以太坊最佳手续费

在EIP-1559之前，ETH上的 gas fee 采用的是简单的竞拍模式，出价高者的交易优先被验证。这样的计算模式会导致 gas fee 因为人为因素（出价）而产生剧烈波动。EIP-1559 就是为了解决 gas fee 难以预测以及波动大的问题。

许多应用程序喜欢为用户提供设置自己的汽油费出价的选项，包括“慢速”、“平均”和“快速”选项。在本文中，我们将了解如何使用伦敦分叉后的 API 构建这些选项。

1.Gas Fee 的计算公式

units of gas used * (base fee + priority fee)

units of gas used 交易花费的gas单位数量，标准的一次简单ETH转账交易需要21000单位的gas

base fee 基础gas费，单位是gwei (1gwei = 0.000000001 ETH)

priority fee 优先费或者说是给矿工的小费和base fee 一样，单位为gwei

2.base fee | 基础费

首先，base fee 是根据上一个区块的base fee而来的。由eth节点自动完成计算，本质上区别于之前的竞拍模式。粗略来说上一个区块交易量大，则提高当前区块的base fee，反之降低。这样的动态调整。也就是说这个base fee 完全可以精准的计算出来。

EIP-1559 源代码

上述代码基于golang，我们将他转化为一段简单的python代码:

复制

from web3 import Web3

eth_json_rpc_endpoint = "https://services.tokenview.io/vipapi/nodeservice/eth?apikey=xxx"
ElasticityMultiplier = 2 # EIP-1559 The block size has been expanded, the maximum multiple is 2
BaseFeeChangeDenominator = 8 # The amount the base fee can change between blocks

def calc_base_fee_of_next_block(parent_block):
    parent_gas_target = parent_block.gasLimit // ElasticityMultiplier
    print('parent_gas_target',parent_gas_target)
    print('parent_block.gasUsed',parent_block.gasUsed)
    print('parent_block.baseFeePerGas',parent_block.baseFeePerGas)
    if parent_block.gasUsed == parent_gas_target:
        # parent block's gasUsed is the same as the target, baseFee remains unchanged
        return parent_block.baseFeePerGas
    if parent_block.gasUsed > parent_gas_target:
        # parent block uses gas greater than the target value, baseFee increase
        gas_used_delta = parent_block.gasUsed - parent_gas_target
        x = parent_block.baseFeePerGas * gas_used_delta
        y = x // parent_gas_target
        base_fee_delta = max(
            y // BaseFeeChangeDenominator,
            1
        )
        return parent_block.baseFeePerGas + base_fee_delta
    else:
        # the gas used by the parent block is less than the target value, baseFee reduce
        gas_used_delta = parent_gas_target - parent_block.gasUsed
        x = parent_block.baseFeePerGas * gas_used_delta
        y = x // parent_gas_target
        base_fee_delta = y // BaseFeeChangeDenominator
        return max(
            parent_block.baseFeePerGas - base_fee_delta,
            0
        )
        

def main():
    ethClient = Web3(Web3.HTTPProvider(eth_json_rpc_endpoint))
    block_number = ethClient.eth.block_number
    block = ethClient.eth.get_block(block_number)
    
    base_fee_of_next_block = calc_base_fee_of_next_block(block)
    print(f"Base fee for block {block_number + 1} will be {base_fee_of_next_block}")

3.priority fee | 优先费

与基础费不同，优先费是人为设定的值。对于需要在同一区块中优先执行的交易，需要提供更高的小费。要预测优先费，得扫描过去一段时间的区块，看看其他人都在用多少的费用

我们将会用到这个API：eth_feeHistory

复制

from web3 import Web3

eth_json_rpc_endpoint = "https://services.tokenview.io/vipapi/nodeservice/eth?apikey=xxx"
ethClient = Web3(Web3.HTTPProvider(eth_json_rpc_endpoint))
print(ethClient.eth.fee_history(4,"pending", [25, 50, 75]))

上述调用的意思是，“向我提供从待处理区块开始并向后查看 4 个区块的费用历史信息。对于每个区块，还向我提供该区块中交易的优先费用的第 25、50 和 75 个百分点”。原始结果如下所示:

复制

{
  "oldestBlock": 17913327,    
  "reward": [                
    [39519672,100000000,2000000000],
    [100000000,1000000000,3000000000],
    [100000000,365064718,1000000000],
    100000000,570000000,3000000000]
  ],
  "baseFeePerGas": [        
    21121728416,
    21666906452,
    20742307151,
    19782866894,
    17762883032
  ],
  "gasUsedRatio": [         
    0.6032449666666667,
    0.3293066333333333,
    0.31497906666666664,
    0.09156903333333333
  ]
}

为了让我们的计算更加容易，让我们编写一段格式化方法，将上述结果按块进行分组:

复制

def format_fee_history(result, include_pending):
    block_num = result['oldestBlock']
    index = 0
    blocks = []
    historical_blocks = len(result['reward'])
    while block_num < result['oldestBlock'] + historical_blocks:
       blocks.append({
           'number': block_num,
           'baseFeePerGas': float(result['baseFeePerGas'][index]),
           'gasUsedRatio': float(result['gasUsedRatio'][index]),
           'priorityFeePerGas': [float(x) for x in result['reward'][index]],
       })
       block_num += 1
       index += 1
    if include_pending:
       blocks.append({
           'number': 'pending',
           'baseFeePerGas': float(result['baseFeePerGas'][historical_blocks]),
           'gasUsedRatio': float('nan'),
           'priorityFeePerGas': [],
       })
    return blocks

接下来我们可以得到这样格式的数据:

复制

blocks = format_fee_history(ethClient.eth.fee_history(4,"latest", [25, 50, 75]),False)
print(blocks)

->>
[
  {
    number: 17913335,
    baseFeePerGas: 315777006840,
    gasUsedRatio: 0.9922326809477219,
    priorityFeePerGas: [ 34222993160, 34222993160, 63222993160 ]
  },
  {
    number: 17913336,
    baseFeePerGas: 354635947504,
    gasUsedRatio: 0.22772779167798343,
    priorityFeePerGas: [ 20000000000, 38044555767, 38364052496 ]
  },
  {
    number: 17913337,
    baseFeePerGas: 330496570085,
    gasUsedRatio: 0.8876034775653597,
    priorityFeePerGas: [ 9503429915, 19503429915, 36503429915 ]
  },
  {
    number: 17913338,
    baseFeePerGas: 362521975057,
    gasUsedRatio: 0.9909446241177369,
    priorityFeePerGas: [ 18478024943, 37478024943, 81478024943 ]
  }
]

可读性大大提高！其中 baseFeePerGas、gasUsedRatio 是用于计算base fee的。我们现在只关注priorityFeePerGas

继续我们的需求，要估算出priority fee 我们需要确定两个值:

i.要用多少的历史数据来进行估算

ii.希望自己的交易有多大的优先级?

对于 i ，我们将值设为4个区块，大概一分钟的时长算是比较合理

对于ii ，我们将期望定为25% 50% 75% 对应低，中，高的小费

复制

blocks = format_fee_history(ethClient.eth.fee_history(4,"latest", [25, 50, 75]),False)
low_percential_priority_fees = [b['priorityFeePerGas'][0] for b in blocks]
mid_percential_priority_fees = [b['priorityFeePerGas'][1] for b in blocks]
high_percential_priority_fees = [b['priorityFeePerGas'][2] for b in blocks]
low_average = sum(low_percential_priority_fees) / len(low_percential_priority_fees) 
mid_average = sum(mid_percential_priority_fees) / len(mid_percential_priority_fees) 
high_average = sum(high_percential_priority_fees) / len(high_percential_priority_fees) 
print(low_average,' | ',mid_average,' | ',high_average)

tokenview 使用的算法是简单地将过去4个历史块的priorityFeePerGas求平均值。当然您如果有更好的估算方式，比如将历史区块数量增加到20个块，或者是只考虑最便宜的几个成功交易。您可以构建您自己的gas fee 估算器

感谢您从头到尾阅读 🎉🎉🎉