feat(indexes): make sync-v1 indexes optional

Author: jansegre

hathor/indexes/height_index.py

@@ -13,7 +13,7 @@
 # limitations under the License.
 
 from abc import abstractmethod
-from typing import NamedTuple, Optional
+from typing import TYPE_CHECKING, NamedTuple, Optional
 
 from hathor.indexes.base_index import BaseIndex
 from hathor.indexes.scope import Scope
@@ -22,6 +22,9 @@
 from hathor.types import VertexId
 from hathor.util import not_none
 
+if TYPE_CHECKING:  # pragma: no cover
+    from hathor.transaction.storage import TransactionStorage
+
 SCOPE = Scope(
     include_blocks=True,
     include_txs=False,
@@ -84,6 +87,18 @@ def get(self, height: int) -> Optional[bytes]:
         """
         raise NotImplementedError
 
+    def find_by_timestamp(self, timestamp: float, tx_storage: 'TransactionStorage') -> Optional[Block]:
+        """ This method starts from the tip and advances to the parent until it finds a block with lower timestamp.
+        """
+        # TODO: optimize
+        if timestamp < BLOCK_GENESIS.timestamp:
+            return None
+        block = tx_storage.get_transaction(self.get_tip())
+        assert isinstance(block, Block)
+        while block.timestamp > timestamp:
+            block = block.get_block_parent()
+        return block
+
     @abstractmethod
     def get_tip(self) -> bytes:
         """ Return the best block hash, it returns the genesis when there is no other block

hathor/indexes/manager.py

@@ -52,14 +52,12 @@ class IndexesManager(ABC):
     log = get_logger()
 
     info: InfoIndex
-    all_tips: TipsIndex
-    block_tips: TipsIndex
-    tx_tips: TipsIndex
-
+    all_tips: Optional[TipsIndex]
+    block_tips: Optional[TipsIndex]
+    tx_tips: Optional[TipsIndex]
     sorted_all: TimestampIndex
     sorted_blocks: TimestampIndex
     sorted_txs: TimestampIndex
-
     height: HeightIndex
     deps: Optional[DepsIndex]
     mempool_tips: Optional[MempoolTipsIndex]
@@ -97,6 +95,11 @@ def iter_all_indexes(self) -> Iterator[BaseIndex]:
             self.utxo,
         ])
 
+    @abstractmethod
+    def enable_tips_indexes(self) -> None:
+        """Enable tips indexs. It does nothing if it has already been enabled."""
+        raise NotImplementedError
+
     @abstractmethod
     def enable_address_index(self, pubsub: 'PubSubManager') -> None:
         """Enable address index. It does nothing if it has already been enabled."""
@@ -208,18 +211,21 @@ def add_tx(self, tx: BaseTransaction) -> bool:
 
         # These two calls return False when a transaction changes from
         # voided to executed and vice-versa.
-        r1 = self.all_tips.add_tx(tx)
-        r2 = self.sorted_all.add_tx(tx)
-        assert r1 == r2
+        r1 = self.sorted_all.add_tx(tx)
+        if self.all_tips is not None:
+            r2 = self.all_tips.add_tx(tx)
+            assert r1 == r2
 
         if tx.is_block:
-            r3 = self.block_tips.add_tx(tx)
-            r4 = self.sorted_blocks.add_tx(tx)
-            assert r3 == r4
+            r3 = self.sorted_blocks.add_tx(tx)
+            if self.block_tips is not None:
+                r4 = self.block_tips.add_tx(tx)
+                assert r3 == r4
         else:
-            r3 = self.tx_tips.add_tx(tx)
-            r4 = self.sorted_txs.add_tx(tx)
-            assert r3 == r4
+            r3 = self.sorted_txs.add_tx(tx)
+            if self.tx_tips is not None:
+                r4 = self.tx_tips.add_tx(tx)
+                assert r3 == r4
 
         if self.addresses:
             self.addresses.add_tx(tx)
@@ -249,7 +255,8 @@ def del_tx(self, tx: BaseTransaction, *, remove_all: bool = False, relax_assert:
             # We delete from indexes in two cases: (i) mark tx as voided, and (ii) remove tx.
             # We only remove tx from all_tips and sorted_all when it is removed from the storage.
             # For clarity, when a tx is marked as voided, it is not removed from all_tips and sorted_all.
-            self.all_tips.del_tx(tx, relax_assert=relax_assert)
+            if self.all_tips is not None:
+                self.all_tips.del_tx(tx, relax_assert=relax_assert)
             self.sorted_all.del_tx(tx)
             if self.addresses:
                 self.addresses.remove_tx(tx)
@@ -263,11 +270,13 @@ def del_tx(self, tx: BaseTransaction, *, remove_all: bool = False, relax_assert:
             self.mempool_tips.update(tx, remove=True)
 
         if tx.is_block:
-            self.block_tips.del_tx(tx, relax_assert=relax_assert)
             self.sorted_blocks.del_tx(tx)
+            if self.block_tips is not None:
+                self.block_tips.del_tx(tx, relax_assert=relax_assert)
         else:
-            self.tx_tips.del_tx(tx, relax_assert=relax_assert)
             self.sorted_txs.del_tx(tx)
+            if self.tx_tips is not None:
+                self.tx_tips.del_tx(tx, relax_assert=relax_assert)
 
         if self.tokens:
             self.tokens.del_tx(tx)
@@ -282,12 +291,11 @@ def __init__(self) -> None:
         from hathor.indexes.memory_height_index import MemoryHeightIndex
         from hathor.indexes.memory_info_index import MemoryInfoIndex
         from hathor.indexes.memory_timestamp_index import MemoryTimestampIndex
-        from hathor.indexes.memory_tips_index import MemoryTipsIndex
 
         self.info = MemoryInfoIndex()
-        self.all_tips = MemoryTipsIndex(scope_type=TipsScopeType.ALL)
-        self.block_tips = MemoryTipsIndex(scope_type=TipsScopeType.BLOCKS)
-        self.tx_tips = MemoryTipsIndex(scope_type=TipsScopeType.TXS)
+        self.all_tips = None
+        self.block_tips = None
+        self.tx_tips = None
 
         self.sorted_all = MemoryTimestampIndex(scope_type=TimestampScopeType.ALL)
         self.sorted_blocks = MemoryTimestampIndex(scope_type=TimestampScopeType.BLOCKS)
@@ -303,6 +311,15 @@ def __init__(self) -> None:
         # XXX: this has to be at the end of __init__, after everything has been initialized
         self.__init_checks__()
 
+    def enable_tips_indexes(self) -> None:
+        from hathor.indexes.memory_tips_index import MemoryTipsIndex
+        if self.all_tips is None:
+            self.all_tips = MemoryTipsIndex(scope_type=TipsScopeType.ALL)
+        if self.block_tips is None:
+            self.block_tips = MemoryTipsIndex(scope_type=TipsScopeType.BLOCKS)
+        if self.tx_tips is None:
+            self.tx_tips = MemoryTipsIndex(scope_type=TipsScopeType.TXS)
+
     def enable_address_index(self, pubsub: 'PubSubManager') -> None:
         from hathor.indexes.memory_address_index import MemoryAddressIndex
         if self.addresses is None:
@@ -331,7 +348,6 @@ def enable_deps_index(self) -> None:
 
 class RocksDBIndexesManager(IndexesManager):
     def __init__(self, rocksdb_storage: 'RocksDBStorage') -> None:
-        from hathor.indexes.partial_rocksdb_tips_index import PartialRocksDBTipsIndex
         from hathor.indexes.rocksdb_height_index import RocksDBHeightIndex
         from hathor.indexes.rocksdb_info_index import RocksDBInfoIndex
         from hathor.indexes.rocksdb_timestamp_index import RocksDBTimestampIndex
@@ -340,9 +356,9 @@ def __init__(self, rocksdb_storage: 'RocksDBStorage') -> None:
 
         self.info = RocksDBInfoIndex(self._db)
         self.height = RocksDBHeightIndex(self._db)
-        self.all_tips = PartialRocksDBTipsIndex(self._db, scope_type=TipsScopeType.ALL)
-        self.block_tips = PartialRocksDBTipsIndex(self._db, scope_type=TipsScopeType.BLOCKS)
-        self.tx_tips = PartialRocksDBTipsIndex(self._db, scope_type=TipsScopeType.TXS)
+        self.all_tips = None
+        self.block_tips = None
+        self.tx_tips = None
 
         self.sorted_all = RocksDBTimestampIndex(self._db, scope_type=TimestampScopeType.ALL)
         self.sorted_blocks = RocksDBTimestampIndex(self._db, scope_type=TimestampScopeType.BLOCKS)
@@ -357,6 +373,15 @@ def __init__(self, rocksdb_storage: 'RocksDBStorage') -> None:
         # XXX: this has to be at the end of __init__, after everything has been initialized
         self.__init_checks__()
 
+    def enable_tips_indexes(self) -> None:
+        from hathor.indexes.partial_rocksdb_tips_index import PartialRocksDBTipsIndex
+        if self.all_tips is None:
+            self.all_tips = PartialRocksDBTipsIndex(self._db, scope_type=TipsScopeType.ALL)
+        if self.block_tips is None:
+            self.block_tips = PartialRocksDBTipsIndex(self._db, scope_type=TipsScopeType.BLOCKS)
+        if self.tx_tips is None:
+            self.tx_tips = PartialRocksDBTipsIndex(self._db, scope_type=TipsScopeType.TXS)
+
     def enable_address_index(self, pubsub: 'PubSubManager') -> None:
         from hathor.indexes.rocksdb_address_index import RocksDBAddressIndex
         if self.addresses is None:

hathor/manager.py

@@ -697,6 +697,42 @@ def generate_parent_txs(self, timestamp: Optional[float]) -> 'ParentTxs':
         This method tries to return a stable result, such that for a given timestamp and storage state it will always
         return the same.
         """
+        # return self._generate_parent_txs_from_tips_index(timestamp)
+        # XXX: prefer txs_tips index since it's been tested more
+        assert self.tx_storage.indexes is not None
+        if self.tx_storage.indexes.tx_tips is not None:
+            return self._generate_parent_txs_from_tips_index(timestamp)
+        else:
+            return self._generate_parent_txs_from_mempool_index(timestamp)
+
+    def _generate_parent_txs_from_mempool_index(self, timestamp: Optional[float]) -> 'ParentTxs':
+        # XXX: this implementation is naive, it will return a working result but not necessarily actual tips,
+        #      particularly when the timestamp is in the past it will just return tx parents of a previous block that
+        #      is within the timestamp, this is because we don't need to support that case for normal usage
+        if timestamp is None:
+            timestamp = self.reactor.seconds()
+        assert self.tx_storage.indexes is not None
+        assert self.tx_storage.indexes.height is not None
+        assert self.tx_storage.indexes.mempool_tips is not None
+        tips = [tx for tx in self.tx_storage.indexes.mempool_tips.iter(self.tx_storage) if tx.timestamp < timestamp]
+        max_timestamp = max(tx.timestamp for tx in tips) if tips else 0
+        can_include: list[bytes] = [not_none(tx.hash) for tx in tips]
+        must_include = []
+        if len(can_include) < 2:
+            best_block = self.tx_storage.indexes.height.find_by_timestamp(timestamp, self.tx_storage)
+            assert best_block is not None
+            all_best_block_parent_txs = list(map(self.tx_storage.get_transaction, best_block.parents[1:]))
+            best_block_parent_txs = [tx for tx in all_best_block_parent_txs if tx.timestamp < timestamp]
+            max_timestamp = max(max_timestamp, *list(tx.timestamp for tx in best_block_parent_txs))
+            if len(can_include) < 1:
+                can_include.extend(not_none(tx.hash) for tx in best_block_parent_txs)
+            else:
+                must_include = can_include
+                can_include = [not_none(tx.hash) for tx in best_block_parent_txs]
+        assert len(can_include) + len(must_include) >= 2
+        return ParentTxs(max_timestamp, can_include, must_include)
+
+    def _generate_parent_txs_from_tips_index(self, timestamp: Optional[float]) -> 'ParentTxs':
         if timestamp is None:
             timestamp = self.reactor.seconds()
         can_include_intervals = sorted(self.tx_storage.get_tx_tips(timestamp - 1))

hathor/p2p/manager.py

@@ -191,9 +191,12 @@ def __init__(self,
     def set_manager(self, manager: 'HathorManager') -> None:
         """Set the manager. This method must be called before start()."""
         self.manager = manager
+        assert self.manager.tx_storage.indexes is not None
+        indexes = self.manager.tx_storage.indexes
+        if self.enable_sync_v1 or self.enable_sync_v1_1:
+            self.log.debug('enable sync-v1 indexes')
+            indexes.enable_tips_indexes()
         if self.enable_sync_v2:
-            assert self.manager.tx_storage.indexes is not None
-            indexes = self.manager.tx_storage.indexes
             self.log.debug('enable sync-v2 indexes')
             indexes.enable_deps_index()
             indexes.enable_mempool_index()

hathor/transaction/storage/transaction_storage.py

@@ -625,15 +625,23 @@ def first_timestamp(self) -> int:
         raise NotImplementedError
 
     @abstractmethod
-    def get_best_block_tips(self, timestamp: Optional[float] = None, *, skip_cache: bool = False) -> list[bytes]:
+    def get_best_block_tips(self, *, skip_cache: bool = False) -> list[bytes]:
         """ Return a list of blocks that are heads in a best chain. It must be used when mining.
 
         When more than one block is returned, it means that there are multiple best chains and
         you can choose any of them.
         """
-        if timestamp is None and not skip_cache and self._best_block_tips_cache is not None:
-            return self._best_block_tips_cache[:]
+        # ignoring cache because current implementation is ~O(1)
+        assert self.indexes is not None
+        return [self.indexes.height.get_tip()]
+
+    @abstractmethod
+    def get_past_best_block_tips(self, timestamp: Optional[float] = None) -> list[bytes]:
+        """ Return a list of blocks that are heads in a best chain. It must be used when mining.
 
+        When more than one block is returned, it means that there are multiple best chains and
+        you can choose any of them.
+        """
         best_score = 0.0
         best_tip_blocks: list[bytes] = []
 
@@ -1019,6 +1027,7 @@ def iter_mempool_tips_from_tx_tips(self) -> Iterator[Transaction]:
         This method requires indexes to be enabled.
         """
         assert self.indexes is not None
+        assert self.indexes.tx_tips is not None
         tx_tips = self.indexes.tx_tips
 
         for interval in tx_tips[self.latest_timestamp + 1]:
@@ -1121,8 +1130,11 @@ def remove_cache(self) -> None:
         """Remove all caches in case we don't need it."""
         self.indexes = None
 
-    def get_best_block_tips(self, timestamp: Optional[float] = None, *, skip_cache: bool = False) -> list[bytes]:
-        return super().get_best_block_tips(timestamp, skip_cache=skip_cache)
+    def get_best_block_tips(self, *, skip_cache: bool = False) -> list[bytes]:
+        return super().get_best_block_tips(skip_cache=skip_cache)
+
+    def get_past_best_block_tips(self, timestamp: Optional[float] = None) -> list[bytes]:
+        return super().get_past_best_block_tips(timestamp)
 
     def get_n_height_tips(self, n_blocks: int) -> list[HeightInfo]:
         block = self.get_best_block()
@@ -1141,6 +1153,7 @@ def get_block_tips(self, timestamp: Optional[float] = None) -> set[Interval]:
         if self.indexes is None:
             raise NotImplementedError
         assert self.indexes is not None
+        assert self.indexes.block_tips is not None
         if timestamp is None:
             timestamp = self.latest_timestamp
         return self.indexes.block_tips[timestamp]
@@ -1149,6 +1162,7 @@ def get_tx_tips(self, timestamp: Optional[float] = None) -> set[Interval]:
         if self.indexes is None:
             raise NotImplementedError
         assert self.indexes is not None
+        assert self.indexes.tx_tips is not None
         if timestamp is None:
             timestamp = self.latest_timestamp
         tips = self.indexes.tx_tips[timestamp]
@@ -1166,6 +1180,7 @@ def get_all_tips(self, timestamp: Optional[float] = None) -> set[Interval]:
         if self.indexes is None:
             raise NotImplementedError
         assert self.indexes is not None
+        assert self.indexes.all_tips is not None
         if timestamp is None:
             timestamp = self.latest_timestamp
 

tests/p2p/test_double_spending.py

@@ -82,8 +82,9 @@ def test_simple_double_spending(self):
             self.assertEqual([tx1.hash, tx2.hash], spent_meta.spent_outputs[txin.index])
 
         # old indexes
-        self.assertNotIn(tx1.hash, [x.data for x in self.manager1.tx_storage.get_tx_tips()])
-        self.assertNotIn(tx2.hash, [x.data for x in self.manager1.tx_storage.get_tx_tips()])
+        if self.manager1.tx_storage.indexes.tx_tips is not None:
+            self.assertNotIn(tx1.hash, [x.data for x in self.manager1.tx_storage.get_tx_tips()])
+            self.assertNotIn(tx2.hash, [x.data for x in self.manager1.tx_storage.get_tx_tips()])
 
         # new indexes
         if self.manager1.tx_storage.indexes.mempool_tips is not None:
@@ -112,9 +113,10 @@ def test_simple_double_spending(self):
             self.assertEqual([tx1.hash, tx2.hash, tx3.hash], spent_meta.spent_outputs[txin.index])
 
         # old indexes
-        self.assertNotIn(tx1.hash, [x.data for x in self.manager1.tx_storage.get_tx_tips()])
-        self.assertNotIn(tx2.hash, [x.data for x in self.manager1.tx_storage.get_tx_tips()])
-        self.assertIn(tx3.hash, [x.data for x in self.manager1.tx_storage.get_tx_tips()])
+        if self.manager1.tx_storage.indexes.tx_tips is not None:
+            self.assertNotIn(tx1.hash, [x.data for x in self.manager1.tx_storage.get_tx_tips()])
+            self.assertNotIn(tx2.hash, [x.data for x in self.manager1.tx_storage.get_tx_tips()])
+            self.assertIn(tx3.hash, [x.data for x in self.manager1.tx_storage.get_tx_tips()])
 
         # new indexes
         if self.manager1.tx_storage.indexes.mempool_tips is not None: