reth_trie/

witness.rs

use crate::{
    hashed_cursor::HashedCursorFactory, prefix_set::TriePrefixSetsMut, proof::Proof, proof_v2,
    trie_cursor::TrieCursorFactory, TRIE_ACCOUNT_RLP_MAX_SIZE,
};
use alloy_primitives::{
    keccak256,
    map::{B256Map, HashMap},
    Bytes, B256,
};
use alloy_rlp::{Encodable, EMPTY_STRING_CODE};
use alloy_trie::{nodes::BranchNodeRef, EMPTY_ROOT_HASH};
use reth_execution_errors::{SparseStateTrieErrorKind, StateProofError, TrieWitnessError};
use reth_trie_common::{
    DecodedMultiProofV2, ExecutionWitnessMode, HashedPostState, MultiProofTargetsV2, ProofV2Target,
    TrieNodeV2,
};
use reth_trie_sparse::{LeafUpdate, SparseStateTrie, SparseTrie as _, TrieNodeEpoch};

/// State transition witness for the trie.
#[derive(Debug)]
pub struct TrieWitness<T, H> {
    /// The cursor factory for traversing trie nodes.
    trie_cursor_factory: T,
    /// The factory for hashed cursors.
    hashed_cursor_factory: H,
    /// A set of prefix sets that have changes.
    prefix_sets: TriePrefixSetsMut,
    /// Flag indicating whether the root node should always be included (even if the target state
    /// is empty). This setting is useful if the caller wants to verify the witness against the
    /// parent state root.
    /// Set to `false` by default.
    always_include_root_node: bool,
    /// Controls how the witness is generated.
    mode: ExecutionWitnessMode,
    /// Recorded witness.
    witness: B256Map<Bytes>,
}

impl<T, H> TrieWitness<T, H> {
    /// Creates a new witness generator.
    pub fn new(trie_cursor_factory: T, hashed_cursor_factory: H) -> Self {
        Self {
            trie_cursor_factory,
            hashed_cursor_factory,
            prefix_sets: TriePrefixSetsMut::default(),
            always_include_root_node: false,
            mode: ExecutionWitnessMode::Legacy,
            witness: HashMap::default(),
        }
    }

    /// Set the trie cursor factory.
    pub fn with_trie_cursor_factory<TF>(self, trie_cursor_factory: TF) -> TrieWitness<TF, H> {
        TrieWitness {
            trie_cursor_factory,
            hashed_cursor_factory: self.hashed_cursor_factory,
            prefix_sets: self.prefix_sets,
            always_include_root_node: self.always_include_root_node,
            mode: self.mode,
            witness: self.witness,
        }
    }

    /// Set the hashed cursor factory.
    pub fn with_hashed_cursor_factory<HF>(self, hashed_cursor_factory: HF) -> TrieWitness<T, HF> {
        TrieWitness {
            trie_cursor_factory: self.trie_cursor_factory,
            hashed_cursor_factory,
            prefix_sets: self.prefix_sets,
            always_include_root_node: self.always_include_root_node,
            mode: self.mode,
            witness: self.witness,
        }
    }

    /// Set the prefix sets. They have to be mutable in order to allow extension with proof target.
    pub fn with_prefix_sets_mut(mut self, prefix_sets: TriePrefixSetsMut) -> Self {
        self.prefix_sets = prefix_sets;
        self
    }

    /// Set `always_include_root_node` to true. Root node will be included even in empty state.
    /// This setting is useful if the caller wants to verify the witness against the
    /// parent state root.
    pub const fn always_include_root_node(mut self) -> Self {
        self.always_include_root_node = true;
        self
    }

    /// Set the execution witness generation mode.
    pub const fn with_execution_witness_mode(mut self, mode: ExecutionWitnessMode) -> Self {
        self.mode = mode;
        self
    }
}

impl<T, H> TrieWitness<T, H>
where
    T: TrieCursorFactory + Clone,
    H: HashedCursorFactory + Clone,
{
    /// Compute the state transition witness for the trie. Gather all required nodes
    /// to apply `state` on top of the current trie state.
    ///
    /// # Arguments
    ///
    /// `state` - state transition containing both modified and touched accounts and storage slots.
    pub fn compute(mut self, state: HashedPostState) -> Result<B256Map<Bytes>, TrieWitnessError> {
        let is_state_empty = state.is_empty();
        if is_state_empty && !self.always_include_root_node {
            return Ok(Default::default())
        }

        let proof_targets = if is_state_empty {
            MultiProofTargetsV2 {
                account_targets: vec![ProofV2Target::new(B256::ZERO)],
                ..Default::default()
            }
        } else {
            Self::get_proof_targets(&state)
        };
        let multiproof =
            Proof::new(self.trie_cursor_factory.clone(), self.hashed_cursor_factory.clone())
                .with_prefix_sets_mut(self.prefix_sets.clone())
                .multiproof_v2(proof_targets)?;

        // No need to reconstruct the rest of the trie, we just need to include
        // the root node and return.
        if is_state_empty {
            let (root_hash, root_node) = if let Some(root_node) =
                multiproof.account_proofs.into_iter().find(|n| n.path.is_empty())
            {
                let mut encoded = Vec::new();
                root_node.node.encode(&mut encoded);
                let bytes = Bytes::from(encoded);
                (keccak256(&bytes), bytes)
            } else {
                (EMPTY_ROOT_HASH, Bytes::from([EMPTY_STRING_CODE]))
            };
            return Ok(B256Map::from_iter([(root_hash, root_node)]))
        }

        // Record all nodes from multiproof in the witness.
        self.record_multiproof_nodes(&multiproof);

        let mut sparse_trie = SparseStateTrie::new();
        sparse_trie.reveal_decoded_multiproof_v2(multiproof)?;

        // Build storage leaf updates for all accounts with storage changes, split into
        // removals and upserts. Legacy mode applies removals first to preserve the
        // historical witness shape expected by existing consumers: a removal can collapse
        // a branch and force proof fetches that some consumers still rely on. Canonical
        // mode applies upserts first to avoid those compatibility-only nodes and emit
        // the minimized draft-spec witness.
        let mut storage_removals: B256Map<B256Map<LeafUpdate>> = B256Map::default();
        let mut storage_upserts: B256Map<B256Map<LeafUpdate>> = B256Map::default();
        for (hashed_address, storage) in &state.storages {
            for (&hashed_slot, value) in &storage.storage {
                if value.is_zero() {
                    storage_removals
                        .entry(*hashed_address)
                        .or_default()
                        .insert(hashed_slot, LeafUpdate::Changed(vec![]));
                } else {
                    storage_upserts.entry(*hashed_address).or_default().insert(
                        hashed_slot,
                        LeafUpdate::Changed(alloy_rlp::encode_fixed_size(value).to_vec()),
                    );
                }
            }
        }

        let storage_update_sets = if self.mode.is_canonical() {
            [&mut storage_upserts, &mut storage_removals]
        } else {
            [&mut storage_removals, &mut storage_upserts]
        };

        // Apply storage updates in mode-specific order, fetching additional proofs as needed.
        for storage_updates in storage_update_sets {
            loop {
                let mut targets = MultiProofTargetsV2::default();

                for (&hashed_address, slot_updates) in storage_updates.iter_mut() {
                    if slot_updates.is_empty() {
                        continue;
                    }
                    let storage_trie = sparse_trie
                        .storage_trie_mut(&hashed_address)
                        .expect("storage trie was revealed from multiproof");
                    storage_trie
                        .update_leaves(slot_updates, |key, parent| {
                            targets
                                .storage_targets
                                .entry(hashed_address)
                                .or_default()
                                .push(ProofV2Target::new(key).with_parent(parent));
                        })
                        .map_err(|err| {
                            SparseStateTrieErrorKind::SparseStorageTrie(
                                hashed_address,
                                err.into_kind(),
                            )
                        })?;
                }

                if targets.is_empty() {
                    break;
                }

                let multiproof = Proof::new(
                    self.trie_cursor_factory.clone(),
                    self.hashed_cursor_factory.clone(),
                )
                .with_prefix_sets_mut(self.prefix_sets.clone())
                .multiproof_v2(targets)?;
                self.record_multiproof_nodes(&multiproof);
                sparse_trie.reveal_decoded_multiproof_v2(multiproof)?;
            }
        }

        // Build account leaf updates, split into removals and upserts. Legacy mode keeps
        // removals-first for the same compatibility reason as storage updates, while
        // canonical mode uses upserts-first so account updates follow the minimized
        // draft-spec witness order.
        let mut account_removals: B256Map<LeafUpdate> = B256Map::default();
        let mut account_upserts: B256Map<LeafUpdate> = B256Map::default();
        for &hashed_address in state.accounts.keys().chain(state.storages.keys()) {
            if account_removals.contains_key(&hashed_address) ||
                account_upserts.contains_key(&hashed_address)
            {
                continue;
            }

            let account = state
                .accounts
                .get(&hashed_address)
                .ok_or(TrieWitnessError::MissingAccount(hashed_address))?
                .unwrap_or_default();

            let storage_root =
                if let Some(storage_trie) = sparse_trie.storage_trie_mut(&hashed_address) {
                    storage_trie.root(TrieNodeEpoch::UNMODIFIED)
                } else {
                    let record_root_node = !self.mode.is_canonical() ||
                        state
                            .storages
                            .get(&hashed_address)
                            .is_some_and(|storage| !storage.storage.is_empty());
                    self.account_storage_root(hashed_address, record_root_node)?
                };

            if account.is_empty() && storage_root == EMPTY_ROOT_HASH {
                account_removals.insert(hashed_address, LeafUpdate::Changed(vec![]));
            } else {
                let mut rlp = Vec::with_capacity(TRIE_ACCOUNT_RLP_MAX_SIZE);
                account.into_trie_account(storage_root).encode(&mut rlp);
                account_upserts.insert(hashed_address, LeafUpdate::Changed(rlp));
            }
        }

        let account_update_sets = if self.mode.is_canonical() {
            [&mut account_upserts, &mut account_removals]
        } else {
            [&mut account_removals, &mut account_upserts]
        };

        // Apply account updates in mode-specific order, fetching additional proofs as needed.
        for account_updates in account_update_sets {
            loop {
                let mut targets = MultiProofTargetsV2::default();

                sparse_trie
                    .trie_mut()
                    .update_leaves(account_updates, |key, parent| {
                        targets.account_targets.push(ProofV2Target::new(key).with_parent(parent));
                    })
                    .map_err(SparseStateTrieErrorKind::from)?;

                if targets.is_empty() {
                    break;
                }

                let multiproof = Proof::new(
                    self.trie_cursor_factory.clone(),
                    self.hashed_cursor_factory.clone(),
                )
                .with_prefix_sets_mut(self.prefix_sets.clone())
                .multiproof_v2(targets)?;
                self.record_multiproof_nodes(&multiproof);
                sparse_trie.reveal_decoded_multiproof_v2(multiproof)?;
            }
        }

        if self.mode.is_canonical() {
            // Empty trie nodes carry no useful witness information and are trivially
            // reconstructible from the empty root hash.
            self.witness.retain(|_, value| value.as_ref() != [EMPTY_STRING_CODE]);
        }

        Ok(self.witness)
    }

    /// Record all nodes from a V2 decoded multiproof in the witness.
    fn record_multiproof_nodes(&mut self, multiproof: &DecodedMultiProofV2) {
        let mut encoded = Vec::new();
        for proof_node in &multiproof.account_proofs {
            self.record_witness_node(&proof_node.node, &mut encoded);
        }
        for proof_nodes in multiproof.storage_proofs.values() {
            for proof_node in proof_nodes {
                self.record_witness_node(&proof_node.node, &mut encoded);
            }
        }
    }

    /// Record a single [`TrieNodeV2`] in the witness.
    fn record_witness_node(&mut self, node: &TrieNodeV2, encoded: &mut Vec<u8>) {
        encoded.clear();
        node.encode(encoded);
        let hash = keccak256(encoded.as_slice());
        self.witness.entry(hash).or_insert_with(|| Bytes::copy_from_slice(encoded));

        if let TrieNodeV2::Branch(branch) = node &&
            !branch.key.is_empty()
        {
            encoded.clear();
            BranchNodeRef::new(&branch.stack, branch.state_mask).encode(encoded);
            let hash = keccak256(encoded.as_slice());
            self.witness.entry(hash).or_insert_with(|| Bytes::copy_from_slice(encoded));
        }
    }

    /// Compute the storage root for an account by walking the storage trie using the cursor
    /// factories and trie input prefix sets. Records the root node in the witness when requested.
    fn account_storage_root(
        &mut self,
        hashed_address: B256,
        record_root_node: bool,
    ) -> Result<B256, TrieWitnessError> {
        let storage_trie_cursor = self
            .trie_cursor_factory
            .storage_trie_cursor(hashed_address)
            .map_err(StateProofError::from)?;
        let hashed_storage_cursor = self
            .hashed_cursor_factory
            .hashed_storage_cursor(hashed_address)
            .map_err(StateProofError::from)?;
        let mut calculator = proof_v2::StorageProofCalculator::new_storage(
            storage_trie_cursor,
            hashed_storage_cursor,
        );
        if let Some(prefix_set) = self.prefix_sets.storage_prefix_sets.get(&hashed_address) {
            calculator = calculator.with_prefix_set(prefix_set.clone().freeze());
        }
        let root_node = calculator.storage_root_node(hashed_address)?;
        let root_hash = calculator
            .compute_root_hash(core::slice::from_ref(&root_node))?
            .unwrap_or(EMPTY_ROOT_HASH);
        drop(calculator);
        if record_root_node {
            let mut encoded = Vec::new();
            self.record_witness_node(&root_node.node, &mut encoded);
        }
        Ok(root_hash)
    }

    /// Retrieve proof targets for incoming hashed state.
    /// Aggregates all accounts and slots present in the state.
    fn get_proof_targets(state: &HashedPostState) -> MultiProofTargetsV2 {
        let mut targets = MultiProofTargetsV2::default();
        for &hashed_address in state.accounts.keys() {
            targets.account_targets.push(ProofV2Target::new(hashed_address));
        }
        for (&hashed_address, storage) in &state.storages {
            if !state.accounts.contains_key(&hashed_address) {
                targets.account_targets.push(ProofV2Target::new(hashed_address));
            }
            // Skip accounts with no storage slot changes — an empty target set would produce
            // an empty proof vec which cannot be revealed (no root node).
            if storage.storage.is_empty() {
                continue;
            }
            let storage_keys = storage.storage.keys().map(|k| ProofV2Target::new(*k)).collect();
            targets.storage_targets.insert(hashed_address, storage_keys);
        }
        targets
    }
}