Cached Generation
ci / checks (push) Successful in 2h36m35s

This commit is contained in:
khodowany
2026-09-20 10:39:12 -05:00
parent 7244cebe68
commit 68ded4d8c8
8 changed files with 322 additions and 16 deletions
+151
View File
@@ -0,0 +1,151 @@
import argparse
import copy
import hashlib
import json
from pathlib import Path
import tempfile
import unittest
from unittest.mock import patch
from generator.__main__ import generate_assets
from generator.engraving import EngravingError, Fonts, Style, engrave_tree, layout_score
from generator.engraving.cache import renderer_fingerprint, restore_pages, score_fingerprint
from tests.test_musicxml import parse, xml_note, xml_score
class EngravingCacheTests(unittest.TestCase):
def setUp(self):
self.temporary = tempfile.TemporaryDirectory()
self.addCleanup(self.temporary.cleanup)
self.root = Path(self.temporary.name)
self.cache = self.root / 'cache'
self.fonts = Path('package/assets/fonts/petaluma')
self.document = parse(xml_score(xml_note()))
def write_scores(self, root, documents):
root.mkdir(parents=True, exist_ok=True)
for index, document in enumerate(documents):
(root / f'{index:03}.json').write_text(json.dumps(document), encoding='utf-8')
def files(self, root):
return {p.relative_to(root): p.read_bytes() for p in root.rglob('*') if p.is_file()}
def seed_cache(self):
root = self.root / 'seed'
self.write_scores(root, [self.document])
engrave_tree(root, self.fonts, cache_dir=self.cache)
return next(self.cache.iterdir())
def test_cache_hit_preserves_bundle_bytes_across_output_directories(self):
cold, warm = self.root / 'cold', self.root / 'warm'
self.write_scores(cold, [self.document])
engrave_tree(cold, self.fonts, cache_dir=self.cache)
self.write_scores(warm, [self.document])
with patch('generator.engraving.layout_score', side_effect=AssertionError('cache miss')):
with patch('generator.engraving.rasterize_page', side_effect=AssertionError('cache miss')):
engrave_tree(warm, self.fonts, cache_dir=self.cache)
self.assertEqual(self.files(cold), self.files(warm))
def test_playback_is_not_restored_from_page_cache(self):
self.seed_cache()
changed = copy.deepcopy(self.document)
changed['playback'] = {'new-playback': 'independent of page rendering'}
root = self.root / 'warm'
self.write_scores(root, [changed])
with patch('generator.engraving.layout_score', side_effect=AssertionError('cache miss')):
engrave_tree(root, self.fonts, cache_dir=self.cache)
self.assertEqual(json.loads((root / '000.json').read_text())['playback'], changed['playback'])
def test_changed_music_and_renderer_invalidate_pages(self):
self.seed_cache()
changed = parse(xml_score(xml_note('D')).replace('Fixture', 'Changed'))
self.assertNotEqual(score_fingerprint(self.document, 'renderer'), score_fingerprint(changed, 'renderer'))
self.assertNotEqual(score_fingerprint(self.document, 'renderer'), score_fingerprint(self.document, 'new-renderer'))
root = self.root / 'changed'
self.write_scores(root, [changed])
with patch('generator.engraving.layout_score', wraps=layout_score) as render:
engrave_tree(root, self.fonts, cache_dir=self.cache)
render.assert_called_once()
self.assertEqual(len(list(self.cache.iterdir())), 2)
def test_renderer_fingerprint_tracks_fonts_style_code_and_dependencies(self):
fonts, style = Fonts(self.fonts), Style()
expected = renderer_fingerprint(fonts, style)
self.assertEqual(expected, renderer_fingerprint(fonts, style))
self.assertNotEqual(expected, renderer_fingerprint(fonts, Style(margin_x=29)))
with patch.object(fonts, 'fingerprint', return_value='changed-fonts'):
self.assertNotEqual(expected, renderer_fingerprint(fonts, style))
with patch('generator.engraving.cache.importlib.metadata.version', return_value='new-version'):
self.assertNotEqual(expected, renderer_fingerprint(fonts, style))
read_bytes = Path.read_bytes
with patch.object(Path, 'read_bytes', lambda path: read_bytes(path) + (b'\n' if path.name == 'raster.py' else b'')):
self.assertNotEqual(expected, renderer_fingerprint(fonts, style))
def test_corrupt_or_missing_pages_are_regenerated_and_repaired(self):
entry = self.seed_cache()
page = entry / '0001.png'
original = page.read_bytes()
for corruption in ('missing', 'checksum', 'undecodable', 'manifest', 'page_directory', 'manifest_directory'):
with self.subTest(corruption=corruption):
if corruption == 'missing':
page.unlink()
elif corruption == 'checksum':
page.write_bytes(original + b'changed')
elif corruption == 'undecodable':
page.write_bytes(original[:24])
(entry / 'pages.json').write_text(json.dumps([{'path': page.name, 'sha256': hashlib.sha256(page.read_bytes()).hexdigest()}]))
elif corruption == 'manifest':
(entry / 'pages.json').write_text('invalid json')
else:
target = page if corruption == 'page_directory' else entry / 'pages.json'
target.unlink()
target.mkdir()
root = self.root / corruption
self.write_scores(root, [self.document])
with patch('generator.engraving.layout_score', wraps=layout_score) as render:
engrave_tree(root, self.fonts, cache_dir=self.cache)
render.assert_called_once()
self.assertEqual(page.read_bytes(), original)
self.assertIsNotNone(restore_pages(entry, self.root / 'restored'))
def test_cache_manifest_cannot_escape_entry_or_inject_bundle_fields(self):
entry = self.seed_cache()
original = json.loads((entry / 'pages.json').read_text())
for invalid in ([], {}, [{'path': '../outside.png', 'sha256': original[0]['sha256']}], [{**original[0], 'extra': True}]):
with self.subTest(manifest=invalid):
(entry / 'pages.json').write_text(json.dumps(invalid))
self.assertIsNone(restore_pages(entry, self.root / 'output'))
(entry / 'pages.json').write_text(json.dumps(original))
page = entry / '0001.png'
external = self.root / 'outside.png'
page.rename(external)
page.symlink_to(external)
self.assertIsNone(restore_pages(entry, self.root / 'output'))
alias = self.root / 'alias'
alias.symlink_to(entry, target_is_directory=True)
self.assertIsNone(restore_pages(alias, self.root / 'output'))
def test_parallel_output_and_shared_cache_are_deterministic(self):
documents = [self.document, parse(xml_score(xml_note('D')))] * 4
serial, parallel, warm = (self.root / name for name in ('serial', 'parallel', 'warm'))
self.write_scores(serial, documents)
engrave_tree(serial, self.fonts)
for root in (parallel, warm):
self.write_scores(root, documents)
engrave_tree(root, self.fonts, cache_dir=self.cache, jobs=2)
self.assertEqual(self.files(serial), self.files(root))
self.assertEqual(len(list(self.cache.iterdir())), 2)
self.assertFalse(list(self.cache.glob('.engraving-*')))
def test_invalid_worker_count_and_overlapping_cache_are_rejected(self):
root = self.root / 'assets'
self.write_scores(root, [self.document])
for jobs in (0, -1, 17, True):
with self.assertRaisesRegex(ValueError, 'between 1 and 16'):
engrave_tree(root, self.fonts, jobs=jobs)
for cache in (root, root / 'cache', self.root):
with self.assertRaisesRegex(ValueError, 'must not overlap'):
engrave_tree(root, self.fonts, cache_dir=cache)
def test_parallel_generation_preserves_assets_on_worker_failure(self):
data, output = self.root / 'data', self.root / 'output'
(data / 'songs').mkdir(parents=True)
source = data / '001.xml'
source.write_text(xml_score(xml_note()))
args = argparse.Namespace(data_dir=data, output_dir=output, static_assets_dir=Path('package/assets'), check=False, engraving_cache=self.cache, jobs=2)
generate_assets(args)
original = self.files(output)
generate_assets(args)
self.assertEqual(original, self.files(output))
source.write_text(xml_score(xml_note(extra='<accidental smufl="unavailableGlyph">other</accidental>')))
with self.assertRaisesRegex(EngravingError, 'unavailableGlyph'):
generate_assets(args)
self.assertEqual(original, self.files(output))
self.assertFalse(list(output.parent.glob('.vamp-assets-*')))