hokusai/tests/test_builder.py

570 lines
20 KiB
Python

"""Integration tests for hokusai.builder."""
from __future__ import annotations
from collections.abc import Callable
from pathlib import Path
from typing import override
from unittest.mock import AsyncMock, MagicMock, patch
import pytest
from hokusai.builder import (
_collect_all_deps, # pyright: ignore[reportPrivateUsage]
_collect_dep_files, # pyright: ignore[reportPrivateUsage]
_collect_extra_params, # pyright: ignore[reportPrivateUsage]
_resolve_prompt, # pyright: ignore[reportPrivateUsage]
run_build,
)
from hokusai.config import GenerateTargetConfig, ProjectConfig
from hokusai.providers import Provider
from hokusai.providers.models import Capability, ModelInfo
from hokusai.state import load_state
WriteConfig = Callable[[dict[str, object]], ProjectConfig]
_PROJECT = "project"
_FAKE_TEXT_MODELS = [
ModelInfo(
name="pixtral-large-latest",
provider="Fake",
type="text",
capabilities=[Capability.TEXT_GENERATION, Capability.VISION],
),
]
_FAKE_IMAGE_MODELS = [
ModelInfo(
name="flux-2-pro",
provider="Fake",
type="image",
capabilities=[Capability.TEXT_TO_IMAGE, Capability.REFERENCE_IMAGES],
),
]
class FakeTextProvider(Provider):
"""A text provider that writes a marker file instead of calling an API."""
@staticmethod
@override
def get_provided_models() -> list[ModelInfo]:
return _FAKE_TEXT_MODELS
@override
async def generate(
self,
target_name: str,
target_config: GenerateTargetConfig,
resolved_prompt: str,
resolved_model: ModelInfo,
project_dir: Path,
) -> None:
output = project_dir / target_name
_ = output.write_text(f"generated:{target_name}:{resolved_prompt}")
class FakeImageProvider(Provider):
"""An image provider that writes a marker file instead of calling an API."""
@staticmethod
@override
def get_provided_models() -> list[ModelInfo]:
return _FAKE_IMAGE_MODELS
@override
async def generate(
self,
target_name: str,
target_config: GenerateTargetConfig,
resolved_prompt: str,
resolved_model: ModelInfo,
project_dir: Path,
) -> None:
output = project_dir / target_name
_ = output.write_text(f"generated:{target_name}:{resolved_prompt}")
class FailingTextProvider(Provider):
"""A text provider that always raises."""
@staticmethod
@override
def get_provided_models() -> list[ModelInfo]:
return _FAKE_TEXT_MODELS
@override
async def generate(
self,
target_name: str,
target_config: GenerateTargetConfig,
resolved_prompt: str,
resolved_model: ModelInfo,
project_dir: Path,
) -> None:
msg = f"Simulated failure for {target_name}"
raise RuntimeError(msg)
def _fake_providers() -> list[Provider]:
return [FakeTextProvider(), FakeImageProvider()]
class TestResolvePrompt:
"""Test prompt resolution (file vs inline)."""
def test_inline_prompt(self, project_dir: Path) -> None:
assert _resolve_prompt("Just a string", project_dir) == "Just a string"
def test_file_prompt(self, project_dir: Path, prompt_file: Path) -> None:
result = _resolve_prompt(prompt_file.name, project_dir)
assert result == "This prompt comes from a file"
def test_nonexistent_file_treated_as_inline(self, project_dir: Path) -> None:
result = _resolve_prompt("no_such_file.txt", project_dir)
assert result == "no_such_file.txt"
class TestCollectHelpers:
"""Test dependency collection helpers."""
def test_collect_dep_files(
self, project_dir: Path, write_config: WriteConfig
) -> None:
_ = (project_dir / "input.txt").write_text("data")
_ = (project_dir / "ref.png").write_bytes(b"ref")
config = write_config(
{
"targets": {
"out.png": {
"prompt": "x",
"inputs": ["input.txt"],
"reference_images": ["ref.png"],
"control_images": [],
}
}
}
)
deps = _collect_dep_files("out.png", config, project_dir)
dep_names = [d.name for d in deps]
assert "input.txt" in dep_names
assert "ref.png" in dep_names
def test_collect_extra_params(self, write_config: WriteConfig) -> None:
config = write_config(
{
"targets": {
"out.png": {
"prompt": "x",
"width": 512,
"height": 256,
"reference_images": ["ref.png"],
}
}
}
)
params = _collect_extra_params("out.png", config)
assert params["width"] == 512
assert params["height"] == 256
assert params["reference_images"] == ("ref.png",)
def test_collect_extra_params_empty(self, write_config: WriteConfig) -> None:
config = write_config({"targets": {"out.txt": {"prompt": "x"}}})
assert _collect_extra_params("out.txt", config) == {}
def test_collect_all_deps(self, write_config: WriteConfig) -> None:
config = write_config(
{
"targets": {
"out.png": {
"prompt": "x",
"inputs": ["a.txt"],
"reference_images": ["ref.png"],
"control_images": ["c1.png", "c2.png"],
}
}
}
)
deps = _collect_all_deps("out.png", config)
assert deps == ["a.txt", "ref.png", "c1.png", "c2.png"]
class TestRunBuild:
"""Integration tests for the full build pipeline with fake providers."""
async def test_build_single_text_target(
self, project_dir: Path, simple_text_config: ProjectConfig
) -> None:
with patch("hokusai.builder._create_providers", return_value=_fake_providers()):
result = await run_build(simple_text_config, project_dir, _PROJECT)
assert result.built == ["output.txt"]
assert result.skipped == []
assert result.failed == {}
assert (project_dir / "output.txt").exists()
async def test_build_chain_dependency(
self, project_dir: Path, multi_target_config: ProjectConfig
) -> None:
with patch("hokusai.builder._create_providers", return_value=_fake_providers()):
result = await run_build(multi_target_config, project_dir, _PROJECT)
assert "summary.md" in result.built
assert "final.txt" in result.built
assert "hero.png" in result.built
assert result.failed == {}
assert (project_dir / "summary.md").exists()
assert (project_dir / "final.txt").exists()
assert (project_dir / "hero.png").exists()
async def test_incremental_build_skips_clean_targets(
self, project_dir: Path, simple_text_config: ProjectConfig
) -> None:
with patch("hokusai.builder._create_providers", return_value=_fake_providers()):
result1 = await run_build(simple_text_config, project_dir, _PROJECT)
assert result1.built == ["output.txt"]
result2 = await run_build(simple_text_config, project_dir, _PROJECT)
assert result2.skipped == ["output.txt"]
assert result2.built == []
async def test_rebuild_after_prompt_change(
self, project_dir: Path, write_config: WriteConfig
) -> None:
config1 = write_config({"targets": {"out.txt": {"prompt": "version 1"}}})
with patch("hokusai.builder._create_providers", return_value=_fake_providers()):
r1 = await run_build(config1, project_dir, _PROJECT)
assert r1.built == ["out.txt"]
config2 = write_config({"targets": {"out.txt": {"prompt": "version 2"}}})
r2 = await run_build(config2, project_dir, _PROJECT)
assert r2.built == ["out.txt"]
async def test_rebuild_after_input_change(
self, project_dir: Path, write_config: WriteConfig
) -> None:
_ = (project_dir / "data.txt").write_text("original")
config = write_config(
{"targets": {"out.md": {"prompt": "x", "inputs": ["data.txt"]}}}
)
with patch("hokusai.builder._create_providers", return_value=_fake_providers()):
r1 = await run_build(config, project_dir, _PROJECT)
assert r1.built == ["out.md"]
_ = (project_dir / "data.txt").write_text("modified")
r2 = await run_build(config, project_dir, _PROJECT)
assert r2.built == ["out.md"]
async def test_selective_build_single_target(
self, project_dir: Path, multi_target_config: ProjectConfig
) -> None:
with patch("hokusai.builder._create_providers", return_value=_fake_providers()):
result = await run_build(
multi_target_config, project_dir, _PROJECT, target="summary.md"
)
assert "summary.md" in result.built
assert "hero.png" not in result.built
assert "final.txt" not in result.built
async def test_selective_build_unknown_target_raises(
self, project_dir: Path, simple_text_config: ProjectConfig
) -> None:
with patch("hokusai.builder._create_providers", return_value=_fake_providers()):
with pytest.raises(ValueError, match="Unknown target"):
_ = await run_build(
simple_text_config, project_dir, _PROJECT, target="nonexistent.txt"
)
async def test_failed_target_isolates_independent(
self, project_dir: Path, write_config: WriteConfig
) -> None:
config = write_config(
{
"targets": {
"fail.txt": {"prompt": "will fail"},
"ok.txt": {"prompt": "will succeed"},
}
}
)
fail_provider = FailingTextProvider()
fake_provider = FakeTextProvider()
async def selective_generate(
target_name: str,
target_config: GenerateTargetConfig,
resolved_prompt: str,
resolved_model: ModelInfo,
project_dir: Path,
) -> None:
if target_name == "fail.txt":
await fail_provider.generate(
target_name,
target_config,
resolved_prompt,
resolved_model,
project_dir,
)
else:
await fake_provider.generate(
target_name,
target_config,
resolved_prompt,
resolved_model,
project_dir,
)
routing_provider = FakeTextProvider()
routing_provider.generate = selective_generate # type: ignore[assignment]
with patch(
"hokusai.builder._create_providers",
return_value=[routing_provider, FakeImageProvider()],
):
result = await run_build(config, project_dir, _PROJECT)
assert "fail.txt" in result.failed
assert "ok.txt" in result.built
async def test_failed_dep_cascades(
self, project_dir: Path, write_config: WriteConfig
) -> None:
config = write_config(
{
"targets": {
"base.txt": {"prompt": "base"},
"child.txt": {"prompt": "child", "inputs": ["base.txt"]},
}
}
)
with patch(
"hokusai.builder._create_providers",
return_value=[FailingTextProvider(), FakeImageProvider()],
):
result = await run_build(config, project_dir, _PROJECT)
assert "base.txt" in result.failed
assert "child.txt" in result.failed
assert "Dependency failed" in result.failed["child.txt"]
async def test_missing_provider_records_failure(
self, project_dir: Path, simple_text_config: ProjectConfig
) -> None:
with patch(
"hokusai.builder._create_providers",
return_value=[],
):
result = await run_build(simple_text_config, project_dir, _PROJECT)
assert "output.txt" in result.failed
assert "No provider available" in result.failed["output.txt"]
async def test_state_saved_after_each_generation(
self, project_dir: Path, write_config: WriteConfig
) -> None:
config = write_config(
{
"targets": {
"a.txt": {"prompt": "first"},
"b.txt": {"prompt": "second", "inputs": ["a.txt"]},
}
}
)
with patch("hokusai.builder._create_providers", return_value=_fake_providers()):
_ = await run_build(config, project_dir, _PROJECT)
state = load_state(project_dir, _PROJECT)
assert "a.txt" in state.targets
assert "b.txt" in state.targets
async def test_prompt_file_resolution_in_build(
self, project_dir: Path, prompt_file: Path, write_config: WriteConfig
) -> None:
config = write_config({"targets": {"out.txt": {"prompt": prompt_file.name}}})
with patch("hokusai.builder._create_providers", return_value=_fake_providers()):
result = await run_build(config, project_dir, _PROJECT)
assert result.built == ["out.txt"]
content = (project_dir / "out.txt").read_text()
assert "This prompt comes from a file" in content
async def test_rebuild_after_output_deleted(
self, project_dir: Path, simple_text_config: ProjectConfig
) -> None:
with patch("hokusai.builder._create_providers", return_value=_fake_providers()):
r1 = await run_build(simple_text_config, project_dir, _PROJECT)
assert r1.built == ["output.txt"]
(project_dir / "output.txt").unlink()
r2 = await run_build(simple_text_config, project_dir, _PROJECT)
assert r2.built == ["output.txt"]
async def test_diamond_dependency_all_built(
self, project_dir: Path, write_config: WriteConfig
) -> None:
_ = (project_dir / "root.txt").write_text("root data")
config = write_config(
{
"targets": {
"left.md": {"prompt": "left", "inputs": ["root.txt"]},
"right.md": {"prompt": "right", "inputs": ["root.txt"]},
"merge.txt": {
"prompt": "merge",
"inputs": ["left.md", "right.md"],
},
}
}
)
with patch("hokusai.builder._create_providers", return_value=_fake_providers()):
result = await run_build(config, project_dir, _PROJECT)
assert set(result.built) == {"left.md", "right.md", "merge.txt"}
assert result.failed == {}
class TestDownloadTarget:
"""Tests for download-type targets that fetch files from URLs."""
async def test_download_target_fetches_url(
self, project_dir: Path, write_config: WriteConfig
) -> None:
config = write_config(
{
"targets": {
"fish.png": {"download": "https://example.com/fish.png"},
}
}
)
mock_response = MagicMock()
mock_response.content = b"fake image bytes"
mock_response.raise_for_status = MagicMock()
with (
patch("hokusai.builder._create_providers", return_value=_fake_providers()),
patch("hokusai.builder.httpx.AsyncClient") as mock_client_cls,
):
mock_client = AsyncMock()
mock_client.get = AsyncMock(return_value=mock_response)
mock_client.__aenter__ = AsyncMock(return_value=mock_client)
mock_client.__aexit__ = AsyncMock(return_value=False)
mock_client_cls.return_value = mock_client
result = await run_build(config, project_dir, _PROJECT)
assert result.built == ["fish.png"]
assert (project_dir / "fish.png").read_bytes() == b"fake image bytes"
mock_client.get.assert_called_once_with( # pyright: ignore[reportAny]
"https://example.com/fish.png", follow_redirects=True
)
async def test_download_target_incremental_skip(
self, project_dir: Path, write_config: WriteConfig
) -> None:
config = write_config(
{
"targets": {
"fish.png": {"download": "https://example.com/fish.png"},
}
}
)
mock_response = MagicMock()
mock_response.content = b"fake image bytes"
mock_response.raise_for_status = MagicMock()
with (
patch("hokusai.builder._create_providers", return_value=_fake_providers()),
patch("hokusai.builder.httpx.AsyncClient") as mock_client_cls,
):
mock_client = AsyncMock()
mock_client.get = AsyncMock(return_value=mock_response)
mock_client.__aenter__ = AsyncMock(return_value=mock_client)
mock_client.__aexit__ = AsyncMock(return_value=False)
mock_client_cls.return_value = mock_client
r1 = await run_build(config, project_dir, _PROJECT)
assert r1.built == ["fish.png"]
r2 = await run_build(config, project_dir, _PROJECT)
assert r2.skipped == ["fish.png"]
assert r2.built == []
async def test_download_target_rebuild_on_url_change(
self, project_dir: Path, write_config: WriteConfig
) -> None:
config1 = write_config(
{
"targets": {
"fish.png": {"download": "https://example.com/fish-v1.png"},
}
}
)
mock_response = MagicMock()
mock_response.content = b"v1 bytes"
mock_response.raise_for_status = MagicMock()
with (
patch("hokusai.builder._create_providers", return_value=_fake_providers()),
patch("hokusai.builder.httpx.AsyncClient") as mock_client_cls,
):
mock_client = AsyncMock()
mock_client.get = AsyncMock(return_value=mock_response)
mock_client.__aenter__ = AsyncMock(return_value=mock_client)
mock_client.__aexit__ = AsyncMock(return_value=False)
mock_client_cls.return_value = mock_client
r1 = await run_build(config1, project_dir, _PROJECT)
assert r1.built == ["fish.png"]
config2 = write_config(
{
"targets": {
"fish.png": {"download": "https://example.com/fish-v2.png"},
}
}
)
mock_response.content = b"v2 bytes"
r2 = await run_build(config2, project_dir, _PROJECT)
assert r2.built == ["fish.png"]
assert (project_dir / "fish.png").read_bytes() == b"v2 bytes"
async def test_download_target_as_dependency(
self, project_dir: Path, write_config: WriteConfig
) -> None:
config = write_config(
{
"targets": {
"fish.png": {"download": "https://example.com/fish.png"},
"description.txt": {
"prompt": "Describe the fish",
"inputs": ["fish.png"],
},
}
}
)
mock_response = MagicMock()
mock_response.content = b"fake fish image"
mock_response.raise_for_status = MagicMock()
with (
patch("hokusai.builder._create_providers", return_value=_fake_providers()),
patch("hokusai.builder.httpx.AsyncClient") as mock_client_cls,
):
mock_client = AsyncMock()
mock_client.get = AsyncMock(return_value=mock_response)
mock_client.__aenter__ = AsyncMock(return_value=mock_client)
mock_client.__aexit__ = AsyncMock(return_value=False)
mock_client_cls.return_value = mock_client
result = await run_build(config, project_dir, _PROJECT)
assert "fish.png" in result.built
assert "description.txt" in result.built
assert (project_dir / "fish.png").read_bytes() == b"fake fish image"
assert (project_dir / "description.txt").exists()