Wallet: skip parent fetch in determine_coin_type for locally-known DL coins

[TheLastCicada]

Purpose:

Speed up wallet sync of DataLayer singletons (the CADT-style bulk-subscription case) by classifying tracked DL coins from local metadata before paying the parent get_coin_state + request_puzzle_solution round-trip in WalletStateManager.determine_coin_type.

Current Behavior:

For DataLayer mirror coins and previously-tracked singletons, WalletStateManager._add_coin_states falls through to determine_coin_type, which fetches the parent coin state and the puzzle solution and runs CLVM uncurry. None of the CAT/NFT/DID/Clawback/VC patterns match for these coins; the function returns and a subsequent block in _add_coin_states re-queries dl_store (or compares against MIRROR_PUZZLE_HASH) and re-classifies the coin to the DL wallet. The two RPCs and the CLVM uncurry are wasted work for every DL-classified coin, and the worst case (a heavily-used DL launcher being replayed against a freshly-tracking wallet) is ten of thousands of singleton coin states each paying that tax.

New Behavior:

A new WalletStateManager._data_layer_identifier_for_coin helper performs the same predicate (puzzle_hash == MIRROR_PUZZLE_HASH or dl_store.get_singleton_record(coin.name()) is not None) using only local data. determine_coin_type calls it after the existing pool / farmer reward early-return and returns the DL WalletIdentifier immediately when it matches, skipping the parent fetch and the CLVM uncurry. The post-call DL override block in _add_coin_states is removed because the helper covers the same predicate strictly earlier in the per-coin pipeline.

Invariants unchanged:

• Set of coins classified as DL is identical (same predicate, evaluated in the same transaction).
• coin_data for DL-classified coins is None on this path, matching what the old post-call override produced.
• Pool / farmer reward early-return runs before the new short-circuit.
• Coins not classified by local DL metadata still take the full parent-fetch path through determine_coin_type.
• The request_puzzle_solution and parent get_coin_state round-trip are still performed for all other classifications (CAT, NFT, DID, Clawback, VC, notifications).

Testing Notes:

New unit tests in chia/_tests/wallet/test_wallet_state_manager.py pin each branch of the helper:

• test_data_layer_identifier_for_coin_returns_none_without_dl_wallet
• test_data_layer_identifier_for_coin_mirror_puzzle_hash
• test_data_layer_identifier_for_coin_tracked_singleton
• test_data_layer_identifier_for_coin_no_match_with_dl_wallet

New integration test test_add_coin_states_skips_parent_fetch_for_tracked_dl_singleton drives a tracked DL singleton through the production entry point (WalletStateManager.add_coin_states, the same one used by add_states_from_peer in trusted mode at wallet_node.py:999) with WalletNode.get_coin_state monkeypatched as a recorder. It asserts that parent_fetch_calls == [] and that the coin is owned by the DL wallet.

Mutation-verified locally: temporarily disabling the short-circuit makes the integration test fail with parent_fetch_calls != [], then re-enabling it restores the pass.

Local validation pipeline (skipping benchmarks, will run in CI):

• Phase 1 (ruff check / ruff format --check / mypy on changed files): clean.
• Phase 2 (targeted tests): chia/_tests/wallet/test_wallet_state_manager.py (39 selected), chia/_tests/wallet/db_wallet/test_dl_wallet.py (mirror, lifecycle, reorg, ownership-rollback), chia/_tests/wallet/test_wallet_node.py (41 selected), chia/_tests/wallet/sync/test_wallet_sync.py (sync, long_sync, reorgs, dusted wallet) — 120+ tests passing.
• Phase 5 (pre-commit run --all-files): clean.

[coveralls-official]

Coverage Report for CI Build 25130365079

Coverage increased (+0.03%) to 91.241%

Details

• Coverage increased (+0.03%) from the base build.
• Patch coverage: 95 of 95 lines across 2 files are fully covered (100%).
• 13 coverage regressions across 7 files.

Uncovered Changes

No uncovered changes found.

Coverage Regressions

13 previously-covered lines in 7 files lost coverage.

File	Lines Losing Coverage	Coverage
chia/full_node/full_node_api.py	3	86.47%
chia/timelord/timelord.py	3	72.76%
chia/full_node/full_node.py	2	87.64%
chia/timelord/timelord_launcher.py	2	70.0%
chia/data_layer/data_layer.py	1	85.68%
chia/_tests/core/util/test_file_keyring_synchronization.py	1	96.88%
chia/_tests/simulation/test_simulation.py	1	96.49%

---

Coverage Stats

Relevant Lines:	118125
Covered Lines:	107945
Line Coverage:	91.38%
Relevant Branches:	11749
Covered Branches:	10553
Branch Coverage:	89.82%
Branches in Coverage %:	Yes
Coverage Strength:	1.83 hits per line

---

💛 - Coveralls