fix(testing): Optimize fill (#1804)

* fix(t8n): introduce profiler

* profiler pauses

* fix(testing/base_types): Speed up to_bytes

fix(base_types): unnecessary cast

* fix(testing/base_types): CoerceBytes

* fix(testing): Optimize pre-alloc grouping state hash

* feat(t8n): introduce LazyAlloc

* feat(fixtures): Cached genesis model dump

* refactor: prefer explicit cache over circular reference (#2)

* Update packages/testing/src/execution_testing/client_clis/cli_types.py

Co-authored-by: spencer <[email protected]>

* fix: verify modified gas limit

* refactor: Use generics

* refactor: Add state root cache

* docs: changelog

* feat(client_clis): Add unit tests

---------

Co-authored-by: danceratopz <[email protected]>
Co-authored-by: spencer <[email protected]>

Author: 3 people

docs/CHANGELOG.md

@@ -18,6 +18,7 @@ Test fixtures for use by clients are available for each release on the [Github r
 
 - 🐞 Allow `evmone` to fill Prague and Osaka blockchain tests (mainly modified deposit contract tests) ([#1689](https://github.com/ethereum/execution-specs/pull/1689)).
 - 🐞 Turn off Block-Level Access List related checks when filling tests for Amsterdam ([#1737](https://github.com/ethereum/execution-specs/pull/1737)).
+- ✨ Optimize the filling process by lazily loading the t8n response only when it’s actually needed. Otherwise, pass it verbatim into the next t8n execution, skipping both pydantic validation and model serialization entirely ([#1804](https://github.com/ethereum/execution-specs/pull/1804)).
 
 #### `consume`
 

packages/testing/src/execution_testing/__init__.py

@@ -5,6 +5,7 @@
     Account,
     Address,
     Bytes,
+    CoerceBytes,
     Hash,
     Storage,
     TestAddress,
@@ -199,6 +200,7 @@
     "TransactionType",
     "UndefinedOpcodes",
     "While",
+    "CoerceBytes",
     "Withdrawal",
     "WithdrawalRequest",
     "add_kzg_version",

packages/testing/src/execution_testing/base_types/__init__.py

@@ -6,6 +6,7 @@
     BLSPublicKey,
     BLSSignature,
     Bytes,
+    CoerceBytes,
     FixedSizeBytes,
     ForkHash,
     Hash,
@@ -56,6 +57,7 @@
     "BLSSignature",
     "Bytes",
     "CamelModel",
+    "CoerceBytes",
     "EmptyOmmersRoot",
     "EmptyTrieRoot",
     "EthereumTestBaseModel",

packages/testing/src/execution_testing/base_types/base_types.py

@@ -1,6 +1,7 @@
 """Basic type primitives used to define other types."""
 
 from hashlib import sha256
+from re import sub
 from typing import Annotated, Any, ClassVar, SupportsBytes, Type, TypeVar
 
 from Crypto.Hash import keccak
@@ -57,15 +58,6 @@ def hex(self) -> str:
         """Return the hexadecimal representation of the number."""
         return hex(self)
 
-    @classmethod
-    def or_none(
-        cls: Type[Self], input_number: Self | NumberConvertible | None
-    ) -> Self | None:
-        """Convert the input to a Number while accepting None."""
-        if input_number is None:
-            return input_number
-        return cls(input_number)
-
 
 class Wei(Number):
     """Class that helps represent wei that can be parsed from strings."""
@@ -198,15 +190,6 @@ def hex(self, *args: Any, **kwargs: Any) -> str:
         """Return the hexadecimal representation of the bytes."""
         return "0x" + super().hex(*args, **kwargs)
 
-    @classmethod
-    def or_none(
-        cls, input_bytes: "Bytes | BytesConvertible | None"
-    ) -> "Bytes | None":
-        """Convert the input to a Bytes while accepting None."""
-        if input_bytes is None:
-            return input_bytes
-        return cls(input_bytes)
-
     def keccak256(self) -> "Hash":
         """Return the keccak256 hash of the opcode byte representation."""
         k = keccak.new(digest_bits=256)
@@ -235,6 +218,21 @@ def __get_pydantic_core_schema__(
         )
 
 
+class CoerceBytes(Bytes):
+    """
+    Class that helps represent bytes of variable length in tests and
+    supports removing white spaces anywhere in the string.
+    """
+
+    def __new__(cls, input_bytes: BytesConvertible = b"") -> Self:
+        """Create a new CoerceBytes object."""
+        if type(input_bytes) is cls:
+            return input_bytes
+        if isinstance(input_bytes, str):
+            input_bytes = sub(r"\s+", "", input_bytes)
+        return super(Bytes, cls).__new__(cls, to_bytes(input_bytes))
+
+
 class FixedSizeHexNumber(int, ToStringSchema):
     """
     A base class that helps represent an integer as a fixed byte-length
@@ -347,15 +345,6 @@ def __hash__(self) -> int:
         """Return the hash of the bytes."""
         return super(FixedSizeBytes, self).__hash__()
 
-    @classmethod
-    def or_none(
-        cls: Type[Self], input_bytes: Self | FixedSizeBytesConvertible | None
-    ) -> Self | None:
-        """Convert the input to a Fixed Size Bytes while accepting None."""
-        if input_bytes is None:
-            return input_bytes
-        return cls(input_bytes)
-
     def __eq__(self, other: object) -> bool:
         """Compare two FixedSizeBytes objects to be equal."""
         if other is None:

packages/testing/src/execution_testing/base_types/conversions.py

@@ -1,7 +1,6 @@
 """Common conversion methods."""
 
-from re import sub
-from typing import Any, List, Optional, SupportsBytes, TypeAlias
+from typing import List, SupportsBytes, TypeAlias
 
 BytesConvertible: TypeAlias = str | bytes | SupportsBytes | List[int]
 FixedSizeBytesConvertible: TypeAlias = (
@@ -10,46 +9,20 @@
 NumberConvertible: TypeAlias = str | bytes | SupportsBytes | int
 
 
-def int_or_none(input_value: Any, default: Optional[int] = None) -> int | None:
-    """Convert a value to int or returns a default (None)."""
-    if input_value is None:
-        return default
-    if isinstance(input_value, int):
-        return input_value
-    return int(input_value, 0)
-
-
-def str_or_none(input_value: Any, default: Optional[str] = None) -> str | None:
-    """Convert a value to string or returns a default (None)."""
-    if input_value is None:
-        return default
-    if isinstance(input_value, str):
-        return input_value
-    return str(input_value)
-
-
 def to_bytes(input_bytes: BytesConvertible) -> bytes:
     """Convert multiple types into bytes."""
     if input_bytes is None:
         raise Exception("Cannot convert `None` input to bytes")
 
-    if (
-        isinstance(input_bytes, SupportsBytes)
-        or isinstance(input_bytes, bytes)
-        or isinstance(input_bytes, list)
-    ):
-        return bytes(input_bytes)
-
     if isinstance(input_bytes, str):
         # We can have a hex representation of bytes with spaces for readability
-        input_bytes = sub(r"\s+", "", input_bytes)
         if input_bytes.startswith("0x"):
             input_bytes = input_bytes[2:]
         if len(input_bytes) % 2 == 1:
             input_bytes = "0" + input_bytes
         return bytes.fromhex(input_bytes)
 
-    raise Exception("invalid type for `bytes`")
+    return bytes(input_bytes)
 
 
 def to_fixed_size_bytes(
@@ -85,9 +58,9 @@ def to_fixed_size_bytes(
         )
     if len(input_bytes) < size:
         if left_padding:
-            return bytes(input_bytes).rjust(size, b"\x00")
+            return input_bytes.rjust(size, b"\x00")
         if right_padding:
-            return bytes(input_bytes).ljust(size, b"\x00")
+            return input_bytes.ljust(size, b"\x00")
         raise Exception(
             f"input is too small for fixed size bytes: {len(input_bytes)} < {size}\n"
             "Use `left_padding=True` or `right_padding=True` to allow padding."

packages/testing/src/execution_testing/cli/pytest_commands/plugins/filler/filler.py

@@ -40,6 +40,8 @@
     FixtureFillingPhase,
     LabeledFixtureFormat,
     PreAllocGroup,
+    PreAllocGroupBuilder,
+    PreAllocGroupBuilders,
     PreAllocGroups,
     TestInfo,
 )
@@ -231,7 +233,8 @@ class FillingSession:
     fixture_output: FixtureOutput
     phase_manager: PhaseManager
     format_selector: FormatSelector
-    pre_alloc_groups: PreAllocGroups | None
+    pre_alloc_groups: PreAllocGroups | None = None
+    pre_alloc_group_builders: PreAllocGroupBuilders | None = None
 
     @classmethod
     def from_config(cls, config: pytest.Config) -> "Self":
@@ -263,7 +266,7 @@ def _initialize_pre_alloc_groups(self) -> None:
         """Initialize pre-allocation groups based on the current phase."""
         if self.phase_manager.is_pre_alloc_generation:
             # Phase 1: Create empty container for collecting groups
-            self.pre_alloc_groups = PreAllocGroups(root={})
+            self.pre_alloc_group_builders = PreAllocGroupBuilders(root={})
         elif self.phase_manager.is_fill_after_pre_alloc:
             # Phase 2: Load pre-alloc groups from disk
             self._load_pre_alloc_groups_from_folder()
@@ -326,15 +329,15 @@ def get_pre_alloc_group(self, hash_key: str) -> PreAllocGroup:
 
         return self.pre_alloc_groups[hash_key]
 
-    def update_pre_alloc_group(
-        self, hash_key: str, group: PreAllocGroup
+    def update_pre_alloc_group_builder(
+        self, hash_key: str, group_builder: PreAllocGroupBuilder
     ) -> None:
         """
         Update or add a pre-allocation group.
 
         Args:
             hash_key: The hash of the pre-alloc group.
-            group: The pre-allocation group.
+            group_builder: The pre-allocation group builder.
 
         Raises:
             ValueError: If not in pre-alloc generation phase.
@@ -345,44 +348,19 @@ def update_pre_alloc_group(
                 "Can only update pre-alloc groups in generation phase"
             )
 
-        if self.pre_alloc_groups is None:
-            self.pre_alloc_groups = PreAllocGroups(root={})
+        if self.pre_alloc_group_builders is None:
+            self.pre_alloc_group_builders = PreAllocGroupBuilders(root={})
 
-        self.pre_alloc_groups[hash_key] = group
+        self.pre_alloc_group_builders.root[hash_key] = group_builder
 
     def save_pre_alloc_groups(self) -> None:
         """Save pre-allocation groups to disk."""
-        if self.pre_alloc_groups is None:
+        if self.pre_alloc_group_builders is None:
             return
 
         pre_alloc_folder = self.fixture_output.pre_alloc_groups_folder_path
         pre_alloc_folder.mkdir(parents=True, exist_ok=True)
-        self.pre_alloc_groups.to_folder(pre_alloc_folder)
-
-    def aggregate_pre_alloc_groups(
-        self, worker_groups: PreAllocGroups
-    ) -> None:
-        """
-        Aggregate pre-alloc groups from a worker process (xdist support).
-
-        Args:
-            worker_groups: Pre-alloc groups from a worker process.
-
-        """
-        if self.pre_alloc_groups is None:
-            self.pre_alloc_groups = PreAllocGroups(root={})
-
-        for hash_key, group in worker_groups.items():
-            if hash_key in self.pre_alloc_groups:
-                # Merge if exists (should not happen in practice)
-                existing = self.pre_alloc_groups[hash_key]
-                if existing.pre != group.pre:
-                    raise ValueError(
-                        f"Conflicting pre-alloc groups for hash {hash_key}: "
-                        f"existing={self.pre_alloc_groups[hash_key].pre}, new={group.pre}"
-                    )
-            else:
-                self.pre_alloc_groups[hash_key] = group
+        self.pre_alloc_group_builders.to_folder(pre_alloc_folder)
 
 
 def calculate_post_state_diff(
@@ -1382,7 +1360,7 @@ def __init__(self, *args: Any, **kwargs: Any) -> None:
                     # Use the original update_pre_alloc_groups method which
                     # returns the groups
                     self.update_pre_alloc_groups(
-                        session.pre_alloc_groups, request.node.nodeid
+                        session.pre_alloc_group_builders, request.node.nodeid
                     )
                     return  # Skip fixture generation in phase 1