fix(tests): optimize json_infra fixture reads

tests/json_infra/helpers/__init__.py

@@ -1 +1,30 @@
 """Helpers to load tests from JSON files."""
+
+import json
+from functools import lru_cache
+from typing import Dict
+
+
+@lru_cache(maxsize=100)
+def load_json_file(test_file: str) -> Dict:
+    """
+    Load and cache a JSON fixture file.
+
+    Uses LRU (Least Recently Used) cache to avoid re-reading the same
+    JSON files multiple times during test execution. This is especially
+    important when running tests in parallel, as multiple test cases
+    from the same file would otherwise cause redundant file I/O.
+
+    The cache is bounded to 100 files, which provides a good balance
+    between memory usage and performance. When the cache is full, the
+    least recently accessed file will be evicted.
+
+    Args:
+        test_file: Path to the JSON fixture file
+
+    Returns:
+        Parsed JSON data as a dictionary
+
+    """
+    with open(test_file, "r") as fp:
+        return json.load(fp)

tests/json_infra/helpers/load_blockchain_tests.py

@@ -1,7 +1,6 @@
 """Helpers to load and run blockchain tests from JSON files."""
 
 import importlib
-import json
 import os.path
 from glob import glob
 from typing import Any, Dict, Generator
@@ -19,6 +18,7 @@
 from ethereum_spec_tools.evm_tools.loaders.fixture_loader import Load
 
 from .. import FORKS
+from . import load_json_file
 from .exceptional_test_patterns import exceptional_blockchain_test_patterns
 
 
@@ -34,8 +34,7 @@ def run_blockchain_st_test(test_case: Dict, load: Load) -> None:
     test_file = test_case["test_file"]
     test_key = test_case["test_key"]
 
-    with open(test_file, "r") as fp:
-        data = json.load(fp)
+    data = load_json_file(test_file)
 
     json_data = data[test_key]
 
@@ -140,27 +139,26 @@ def load_json_fixture(test_file: str, json_fork: str) -> Generator:
     # Ex: Extract "world.json" from "path/to/file/world.json"
     # Extract the filename without the extension. Ex: Extract "world" from
     # "world.json"
-    with open(test_file, "r") as fp:
-        data = json.load(fp)
+    data = load_json_file(test_file)
 
-        # Search tests by looking at the `network` attribute
-        found_keys = []
-        for key, test in data.items():
-            if "network" not in test:
-                continue
+    # Search tests by looking at the `network` attribute
+    found_keys = []
+    for key, test in data.items():
+        if "network" not in test:
+            continue
 
-            if test["network"] == json_fork:
-                found_keys.append(key)
+        if test["network"] == json_fork:
+            found_keys.append(key)
 
-        if not any(found_keys):
-            raise NoTestsFoundError
+    if not any(found_keys):
+        raise NoTestsFoundError
 
-        for _key in found_keys:
-            yield {
-                "test_file": test_file,
-                "test_key": _key,
-                "json_fork": json_fork,
-            }
+    for _key in found_keys:
+        yield {
+            "test_file": test_file,
+            "test_key": _key,
+            "json_fork": json_fork,
+        }
 
 
 def fetch_blockchain_tests(
@@ -202,18 +200,24 @@ def fetch_blockchain_tests(
                     + ")"
                 )
                 _test_case["eels_fork"] = eels_fork
+
+                # Build marks list with xdist_group for all tests
+                marks = [pytest.mark.xdist_group(_test_case["test_file"])]
+
                 if any(
                     x.search(_identifier) for x in test_patterns.expected_fail
                 ):
                     continue
                 elif any(x.search(_identifier) for x in test_patterns.slow):
-                    yield pytest.param(_test_case, marks=pytest.mark.slow)
+                    marks.append(pytest.mark.slow)
+                    yield pytest.param(_test_case, marks=marks)
                 elif any(
                     x.search(_identifier) for x in test_patterns.big_memory
                 ):
-                    yield pytest.param(_test_case, marks=pytest.mark.bigmem)
+                    marks.append(pytest.mark.bigmem)
+                    yield pytest.param(_test_case, marks=marks)
                 else:
-                    yield _test_case
+                    yield pytest.param(_test_case, marks=marks)
         except NoTestsFoundError:
             # file doesn't contain tests for the given fork
             continue

tests/json_infra/helpers/load_state_tests.py

@@ -16,6 +16,7 @@
 from ethereum_spec_tools.evm_tools.t8n import T8N
 
 from .. import FORKS
+from . import load_json_file
 from .exceptional_test_patterns import exceptional_state_test_patterns
 
 parser = create_parser()
@@ -53,10 +54,14 @@ def fetch_state_tests(json_fork: str) -> Generator:
                 "json_fork": json_fork,
             }
 
+            # Build marks list with xdist_group for all tests
+            marks = [pytest.mark.xdist_group(test_case_dict["test_file"])]
+
             if any(x.search(test_case.key) for x in test_patterns.slow):
-                yield pytest.param(test_case_dict, marks=pytest.mark.slow)
+                marks.append(pytest.mark.slow)
+                yield pytest.param(test_case_dict, marks=marks)
             else:
-                yield test_case_dict
+                yield pytest.param(test_case_dict, marks=marks)
 
 
 def idfn(test_case: Dict) -> str:
@@ -79,8 +84,7 @@ def run_state_test(test_case: Dict[str, str]) -> None:
     test_key = test_case["test_key"]
     index = test_case["index"]
     json_fork = test_case["json_fork"]
-    with open(test_file) as f:
-        tests = json.load(f)
+    tests = load_json_file(test_file)
 
     env = tests[test_key]["env"]
     try:

tox.ini

@@ -57,6 +57,7 @@ commands =
         --cov-branch \
         --ignore-glob='tests/json_infra/fixtures/*' \
         --basetemp="{temp_dir}/pytest" \
+        --dist=loadgroup \
         tests/json_infra
 
 [testenv:py3]
@@ -105,6 +106,7 @@ commands =
         --ignore-glob='eest_tests/*' \
         --basetemp="{temp_dir}/pytest" \
         --optimized \
+        --dist=loadgroup \
         tests/json_infra
 
 [testenv:doc]