[REFACTOR](build) Migrate workspace to uv_build and PEP 420 namespace packages

Makefile

@@ -1,4 +1,4 @@
-.PHONY: default uv-sync clean-pyspark generate-pyspark check test-all test test-only docformat docformat-only doctest doctest-only mypy mypy-only lint-only update-baselines
+.PHONY: default uv-sync clean-pyspark generate-pyspark check check-namespace test-all test test-only docformat docformat-only doctest doctest-only mypy mypy-only lint-only update-baselines
 
 TESTMON ?= --testmon
 
@@ -23,7 +23,23 @@ generate-pyspark: uv-sync clean-pyspark
 	@uv run ruff format --quiet $(PYSPARK_EXPRESSIONS) $(PYSPARK_GENERATED_TESTS)
 
 check: uv-sync generate-pyspark
-	@$(MAKE) -j test-only docformat-only doctest-only lint-only mypy-only
+	@$(MAKE) -j test-only docformat-only doctest-only lint-only mypy-only check-namespace
+
+# Guard the PEP 420 namespace roots. overture/ and overture/schema/ are
+# implicit namespace packages -- no __init__.py -- so every workspace package
+# contributes its own leaf under the shared namespace. An __init__.py at either
+# root turns that package into a regular package that shadows every other
+# distribution's contribution at import time. -path patterns (not shell globs)
+# keep this shell-independent and match only the two roots, never the leaves.
+check-namespace:
+	@found=$$(find packages -type f -path '*/src/overture/__init__.py' \
+	              -o -type f -path '*/src/overture/schema/__init__.py'); \
+	if [ -n "$$found" ]; then \
+		echo "ERROR: __init__.py at PEP 420 namespace root(s) -- delete these:" >&2; \
+		echo "$$found" >&2; \
+		echo "overture/ and overture/schema/ must stay implicit namespace packages." >&2; \
+		exit 1; \
+	fi
 
 # test-all is the unconditional full run -- testmon-independent, unlike the
 # incremental test/test-only targets -- so data-only changes (golden JSON,

packages/overture-schema-cli/changelog.d/618.misc.md

@@ -0,0 +1 @@
+Migrated the build backend from hatchling to uv_build and made `overture.schema` a PEP 420 namespace package.

packages/overture-schema-cli/pyproject.toml

@@ -28,8 +28,8 @@ overture-schema-common = { workspace = true }
 overture-schema-system = { workspace = true }
 
 [build-system]
-build-backend = "hatchling.build"
-requires = ["hatchling"]
+requires = ["uv_build>=0.11.32,<0.13"]
+build-backend = "uv_build"
 
 [dependency-groups]
 dev = [
@@ -38,8 +38,8 @@ dev = [
     "mypy>=1.17.0",
 ]
 
-[tool.hatch.build.targets.wheel]
-packages = ["src/overture"]
+[tool.uv.build-backend]
+module-name = "overture.schema.cli"
 
 [project.scripts]
 overture-schema = "overture.schema.cli:cli"

packages/overture-schema-cli/tests/test_cli_commands.py

@@ -6,6 +6,7 @@
 import pytest
 from click.testing import CliRunner
 from conftest import build_feature
+
 from overture.schema.cli.commands import cli
 
 

packages/overture-schema-cli/tests/test_cli_functions.py

@@ -9,6 +9,8 @@
 import yaml
 from click.exceptions import UsageError
 from conftest import build_feature
+from pydantic import BaseModel, ValidationError
+
 from overture.schema.cli.commands import (
     _best_fit_model,
     _revalidate_undiscriminatable_items,
@@ -19,7 +21,6 @@
 from overture.schema.cli.type_analysis import get_item_index
 from overture.schema.cli.types import ValidationErrorDict
 from overture.schema.system.discovery import TagSelector
-from pydantic import BaseModel, ValidationError
 
 
 class TestLoadInput:

packages/overture-schema-cli/tests/test_data_display.py

@@ -3,14 +3,15 @@
 from io import StringIO
 
 from conftest import build_feature
+from rich.console import Console
+from rich.panel import Panel
+
 from overture.schema.cli.data_display import (
     create_feature_display,
     extract_feature_data,
     format_field_value,
     select_context_fields,
 )
-from rich.console import Console
-from rich.panel import Panel
 
 
 class TestExtractFeatureData:

packages/overture-schema-cli/tests/test_error_formatting.py

@@ -6,6 +6,9 @@
 
 import pytest
 from click.testing import CliRunner
+from pydantic import BaseModel, Field, TypeAdapter, ValidationError
+from rich.console import Console
+
 from overture.schema.cli.commands import cli
 from overture.schema.cli.error_formatting import (
     format_path,
@@ -14,8 +17,6 @@
 )
 from overture.schema.cli.type_analysis import introspect_union
 from overture.schema.cli.types import ErrorLocation, ValidationErrorDict
-from pydantic import BaseModel, Field, TypeAdapter, ValidationError
-from rich.console import Console
 
 
 class TestErrorGrouping:

packages/overture-schema-cli/tests/test_heterogeneous_collections.py

@@ -6,6 +6,7 @@
 import pytest
 from click.testing import CliRunner
 from conftest import build_feature
+
 from overture.schema.cli.commands import cli
 
 

packages/overture-schema-cli/tests/test_resolve_types.py

@@ -9,6 +9,7 @@
 from unittest.mock import patch
 
 import pytest
+
 from overture.schema.cli.commands import resolve_types
 from overture.schema.system.discovery import ModelKey, TagSelector
 

packages/overture-schema-cli/tests/test_type_analysis.py

@@ -3,14 +3,15 @@
 from typing import Annotated, Literal
 
 import pytest
+from pydantic import BaseModel, Discriminator, Field
+
 from overture.schema.cli.type_analysis import (
     StructuralTuple,
     create_structural_tuple,
     extract_discriminator_path,
     get_or_create_structural_tuple,
     introspect_union,
 )
-from pydantic import BaseModel, Discriminator, Field
 
 
 class TestStructuralTuples:

packages/overture-schema-codegen/changelog.d/618.misc.md

@@ -0,0 +1 @@
+Migrated the build backend to uv_build and stopped emitting `__init__.py` in the generated PySpark trees, making them PEP 420 namespace packages.

packages/overture-schema-codegen/pyproject.toml

@@ -1,6 +1,6 @@
 [build-system]
-build-backend = "hatchling.build"
-requires = ["hatchling"]
+requires = ["uv_build>=0.11.32,<0.13"]
+build-backend = "uv_build"
 
 [project]
 dependencies = [
@@ -42,5 +42,5 @@ test = [
     "overture-schema-theme-transportation",
 ]
 
-[tool.hatch.build.targets.wheel]
-packages = ["src/overture"]
+[tool.uv.build-backend]
+module-name = "overture.schema.codegen"

packages/overture-schema-codegen/src/overture/schema/codegen/pyspark/pipeline.py

@@ -7,7 +7,7 @@
 
 from __future__ import annotations
 
-from collections.abc import Iterable, Sequence
+from collections.abc import Sequence
 from dataclasses import dataclass
 from pathlib import PurePosixPath
 
@@ -48,12 +48,10 @@ class GeneratedModule:
 class PipelineOutput:
     """PySpark modules emitted by the pipeline, split by output tree.
 
-    The source and test trees are written to separate directories and
-    mirror the same relative layout, so a path is meaningful only
-    relative to its tree. Splitting at the boundary keeps each tree
-    self-contained -- in practice the overlap today is just
-    `__init__.py`, but any path duplicated between trees would be
-    ambiguous in a single flat list.
+    Source and test modules write to separate directories
+    (`--output-dir` and `--test-output-dir`), so they travel as two
+    lists rather than one. Both trees mirror the same relative layout,
+    so a path is meaningful only relative to its own tree.
     """
 
     source: list[GeneratedModule]
@@ -62,21 +60,6 @@ class PipelineOutput:
 
 _OUTPUT_PACKAGE = "overture.schema.pyspark.expressions.generated"
 
-# Dots in `from ...x import y` from a generated test module to reach
-# `tests/`: one to leave the file's package, one to leave `generated/`.
-# Each additional directory component under `generated/` adds another.
-_DOTS_FROM_TEST_TO_TESTS_ROOT = 2
-
-
-def _support_prefix(directory: PurePosixPath) -> str:
-    """Relative-import prefix used by generated test modules to reach `_support`.
-
-    Each leading dot climbs one package level; the first two dots step
-    out of `tests/generated/` to `tests/`, and an extra dot is appended
-    for every component of *directory* under `generated/`.
-    """
-    return "." * (len(directory.parts) + _DOTS_FROM_TEST_TO_TESTS_ROOT)
-
 
 def _require_entry_point(spec: ModelSpec) -> str:
     """Return *spec*'s entry point or raise if it's missing."""
@@ -117,21 +100,6 @@ def _extract_geometry_types(
     return tuple(sorted(seen, key=lambda g: g.name))
 
 
-def _init_modules(paths: Iterable[PurePosixPath]) -> list[GeneratedModule]:
-    """Emit empty `__init__.py` for every directory of `paths`.
-
-    Includes the output root so the top-level package exists after a
-    full `rm -rf` of the generated tree.
-    """
-    paths = list(paths)
-    if not paths:
-        return []
-    dirs: set[PurePosixPath] = set()
-    for path in paths:
-        dirs.update(path.parents)
-    return [GeneratedModule(content="", path=d / "__init__.py") for d in sorted(dirs)]
-
-
 def generate_pyspark_module(spec: ModelSpec) -> GeneratedModule:
     """Generate a PySpark validation module from a model spec.
 
@@ -162,16 +130,14 @@ def generate_pyspark_modules(
     Returns
     -------
     PipelineOutput
-        Source-tree model modules and test-tree modules. Each tree
-        includes the `__init__.py` files needed for its package layout.
+        Source-tree model modules and test-tree modules. The generated
+        tree is PEP 420, so no `__init__.py` files are emitted.
     """
     items = [(spec, build_checks(spec)) for spec in model_specs]
     source = [_render_module(spec, checks) for spec, checks in items]
     test: list[GeneratedModule] = []
     for spec, checks in items:
         test.extend(_render_test_modules(spec, checks))
-    source.extend(_init_modules(m.path for m in source))
-    test.extend(_init_modules(m.path for m in test))
     return PipelineOutput(source=source, test=test)
 
 
@@ -232,7 +198,6 @@ def _render_test_modules(
     field_checks, model_checks = checks
     directory, model_name = _directory_and_model_name(spec)
     expression_import = ".".join([_OUTPUT_PACKAGE, *directory.parts, model_name])
-    support_prefix = _support_prefix(directory)
 
     modules: list[GeneratedModule] = []
     for arm, (base_row_sparse, base_row_populated) in _select_arm_rows(spec).items():
@@ -248,7 +213,6 @@ def _render_test_modules(
                     arm=arm,
                     spec=spec,
                     expression_import=expression_import,
-                    support_prefix=support_prefix,
                 ),
                 path=directory / f"test_{model_name}{suffix}.py",
             )

packages/overture-schema-codegen/src/overture/schema/codegen/pyspark/templates/test_module.py.jinja2

@@ -11,17 +11,17 @@ from {{ expression_import }} import (
 )
 from pyspark.sql import SparkSession
 
-from {{ support_prefix }}_support.harness import (
+from _support.harness import (
     ValidationResults,
     run_validation_pipeline,
 )
 {% if mutation_imports %}
-from {{ support_prefix }}_support.mutations import {{ mutation_imports | join(", ") }}
+from _support.mutations import {{ mutation_imports | join(", ") }}
 {% endif %}
 {% if needs_set_at_path %}
-from {{ support_prefix }}_support.helpers import set_at_path
+from _support.helpers import set_at_path
 {% endif %}
-from {{ support_prefix }}_support.scenarios import Scenario
+from _support.scenarios import Scenario
 
 BASE_ROW_SPARSE: dict = {{ base_row_sparse }}
 

packages/overture-schema-codegen/src/overture/schema/codegen/pyspark/test_renderer.py

@@ -141,7 +141,6 @@ def render_test_module(
     model_checks: list[ModelCheck],
     *,
     expression_import: str,
-    support_prefix: str,
     base_row_sparse: dict[str, Any] | None = None,
     base_row_populated: dict[str, Any] | None = None,
     arm: str | None = None,
@@ -187,7 +186,6 @@ def render_test_module(
         base_row_populated=populated_repr,
         scenarios=all_scenarios,
         expression_import=expression_import,
-        support_prefix=support_prefix,
     )
 
 

packages/overture-schema-codegen/tests/codegen_test_support.py

@@ -13,6 +13,8 @@
 
 import pytest
 from annotated_types import MinLen
+from pydantic import BaseModel, EmailStr, Field, HttpUrl, RootModel
+
 from overture.schema.codegen.extraction.field import LiteralScalar, Primitive
 from overture.schema.codegen.extraction.field_walk import terminal_of
 from overture.schema.codegen.extraction.model_extraction import extract_model
@@ -55,7 +57,6 @@
 from overture.schema.system.numeric import float64, int32
 from overture.schema.system.ref import Id, Identified, Reference, Relationship
 from overture.schema.system.string import HexColor, LanguageTag, StrippedString
-from pydantic import BaseModel, EmailStr, Field, HttpUrl, RootModel
 
 STR_TYPE = Primitive(base_type="str")
 

packages/overture-schema-codegen/tests/conftest.py

@@ -1,10 +1,12 @@
 """Shared pytest fixtures for overture-schema-codegen tests."""
 
-import overture.schema.system.geometric as _system_geometric
-import overture.schema.system.numeric as _system_numeric
 import pytest
 from click.testing import CliRunner
 from codegen_test_support import find_model_class
+from pydantic import BaseModel
+
+import overture.schema.system.geometric as _system_geometric
+import overture.schema.system.numeric as _system_numeric
 from overture.schema.codegen.extraction.model_extraction import extract_model
 from overture.schema.codegen.extraction.numeric_extraction import extract_numerics
 from overture.schema.codegen.extraction.specs import RecordSpec
@@ -17,7 +19,6 @@
 )
 from overture.schema.system.discovery import discover_models
 from overture.schema.system.geometric import GeometryType
-from pydantic import BaseModel
 
 
 def pytest_addoption(parser: pytest.Parser) -> None:

packages/overture-schema-codegen/tests/test_cli.py

@@ -6,6 +6,7 @@
 
 import pytest
 from click.testing import CliRunner
+
 from overture.schema.codegen.cli import cli
 from overture.schema.codegen.extraction.specs import RecordSpec
 

packages/overture-schema-codegen/tests/test_constraint_description.py

@@ -3,6 +3,7 @@
 import re
 
 from annotated_types import Ge, Gt, Interval, Le, Lt, MultipleOf
+
 from overture.schema.codegen.extraction.field_constraints import (
     constraint_display_text,
     describe_field_constraint,

packages/overture-schema-codegen/tests/test_enum_extraction.py

@@ -3,6 +3,7 @@
 from enum import Enum
 
 from codegen_test_support import find_member
+
 from overture.schema.codegen.extraction.enum_extraction import extract_enum
 from overture.schema.codegen.extraction.specs import EnumMemberSpec, EnumSpec
 from overture.schema.system.doc import DocumentedEnum