Multihop feature request

Hey I've noticed that the prices are always displayed for direct routing, in many cases this returns very bad prices with high slippages.
Is there already a method to get the most efficient routing or could you implement it? Thanks!

@shanefontaine

You're right that it's not supported (and that it should be). Would require modifications to:

https://github.com/shanefontaine/uniswap-python/blob/e58f3fb91dfca181630fb2732910a65a406b40db/uniswap/uniswap.py#L552

https://github.com/shanefontaine/uniswap-python/blob/e58f3fb91dfca181630fb2732910a65a406b40db/uniswap/uniswap.py#L669

https://github.com/shanefontaine/uniswap-python/blob/e58f3fb91dfca181630fb2732910a65a406b40db/uniswap/uniswap.py#L316

https://github.com/shanefontaine/uniswap-python/blob/e58f3fb91dfca181630fb2732910a65a406b40db/uniswap/uniswap.py#L357

Specifically, the lines looking like this:

https://github.com/shanefontaine/uniswap-python/blob/e58f3fb91dfca181630fb2732910a65a406b40db/uniswap/uniswap.py#L703

Should be pretty easy to fix by adding an optional route kwarg (with the desired route) or similar for all the mentioned functions.

Not sure how you'd get the most efficient routing though (without brute force search). @shanefontaine do you know?

I believe the Uniswap router logic (off-chain) that chooses the path is closed source, unfortunately.

With that said, I cannot find any data that confirms or denies this. I will try to figure out if it is open source today.

Thanks for the quick feedback guys!
I'd think it would be open source since all the clones & forks also have the routing implemented (Quickswap and Sushiswap for example).

I suppose that multihop feature is implemented here

public static bestTradeExactIn
adn next one method bestTradeExactOut

@ErikBjare

@ErikBjare @shanefontaine Are you guys there?

Yup, waiting for someone who wants this to make a PR!

To my regret I can't make pr myself because my implementation differs too much but here is sort of the summary.

Let's say we need the optimal route for swap (token0,token1). Please consider that any of them might be eth/weth.

First, uniswap engine calls useAllCommonPairs(...) to get the list of possible routes. It doesn't contain EVERY possible route, just common ones. Key tokens are listed in BASES_TO_CHECK_TRADES_AGAINST. So, after that call we have list of the routes looking like [token0, basetoken0,token1], [token0, basetoken1,token1] etc. Please note that increasing hops count lead to dramatic growth of that list.

Second, we pass the list to bestTradeExactIn/Out(...) accordingly that analizes routes one by one via recursive calls and returns list of N routes sorted from the best to to worst.

Sure, this is not too easy to implement but also not THAT hard. Also remember that BASES_TO_CHECK_TRADES_AGAINST list varies from one network to another if you need alternate networks support like BSC/Heco/whatever.

Hope that helps, regards.

To my regret I can't make pr myself because my implementation differs too much but here is sort of the summary.

Let's say we need the optimal route for swap (token0,token1). Please consider that any of them might be eth/weth.

First, uniswap engine calls useAllCommonPairs(...) to get the list of possible routes. It doesn't contain EVERY possible route, just common ones. Key tokens are listed in BASES_TO_CHECK_TRADES_AGAINST. So, after that call we have list of the routes looking like [token0, basetoken0,token1], [token0, basetoken1,token1] etc. Please note that increasing hops count lead to dramatic growth of that list.

Second, we pass the list to bestTradeExactIn/Out(...) accordingly that analizes routes one by one via recursive calls and returns list of N routes sorted from the best to to worst.

Sure, this is not too easy to implement but also not THAT hard. Also remember that BASES_TO_CHECK_TRADES_AGAINST list varies from one network to another if you need alternate networks support like BSC/Heco/whatever.

Hope that helps, regards.

thanks, I will think

I added initial (largely untested) support for this in #58.

Closing this issue, opened #93 for finding the optimal trade route.

My solution is here. +- 1 in 1000 margin of error.

If a PR is requested, I can write a more professional one. But for me this is enough.

from concurrent.futures import ThreadPoolExecutor
from collections import defaultdict
from itertools import permutations
from uniswap import Uniswap
from web3 import Web3


class Token:
    def __init__(self, address, name, symbol, decimals) -> None:
        self.address = Web3.toChecksumAddress(address)
        self._raw_address = address
        self.name = name
        self.symbol = symbol
        self.decimals = decimals
        self.qty = 10 ** self.decimals


class UniswapPrice:
    sides = {
        'sell': 'get_price_output',
        'buy': 'get_price_input'
    }

    percentages = [50, 100]
    fees = [100, 300, 3000, 10000]

    def __init__(self, v2, v3):
        self.swap = {
            'v2': {
                'exchange': v2,
                'percentages': self.percentages,
                'fees': [3000]
            },
            'v3': {
                'exchange': v3,
                'percentages': self.percentages,
                'fees': self.fees
            }
        }

    def find_best(self):
        prices = [perm[0] + perm[1] for perm in permutations(
            self.pool_prices('different'), 2)] + self.pool_prices('single')
        if len(prices) == 0:
            return
        if self.side == 'sell':
            return min(prices)
        elif self.side == 'buy':
            return max(prices)

    def get_price(self, exchange, fee, qty):
        queue = [self.t0.address, self.t1.address] if self.side == 'buy' else [
            self.t1.address, self.t0.address]
        price_func = getattr(exchange, self.sides[self.side])
        return price_func(*queue, qty, fee=fee) / self.t1.qty

    def pool_prices(self, price_type):
        if price_type == 'single':
            perc = 100
        elif price_type == 'different':
            perc = 50

        return [
            fee[perc]
            for exchanges in self.futures.values()
            for fee in exchanges.values()
            if isinstance(fee[perc], float)
        ]

    def get_best_price(self, t0, t1, side):
        self.t0 = t0
        self.t1 = t1
        self.side = side

        self.get_prices()
        return self.find_best()

    def get_prices(self):
        self.futures = defaultdict(dict)

        with ThreadPoolExecutor(max_workers=10) as executor:
            for exchange_name, exchange_data in self.swap.items():
                self.futures[exchange_name] = defaultdict(dict)
                for fee in exchange_data['fees']:
                    self.futures[exchange_name][fee] = defaultdict(dict)
                    for percentage in exchange_data['percentages']:
                        self.futures[exchange_name][fee][percentage] = defaultdict(
                            dict)
                        qty = self.t0.qty * percentage // 100
                        self.futures[exchange_name][fee][percentage] = executor.submit(
                            self.get_price, exchange_data['exchange'], fee, qty
                        )

            for exchange_name, exchange_fees in self.futures.items():
                self.futures[exchange_name] = dict(exchange_fees)
                for fee, percentage_data in exchange_fees.items():
                    self.futures[exchange_name][fee] = dict(percentage_data)
                    for percentage, future in percentage_data.items():
                        try:
                            self.futures[exchange_name][fee][percentage] = future.result(
                            )
                        except:
                            pass


web3 = Web3(Web3.HTTPProvider("YOUR-JSON-RPC-URL"))

uni_v2 = Uniswap(
    address=None,
    private_key=None,
    version=2,
    web3=web3
)

uni_v3 = Uniswap(
    address=None,
    private_key=None,
    version=3,
    web3=web3
)

t0 = Token(**{
               "symbol": "ETH",
               "name": "Wrapped Ethereum",
               "decimals": 18,
               "address": "0xC02aaA39b223FE8D0A0e5C4F27eAD9083C756Cc2"
           }
)


t1 = Token(**{
               "symbol": "WBTC",
               "name": "Wrapped Bitcoin",
               "decimals": 8,
               "address": "0x2260fac5e5542a773aa44fbcfedf7c193bc2c599"
           }
)


price_calculator = UniswapPrice(uni_v2, uni_v3)

print(price_calculator.get_best_price(t0, t1, 'buy'),
      price_calculator.get_best_price(t0, t1, 'sell'))

Navigation Menu

Repository navigation

Multihop feature request #69

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Multihop feature request #69

Description

Activity

ErikBjare commented on Mar 7, 2021

shanefontaine commented on Mar 8, 2021

MoneroCrusher commented on Mar 9, 2021

liquid-8 commented on Mar 11, 2021

MoneroCrusher commented on Mar 19, 2021

ErikBjare commented on Mar 20, 2021

liquid-8 commented on Mar 22, 2021

dustiwe commented on Mar 23, 2021

ErikBjare commented on May 4, 2021

qraxiss commented on Mar 11, 2023

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions