chore: merge audit fixes (site)

# Conflicts:
#	wiki/testing-guide.md
This commit is contained in:
Abdessamad Derraz committed 2026-10-21 05:53:15 +02:00
commit 734643b8fc
12 files changed
+1000 -147

No files matched your search

+4 -3
View File
@@ -4175,8 +4175,9 @@ class AFileCardSaysAllItKnows(unittest.TestCase):
text = "\n".join(_render_emulator_file(
entry, {"files": [entry]}, None, [entry], lambda f: True, "file-neogeo-zip",
))
self.assertIn('id="file-neogeo-zip"', text)
self.assertIn('href="#file-neogeo-zip"', text)
# The card is a heading with this id, linked by the heading's own
# permalink.
self.assertIn("{ #file-neogeo-zip .rb-file-title }", text)
self.assertIn("Standalone path: `Firmware/neogeo.zip`", text)
self.assertIn("`m11.bin`", text)
self.assertIn("CRC32 `0000000b`", text)
@@ -4894,7 +4895,7 @@ class VersionAnchorsFollowContent(unittest.TestCase):
"raze", {"emulator": "Raze", "type": "standalone", "files": files},
{"files": {}, "indexes": {}},
)
return re.findall(r'id="(file-[^"]+)"', page)
return re.findall(r"\{ #(file-[^ ]+) \.rb-file-title \}", page)
def test_inserting_a_version_keeps_the_others(self):
a = {"name": "DUKE3D.GRP", "path": "DUKE3D.GRP", "crc32": "bbc9ce44"}
+25
View File
@@ -44,5 +44,30 @@ class SiteCoreCounts(unittest.TestCase):
self.assertEqual(shown[cov["platform"]], expected)
class AllLibretroLabel(unittest.TestCase):
"""RetroPie and RetroDECK pull in every libretro core by naming retroarch
in a list, and the label read "(all libretro)" for RetroArch and Lakka
only, the two whose cores: is the string."""
PROFILES = {
"a": {"type": "libretro"},
"b": {"type": "standalone + libretro"},
"sa": {"type": "standalone"},
}
def test_a_list_naming_retroarch_is_labelled(self):
note = generate_site._core_scope_note(
{"cores": ["retroarch", "sa"]}, self.PROFILES)
self.assertEqual(note, " (all libretro, plus 1 other core)")
def test_the_string_form_is_labelled(self):
note = generate_site._core_scope_note({"cores": "all_libretro"}, self.PROFILES)
self.assertEqual(note, " (all libretro)")
def test_a_selection_is_not(self):
self.assertEqual(
generate_site._core_scope_note({"cores": ["a", "sa"]}, self.PROFILES), "")
if __name__ == "__main__":
unittest.main()
+415
View File
@@ -0,0 +1,415 @@
"""Renderings shared by the site pages, and the icon cache they read.
No network: icon checks are answered by a stub.
"""
from __future__ import annotations
import json
import sys
import tempfile
import unittest
import urllib.error
from pathlib import Path
from unittest import mock
REPO_ROOT = Path(__file__).resolve().parents[1]
sys.path.insert(0, str(REPO_ROOT / "scripts"))
import siterender # noqa: E402
class _Served:
status = 200
def __enter__(self) -> "_Served":
return self
def __exit__(self, *exc: object) -> None:
return None
class IconCacheIsReplacedWhole(unittest.TestCase):
"""The cache was truncated then dumped in place: a concurrent reader saw
half a file, took it for empty, and an offline run then wrote ``{}``
over every recorded icon."""
def setUp(self) -> None:
tmp = tempfile.TemporaryDirectory()
self.addCleanup(tmp.cleanup)
self.path = Path(tmp.name) / "system_icons.json"
patcher = mock.patch.object(siterender, "ICON_CACHE_PATH", self.path)
patcher.start()
self.addCleanup(patcher.stop)
saved = dict(siterender._icon_available)
self.addCleanup(lambda: (siterender._icon_available.clear(),
siterender._icon_available.update(saved)))
def test_an_offline_run_leaves_an_unreadable_cache_untouched(self):
self.path.write_text('{"Sega - Mega Drive": tr')
offline = mock.patch.object(
siterender.urllib.request, "urlopen",
side_effect=urllib.error.URLError("offline"))
with offline, mock.patch("sys.stderr"), mock.patch("builtins.print"):
siterender.prime_system_icons({"Sega - Mega Drive"})
self.assertEqual(self.path.read_text(), '{"Sega - Mega Drive": tr')
def test_names_another_run_recorded_meanwhile_are_kept(self):
self.path.write_text(json.dumps({"A - One": True}))
def answer(*_args: object, **_kwargs: object) -> _Served:
# A second build records its own check while this one runs.
self.path.write_text(json.dumps({"A - One": True, "C - Three": False}))
return _Served()
with mock.patch.object(siterender.urllib.request, "urlopen", side_effect=answer), \
mock.patch("builtins.print"):
siterender.prime_system_icons({"A - One", "B - Two"})
self.assertEqual(json.loads(self.path.read_text()),
{"A - One": True, "B - Two": True, "C - Three": False})
self.assertEqual(sorted(p.name for p in self.path.parent.iterdir()),
["system_icons.json"])
def test_the_write_goes_through_a_scratch_file(self):
with mock.patch.object(siterender.urllib.request, "urlopen",
return_value=_Served()), \
mock.patch.object(siterender, "write_text_atomic",
wraps=siterender.write_text_atomic) as atomic, \
mock.patch("builtins.print"):
siterender.prime_system_icons({"B - Two"})
atomic.assert_called_once()
class EveryPageAddressesPlatformsAsPeers(unittest.TestCase):
"""The exports page was held to neutral wording, and the gaps page next
to it still called each platform's file a "corrected list". Reads the
text every site generator writes, docstrings and bare identifiers aside."""
WORDS = ("their file", "theirs", "of ours", "corrected", "left out", "drift")
SOURCES = ("generate_site.py", "siterender.py", "sitesources.py", "siteexports.py")
def test_no_page_text_judges_another_project(self):
import ast
import re
found = []
for name in self.SOURCES:
tree = ast.parse((REPO_ROOT / "scripts" / name).read_text(encoding="utf-8"))
docstrings = {
id(node.body[0].value) for node in ast.walk(tree)
if isinstance(node, (ast.Module, ast.ClassDef, ast.FunctionDef,
ast.AsyncFunctionDef))
and ast.get_docstring(node, clean=False) is not None
}
for node in ast.walk(tree):
if (not isinstance(node, ast.Constant) or not isinstance(node.value, str)
or id(node) in docstrings or re.fullmatch(r"\w+", node.value)):
continue
text = node.value.lower()
found += [f"{name}:{node.lineno} {word!r}"
for word in self.WORDS if word in text]
self.assertEqual(found, [])
class BundledFilesAreNotToSource(unittest.TestCase):
"""The gaps page summed the report's gap_missing, bundled files included:
1,356 files to source where 609 were, and a red badge of 13 on ZQuest
Classic, whose 13 missing files all come with its own package."""
def test_the_summary_and_the_row_count_only_what_must_be_found(self):
import generate_site
gap = {"required": True, "source_ref": "x", "in_repo": False}
report = {"zq": {
"emulator": "ZQuest Classic", "gaps": 4, "gap_missing": 3, "gap_bios": 1,
"unsourceable": [],
"gap_details": [
{**gap, "name": "a.dll", "source": "missing", "bundled": True},
{**gap, "name": "b.dll", "source": "missing", "bundled": True},
{**gap, "name": "c.bin", "source": "missing"},
{**gap, "name": "d.bin", "source": "bios", "in_repo": True},
],
}}
page = generate_site.generate_gap_analysis(
{"zq": {"emulator": "ZQuest Classic", "files": []}}, {},
{"files": {}, "indexes": {}}, gap_report=report)
self.assertIn("2 installed with their emulator, 1 to source.", page)
self.assertIn("| missing | 1 | Not available, needs sourcing |", page)
self.assertIn("| [ZQuest Classic](emulators/zq.md) | 4 | 1 | 0 | 0 | 2 | "
'<span class="rb-badge rb-badge-danger">1</span> |', page)
class StructuredBlocks(unittest.TestCase):
PIN = "1e0de94dc7e669c0925a22c0fccf6cdc837af0a0"
def profile(self, block: str, value: object) -> dict:
return {"source": "https://github.com/libretro/blastem",
"source_commit": self.PIN, block: value}
def render(self, profile: dict) -> str:
import generate_site
return "\n".join(generate_site._render_structured_blocks(profile))
def test_a_source_ref_in_a_block_is_a_pinned_link(self):
"""186 citations in analysis and firmware blocks were plain code."""
page = self.render(self.profile("analysis", {"tmss": {
"source_ref": "libblastem.c:355-359"}}))
self.assertIn(
f"[`libblastem.c:355-359`](https://github.com/libretro/blastem/blob/"
f"{self.PIN}/libblastem.c#L355-L359)", page)
def test_a_list_inside_a_list_item_is_shown(self):
"""cemu's firmware titles lost the files each title carries."""
page = self.render(self.profile("firmware_titles", [
{"id": "0005001b-10042400", "name": "Shared fonts",
"content": ["CafeCn.ttf", "CafeKr.ttf"]}]))
self.assertIn(" - id: 0005001b-10042400, name: Shared fonts\n", page)
self.assertIn(" - **content:**", page)
self.assertIn(" - CafeKr.ttf", page)
def test_a_folded_note_stays_inside_its_block(self):
page = self.render(self.profile("analysis", {"tmss": {
"note": "The core points\nsystem/tmss_path at the system directory."}}))
self.assertIn("- **note:** The core points system/tmss_path at the", page)
body = page.split("\n")[1:]
self.assertEqual([line for line in body if line and not line.startswith(" ")], [])
def test_a_long_note_is_not_cut(self):
note = "word " * 80 + "end."
page = self.render(self.profile("analysis", {"x": {"note": note}}))
self.assertIn("end.", page)
class FileCardsLinkTheReadingEntry(unittest.TestCase):
def test_an_emulator_link_lands_on_its_entry(self):
"""The platform links landed on their section, the emulator links on
the top of a page of a hundred entries."""
import generate_site
from cross_reference import FileConsumers
consumers = FileConsumers({"files": {}})
files = [{"name": "scph1001.bin"}, {"name": "scph5501.bin"},
{"name": "tmss.md", "sha1": "d0fd0c16" + "0" * 32},
{"name": "tmss.md", "sha1": "aa" * 20}]
consumers.emulators = {"s": {"duckstation", "other"}, "t": {"blastem"}}
consumers.emulator_entries = {"s": {"duckstation": 1}, "t": {"blastem": 2}}
consumers.emulator_files = {"duckstation": files, "blastem": files}
self.assertEqual(generate_site._emulator_links("s", consumers), [
"[duckstation](../emulators/duckstation.md#file-scph5501-bin)",
"[other](../emulators/other.md)",
])
self.assertEqual(generate_site._emulator_links("t", consumers),
["[blastem](../emulators/blastem.md#file-tmss-md-d0fd0c16)"])
class DirectoryMembersNameTheirEmulator(unittest.TestCase):
"""The site resolved a ``type: directory`` entry by its own name: 11,976
files under alephone's Plugins/, cdogs' and thextech's trees had no
emulator on their card, and the directory card linked none of them."""
SHA1 = "c63078a6c9ccaca7339a726a5d39410983933070"
@classmethod
def setUpClass(cls) -> None:
from common import build_zip_contents_index, load_database, load_emulator_profiles
from cross_reference import FileConsumers
cls.db = load_database(str(REPO_ROOT / "database.json"))
if cls.SHA1 not in cls.db.get("files", {}):
raise unittest.SkipTest("the collection does not hold the alephone plugin")
cls.profile = load_emulator_profiles(str(REPO_ROOT / "emulators"))["alephone"]
cls.consumers = FileConsumers(cls.db, build_zip_contents_index(cls.db))
cls.consumers.add_emulators({"alephone": cls.profile})
def test_a_member_names_the_emulator_and_its_directory_entry(self):
import generate_site
files = self.profile["files"]
index = self.consumers.emulator_entries[self.SHA1]["alephone"]
self.assertEqual(files[index].get("type"), "directory")
self.assertEqual(generate_site._emulator_links(self.SHA1, self.consumers),
["[alephone](../emulators/alephone.md#file-marathon-plugins)"])
def test_the_directory_card_points_at_its_copies(self):
import generate_site
entry = next(f for f in self.profile["files"]
if f.get("type") == "directory" and f.get("path") == "Marathon/Plugins/")
line = generate_site._directory_copies(entry, self.consumers)[0]
self.assertRegex(line, r"^Collection copies: \d+ files under "
r"\[`bios/Bungie/Marathon/Plugins/`\]\(\.\./systems/bungie\.md#")
class AnAliasTreeIsListedOnItsParentsPage(unittest.TestCase):
"""The 90 FreeIntv overlays were linked to the page of the alias that
declares them, which is never generated: every permalink was a 404."""
PROFILES = {
"freeintv": {"type": "libretro"},
"freeintv_ts_overlay": {
"type": "alias", "alias_of": "freeintv",
"data_directories": [{"ref": "freeintv-overlays",
"destination": "freeintv_overlays"}],
},
}
def test_the_permalink_goes_to_a_page_that_exists(self):
import generate_site
members = {"freeintv-overlays": {"a.png": {"sha1": "02" * 20}}}
items = generate_site.file_location_items({"files": {}}, self.PROFILES, members)
self.assertEqual(items, [{"sha1": "02" * 20,
"url": "emulators/freeintv/#sha1-020202020202"}])
def test_the_parent_page_lists_the_tree(self):
import generate_site
trees = generate_site.page_data_directories("freeintv", self.PROFILES)
self.assertEqual([t["ref"] for t in trees], ["freeintv-overlays"])
class ForgesUpstreamDoesNotRead(unittest.TestCase):
"""Profiles pinned on SourceForge, voidpoint.io, Bitbucket or Software
Heritage showed no revision, and their citations were plain code."""
def links(self, url: str, ref: str, **extra: str) -> tuple[str, str]:
import sitesources
profile = {"source": url, "source_commit": "f6c00044" + "0" * 32, **extra}
return (sitesources.revision_markdown(profile),
sitesources._source_ref_markdown(profile, ref))
def test_bitbucket(self):
revision, ref = self.links("https://bitbucket.org/linuxwolf6/splitwolf",
"id_lwmp.cpp:1831-1868", source_branch="scrubbed")
pin = "f6c00044" + "0" * 32
self.assertEqual(revision, f"[linuxwolf6/splitwolf@{pin[:12]}]"
f"(https://bitbucket.org/linuxwolf6/splitwolf/commits/{pin}), "
"branch `scrubbed`")
self.assertIn(f"(https://bitbucket.org/linuxwolf6/splitwolf/src/{pin}/"
"id_lwmp.cpp#lines-1831:1868)", ref)
def test_a_gitlab_instance(self):
_revision, ref = self.links("https://voidpoint.io/terminx/eduke32",
"source/duke3d/src/common.cpp:36")
self.assertIn("https://voidpoint.io/terminx/eduke32/-/blob/f6c00044", ref)
self.assertIn("/source/duke3d/src/common.cpp#L36)", ref)
def test_sourceforge(self):
revision, ref = self.links("https://sourceforge.net/p/dgen/dgen", "rc.cpp:12")
self.assertIn("(https://sourceforge.net/p/dgen/dgen/ci/f6c00044", revision)
self.assertRegex(ref, r"\(https://sourceforge\.net/p/dgen/dgen/ci/f6c00044\w+/tree/rc\.cpp\)")
def test_software_heritage(self):
revision, ref = self.links(
"https://archive.softwareheritage.org/browse/revision/f6c00044/",
"app/src/os.cpp:35-36")
self.assertTrue(revision.startswith("[Software Heritage@f6c000440000]"))
self.assertIn("/browse/revision/f6c00044" + "0" * 32 + "/?path=app/src/os.cpp#L35-L36)", ref)
class TheDatHoldsEveryCopy(unittest.TestCase):
"""The DAT held one path per content, the database's: auditing bios/
with it flagged the 2,502 other copies (RPG Maker, engine trees, a dump
filed under two consoles) as unknown, and a fix moved them away."""
def test_a_second_copy_is_a_rom_of_its_own_folder(self):
import generate_site
sha1 = "9526ef63" + "0" * 32
with tempfile.TemporaryDirectory() as tmp:
for rel in ("Atari/ST/tos104.img", "Atari/ST/tos104uk.img"):
(Path(tmp) / rel).parent.mkdir(parents=True, exist_ok=True)
(Path(tmp) / rel).write_bytes(b"x")
db = {
"bios_dir": tmp,
"files": {sha1: {"path": "bios/Atari/ST/tos104.img", "sha1": sha1,
"size": 1, "name": "tos104.img"}},
"indexes": {"by_path_suffix": {
"Atari/ST/tos104.img": [sha1], "ST/tos104.img": [sha1],
"Atari/ST/tos104uk.img": [sha1], "ST/tos104uk.img": [sha1],
}},
}
dat = generate_site.collection_dat(db, "2026-10-21T00:00:00Z")
self.assertIn('<rom name="tos104.img" size="1"', dat)
self.assertIn('<rom name="tos104uk.img" size="1"', dat)
self.assertEqual(dat.count("<rom "), 2)
class TheSearchLandsOnTheCard(unittest.TestCase):
"""The site search splits a page at its headings. Emulator cards were
titled in bold and platform systems folded into blocks under one
heading, so a pasted hash led to the page top or to the systems
overview of a 100-block page."""
def test_a_platform_system_is_a_heading(self):
import generate_site
by_system = {f"system-{i:02d}": [{"name": f"f{i}.bin", "status": "ok"}]
for i in range(11)}
lines = generate_site._render_platform_systems(by_system, {}, {})
headings = [line for line in lines if line.startswith("### ")]
self.assertEqual(len(headings), 11)
self.assertTrue(headings[0].endswith("{ #system-00 }"))
self.assertNotIn('<a id="system-00"></a>', lines)
self.assertIn('??? note "1 file (OK)"', lines)
def test_an_emulator_card_is_a_heading_with_its_anchor(self):
import generate_site
entry = {"name": "scph5501.bin", "required": True}
lines = generate_site._render_emulator_file(
entry, {"files": [entry]}, None, [entry], lambda f: True, "file-scph5501-bin")
self.assertTrue(lines[2].startswith("#### `scph5501.bin` "))
self.assertTrue(lines[2].endswith("{ #file-scph5501-bin .rb-file-title }"))
def test_a_system_card_is_a_heading_with_its_anchor(self):
import generate_site
from cross_reference import FileConsumers
entry = {"name": "a.bin", "sha1": "a" * 40, "md5": "b" * 32, "size": 3}
lines = generate_site._system_file_card(entry, FileConsumers({"files": {}}))
self.assertTrue(lines[2].startswith("#### `a.bin` "))
self.assertTrue(lines[2].endswith("{ #sha1-aaaaaaaaaaaa .rb-file-title }"))
class TheSearchSkipsLinkLists(unittest.TestCase):
"""The search index was 24.7 MB, read and indexed before the first
result. Lists of platforms, copies and source citations on every card
made 6.6 MB of it, names already indexed on their own pages."""
def test_a_system_card_keeps_its_hashes_and_skips_its_platforms(self):
import generate_site
from cross_reference import FileConsumers
consumers = FileConsumers({"files": {}})
sha1 = "a" * 40
consumers.platforms = {sha1: {"retroarch"}}
consumers.installed = {sha1: {"batocera": {"bios/a.bin"}}}
consumers.emulators = {sha1: {"duckstation"}}
entry = {"name": "a.bin", "sha1": sha1, "md5": "b" * 32, "size": 3}
text = "\n".join(generate_site._system_file_card(entry, consumers))
searched, skipped = text.split('<div class="rb-search-skip" data-search-exclude markdown>')
self.assertIn("- MD5:", searched)
self.assertIn("- Emulators:", searched)
self.assertIn("- Platforms:", skipped)
self.assertIn("- Also installed by:", skipped)
self.assertEqual(text.count("</div>"), 2)
def test_runs_alternate_in_card_order(self):
import generate_site
lines = ["- SHA1: x", "- Source: `a.c:1`", "- Platforms: p",
"- Destination: `d`", "- Contents (1 members):", " - `m`"]
out = [line for line in generate_site._search_skipped_apart(lines) if line]
self.assertEqual(out, [
"- SHA1: x",
'<div class="rb-search-skip" data-search-exclude markdown>',
"- Source: `a.c:1`", "- Platforms: p", "</div>",
"- Destination: `d`", "- Contents (1 members):", " - `m`",
])
if __name__ == "__main__":
unittest.main()
+23 -14
View File
@@ -69,31 +69,40 @@ class OtherSpellingsOfOneSystem(unittest.TestCase):
self.assertEqual(gs._system_emulators("sega-mega-drive", index), ["genesis_plus_gx"])
self.assertEqual(gs._system_emulators("nintendo-snes", index), [])
def _install(self, mapping: dict) -> None:
import siterender
saved = dict(siterender._system_page_map)
self.addCleanup(siterender.set_system_page_map, saved)
siterender.set_system_page_map(mapping)
def test_a_profile_spelling_links_to_the_platform_page(self):
import siterender
saved = dict(siterender._system_page_map)
self.addCleanup(lambda: (siterender._system_page_map.clear(),
siterender._system_page_map.update(saved)))
siterender._system_page_map.clear()
siterender._system_page_map["sega-mega-drive"] = ("sega", "Mega Drive")
self._install({"sega-mega-drive": ("sega", "Mega Drive")})
self.assertIn("systems/sega.md#mega-drive", siterender._system_link("sega-megadrive"))
siterender._system_page_map["megadrive"] = ("other", "Elsewhere")
self._install({"sega-mega-drive": ("sega", "Mega Drive"),
"megadrive": ("other", "Elsewhere")})
self.assertEqual(siterender._system_link("sega-megadrive"), "sega-megadrive")
def test_a_system_only_profiles_name_is_placed_by_their_files(self):
def test_a_table_of_the_same_size_replaces_the_other_spellings(self):
"""The index of other spellings was kept while the table kept its
length, so a second table of one entry linked to the first one's page."""
import siterender
saved = dict(siterender._system_page_map)
self.addCleanup(lambda: (siterender._system_page_map.clear(),
siterender._system_page_map.update(saved)))
siterender._system_page_map.clear()
self._install({"sega-mega-drive": ("sega", "Mega Drive")})
self.assertIn("systems/sega.md#", siterender._system_link("sega-megadrive"))
self._install({"sega-mega-drive": ("other", "Elsewhere")})
self.assertEqual(siterender._system_link("sega-megadrive"),
"[sega-megadrive](systems/other.md#elsewhere)")
def test_a_system_only_profiles_name_is_placed_by_their_files(self):
manufacturers = {"Commodore": {"PET": [{"name": "basic4.bin"}]}}
profiles = {"vice_xpet": {"systems": ["commodore-pet"],
"files": [{"name": "basic4.bin"}]}}
gs._build_system_page_map_from_data(
pages = gs._system_pages_from_data(
manufacturers, {}, {"files": {}, "indexes": {}}, profiles)
self.assertEqual(siterender._system_page_map["commodore-pet"], ("commodore", "PET"))
self.assertEqual(pages["commodore-pet"], ("commodore", "PET"))
class ACopyIsOnItsOwnFoldersPage(unittest.TestCase):
@@ -134,7 +143,7 @@ class AnIdNamesOneCardPerPage(unittest.TestCase):
"path": "bios/Nintendo/DS/a.bin"}
page = gs.generate_system_page(
"Nintendo", {"DS": [entry], "DSi": [entry]}, FileConsumers({"files": {}}))
self.assertEqual(page.count('id="sha1-aaaaaaaaaaaa"'), 1)
self.assertEqual(page.count("#sha1-aaaaaaaaaaaa .rb-file-title"), 1)
self.assertEqual(page.count("rb-sys-file"), 2)