mirror of
https://github.com/Abdess/retroarch_system.git
synced 2026-10-20 10:23:16 -05:00
Abdessamad Derraz
4231902a6c
Profiles, the wiki, comments, docstrings, command output and schema descriptions carried em-dashes, clause-joining semicolons, unicode arrows, box-drawing banners and check-mark glyphs. Each sentence was rewritten rather than having its character swapped. Hashes, paths, source_ref values and every cited file:line are unchanged, which a comparison of each parsed profile against the previous revision confirms. validate_pr comments now say PASS and FAIL in words, and source_ref values in a profile's structured blocks render as code. test_prose_style fails if any of these comes back, in tracked files or in the generated site.
241 lines
9.6 KiB
Python
241 lines
9.6 KiB
Python
#!/usr/bin/env python3
|
|
"""Rename detection in the truth-vs-platform diff.
|
|
|
|
A platform may call a file whatever it likes: Batocera ships the same Sega CD
|
|
ROM as ROM1. A name that matches nothing in the profile is therefore not yet a
|
|
gap, and reporting it as both a missing file and an extra one would invent a
|
|
discrepancy that does not exist. Content decides, as everywhere else here.
|
|
"""
|
|
|
|
from __future__ import annotations
|
|
|
|
import pathlib
|
|
import subprocess
|
|
import sys
|
|
import tempfile
|
|
import unittest
|
|
from pathlib import Path
|
|
|
|
REPO_ROOT = Path(__file__).resolve().parent.parent
|
|
sys.path.insert(0, str(REPO_ROOT / "scripts"))
|
|
|
|
from exporter.baseline import pair_truth_entry # noqa: E402
|
|
from truth import ( # noqa: E402
|
|
_diff_system,
|
|
_merge_file_into_system,
|
|
generate_platform_truth,
|
|
)
|
|
|
|
|
|
def _entry(name: str, **hashes) -> dict:
|
|
return {"name": name, **hashes}
|
|
|
|
|
|
class ATargetedModelIsItsOwnArtifact(unittest.TestCase):
|
|
"""A narrowed truth model must not take the full model's place.
|
|
|
|
generate_truth wrote dist/truth/<platform>.yml whatever --target said, so
|
|
a targeted run overwrote the full model and diff_truth then compared a
|
|
narrowed model against the whole scrape.
|
|
"""
|
|
|
|
def test_a_target_writes_beside_the_full_model(self):
|
|
repo = pathlib.Path(__file__).resolve().parent.parent
|
|
with tempfile.TemporaryDirectory(dir=str(repo / "tmp")) as directory:
|
|
for extra in ([], ["--target", "browser"]):
|
|
proc = subprocess.run(
|
|
[sys.executable, "scripts/generate_truth.py", "--platform",
|
|
"romm", *extra, "--output-dir", directory],
|
|
capture_output=True, check=False, text=True, cwd=str(repo), timeout=400,
|
|
)
|
|
self.assertEqual(proc.returncode, 0, proc.stdout + proc.stderr)
|
|
produced = sorted(
|
|
str(p.relative_to(directory))
|
|
for p in pathlib.Path(directory).rglob("*.yml")
|
|
)
|
|
self.assertEqual(
|
|
produced,
|
|
["browser/romm.yml", "romm.yml"],
|
|
"the targeted model did not land beside the full one",
|
|
)
|
|
|
|
|
|
class ADiffThatComparedNothingFails(unittest.TestCase):
|
|
"""diff_truth --all skipped every platform without a model and exited 0."""
|
|
|
|
def test_an_empty_truth_directory_is_an_error(self):
|
|
repo = pathlib.Path(__file__).resolve().parent.parent
|
|
with tempfile.TemporaryDirectory(dir=str(repo / "tmp")) as directory:
|
|
for selection in (["--all"], ["--platform", "retroarch"]):
|
|
proc = subprocess.run(
|
|
[sys.executable, "scripts/diff_truth.py", *selection,
|
|
"--truth-dir", directory, "--json"],
|
|
capture_output=True, check=False, text=True, cwd=str(repo), timeout=120,
|
|
)
|
|
self.assertEqual(proc.returncode, 1, selection)
|
|
self.assertIn("skip", proc.stderr)
|
|
|
|
|
|
class OneNameSeveralFiles(unittest.TestCase):
|
|
"""Dolphin's three IPL.bin differ only by path, and the model kept one."""
|
|
|
|
REGIONS = {"USA": "a" * 32, "EUR": "b" * 32, "JAP": "c" * 32}
|
|
|
|
def _system(self) -> dict:
|
|
system: dict = {}
|
|
for core in ("dolphin", "ishiiruka"):
|
|
for region, md5 in self.REGIONS.items():
|
|
_merge_file_into_system(
|
|
system,
|
|
{"name": "IPL.bin", "path": f"GC/{region}/IPL.bin", "md5": md5},
|
|
core,
|
|
None,
|
|
)
|
|
_merge_file_into_system(system, {"name": "IPL.bin"}, "other", None)
|
|
return system
|
|
|
|
def test_a_core_declaring_a_name_twice_declares_two_files(self):
|
|
files = self._system()["files"]
|
|
self.assertEqual(
|
|
sorted(f["path"] for f in files),
|
|
["GC/EUR/IPL.bin", "GC/JAP/IPL.bin", "GC/USA/IPL.bin"],
|
|
)
|
|
for entry in files:
|
|
self.assertEqual(entry["_cores"] - {"other"}, {"dolphin", "ishiiruka"})
|
|
|
|
def test_revisions_under_one_name_and_no_path_stay_one_file(self):
|
|
system: dict = {}
|
|
for md5 in ("a" * 32, "b" * 32):
|
|
_merge_file_into_system(
|
|
system, {"name": "MT32_CONTROL.ROM", "md5": md5}, "dosbox", None
|
|
)
|
|
self.assertEqual(len(system["files"]), 1)
|
|
|
|
def test_the_diff_pairs_each_by_destination(self):
|
|
scraped = {"files": [
|
|
{"name": "IPL.bin", "destination": f"dolphin-emu/Sys/GC/{region}/IPL.bin",
|
|
"md5": md5}
|
|
for region, md5 in self.REGIONS.items()
|
|
]}
|
|
self.assertEqual(_diff_system(self._system(), scraped), {})
|
|
|
|
|
|
class DifferentContentsStayApart(unittest.TestCase):
|
|
def test_two_cores_same_name_different_contents(self):
|
|
"""beebem's BBC BIOS.rom and np2kai's PC-98 bios.rom merged into one."""
|
|
system: dict = {}
|
|
_merge_file_into_system(system, {"name": "bios.rom", "path": "np2kai/bios.rom",
|
|
"md5": "a" * 32}, "np2kai", None)
|
|
_merge_file_into_system(system, {"name": "bios.rom", "path": "BeebFile/BIOS.rom",
|
|
"md5": "b" * 32, "size": 16384}, "beebem", None)
|
|
self.assertEqual(len(system["files"]), 2)
|
|
|
|
|
|
class RenameMatching(unittest.TestCase):
|
|
"""A platform may call a file anything (Batocera ships a ROM as ROM1):
|
|
a shared hash pairs the two names, through the rule export_native uses."""
|
|
|
|
@staticmethod
|
|
def _pair(truth, scraped, claimed=None):
|
|
return pair_truth_entry(truth, scraped, claimed or set())
|
|
|
|
def test_a_shared_sha1_pairs_the_two_names(self):
|
|
self.assertEqual(
|
|
self._pair(_entry("bios_CD_U.bin", sha1="a" * 40),
|
|
[_entry("ROM1.bin", sha1="a" * 40)]),
|
|
0,
|
|
)
|
|
|
|
def test_md5_crc32_and_sha256_pair_them_too(self):
|
|
for algorithm, value in (("md5", "b" * 32), ("crc32", "b" * 8), ("sha256", "c" * 64)):
|
|
with self.subTest(algorithm=algorithm):
|
|
self.assertEqual(
|
|
self._pair(_entry("a.bin", **{algorithm: value}),
|
|
[_entry("b.bin", **{algorithm: value})]),
|
|
0,
|
|
)
|
|
|
|
def test_hash_comparison_ignores_case(self):
|
|
self.assertEqual(
|
|
self._pair(_entry("a.bin", sha1="AbCdEf" + "0" * 34),
|
|
[_entry("b.bin", sha1="abcdef" + "0" * 34)]),
|
|
0,
|
|
)
|
|
|
|
def test_different_content_is_not_a_rename(self):
|
|
self.assertIsNone(
|
|
self._pair(_entry("a.bin", sha1="a" * 40), [_entry("b.bin", sha1="b" * 40)])
|
|
)
|
|
|
|
def test_files_with_no_hashes_never_pair(self):
|
|
"""Two unknowns are not evidence of the same file."""
|
|
self.assertIsNone(self._pair(_entry("a.bin"), [_entry("b.bin")]))
|
|
|
|
def test_a_hash_yaml_made_a_number_reads_as_its_digits(self):
|
|
# A digit-only crc32 written unquoted: the export writes it back.
|
|
self.assertEqual(
|
|
self._pair(_entry("a.bin", crc32=12345678), [_entry("b.bin", crc32="12345678")]),
|
|
0,
|
|
)
|
|
|
|
def test_a_claimed_entry_never_answers_twice(self):
|
|
scraped = [_entry("b.bin", sha1="a" * 40)]
|
|
self.assertIsNone(self._pair(_entry("a.bin", sha1="a" * 40), scraped, {0}))
|
|
|
|
def test_the_diff_reads_no_gap_for_a_rename(self):
|
|
truth = {"files": [_entry("a.bin", sha1="a" * 40)]}
|
|
scraped = {"files": [_entry("b.bin", sha1="a" * 40)]}
|
|
self.assertEqual(_diff_system(truth, scraped), {})
|
|
|
|
|
|
|
|
class PlatformMultiHash(unittest.TestCase):
|
|
"""Recalbox and RetroDECK write accepted md5s as one comma string.
|
|
|
|
Read as one value, scph1001.bin was a mismatch the export (which splits
|
|
them) never corrected: diff_truth reported 15 for Recalbox, 9 for
|
|
RetroDECK, against the 4 and 3 the content supports.
|
|
"""
|
|
|
|
TRUTH = "924e392ed05558ffdb115408c263dccf"
|
|
SCRAPED = "dc2b9bf8da62ec93e868cfd29f0d067d,924e392ed05558ffdb115408c263dccf"
|
|
|
|
def test_a_shared_value_is_no_mismatch(self):
|
|
truth = {"files": [{"name": "scph1001.bin", "md5": self.TRUTH}]}
|
|
scraped = {"files": [{"name": "scph1001.bin", "md5": self.SCRAPED}]}
|
|
self.assertEqual(_diff_system(truth, scraped), {})
|
|
|
|
def test_a_rename_is_found_through_the_list(self):
|
|
truth = [_entry("scph1001.bin", md5=self.TRUTH)]
|
|
scraped = {"psx.bin": _entry("psx.bin", md5=self.SCRAPED)}
|
|
self.assertEqual(pair_truth_entry(truth[0], list(scraped.values()), set()), 0)
|
|
|
|
|
|
class OnlySystemDirectoryEntries(unittest.TestCase):
|
|
"""The truth kept files read from the save, content or working directory:
|
|
diff_truth listed Cemu's mlc01 certificates, panda3ds's seeddb.bin and
|
|
neo-emu's neogeo.zip as missing from platform lists that the pack and
|
|
verify deliberately leave them out of."""
|
|
|
|
def test_a_file_read_elsewhere_is_not_asked_of_the_platform(self):
|
|
profiles = {
|
|
"core": {
|
|
"emulator": "Core",
|
|
"type": "libretro",
|
|
"systems": ["sys"],
|
|
"files": [
|
|
{"name": "bios.bin", "sha1": "a" * 40},
|
|
{"name": "explicit.bin", "sha1": "b" * 40, "load_from": "system_dir"},
|
|
{"name": "cert.der", "sha1": "c" * 40, "load_from": "save_dir"},
|
|
{"name": "game.zip", "sha1": "d" * 40, "load_from": "content_dir"},
|
|
],
|
|
}
|
|
}
|
|
result = generate_platform_truth("p", {"cores": ["core"]}, {}, profiles)
|
|
names = sorted(f["name"] for f in result["systems"]["sys"]["files"])
|
|
self.assertEqual(names, ["bios.bin", "explicit.bin"])
|
|
|
|
|
|
if __name__ == "__main__":
|
|
unittest.main()
|