"""Regression tests for the public website data and citation contracts.""" from __future__ import annotations import hashlib import json import sqlite3 import sys import tempfile import unittest from unittest import mock from pathlib import Path import yaml ROOT = Path(__file__).resolve().parent.parent sys.path.insert(0, str(ROOT / "scripts")) from generate_site import ( # noqa: E402 _admonition_body, _browser_title, _forge_sources, _source_ref_markdown, decorate_markdown_pages, generate_data_exports, ) class SiteReferenceContracts(unittest.TestCase): def setUp(self) -> None: scratch = ROOT / "tmp" / "tests" scratch.mkdir(parents=True, exist_ok=True) self.temp = tempfile.TemporaryDirectory(dir=scratch) self.docs = Path(self.temp.name) / "docs" self.docs.mkdir() def tearDown(self) -> None: self.temp.cleanup() @staticmethod def _fixtures() -> tuple[dict, dict, dict, dict]: payload = b"firmware" sha1 = hashlib.sha1(payload).hexdigest() md5 = hashlib.md5(payload).hexdigest() sha256 = hashlib.sha256(payload).hexdigest() db = { "schema_version": 1, "generated_at": "2026-08-09T00:00:00Z", "total_files": 1, "total_size": len(payload), "files": { sha1: { "path": "bios/Test/fw.bin", "name": "fw.bin", "size": len(payload), "sha1": sha1, "md5": md5, "sha256": sha256, "crc32": "0f00ba11", "adler32": "0f00ba11", } }, "indexes": { "by_md5": {md5: sha1}, "by_name": {"fw.bin": [sha1]}, "by_crc32": {"0f00ba11": sha1}, "by_sha256": {sha256: sha1}, "by_path_suffix": {"fw.bin": [sha1]}, }, } coverages = { "test": { "platform": "Test Platform", "config": { "platform": "Test Platform", "cores": ["testemu"], "systems": { "test-system": { "files": [ { "name": "fw.bin", "destination": "fw.bin", "required": True, "sha1": sha1, "md5": "C87A52C277F7952B41C639FC7BF0A43B," "d0ed892a5e8d47e0cb5f8aa5f4dd7c5a", "crc32": "3F9F95A4", "size": len(payload), } ] } }, }, "total": 1, "present": 1, "verified": 1, "untested": 0, "missing": 0, "core_present": 1, "core_missing": 0, "core_unsourceable": 0, "details": [], } } profiles = { "testemu": { "emulator": "Test Emulator", "type": "libretro", "source": "https://github.com/example/emulator", "source_commit": "a" * 40, "cores": ["testemu"], "systems": ["test-system"], "files": [ { "name": "fw.bin", "system": "test-system", "required": True, "md5": ["5c124e7948ada85da603a522782940d0", "4921d5f2165dee6e2496f4388c4c81b2"], "source_ref": "src/firmware.cpp:10-12", } ], } } stats = { "schema_version": 1, "generated_at": "2026-08-09T00:00:00Z", "files": 1, "size_bytes": len(payload), "platforms": 1, "emulators": 1, "systems": 1, "catalog_matched": 0, "source": "https://github.com/Abdess/retrobios", "downloads": "https://github.com/Abdess/retrobios/releases/tag/v2026.08.06", "composition": { "systems": {"files": 1, "size_bytes": len(payload)}, "arcade": {"files": 0, "size_bytes": 0}, "game_data": {"files": 0, "size_bytes": 0}, }, } return db, coverages, profiles, stats @mock.patch("sitesources.tree_holds", lambda *_args: True) # a path the record confirms def test_source_references_are_revision_pinned_links(self): _, _, profiles, _ = self._fixtures() rendered = _source_ref_markdown( profiles["testemu"], "src/firmware.cpp:10-12" ) self.assertIn("/blob/" + "a" * 40 + "/src/firmware.cpp#L10-L12", rendered) def test_ambiguous_two_repository_path_is_never_linked(self): """A path that could live in either declared repository stays plain. The libretro port and the original emulator are two different trees. Linking an unattributable path to whichever came first publishes a citation that resolves to nothing. """ profile = { "source": "https://github.com/libretro/example-libretro", "source_commit": "a" * 40, "upstream": "https://gitlab.com/vendor/example", "upstream_commit": "b" * 40, } rendered = _source_ref_markdown(profile, "jg.c:213-214") self.assertEqual(rendered, "`jg.c:213-214`") self.assertNotIn("http", rendered) # A first segment equal to a repository's name proves nothing: PCSX2 # keeps its sources in pcsx2/, hatariB vendors Hatari in hatari/. # Stripped, 470 links pointed at nothing or at another file. prefixed = _source_ref_markdown(profile, "example/jg.c:213-214") self.assertEqual(prefixed, "`example/jg.c:213-214`") @mock.patch("sitesources.tree_holds", lambda *_args: True) # a path the record confirms def test_a_single_repository_links_the_path_as_written(self): profile = {"source": "https://github.com/PCSX2/pcsx2", "source_commit": "c" * 40} rendered = _source_ref_markdown(profile, "pcsx2/ImGui/FullscreenUI.cpp:755-759") self.assertIn("/blob/" + "c" * 40 + "/pcsx2/ImGui/FullscreenUI.cpp#L755-L759", rendered) def test_preprocessor_tokens_cannot_become_page_headings(self): rendered = _admonition_body("First paragraph.\n\n#if FEATURE\nEnabled") self.assertIn("\\#if FEATURE", rendered) self.assertNotIn("\n #if FEATURE", rendered) def test_browser_titles_distinguish_emulators_from_system_pages(self): from unittest import mock # noqa: PLC0415 import generate_site # noqa: PLC0415 pages = {"systems/qemu.md": {}, "emulators/qemu.md": {"hasPart": [{"name": "a"}]}} with mock.patch.dict(generate_site.PAGE_DATA, pages, clear=True): self.assertEqual(_browser_title(Path("emulators/qemu.md"), "QEMU"), "QEMU emulator BIOS files") self.assertEqual(_browser_title(Path("systems/qemu.md"), "QEMU"), "QEMU BIOS files") ACCEPTED = ( ("md5", "c87a52c277f7952b41c639fc7bf0a43b"), ("md5", "d0ed892a5e8d47e0cb5f8aa5f4dd7c5a"), ("crc32", "3f9f95a4"), ("md5", "4921d5f2165dee6e2496f4388c4c81b2"), ) def test_exports_have_catalog_hashes_and_queryable_sqlite(self): db, coverages, profiles, stats = self._fixtures() exports = generate_data_exports( self.docs, db, coverages, profiles, stats ) catalog = json.loads((self.docs / "api/v1/catalog.json").read_text()) self.assertEqual(catalog["count"], len(exports)) # Every reference resolves against the catalog's own URL, as an # RFC 3986 client resolves it: the url was relative to the site root # and the schema to api/v1/, so neither base resolved both. import urllib.parse # noqa: PLC0415 site = "https://example.org/site/" base = site + "api/v1/catalog.json" for item in catalog["items"]: for key in ("url", "schema"): if key in item: target = urllib.parse.urljoin(base, item[key]).removeprefix(site) self.assertTrue((self.docs / target).is_file(), (key, item[key])) path = self.docs / urllib.parse.urljoin(base, item["url"]).removeprefix(site) self.assertEqual(hashlib.sha256(path.read_bytes()).hexdigest(), item["sha256"]) with sqlite3.connect(self.docs / "downloads/retrobios.sqlite") as connection: # One lowercase row per accepted value: a comma-joined Recalbox # md5, a JSON array of accepted revisions and an uppercase crc32 # each missed an exact lookup. for algorithm, value in self.ACCEPTED: with self.subTest(algorithm=algorithm, value=value): self.assertTrue(connection.execute( "SELECT count(*) FROM accepted_hashes WHERE algorithm = ? AND value = ?", (algorithm, value)).fetchone()[0]) self.assertEqual(connection.execute("SELECT count(*) FROM files").fetchone()[0], 1) self.assertEqual(connection.execute("SELECT count(*) FROM platforms").fetchone()[0], 1) self.assertEqual(connection.execute("SELECT count(*) FROM emulators").fetchone()[0], 1) def test_page_metadata_is_specific_and_idempotent(self): page = self.docs / "gaps.md" page.write_text("# Gap Analysis - RetroBIOS\n\nCurrent unresolved entries.\n") decorate_markdown_pages(self.docs) decorate_markdown_pages(self.docs) content = page.read_text() self.assertEqual(content.count("type: application/ld+json"), 1) self.assertIn("Current RetroBIOS verification gaps", content) self.assertIn('"@type":"TechArticle"', content) class PerModeSourcePins(unittest.TestCase): """A profile whose builds live in separate repositories. `source`, `upstream` and `source_commit` may each be keyed by build mode. Reading the URL from one mode and the revision from another produces a permalink into a tree that never held that line, and binding the object form straight into SQLite fails outright. """ PROFILE = { "emulator": "twobuilds", "source": { "standalone": "https://github.com/vendor/emu", "libretro": "https://github.com/porter/emu-libretro", }, "upstream": "https://github.com/vendor/emu", "source_commit": {"standalone": "a" * 40, "libretro": "b" * 40}, } def test_each_repository_keeps_its_own_revision(self): pairs = { (repo.slug, pin) for repo, pin in _forge_sources(self.PROFILE) } self.assertIn(("vendor/emu", "a" * 40), pairs) self.assertIn(("porter/emu-libretro", "b" * 40), pairs) # The libretro fork must never be pinned to the standalone revision. self.assertNotIn(("porter/emu-libretro", "a" * 40), pairs) self.assertNotIn(("vendor/emu", "b" * 40), pairs) def test_a_build_mode_label_is_resolved_first(self): repo, pin = _forge_sources(self.PROFILE, "libretro")[0] self.assertEqual((repo.slug, pin), ("porter/emu-libretro", "b" * 40)) def test_the_plain_string_form_is_unchanged(self): profile = { "source": "https://github.com/vendor/emu", "source_commit": "c" * 40, } self.assertEqual( [(repo.slug, pin) for repo, pin in _forge_sources(profile)], [("vendor/emu", "c" * 40)], ) def test_sqlite_export_accepts_a_per_mode_commit(self): from generate_site import _json_text self.assertEqual( _json_text(self.PROFILE["source_commit"]), '{"libretro": "' + "b" * 40 + '", "standalone": "' + "a" * 40 + '"}', ) self.assertEqual(_json_text("c" * 40), "c" * 40) class AnEmulatorPageNamesItsRevision(unittest.TestCase): """The page linked the repositories unpinned. The commit every citation refers to sat in the YAML alone, for 178 profiles.""" def test_each_repository_links_its_pinned_commit(self): import generate_site # noqa: PLC0415 profile = { "emulator": "Caprice32", "type": "libretro", "source": "https://github.com/libretro/libretro-cap32", "upstream": "https://github.com/ColinPitrat/caprice32", "source_commit": "c" * 40, "upstream_commit": "6" * 40, "source_branch": "main", "systems": ["amstrad-cpc"], "files": [], } page = generate_site.generate_emulator_page("cap32", profile, {"files": {}}) self.assertIn( "| Revision | [libretro/libretro-cap32@cccccccccccc]" "(https://github.com/libretro/libretro-cap32/commit/" + "c" * 40 + "), " "[ColinPitrat/caprice32@666666666666]" "(https://github.com/ColinPitrat/caprice32/commit/" + "6" * 40 + "), " "branch `main` |", page, ) class APlatformPageSaysWhereEachFileGoes(unittest.TestCase): """Three GameCube IPL.bin were listed by name alone: nothing told which goes to GC/USA, GC/EUR or GC/JAP.""" def test_same_named_files_show_their_destinations(self): import generate_site # noqa: PLC0415 regions = ("USA", "EUR", "JAP") declared = [{"name": "IPL.bin", "destination": f"dolphin-emu/Sys/GC/{r}/IPL.bin", "sha1": str(i) * 40} for i, r in enumerate(regions)] files = [{"name": "IPL.bin", "status": "ok"} for _ in regions] lines = generate_site._render_platform_systems( {"nintendo-gamecube": files}, {("nintendo-gamecube", "IPL.bin"): declared}, {}) for region in regions: self.assertIn(f"- `IPL.bin` at `dolphin-emu/Sys/GC/{region}/IPL.bin` - OK", "\n".join(lines)) class AVariantShowsWhoPinsIt(unittest.TestCase): """A .variants/ file showed a name and a truncated MD5: the advision.zip Batocera and RetroBat pin named no platform.""" def test_the_variant_card_names_its_readers(self): import types # noqa: PLC0415 import generate_site # noqa: PLC0415 primary = {"name": "advision.zip", "path": "bios/Entex/Adventure Vision/advision.zip", "sha1": "a" * 40, "md5": "b" * 32, "size": 3000} variant = {"name": "advision.zip", "path": "bios/Entex/Adventure Vision/.variants/advision.zip.1c086064", "sha1": "c" * 40, "md5": "1c086064" + "0" * 24, "size": 3000, "provenance": {"tosec": {"name": "x"}}} consumers = types.SimpleNamespace( platforms={"c" * 40: {"batocera", "retrobat"}}, emulators={"a" * 40: {"mame"}}) page = generate_site.generate_system_page( "Entex", {"Adventure Vision": [primary, variant]}, consumers) # Another revision of a set sits open beside it (round 5, 6.3). block = page[page.index('id="sha1-' + "c" * 12 + '"'):] self.assertIn("
bios/Entex/Adventure Vision/"
".variants/advision.zip.1c086064", block)
self.assertIn('Batocera', block)
self.assertIn('RetroBat', block)
self.assertIn('{md5}', text)
class AMismatchNamesTheDumpTheCollectionHolds(unittest.TestCase):
"""The gaps table printed the hash an emulator accepts and stopped, while
the collection held that exact dump for every row."""
def test_the_held_copy_is_named(self):
from generate_site import _held_accepted
db = {"files": {"s1": {"path": "bios/Microsoft/MSX/CLK/msx-japanese.rom"}},
"indexes": {"by_sha256": {"ab" * 32: "s1"}, "by_crc32": {"84f68bf9": "s1"}}}
text = ("RetroBat says OK but ares says sha256 mismatch: got " + "cd" * 32
+ ", accepted [" + "ab" * 32 + "]")
self.assertEqual(_held_accepted(text, db), "bios/Microsoft/MSX/CLK/msx-japanese.rom")
self.assertEqual(_held_accepted("xroar says crc32 mismatch: got 1, accepted [84f68bf9]", db),
"bios/Microsoft/MSX/CLK/msx-japanese.rom")
self.assertEqual(_held_accepted("size mismatch: got 1, accepted [256]", db), "")
class ADataDirectoryFileCanBeLookedUp(unittest.TestCase):
"""2,534 hashes the packs ship from data directories were on no page:
the emulator page gave only a URL whose content moves."""
def test_a_member_resolves_to_its_emulator_page(self):
import json as _json
from generate_site import data_directory_members, file_location_items
with tempfile.TemporaryDirectory(dir=ROOT / "tmp") as tmp:
Path(tmp, "retroarch.json").write_text(_json.dumps({"data_directories": [
{"ref": "dolphin-sys", "files": {"GC/font.bin": {"sha1": "c" * 40, "size": 3}}},
]}))
members = data_directory_members(Path(tmp))
items = file_location_items(
{"files": {}}, {"dolphin": {"data_directories": [{"ref": "dolphin-sys"}]}}, members)
self.assertEqual(items, [{"sha1": "c" * 40,
"url": "/retrobios/emulators/dolphin/#sha1-cccccccccccc"}])
class ACardSaysWhyAFileIsLeftOut(unittest.TestCase):
"""The badge said "bundled" and nothing about which build, so a reader
could not tell why a RetroArch pack carried PPSSPP's assets."""
def test_the_build_and_the_platforms_are_named(self):
from generate_site import _file_badges
badges = " ".join(_file_badges({"name": "a", "bundled": True, "mode": "standalone"}, True))
self.assertIn("installed with the standalone emulator", badges)
class AnEmulatorPageOffersItsInstall(unittest.TestCase):
"""An emulator page linked each file one by one and offered no way to
install them together."""
def test_the_command_names_the_emulator(self):
from generate_site import _install_block
profile = yaml.safe_load((ROOT / "emulators" / "pcem.yml").read_text(encoding="utf-8"))
text = "\n".join(_install_block("pcem", profile))
self.assertIn("install.sh | sh -s -- --emulator pcem", text)
self.assertIn("install.ps1))) --emulator pcem", text)
self.assertEqual(_install_block("x", {"type": "alias"}), [])
def test_the_installer_takes_the_option(self):
source = (ROOT / "install.py").read_text(encoding="utf-8")
self.assertIn('"--emulator"', source)
class NestedProfileBlocksRenderAsLists(unittest.TestCase):
"""The clownmdemu page showed "- **external_required:** False" in a code
block: a key written as a paragraph left its children indented eight
spaces inside the admonition, which Markdown reads as code."""
def _html(self, value: object) -> str:
try:
import markdown
except ImportError:
self.skipTest("markdown is not installed")
from siterender import _render_yaml_value
# The page opens the block with ???+ (pymdownx.details). An admonition
# nests its body by the same four spaces, and needs nothing CI lacks.
lines = ['!!! abstract "Analysis"', ""]
_render_yaml_value(lines, value, 4)
return markdown.markdown("\n".join(lines), extensions=["admonition"])
def test_a_nested_mapping_is_a_nested_list(self):
html = self._html({"megacd_bios": {"external_required": False, "source_ref": "a.c:1"},
"tmss": {"supported": True}})
self.assertNotIn("", html)
self.assertNotIn("**", html)
self.assertIn("external required: no", html)
self.assertIn("a.c:1", html)
def test_platform_details_are_a_list(self):
import generate_site
lines = generate_site._render_platform_details(
{"platform_details": {"megacd": {"root": "x"}, "note": "y"}}
)
try:
import markdown
except ImportError:
self.skipTest("markdown is not installed")
# The block's body, read as the engine reads Markdown inside HTML.
body = lines[lines.index("Platform details ") + 1:
lines.index("")]
html = markdown.markdown("\n".join(body))
self.assertIn("", html)
self.assertIn("name: a.bin, size: 2", html)
class TheOneLinerIsWhereUsersLand(unittest.TestCase):
"""The download page had no Android line and no Windows form with
options, and the platform and system pages offered only ZIP parts."""
def test_download_page(self):
import generate_site # noqa: PLC0415
page = generate_site.generate_which_pack()
self.assertIn("Android (Termux", page)
self.assertIn("install.ps1))) --check", page)
self.assertIn("--platform retroarch --dest", page)
def test_platform_and_system_pages(self):
import generate_site # noqa: PLC0415
from cross_reference import FileConsumers # noqa: PLC0415
cov = {"platform": "Batocera", "mode": "md5", "present": 0, "total": 0,
"verified": 0, "missing": 0, "details": [], "config": {"systems": {}}}
page = generate_site.generate_platform_page("batocera", cov)
self.assertIn("install.sh | sh -s -- --platform batocera", page)
# Batocera runs the command on the device itself, so its page opens on
# the computer route. A platform detected on Windows gets the
# PowerShell form, and its command comes before the ZIP.
cov = dict(cov, platform="RetroArch", mode="existence")
page = generate_site.generate_platform_page("retroarch", cov)
self.assertIn("install.ps1))) --platform retroarch", page)
self.assertIn("install.sh | sh -s -- --platform retroarch", page)
self.assertLess(page.index("--platform retroarch"), page.index("rb-button"))
entry = {"name": "a.bin", "sha1": "a" * 40, "md5": "b" * 32, "size": 3,
"path": "bios/Recalbox/MSX/a.bin"}
page = generate_site.generate_system_page(
"Microsoft", {"MSX": [entry]}, FileConsumers({"files": {}}),
{"MSX": [("recalbox", "msx1,msx2,msxturbor"), ("batocera", "msx1")]})
# The whole system id, up to the end of the command.
self.assertRegex(page, r"--platform recalbox --system msx1,msx2,msxturbor(\n|)")
self.assertIn("--platform batocera --system msx1", page)
def test_a_system_is_offered_only_where_the_manifest_records_it(self):
import generate_site # noqa: PLC0415
import siterender # noqa: PLC0415
saved = dict(siterender._system_page_map)
self.addCleanup(siterender.set_system_page_map, saved)
siterender.set_system_page_map({"sony-playstation": ("sony", "PlayStation"),
"nowhere": ("sony", "PlayStation")})
cov = {"retroarch": {"config": {"systems": {"sony-playstation": {}, "nowhere": {}}}}}
choices = generate_site.system_install_choices(cov)
self.assertEqual(choices["sony"]["PlayStation"], [("retroarch", "sony-playstation")])
class TheGapPageOpensOnCounts(unittest.TestCase):
"""The page opened on a slogan and on tiles reading "Verified (100%)"
for 5,437 of 5,438 files and "Missing 0", above a table of red cells."""
def test_every_count_is_said(self):
import generate_site # noqa: PLC0415
text = generate_site._platform_lists_sentence(5438, 5437, 1, 0)
self.assertIn("Of the 5,438 files", text)
self.assertIn("5,437 are held", text)
self.assertIn("1 is held but not verified", text)
self.assertIn("0 are missing", text)
self.assertNotIn("%", text)
source = (Path(__file__).resolve().parents[1] / "scripts" / "generate_site.py").read_text()
self.assertNotIn("Unified view of BIOS verification", source)
class WhichPackDescribesNoPlatformByOpinion(unittest.TestCase):
"""The page called RetroPie the largest community with the most guides,
Batocera an easy setup and Recalbox good for a first build: claims
nothing in the repository measures, ranking platforms against each
other."""
def test_no_ranking_words(self):
import re
import generate_site # noqa: PLC0415
page = generate_site.generate_which_pack()
rows = [line for line in page.splitlines() if line.startswith("| ")]
found = [
row for row in rows
if re.search(r"\b(largest|most|popular|easy|best|classic|lightweight|"
r"one-click|plug-and-play|first build|accuracy-focused)\b",
row, re.IGNORECASE)
]
self.assertEqual(found, [])
class EveryPageLinksItsOwnData(unittest.TestCase):
"""An emulator page linked its YAML and JSON. A platform page linked
nothing, and a tool had to download platforms.json (2.8 MB) and filter
it. A system page linked nothing either."""
def test_platform_and_system_documents(self):
import generate_site # noqa: PLC0415
from cross_reference import FileConsumers # noqa: PLC0415
cov = {"platform": "Batocera", "mode": "md5", "present": 0, "total": 0,
"verified": 0, "missing": 0, "details": [], "config": {"systems": {}}}
page = generate_site.generate_platform_page("batocera", cov)
self.assertIn("[JSON](../api/v1/platforms/batocera.json)", page)
self.assertIn("/blob/main/platforms/batocera.yml)", page)
entry = {"name": "a.bin", "sha1": "a" * 40, "md5": "b" * 32, "size": 3,
"path": "bios/Sony/PlayStation/a.bin"}
page = generate_site.generate_system_page(
"Sony", {"PlayStation": [entry]}, FileConsumers({"files": {}}))
self.assertIn(" as JSON](../api/v1/systems/sony.json)", page)
item = generate_site.system_export_item("sony", "Sony", {"PlayStation": [entry]})
self.assertEqual(item["consoles"]["PlayStation"][0]["sha1"], "a" * 40)
with tempfile.TemporaryDirectory(dir=Path(__file__).resolve().parents[1] / "tmp") as tmp:
path = Path(tmp) / "sony.json"
generate_site.write_single_item(path, "systems", item, "2026-10-21T00:00:00Z")
document = json.loads(path.read_text())
self.assertEqual(document["count"], 1)
try:
import jsonschema # noqa: PLC0415
except ImportError:
return
root = Path(__file__).resolve().parents[1]
envelope = json.loads((root / "schemas" / "site-api-envelope.schema.json").read_text())
jsonschema.validate(document, envelope)
if __name__ == "__main__":
unittest.main()