add "north star" benchmark

.gitignore

@@ -26,6 +26,7 @@ _build/
 
 # Benchmark and test files
 /benchmarks/*.json
+/tests/benchmarks/*.json
 /htmlcov/
 /codecov.sh
 /coverage.lcov

Makefile

@@ -64,6 +64,10 @@ test-pyright: .pdm
 test: .pdm
 	pdm run coverage run -m pytest --durations=10
 
+.PHONY: benchmark  ## Run all benchmarks
+benchmark: .pdm
+	pdm run coverage run -m pytest --durations=10 --benchmark-enable tests/benchmarks
+
 .PHONY: testcov  ## Run tests and generate a coverage report, skipping the type-checker integration tests
 testcov: test
 	@echo "building coverage html"

pyproject.toml

@@ -106,6 +106,8 @@ testing = [
     "pytest-mock",
     "pytest-pretty",
     "pytest-examples",
+    "faker>=18.13.0",
+    "pytest-benchmark>=4.0.0",
 ]
 mypy = [
     "mypy",
@@ -151,6 +153,12 @@ filterwarnings = [
     # Work around https://github.com/pytest-dev/pytest/issues/10977 for Python 3.12
     'ignore:(ast\.Str|ast\.NameConstant|ast\.Num|Attribute s) is deprecated and will be removed.*:DeprecationWarning:',
 ]
+addopts = [
+    '--benchmark-columns', 'min,mean,stddev,outliers,rounds,iterations',
+    '--benchmark-group-by', 'group',
+    '--benchmark-warmup', 'on',
+    '--benchmark-disable',  # this is enable by `make benchmark` when you actually want to run benchmarks
+]
 
 # configuring https://github.com/pydantic/hooky
 [tool.hooky]

tests/benchmarks/generate_north_star_data.py

@@ -0,0 +1,119 @@
+from typing import Any, Callable, List, TypeVar, Union
+
+from faker import Faker
+
+f = Faker()
+Faker.seed(0)
+
+
+T = TypeVar('T')
+
+## Helper functions
+
+
+def one_of(*callables: Callable[[], Any]) -> Any:
+    return f.random.choice(callables)()
+
+
+def list_of(callable: Callable[[], T], max_length: int) -> List[T]:
+    return [callable() for _ in range(f.random_int(max=max_length))]
+
+
+def lax_int(*args: Any, **kwargs: Any) -> Union[int, float, str]:
+    return f.random.choice((int, float, str))(f.random_int(*args, **kwargs))
+
+
+def lax_float(*args: Any, **kwargs: Any) -> Union[int, float, str]:
+    return f.random.choice((int, float, str))(f.pyfloat(*args, **kwargs))
+
+
+def time_seconds() -> int:
+    dt = f.date_time()
+    midnight = dt.replace(hour=0, minute=0, second=0, microsecond=0)
+    return (dt - midnight).total_seconds()
+
+
+def time_microseconds() -> float:
+    return float(time_seconds()) + (f.random_int(max=999999) * 1e-6)
+
+
+def time_string() -> str:
+    return f.time()
+
+
+def lax_time() -> Union[int, float, str]:
+    return one_of(time_seconds, time_microseconds, time_string)
+
+
+def datetime_timestamp() -> int:
+    dt = f.date_time()
+    midnight = dt.replace(hour=0, minute=0, second=0, microsecond=0)
+    return (dt - midnight).total_seconds()
+
+
+def datetime_microseconds() -> float:
+    return float(datetime_timestamp()) + (f.random_int(max=999999) * 1e-6)
+
+
+def datetime_str() -> str:
+    return f.date_time().isoformat()
+
+
+def lax_datetime() -> Union[int, float, str]:
+    return one_of(datetime_timestamp, datetime_microseconds, datetime_str)
+
+
+## Sample data generators
+
+
+def blog() -> dict:
+    return {
+        'type': 'blog',
+        'title': f.text(max_nb_chars=40),
+        'post_count': lax_int(),
+        'readers': lax_int(),
+        'avg_post_rating': lax_float(min_value=0, max_value=5),
+        'url': f.url(),
+    }
+
+
+def social_profile() -> dict:
+    return {
+        'type': 'profile',
+        'username': f.user_name(),
+        'join_date': f.date().format('%Y-%m-%d'),
+        **one_of(facebook_profile, twitter_profile, linkedin_profile),
+    }
+
+
+def facebook_profile() -> dict:
+    return {'network': 'facebook', 'friends': lax_int()}
+
+
+def twitter_profile() -> dict:
+    return {'network': 'twitter', 'followers': lax_int()}
+
+
+def linkedin_profile() -> dict:
+    return {'network': 'linkedin', 'connections': min(f.random_int(), 500)}
+
+
+def website() -> dict:
+    return one_of(blog, social_profile)
+
+
+def person() -> dict:
+    return {
+        'id': f.uuid4(),
+        'name': f.name(),
+        'email': f.safe_email(),
+        'height': str(f.pydecimal(min_value=1, max_value=2, right_digits=2)),
+        'entry_created_date': f.date().format('%Y-%m-%d'),
+        'entry_created_time': lax_time(),
+        'entry_updated_at': lax_datetime(),
+        'websites': list_of(website, max_length=5),
+    }
+
+
+def person_data(length: int) -> List[dict]:
+    return [person() for _ in range(length)]