feat(solana): add Drift Protocol visualizer preset

[prasanna-anchorage]

Summary

• Adds IDL-based visualizer preset for Drift Protocol v2 (dRiftyHA39MWEi3m9aunc5MzRF1JYuBsbn6VPcn33UH)
• 249 instructions decoded via Anchor IDL from drift-labs/protocol-v2
• Uses LazyLock to cache parsed 438KB IDL (parsed once, not on every call)
• Includes 4 unit tests: IDL loading, discriminator validation, unknown discriminator error, short data error

Test plan

• cargo clippy -p visualsign-solana --all-targets -- -D warnings
• cargo test -p visualsign-solana (all 108 tests pass)
• make -C src test (full workspace)

🤖 Generated with Claude Code

[Copilot]

Pull request overview

Adds a new Solana preset visualizer that decodes Drift Protocol v2 instructions via an embedded Anchor IDL, producing structured preview fields (with a raw-data fallback) and caching the parsed IDL for performance.

Changes:

• Registered a new drift preset module in the Solana presets registry.
• Implemented DriftVisualizer with IDL-backed instruction parsing + account labeling + raw-data fallback.
• Added a Drift integration config to route Drift program instructions to the new visualizer.

Reviewed changes

Copilot reviewed 3 out of 4 changed files in this pull request and generated 3 comments.

File	Description
src/chain_parsers/visualsign-solana/src/presets/mod.rs	Exposes the new drift preset module so it can be instantiated by the build-time visualizer registry.
src/chain_parsers/visualsign-solana/src/presets/drift/mod.rs	Implements Drift instruction visualization using an embedded IDL with LazyLock caching and adds unit tests.
src/chain_parsers/visualsign-solana/src/presets/drift/config.rs	Adds the config mapping for the Drift program ID so the visualizer can handle Drift instructions.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

visualize_tx_commands discards the concrete parsing error (Err(_)) and always falls back to an "Unknown instruction type" display. This makes failures like short data, IDL load failures, or decode errors indistinguishable from an actually-unknown discriminator. Consider capturing the error and surfacing it (e.g., in a Status field and/or via tracing::warn!) so users and logs reflect the real failure reason.

Suggested change

	Err(_) => build_fallback_fields(&instruction.program_id.to_string()),
	Err(err) => {
	tracing::warn!(
	program_id = %instruction.program_id,
	error = %err,
	"Failed to parse Drift instruction"
	);
	let status_text = format!("Failed to parse Drift instruction: {err}");
	let (title, mut condensed_fields, mut expanded_fields) =
	build_fallback_fields(&instruction.program_id.to_string());
	condensed_fields.push(create_text_field("Status", &status_text));
	expanded_fields.push(create_text_field("Status", &status_text));
	(title, condensed_fields, expanded_fields)
	}

[Copilot]

DriftParsedInstruction stores named_accounts separately even though SolanaParsedInstructionData already supports named accounts (see the existing population pattern in presets/unknown_program/mod.rs). Keeping a parallel map increases maintenance overhead and risks divergence between the parsed payload and displayed account labels. Consider populating parsed.named_accounts directly and using it when building fields, removing the extra wrapper struct if possible.

[Copilot]

The expanded view relies on build_named_accounts to label accounts, but current tests only cover IDL loading/discriminators and generic error cases. Add a focused unit test that exercises build_named_accounts (or the end-to-end visualize path) for at least one known discriminator to ensure account names are mapped correctly and remain stable as the embedded IDL changes.

Suggested change

	#[cfg(test)]
	mod tests {
	use super::*;
	#[test]
	fn build_named_accounts_maps_initialize_user_accounts_by_name() {
	let idl = get_drift_idl().expect("Drift IDL should be available for tests");
	let instruction = idl
	.instructions
	.iter()
	.find(|inst| inst.name == "initializeUser")
	.expect("expected initializeUser instruction in embedded Drift IDL");
	let discriminator = instruction
	.discriminator
	.clone()
	.expect("initializeUser should have a discriminator");
	let expected_account_names = [
	"user",
	"userStats",
	"state",
	"authority",
	"payer",
	"rent",
	"systemProgram",
	];
	let actual_account_names: Vec<&str> = instruction
	.accounts
	.iter()
	.take(expected_account_names.len())
	.map(|account| account.name.as_str())
	.collect();
	assert_eq!(
	actual_account_names,
	expected_account_names,
	"initializeUser accounts changed in the embedded Drift IDL; update this test if the change is intentional"
	);
	let accounts: Vec<solana_sdk::instruction::AccountMeta> = expected_account_names
	.iter()
	.enumerate()
	.map(|(index, _)| {
	let pubkey = solana_sdk::pubkey::Pubkey::new_from_array([index as u8 + 1; 32]);
	solana_sdk::instruction::AccountMeta::new(pubkey, false)
	})
	.collect();
	let named_accounts = build_named_accounts(&discriminator, idl, &accounts);
	assert_eq!(named_accounts.len(), expected_account_names.len());
	for (index, account_name) in expected_account_names.iter().enumerate() {
	let expected_pubkey =
	solana_sdk::pubkey::Pubkey::new_from_array([index as u8 + 1; 32]).to_string();
	assert_eq!(
	named_accounts.get(*account_name),
	Some(&expected_pubkey),
	"account name {account_name} should map to the corresponding account meta pubkey"
	);
	}
	}
	}

[shahan-khatchadourian-anchorage]

Thanks for this — solid addition, and the structure is right. A few things to call out before it can land.

What's good

• Follows the IDL-preset pattern cleanly: include_str! + LazyLock to parse the 438KB IDL exactly once is the correct approach, and the file layout (mod.rs / config.rs / drift.json) matches the other presets.
• Graceful "Unknown Instruction" fallback on decode failure is consistent with the recent spl_token degrade-gracefully behaviour rather than erroring.
• Uses the field_builders helpers throughout instead of hand-building field structs, and the Current/New labeling on account/arg name collisions is a thoughtful touch.
• Sensible unit tests for IDL loading, discriminator shape (8 bytes), and the short-data / unknown-discriminator error paths.

Blockers (all from the branch being ~168 commits behind main)
This branch predates two refactors on main, so the test-plan claims hold against the old base but not current main. After rebasing:

1. config.rs won't compile — SolanaIntegrationConfigData.programs is now BTreeMap, not HashMap (changed for deterministic serialization). Switch the two HashMap::new()s to BTreeMap, matching onre_app / exponent_finance.
2. Drop the presets/mod.rs edit entirely — since #278 that file is auto-generated by build.rs from the directory listing (header says "do NOT edit"). Once rebased, drift/ is auto-discovered and DriftVisualizer registers with no manual wiring.
3. Test module needs #[allow(clippy::unwrap_used, clippy::expect_used, clippy::panic)] — every other preset test module has it; without it clippy -D warnings fails CI.

Worth folding in (overlaps with Copilot's open threads)

• named_accounts as a HashMap gives non-deterministic field ordering in the output, which breaks stable metadata hashing — the fields aren't re-sorted downstream. Copilot also flagged this struct from a maintainability angle (it duplicates SolanaParsedInstructionData::named_accounts). Populating the existing parsed.named_accounts directly resolves both at once.
• A real end-to-end fixture test (decode one known instruction, e.g. initializeUser / placePerpOrder, to expected fields) would lock in the account-name mapping as the IDL evolves — currently nothing exercises the decode path against a real tx.

Also flagging for cross-PR awareness: #350/#351/#360 are other in-flight Solana presets built on current main (none touch presets/mod.rs), and #289 reworks the solana-add-idl scaffolding — worth rebasing onto and regenerating from, so this preset matches the current template.

Happy to take the mechanical fixes (1–3 + the named_accounts swap) on a rebased branch if useful.

[shahan-khatchadourian-anchorage]

I've opened #366, which rebases this work onto current main and gets it building/linting/testing green. This branch is ~168 commits behind and no longer compiles against main (the HashMap→BTreeMap switch in SolanaIntegrationConfigData, the auto-generated presets/mod.rs from #278, and the removed VisualizerContext::current_instruction() API all broke it).

#366 keeps your implementation intact and applies only the mechanical fixes needed to land it — full details in its description. Leaving this PR open for now; feel free to review #366 or pull its changes back here, whichever you prefer.