
(
Tested
) Unit-tests are written, and contract has been debugged. -
(
Optimized
) Contract has been optimized. -
(
Stable
) Contract is production ready.
The tests are done over RPC calls from Node.js, using web3
- the official Ethereum JavaScript API. They require a running development node. Any Ethereum client should work in theory, but the tests has only been run against a geth
node.
Latest compilation and testing was done using: 0.3.4-7dab8902/Release-Emscripten/clang/Interpreter
To run tests, cd into the project root and type:
npm install
Start an ethereum development node, for example using the -dev
flag with geth
(not a testnet node). The tests assumes that an Ethereum node will answer to RPC calls on port 8545
. It will automatically use the first account in the accounts list (accounts[0]
) as sender. Some test contracts require a lot of gas, so you may have to crank up the gas limit and make sure the active account has a lot of gas.
When all is set up, run: mocha integration_tests
Test-data is sometimes generated using Sage 7.0. The generation scripts are normally included so that the process can be replicated.
The contracts will not be fully optimized until it's clear how formal verification will work, since that is a requirement for many of these contracts. Assembly might be removed from some contracts and functions.
Contracts for working with bits.
Bits
is a library with methods for reading, writing, and manipulating individual or groups of bits. The methods operates on uint
variables.
uint u = 10;
/* Get the bit at the given index (0 to 255 inclusive) */
u.bit(0) // 0 (1)
u.bit(1) // 1 (2)
u.bit(2) // 0 (4)
u.bit(3) // 1 (8)
/* Check if a bit is set */
u.bitSet(0) // false
u.bitSet(3) // true
/* Set the bit at the given index */
u = u.setBit(2) // u = 14
/* bitsSet(i, len) checks if the bits i, i + 1, ... , i + len - 1 are all set */
u.bitsSet(2, 2) // true (bits 2 and 3 are both set)
u.bitsSet(0, 3) // false (bit 0, 1 and 2 are not all set)
/* Toggle a bit */
u = u.bitToggle(0) // u = 15
/* Clear a bit */
u = u.bitClear(3) // u = 7
Assuming the indices and lengths are within bounds:
uint(~0).get(i) = 1;
uint u = uint(0).set(i); u == 2**i;
u.bit(i) == u.bits(i, 1); // Same for bitEqual, bitAnd, bitOr, and bitXor.
u.bitEqual(v, i) == (u.bit(i) == v.bit(i))
u.bitAnd(v, i) == u.bit(i) & v.bit(i) // etc.
u.bitsEqual(v, i, len) == (u.bits(i, len) == v.bits(i, len)) // Same as above.
Contracts used for working with bytes.
A ByteSlice
is created from memory variables of type bytes
. It is possible to slice slices, access bytes by index, and to extract a bytes
variable from the slice.
Since dynamic arrays and internal types are used in function input and output, this contract is meant to be extended
. Later it will be possible to use as a library, when some (planned) upgrades has been made to libraries.
Strings are utf encoded, so it is not safe to just cast a string into bytes or vice versa, unless it is known in advance that it is an ASCII string. Then it is safe.
The Slice
struct has two members of type uint
:
_unsafe_memPtr
- a pointer to the memory address where the bytes are stored.
_unsafe_len
- the length (number of bytes).
Extracting the bytes out of a slice is done using the toBytes
function, which creates a new bytes memory
variable, sets its size to len
, and copies all the bytes from memory address memPtr
to memPtr + len - 1
into the new variable.
bytes memory bts = "abcdefg"; // Create a new 'bytes' variable in memory.
/* Slicing bytes */
var s = bts.slice(); // Create a slice on "abcdefg".
var sa = bts.slice(uint(2)); // Create a slice on "cdefg"
var sb = bts.slice(-3); // Create a slice on "efg"
var sc = bts.slice(uint(1), 2); // Create a slice on "b"
/* Slicing slices */
var sCopy = s.slice(); // slice on "abcdefg"
var s2 = s.slice(uint(5)); // slice on "fg"
var s3 = s.slice(uint(5), 6); // slice on "f"
var s4 = s.slice(-2); // slice on "fg"
var s5 = s.slice(-4, -1); // slice on "def"
var s6 = s5.slice(uint(1), 2); // slice on "e"
var sEmpt = s.slice(uint(7), 7); // The empty slice.
var sEmpt2 = (new bytes(0)).slice(); // The empty slice.
var sWillThrow = s.slice(uint(9));
var sWillThrowToo = s.slice(uint(0), 8);
/* length */
var len = s.len(); // 7
var len2 = s2.len(); // 2
var len3 = sEmpt.len(); // 0
/* Index access */
var b = s.at(uint(2)); // 'c'
var b2 = s.at(-2); // 'f'
var bWillThrow = s.at(uint(7)); // Index out of bounds.
s.set(uint(2), 'x'); // s.at(uint(2)) == 'x'
s.set(-2, 'y'); // s.at(-2) == s.at(uint(4)) == 'y'
s.set(uint(7), 'z'); // Will throw
/* Converting to 'bytes memory' */
var bts2 = s.toBytes(); // "abcdefg". Copies 7 bytes.
var bts3 = s.slice(-2).toBytes(); // "fg". Copies 2 bytes.
/* Equals */
var eq = s.equal(s); // true
s.equal(sEmpt); // false
delete s;
s.equal(sEmpt); // true
Contracts used for various different types of encoding and decoding.
Used to parse and decode RLP encoded data. The encoded bytes is converted into an RLPItem
which keeps track of the bytes in memory, and has methods for decoding and extracting new items from lists.
Working with items will never copy any of the raw RLP encoded bytes; only new RLPItem
s and Iterator
s. Those are both simple items that takes 2 and 3 words of memory space.
Decoding an item into a reference type (i.e. bytes
or string
) will copy the required bytes, so use with care.
Encoding is not (yet) supported.
bytes memory data; // input
/* data = "0x880102030405060708" */
var itm = data.toRLPItem(); // Create a new RLP Item
item.isData(); // true
item.isList(); // false
item.items(); // 0
item.toBytes(); // "0x880102030405060708" - the raw RLP-encoded bytes.
item.toData(); // "0x0102030405060708" - The data.
/* data = "0x11" */
var itm = data.toRLPItem();
uint u = itm.toUint(); // 17
address a = itm.toAddress(); // 0x0000000000000000000000000000000000000011
/* data = "0x80" */
var itm = data.toRLPItem();
itm.isEmpty(); // false
itm.isData(); // true
/* data = "0x01" */
var itm = data.toRLPItem();
itm.toBool(); // True
// This is the RLP encoding of [[1, 2], 1, [1, 2, 3]]
/* data = "0xC8C2010201C3010203" */
var itm = rlpList.toRLPItem();
itm.isList(); // true
itm.items(); // 3
itm.toList(); // RLPItem[] of length 3 (one for each list-item)
// Using iterators.
var subItm = itm.iterator().next(); // RLPItem for [1, 2]
var it2 = subItem.iterator();
while(it2.hasNext())
it2.next().toUint() // 1, 2
it2.next(); // will throw
/* data = "0x" */
var itm = nullBts.toRLPItem();
itm.isNull(); // true
itm.isData(); // false
itm.isList(); // false
itm.isEmpty() // false
// Strict mode (costs more, but useful when RLP data may be malformed).
/* data = "0x81" */
var itm = data.toRLPItem(true); // will throw
/* data = "0x8101" */
var itm = data.toRLPItem(true); // will throw
/* data = "0xC211 */
var itm = data.toRLPItem(true); // will throw
Used to convert ECC data between different formats.
Contracts used for cryptographic operations.
Hashing primitives are not implemented, because Solidity already provide some.
Warning: Crypto operations can be very expensive, so should not be done on the public chain. This includes the ECC math functions as well.
Curve is an interface for elliptic curves. The cryptographic properties (e.g. defined over finite fields) of these curves are implied.
The only functions that may be implemented are those that does not involve passing a private key to the contract, meaning functions like 'sign' and 'private-to-public' are not included. There is also no public key recovery function in the interface yet, because there is a built in function for secp256k1 (and for some other reasons).
The contracts use the mathematical representations of coordinates and points, and is thus key-format agnostic. Contracts that require keys to be encoded in a particular way would manage the encoding/decoding themselves; and delegate the actual curve operations to whatever implementation is used (only secp256k1
is provided at this point).
Details can be found here:
SEC 1: Elliptic Curve Cryptography, Version 2.0
SEC 2: Recommended Elliptic Curve Domain Parameters, Version 2.0
Lower-s signatures (requirement for homestead signatures).
Check if a point P
(given in affine coordinates) is on the curve.
Check if a point P
(given in affine coordinates) is a valid public key.
Get the compressed form of a point Q = (Qx, Qy)
, C(Q) = (Qx, Qy % 2)
.
For a given Px
and a yBit = Py % 2
, returns the point P = (Px, Py)
.
Validate the signature (r, s)
of a 32 byte hash msg
against a public key Q
. Returns true
if the signature is valid. Will only validate signatures on lower-s form.
Math helper. Naming is trying to be consistent with similar solidity functions, e.g. 'invmod' and 'expmod' is like 'addmod', 'mulmod', etc.
uint x = invmod(uint a, uint p)
is used to compute the modular inverse ax = 1 (mod p)
.
Constraints: a
and p
must be coprime.
Uses the euclidean algorithm to find the modular inverse.
uint x = expmod(b, e, m)
is used to compute the number b**e % m
The function is a Solidity adaptation of the exponentiation formula found in the Serpent examples.
Used internally to transform an arbitrary (Jacobian) point P = (Px, Py, Pz)
to P' = (Px', Py', 1)
. (Px, Py)
is the affine coordinates. The second variety (with 4 parameters) is used in mul
during the Montgomery inversion.
Library functions for the secp256k1
curve. Includes internal versions of the Curve
functions but also a number of internal functions for doing point arithmetic.
The output of all function is in Jacobian coordinates.
NOTE: You may notice weird arrays and other types where it looks as if they shouldn't be needed. That's mainly to work around the stack item limitations and some optimizer issues. This is a problem, and I will replace the code gradually as it becomes possible.
Point-addition of two points P
and Q
. Both input points must be expressed in Jacobian coordinates.
Same as add
except the second point is expressed in affine coordinates.
Same as addMixed
but mutates the first point rather then creating a new one.
Doubles a point P
. The point must be expressed in Jacobian coordinates.
Same as double
but mutates the point.
Multiplication of a scalar d
and a point P
. The point must be expressed in affine coordinates.
Implementation of Curve
based on the secp256k1
library.
Keys generated using OpenSSL.
The ephemeral k
values used for signatures are just random numbers between 1 and n - 1
(inclusive), where n
is the order of the generator point.
To generate the test signatures from k
, the following Sagemath script was used:
# ('k', 'private key')
ks = [
(0xb578a05106a1e1975d2082add46aad13d76e3ce3b9d2dcabc5f651c9a29fa822, 0x2569d5e120b10d696933074eecb9182719fb3a89b7714bc99037d8416b391436),
(0x7a12fac8796af10ce7dbfc68271d87ca685950a27ab4222c8b720210fe2c21c5, 0xe370174cb9fea3b5c1ef39bcdf36d321cefb06e0e8a6cb856e1bb3e9be3c4385),
(0x13018a36dc981d6ef697ebcb809dc8966773748284281f2338752d2a889fc798, 0xa5378a4c6db7a076af65cd728f6fe3cf59f416a3d8f74b31b0d76f666a868d09),
(0x534813f6aded11b19e45ffc3cd77841783f2928ddd427d4e793808f476c305dd, 0x40a62a62984797e129b9f84fb0202a10c37b52ad83da2d4f3f11c6a8587999f6),
(0xa2d19750e168a6faf9f6b406ed2484f990e90bca4c139d6a8a5d298a687286b2, 0x32e3506d0cf7c8e73c5692f67e6b4c2377a5c0728b4719ff948f420609a3bcdf),
(0x4bb2d83327145d0fa4e5aa9a28dc3d4cdf87cf13cdf6505109d07b770be0559f, 0x1b3c26da0280b299ecfbdc8d9b6e69677094f1c034010a9a27ba482185355ec2),
(0x1f7a57c5f5820236e274ec33802a8c9171f76e5a246bbc71c1dc8afcf51d0f33, 0x9a2e734e0057dc4cbbf8418f9e3d7b9a638e121f1e217884579284b802d0fab3),
(0xb42ae0689c162dcd8de571a7e5caa2ea24d5336eaeae3e9fbcf6946613165878, 0xb79a54eb34d1a2f75d956f593dbd08fb75d307847838a64aec9adc0ae74c3bb0),
(0xc908c6aad4690e029b66333fce8660ccb7b9953890ef566734a67be156f5a2fd, 0xb15b46827e551c12f8bfe066a634aacfd907be42186ec3e4b3d69baf8efa5bde),
(0x22a2b8da7df2da2ced4b4b1ab59cd1f877147cad336ff0d2d01384ddac86c38f, 0xd83a52c3d1965d9ace37fde20da4eb4c98184c6cfc8b33b1b0c56be36f3fc5c0),
(0x72407c35a1e2db22521847e9e78f5227bccef8243df1462affe4013e437a875b, 0x013263a84e05ca597f3c6cdeae5b33349c618308ac5ef844de0aa5b788d0a3a8),
(0x17a332b7124cbf26c4ee6a49d7f68955b23735e4f10acdd27582a653518609b8, 0x6ac0b7ddefa89ca0f129cba3ddda3b4a6e126d43363b4509bd6c9334dd81e0fa),
(0x5f87ee0a303eaf5b88c5b84c307194dc896e462fd0242f512ded874fce5a2b25, 0xba87e08dc57311d5518c7d46958f3e647baed528313952cc5cf34e14f3b523af),
(0xa3622fc8d2f1dac4d29e0fddc90d2b99b8eb9d13acec78dfeead2e66452122af, 0xd8c21bd40cdd0603666fa5c6199d8058050872913fa0c68feb53fed3df35695e),
(0x950a753f782bb0a103359b4d54101df4c06910a72338a5592f2a17f2e5868659, 0x839793d1f796da91096cd1641d06593b6f6c3dadc12e0ccd7eaaf1f10c752c25),
(0xe6f0e82e0bed605dcba808164c809a50da65db8b39ef68658c62c32ba715c069, 0x5c745e065ad28124ca941b02abd0ebdb4502d0c70058ddd7dfa9b659165e76a4),
(0xe9d62c056727ccca3946a11f8822df06c14e830d36340181de2075146501091e, 0x03eebcbd7958778591670c9d47c09fccd94a6dd05f13ee6b6f5f4d5a640f293c),
(0xa4d0c416ee4a8bd8337455e00f9e9d888d644cf9b0db343f0de8521606c14eda, 0xa46f162d79d24ceaf81ef56d118d904067d1e78d9a17ef91f3542183dfb44665)
]
p = 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFEFFFFFC2F
n = 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFEBAAEDCE6AF48A03BBFD25E8CD0364141
msg = 0x590910812c6348ea5bb6a5d503200ffee73d337801843b857f74e3b5a6a2229d
lowSmax = 0x7FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF5D576E7357A4501DDFE92F46681B20A0
ZZn = Integers(n)
secp256k1 = EllipticCurve(FiniteField(p), (0, 7))
G = secp256k1.point((
0x79BE667EF9DCBBAC55A06295CE870B07029BFCDB2DCE28D959F2815B16F81798,
0x483ADA7726A3C4655DA4FBFC0E1108A8FD17B448A68554199C47D08FFB10D4B8
))
def sign(k, dQ):
R = k*G
r = ZZn(R[0])
s = ZZn((msg + r*dQ)/k)
if s > lowSmax:
s = n - s
return ("0x" + format(Integer(r), '064x'), "0x" + format(Integer(s), '064x'))
[sign(k, dQ) for (k, dQ) in ks]
Script for generating random curve-points.
tohex = lambda s: "0x" + format(Integer(s), '064x')
tohex3 = lambda (x, y, z): (tohex(x), tohex(y), tohex(z))
randpoints = [tohex3(secp256k1.random_point()) for i in range(0, 40)]
Script for calculating sums from randpoints (each element plus the next one).
sums = []
for i in range(1, 40):
P = secp256k1.point((Integer(randpoints[i - 1][0]), Integer(randpoints[i - 1][1])));
Q = secp256k1.point((Integer(randpoints[i][0]), Integer(randpoints[i][1])));
sums.append(tohex2(P + Q))
Script for doubling each point.
dbls = []
for i in range(0, 40):
P = secp256k1.point((Integer(randpoints[i][0]), Integer(randpoints[i][1])));
dbls.append(tohex2(P + P))
Script for generating a series of random elements in Z/pZ.
ZZp = Integers(p)
randints = [tohex(ZZp.random_element()) for i in range(0, 40)]
Script for doing point multiplication of random integers and points.
prods = []
for i in range(0, 40):
P = secp256k1.point((Integer(randpoints[i][0]), Integer(randpoints[i][1])));
prods.append(tohex2(Integer(randints[i]) * P))
Script used to generate some numbers for modular inverse:
testVals = [
(0x56, 0x55),
(0x2, 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF),
(0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFE, 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF),
(0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFEFFFFFC2F, 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF),
(0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF, 0xFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFEFFFFFC2F)
Output:
['0x0000000000000000000000000000000000000000000000000000000000000001',
'0x8000000000000000000000000000000000000000000000000000000000000000',
'0xfffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffffe',
'0x41bce9245fc72552d8c1b4259d8130978376be5acb4a45d8f4d5b4da913e7eb6',
'0xbe4316dba038daad273e4bda627ecf687c8941a534b5ba270b2a4b24b07e6798']
]
im = lambda a, p: "0x" + format(inverse_mod(a, p), '064x')