use backend file when compacting the output pos index (#3226)

chain/src/chain.rs

@@ -1109,9 +1109,6 @@ impl Chain {
 			txhashset.compact(&horizon_header, &mut batch)?;
 		}
 
-		// Rebuild our output_pos index in the db based on current UTXO set.
-		txhashset.rebuild_height_pos_index(&header_pmmr, &mut batch)?;
-
 		// If we are not in archival mode remove historical blocks from the db.
 		if !self.archive_mode {
 			self.remove_historical_blocks(&header_pmmr, &mut batch)?;

chain/src/store.rs

@@ -229,6 +229,11 @@ impl<'a> Batch<'a> {
 		Ok(())
 	}
 
+	/// Low level function to delete directly by raw key.
+	pub fn delete(&self, key: &[u8]) -> Result<(), Error> {
+		self.db.delete(key)
+	}
+
 	/// Delete a full block. Does not delete any record associated with a block
 	/// header.
 	pub fn delete_block(&self, bh: &Hash) -> Result<(), Error> {
@@ -269,6 +274,12 @@ impl<'a> Batch<'a> {
 		)
 	}
 
+	/// Iterator over the output_pos index.
+	pub fn output_pos_iter(&self) -> Result<SerIterator<(u64, u64)>, Error> {
+		let key = to_key(COMMIT_POS_HGT_PREFIX, &mut "".to_string().into_bytes());
+		self.db.iter(&key)
+	}
+
 	/// Get output_pos from index.
 	/// Note:
 	///     - Original prefix 'COMMIT_POS_PREFIX' is not used for normal case anymore, refer to #2889 for detail.

chain/src/txhashset/txhashset.rs

@@ -383,6 +383,9 @@ impl TxHashSet {
 			.backend
 			.check_compact(horizon_header.output_mmr_size, &rewind_rm_pos)?;
 
+		debug!("txhashset: compact height pos index...");
+		self.compact_height_pos_index(batch)?;
+
 		debug!("txhashset: ... compaction finished");
 
 		Ok(())
@@ -390,7 +393,6 @@ impl TxHashSet {
 
 	/// Rebuild the index of block height & MMR positions to the corresponding UTXOs.
 	/// This is a costly operation performed only when we receive a full new chain state.
-	/// Note: only called by compact.
 	pub fn rebuild_height_pos_index(
 		&self,
 		header_pmmr: &PMMRHandle<BlockHeader>,
@@ -446,6 +448,32 @@ impl TxHashSet {
 		);
 		Ok(())
 	}
+
+	fn compact_height_pos_index(&self, batch: &Batch<'_>) -> Result<(), Error> {
+		let now = Instant::now();
+		let output_pmmr =
+			ReadonlyPMMR::at(&self.output_pmmr_h.backend, self.output_pmmr_h.last_pos);
+		let last_pos = output_pmmr.unpruned_size();
+
+		let deleted = batch
+			.output_pos_iter()?
+			.filter(|(_, (pos, _))| {
+				// Note we use get_from_file() here as we want to ensure we have an entry
+				// in the index for *every* output still in the file, not just the "unspent"
+				// outputs. This is because we need to support rewind to handle fork/reorg.
+				// Rewind may "unspend" recently spent, but not yet pruned outputs, and the
+				// index must be consistent in this situation.
+				*pos <= last_pos && output_pmmr.get_from_file(*pos).is_none()
+			})
+			.map(|(key, _)| batch.delete(&key))
+			.count();
+		debug!(
+			"compact_output_pos_index: deleted {} entries from the index, took {}s",
+			deleted,
+			now.elapsed().as_secs(),
+		);
+		Ok(())
+	}
 }
 
 /// Starts a new unit of work to extend (or rewind) the chain with additional

core/src/core/pmmr/readonly_pmmr.rs

@@ -86,6 +86,16 @@ where
 		}
 	}
 
+	/// Get the hash from the underlying MMR file, ignoring the leafset.
+	/// Some entries may have been removed from the leafset but not yet pruned from the file.
+	pub fn get_from_file(&self, pos: u64) -> Option<Hash> {
+		if pos > self.last_pos {
+			None
+		} else {
+			self.backend.get_from_file(pos)
+		}
+	}
+
 	/// Iterator over current (unpruned, unremoved) leaf positions.
 	pub fn leaf_pos_iter(&self) -> impl Iterator<Item = u64> + '_ {
 		self.backend.leaf_pos_iter()