Remove CoW AMM indexer from driver

[squadgazzz]

Description

Currently, both autopilot and driver use the same CoW AMM indexer functionality in parallel to serve different purposes: autopilot needs to fill out surplus capturing jit order owners in the auction and driver uses it to fetch tradable tokens and then prepare CoW AMM template orders. Since autopilot already sends CoW AMM addresses within each auction, the only missing part is the CoW AMM helper SC address. It can be received by fetching the CoW AMM factory address and then using a config mapping to find the helper address. This should also reduce the RPC traffic.

Changes

• Driver now contains only the mapping between AMM factory and helper SC.
• When it receives a CoW AMM in the auction, it tries to fetch the AMM factory address by calling the FACTORY function. Since we can now completely deprecate legacy CoW AMMs, this logic is simplified. I also used a simple abi to create rust bindings for this interface. The fetched data is stored in the memory cache.
• Using the factory address, it gets the helper SC instance based on the config.
• Simplify the e2e tests config.
• Adjusted a forked e2e test to start using the new factory address with some caveats(the test with its comments should be self-descriptive).

How to test

Adjusted forked e2e test.

[github-actions]

Reminder: Please consider backward compatibility when modifying the API specification.
If breaking changes are unavoidable, ensure:

• You explicitly pointed out breaking changes.
• You communicate the changes to affected teams (at least Frontend team and SAFE team).
• You provide proper versioning and migration mechanisms.

---

Resolved

[Copilot]

Pull Request Overview

Removes the CoW AMM indexer from the driver by having autopilot send CoW AMM to helper address mappings directly in auction data. This eliminates the need for driver to index CoW AMM deployment events from blockchain data.

• Replaces old Vec<Address> CoW AMM list with HashMap<Address, Vec<Address>> mapping helpers to AMMs
• Removes CoW AMM configuration and indexing infrastructure from driver
• Creates in-memory cache for CoW AMM instances to avoid re-indexing on restarts

Reviewed Changes

Copilot reviewed 36 out of 36 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
crates/e2e/src/setup/deploy.rs	Removes CoW AMM helper contract deployment and configuration
crates/e2e/src/setup/colocation.rs	Removes CoW AMM configuration generation for test setup
crates/driver/src/tests/setup/solver.rs	Updates test setup to use new helper-to-owners mapping structure
crates/driver/src/tests/setup/mod.rs	Changes surplus capturing owner storage from Vec to HashMap by helper
crates/driver/src/tests/setup/driver.rs	Updates API request field name for new data structure
crates/driver/src/tests/cases/jit_orders.rs	Updates test case to use new helper-based owner mapping
crates/driver/src/run.rs	Removes archive node URL parameter from Ethereum initialization
crates/driver/src/infra/solver/mod.rs	Updates method call to use new helper-based owner mapping
crates/driver/src/infra/solver/dto/solution.rs	Updates method call for surplus capturing owner access
crates/driver/src/infra/solver/dto/auction.rs	Changes surplus owner extraction to use HashMap keys
crates/driver/src/infra/observe/mod.rs	Updates function signature to accept HashMap instead of HashSet
crates/driver/src/infra/config/mod.rs	Removes archive node URL and CoW AMM configuration fields
crates/driver/src/infra/config/file/mod.rs	Removes CoW AMM configuration structures and archive node field
crates/driver/src/infra/config/file/load.rs	Removes CoW AMM and archive node configuration loading
crates/driver/src/infra/blockchain/mod.rs	Removes archive node parameter from Ethereum constructor
crates/driver/src/infra/blockchain/contracts.rs	Removes CoW AMM registry and related indexing infrastructure
crates/driver/src/infra/api/routes/solve/dto/solve_request.rs	Updates API to accept HashMap and convert to owner-to-helper mapping
crates/driver/src/domain/quote.rs	Updates method calls to use new mapping structure
crates/driver/src/domain/mod.rs	Adds cow_amm module to domain
crates/driver/src/domain/cow_amm.rs	Implements new CoW AMM cache for on-demand AMM instance creation
crates/driver/src/domain/competition/solution/settlement.rs	Updates scoring method signature for new mapping
crates/driver/src/domain/competition/solution/mod.rs	Updates solution methods to use HashMap instead of HashSet
crates/driver/src/domain/competition/pre_processing.rs	Integrates CoW AMM cache and updates AMM order generation
crates/driver/src/domain/competition/mod.rs	Updates solution filtering and scoring calls
crates/driver/src/domain/competition/auction.rs	Changes surplus capturing owner storage to HashMap
crates/driver/openapi.yml	Updates API schema to reflect new HashMap structure
crates/driver/example.toml	Removes CoW AMM configuration example
crates/cow-amm/src/amm.rs	Makes Amm constructor public and adds helper address accessor
crates/autopilot/src/solvable_orders.rs	Groups CoW AMMs by helper address and updates fee calculation
crates/autopilot/src/run_loop.rs	Updates auction data structure usage throughout
crates/autopilot/src/infra/solvers/dto/solve.rs	Sends both old and new data structures for backward compatibility
crates/autopilot/src/infra/persistence/mod.rs	Updates persistence to handle new HashMap structure
crates/autopilot/src/infra/persistence/dto/auction.rs	Updates auction DTO to use HashMap for surplus owners
crates/autopilot/src/domain/fee/mod.rs	Updates fee calculation to work with new HashMap structure
crates/autopilot/src/domain/competition/winner_selection.rs	Updates winner selection to extract owners from HashMap
crates/autopilot/src/domain/auction/mod.rs	Changes auction data structure to use HashMap

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[Copilot]

The comment starting at line 358 appears to be incomplete or cut off. The comment should be completed to explain what happens when reference prices are not provided.

[MartinquaXD]

You can avoid sending any additional data in the auction by looking up each cow amms factory address with .FACTORY. The driver already has a mapping from factory to helper contract.

[squadgazzz]

You can avoid sending any additional data in the auction by looking up each cow amms factory address with .FACTORY. The driver already has a mapping from factory to helper contract.

Sorry, which factory? The proposal fully drops the driver's cow amm config.

[MartinquaXD]

The cow amm contracts have a public member variable FACTORY that you can query with an eth_call.
I think rather than effectively moving the config into the autopilot by distributing the addresses in the auction we should keep the cowAmmFactory -> helperContract mapping in the driver config and have the driver look up the factory from the cow amm address.

[jmg-duarte]

LGTM, I left some nits and one point for discussion before approval

[MartinquaXD]

Nice that you already thought of removing this. 👍

[MartinquaXD]

Why do you prefer using a hardcoded value here instead of computing it using information queried on the pool?
If someone were to update the WETH we send to the pool they would also have to redo this math.

[squadgazzz]

Makes sense, updated.

[MartinquaXD]

Sorry for flip flopping on this maybe I didn't fully understand the need to make the imbalance a specific value. Judging by the comment it seems like as long as the imbalance on the pool is less than 50% we are good. If we don't actually have any test asserting properties on the order prices based on that imbalance we could also just imbalance the pool using some hardcoded amount and add an assertion + comment to make sure the pool is less than 50% imbalanced (whatever that means).
Maybe you could elaborate a bit more why it's now useful to unbalance the pool to a very specific amount. 🤔

[squadgazzz]

Maybe you could elaborate a bit more why it's now useful to unbalance the pool to a very specific amount. 🤔

Maybe i need to rephrase the code comment. The idea is to imbalance by 30% or so and that is all. I just funded it with 10 WETH, then we need to calculate a proper USDC value to imbalance it enough, so the driver can create a CoW AMM order.

[squadgazzz]

At the time of finalization, the AMM it empty if I didn't miss anything, so we need at least something there.

[MartinquaXD]

At the block you have configured for the forked test the pool had 1925 USDC and 0.77 WETH in a 50/50 weighting. That means the amount you imbalance the pool with is relatively small (originally 2500 USDC / WETH and now with 2551 USDC / WETH). I think simply doing a small one sided deposit should be sufficient and avoid a bunch of math and details.

[squadgazzz]

Ah, true, i somehow missed the correct balances. Simplified the test.

[MartinquaXD]

Could you also update this part to check that the driver was actually able to generate a cow amm order for the desired pool?

[squadgazzz]

Done

[squadgazzz]

Btw, the amounts are pretty weird. Sell/buy amounts are about $10...

Order {
    uid: [81, 95, 167, 62, 197, 127, 127, 51, 77, 255, 175, 153, 213, 48, 132, 162, 19, 231, 229, 129, 4, 104, 182, 49, 246, 123, 172, 231, 142, 82, 222, 32, 240, 141, 77, 234, 54, 156, 69, 109, 38, 163, 22, 143, 240, 2, 75, 144, 79, 45, 139, 145, 102, 179, 104, 168],
    sell_token: 0xc02aaa39b223fe8d0a0e5c4f27ead9083c756cc2,
    buy_token: 0xa0b86991c6218b36c1d19d4a2e9eb0ce3606eb48,
    sell_amount: 4365551272787746,
    full_sell_amount: 4365551272787746,
    buy_amount: 10800057,
    full_buy_amount: 10800057,
    fee_policies: None,
    valid_to: 1723033768,
    kind: Sell,
    receiver: Some(0x0000000000000000000000000000000000000000),
    owner: 0xf08d4dea369c456d26a3168ff0024b904f2d8b91,
    partially_fillable: true,
    pre_interactions: [InteractionData {
        target: 0xf08d4dea369c456d26a3168ff0024b904f2d8b91,
        value: 0,
        call_data: [241, 79, 203, 200, 81, 95, 167, 62, 197, 127, 127, 51, 77, 255, 175, 153, 213, 48, 132, 162, 19, 231, 229, 129, 4, 104, 182, 49, 246, 123, 172, 231, 142, 82, 222, 32]
    }],
    post_interactions: [],
    sell_token_source: Erc20,
    buy_token_destination: Erc20,
    class: Limit,
    app_data: 0x362e5182440b52aa8fffe70a251550fbbcbca424740fe5a14f59bf0c1b06fe1d,
    flashloan_hint: None,
    signing_scheme: Eip1271,
    signature: [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 192, 42, 170, 57, 178, 35, 254, 141, 10, 14, 92, 79, 39, 234, 217, 8, 60, 117, 108, 194, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 160, 184, 105, 145, 198, 33, 139, 54, 193, 209, 157, 74, 46, 158, 176, 206, 54, 6, 235, 72, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 15, 130, 114, 28, 220, 255, 34, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 164, 203, 185, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 102, 179, 104, 168, 54, 46, 81, 130, 68, 11, 82, 170, 143, 255, 231, 10, 37, 21, 80, 251, 188, 188, 164, 36, 116, 15, 229, 161, 79, 89, 191, 12, 27, 6, 254, 29, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 243, 178, 119, 114, 139, 63, 238, 116, 148, 129, 235, 62, 11, 59, 72, 152, 13, 187, 171, 120, 101, 143, 196, 25, 2, 92, 177, 110, 238, 52, 103, 117, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 90, 40, 233, 54, 59, 185, 66, 182, 57, 39, 0, 98, 170, 107, 178, 149, 244, 52, 188, 223, 196, 44, 151, 38, 123, 240, 3, 242, 114, 6, 13, 201, 90, 40, 233, 54, 59, 185, 66, 182, 57, 39, 0, 98, 170, 107, 178, 149, 244, 52, 188, 223, 196, 44, 151, 38, 123, 240, 3, 242, 114, 6, 13, 201]
}

[MartinquaXD]

What is the native price for USDC in the test? The helper contract computes a suggested order based on the price difference between native price and the pool price. If they are close together the order would be small.

[MartinquaXD]

The logic change looks solid to me. The remaining open questions are mostly about how to unbalance the pool to make sense of this. But given that we don't want to test the logic of the helper contract and already have an assertion that we are generating SOME suggested order I'm fine with merging this.
Although simplifying the unbalancing pool to do a single one sided deposit might be nicer.