[Draft code] average gas improvements

queries/num-trades-per-batch.sql

@@ -0,0 +1,6 @@
+SELECT 
+    num_trades,
+    CONCAT('0x', ENCODE(tx_hash, 'hex')) as txHash 
+FROM gnosis_protocol_v2."batches" b
+where block_time > now() - interval '5 hour'
+ORDER BY block_time DESC;

src/gas_saved_2.py

@@ -0,0 +1,63 @@
+import os
+
+import click
+import pandas as pd
+from dotenv import load_dotenv
+from web3 import Web3
+
+from src.db.pg_client import pg_engine
+
+from duneapi.api import DuneAPI
+from duneapi.types import DuneQuery, Network
+from duneapi.util import open_query
+
+
+def fetch_trades_per_batch(dune: DuneAPI):
+    query = DuneQuery.from_environment(
+        raw_sql=open_query("./queries/num-trades-per-batch.sql"),
+        name="num-trades-per-batch",
+        network=Network.MAINNET,
+        parameters=[],
+    )
+    results = dune.fetch(query)
+    return results
+
+
+def get_percentage_gas_used_of_estimate(batch_tx_hash):
+    db_engine = pg_engine()
+    GAS_QUOTES_QUERY = f"""
+    SELECT orders.id::bytea, order_quotes.gas_amount, order_quotes.gas_price
+    FROM
+        solver_competitions,
+        jsonb_to_recordset(solver_competitions.json->'solutions'->-1->'orders') AS orders(id text)
+    LEFT JOIN order_quotes ON order_quotes.order_uid = ('\\' || LTRIM(orders.id::text, '0'))::bytea
+    WHERE
+      solver_competitions.tx_hash = '\\{batch_tx_hash[1:]}'
+    """
+    df_quotes = pd.read_sql(GAS_QUOTES_QUERY, db_engine)
+    # subtract settlement_overhead from price estimation
+    # Ref: https://github.com/cowprotocol/services/blob/fd5f7cf47a6afdff89b310b60b869dfc577ac7a7/crates/shared/src/price_estimation/gas.rs#L37
+    df_quotes["gas_amount"] = df_quotes["gas_amount"].apply(lambda x: x - 106391)
+    load_dotenv()
+    w3 = Web3(
+        Web3.HTTPProvider(f"https://mainnet.infura.io/v3/{os.environ['INFURA_KEY']}")
+    )
+    tx = w3.eth.get_transaction_receipt(batch_tx_hash)
+    if df_quotes['gas_amount'].sum() == 0:
+        return -1
+    print(batch_tx_hash)
+    print(tx.gasUsed/df_quotes['gas_amount'].sum())
+    return tx.gasUsed/df_quotes['gas_amount'].sum()
+
+
+
+if __name__ == "__main__":
+    dune_conn = DuneAPI.new_from_environment()
+    df = fetch_trades_per_batch(dune_conn)
+    data = {'tx_hash': [], 'num_trades': [], 'gas_percentage': []}
+    df_result = pd.DataFrame(data)
+    for i in df:
+        new_row = {'tx_hash':i['txhash'], 'num_trades':i['num_trades'], 'gas_percentage':get_percentage_gas_used_of_estimate(i['txhash'])}
+        df_result = df_result.append(new_row, ignore_index=True)
+
+    print(df_result[df_result.gas_percentage > 0].groupby('num_trades')['gas_percentage'].mean())