#!/usr/bin/env python3 """Platform-native BIOS verification engine. Replicates the exact verification logic of each platform: - RetroArch/Lakka/RetroPie: file existence only (core_info.c path_is_valid) - Batocera: MD5 + checkInsideZip, no required distinction (batocera-systems:1062-1091) - Recalbox: MD5 + mandatory/hashMatchMandatory, 3-color severity (Bios.cpp:109-130) - RetroBat: same as Batocera - EmuDeck: MD5 whitelist per system - RetroDECK: MD5 per file via component manifests - RomM: size + any hash, no ZIP inspection (firmware.py verify_file_hashes) - ROCKNIX: MD5 + checkInsideZip with altmd5 (rocknix-systems checkBios) - MiSTer FPGA: MD5 of the file at its destination, no ZIP inspection - BizHawk: SHA1 firmware hash verification Cross-references emulator profiles to detect undeclared files used by available cores. Usage: python scripts/verify.py --all python scripts/verify.py --platform batocera python scripts/verify.py --all --include-archived python scripts/verify.py --all --json """ from __future__ import annotations import argparse import contextlib import json import os import sys from collections.abc import Mapping from pathlib import Path import slots sys.path.insert(0, os.path.dirname(__file__)) from hashing import parse_md5_list from common import ( platform_arg, target_listing_line, suggest_names, require_platform, archive_entry, archive_members, profiles_for_systems, PROFILE_IDENTITY_FIELDS, list_available_targets, list_platform_system_ids, list_registered_platforms, build_target_cores_cache, build_zip_contents_index, check_inside_zip, compute_hashes, expand_directory_entries, platform_declarations, filter_systems_by_target, group_identical_platforms, list_emulator_profiles, list_system_ids, load_data_dir_registry, load_database, load_emulator_profiles, load_platform_config, resolve_system_arg, select_systems, md5sum, require_yaml, resolve_local_file, ProfileSelectionError, resolve_platform_cores, platform_layouts, runs_standalone, standalone_set as platform_standalone_set, entry_serves_systems, entry_system_ids, sanitize_pack_path, select_emulator_profiles, system_dir_path, ) yaml = require_yaml() from exporter.candidates import BIOS, EntryCategories from launchers import LauncherLayer, declared_destinations from nativemode import ( SIZE_AND_ANY_HASH, digest_algorithm, hash_mismatch_excludes_file, normalize as normalize_mode, reads_file_contents, ) from validation import ( baseline_verdict, case_folded_slots, declaration_settles, frontend_digest_matches, existence_discrepancy, destination_owners, validated_choice, agnostic_substitute, _build_validation_index, _parse_validation, build_ground_truth, check_file_validation, deliverable, installed_with_emulator, unread_on, outside_gap_scope, with_archive_members, core_size_refusal, same_slot_entries, ) DEFAULT_DB = "database.json" DEFAULT_PLATFORMS_DIR = "platforms" DEFAULT_BIOS_DIR = "bios" # The repository's platforms wherever the script runs from, for library calls # that pass no directory. _REPO_PLATFORMS = os.path.join(os.path.dirname(__file__), "..", "platforms") DEFAULT_EMULATORS_DIR = "emulators" REPO_ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__))) def work_from_repo_root( args: argparse.Namespace, parser: argparse.ArgumentParser, path_flags: list[str] ) -> None: """Run from the repository root, whichever directory the command started in. database.json records every file from the root and the data registry names its caches the same way, so a run from scripts/ ended on a FileNotFoundError traceback. A path the caller typed is read from where they typed it, a default from the root. """ if os.path.realpath(os.getcwd()) == os.path.realpath(REPO_ROOT): return for flag in path_flags: value = getattr(args, flag, None) if value and value != parser.get_default(flag): setattr(args, flag, os.path.abspath(value)) os.chdir(REPO_ROOT) if getattr(args, "platform", None): # Read before the move, the registry was not there to map # "RetroArch" to its slug. args.platform = platform_arg(args.platform) # Status model: aligned with Batocera BiosStatus (batocera-systems:967-969) class Status: OK = "ok" UNTESTED = "untested" # file present, hash not confirmed: a warning, never --strict MISSING = "missing" # A declaration whose file is a release asset the cache does not hold yet: # an online build fetches it, verify reads the database record instead. RELEASE_ASSET_PENDING = "release_asset_pending" class Severity: CRITICAL = "critical" # required file missing or bad hash (Recalbox RED) WARNING = "warning" # optional missing or hash mismatch (Recalbox YELLOW) INFO = "info" # optional missing on existence-only platform OK = "ok" # file verified _STATUS_ORDER = {Status.OK: 0, Status.UNTESTED: 1, Status.MISSING: 2} _SEVERITY_ORDER = { Severity.OK: 0, Severity.INFO: 1, Severity.WARNING: 2, Severity.CRITICAL: 3, } # Verification functions def verify_entry_existence( file_entry: dict, local_path: str | None, validation_index: dict[str, dict] | None = None, db: dict | None = None, destination: str = "", owners: dict | None = None, resolve_status: str = "", platform_display: str = "", ) -> dict: """RetroArch verification: path_is_valid(), a file that exists is OK.""" name = file_entry.get("name", "") required = file_entry.get("required", True) if not local_path: return {"name": name, "status": Status.MISSING, "required": required} result = {"name": name, "status": Status.OK, "required": required} if resolve_status == "hash_mismatch": # The frontend reads no bytes, so the file stays, and the declared hash # it contradicts is reported, as the pack does. result["discrepancy"] = existence_discrepancy( file_entry, local_path, platform_display ) return result if not validation_index: return result if db: # The pack ships the variant this returns, so a disagreement it # settles is not reported. _chosen, disagreement = validated_choice( file_entry, local_path, db, validation_index, "bios", None, destination, owners, ) else: check = check_file_validation(local_path, name, validation_index) disagreement = f"{', '.join(check[1])} says {check[0]}" if check else None if disagreement: result["discrepancy"] = f"file present (OK) but {disagreement}" return result def verify_entry_md5( file_entry: dict, local_path: str | None, resolve_status: str = "", hash_rule: str = "", ) -> dict: """MD5 verification -Batocera md5sum + Recalbox multi-hash + Md5Composite. Under SIZE_AND_ANY_HASH (RomM) the size and any one declared hash decide. """ name = file_entry.get("name", "") expected_md5 = file_entry.get("md5", "") zipped_file = file_entry.get("zipped_file") required = file_entry.get("required", True) base = {"name": name, "required": required} # One parse for verify, the builder and the pack check: Recalbox writes # md5 in uppercase, and a copy that lowercased only comma lists judged # a single uppercase value differently from the others. md5_list = parse_md5_list(expected_md5) if not local_path: return {**base, "status": Status.MISSING} if hash_rule == SIZE_AND_ANY_HASH: if frontend_digest_matches(file_entry, local_path, "md5", hash_rule): return {**base, "status": Status.OK, "path": local_path} return { **base, "status": Status.UNTESTED, "path": local_path, "reason": "the declared size and at least one declared hash must match", } if zipped_file: found_in_zip = False had_error = False for md5_candidate in md5_list or [""]: result = check_inside_zip(local_path, zipped_file, md5_candidate) if result == Status.OK: return {**base, "status": Status.OK, "path": local_path} if result == "error": had_error = True elif result != "not_in_zip": found_in_zip = True if had_error and not found_in_zip: return { **base, "status": Status.UNTESTED, "path": local_path, "reason": f"{local_path} read error", } if not found_in_zip: return { **base, "status": Status.UNTESTED, "path": local_path, "reason": f"{zipped_file} not found inside ZIP", } return { **base, "status": Status.UNTESTED, "path": local_path, "reason": f"{zipped_file} MD5 mismatch inside ZIP", } if not md5_list: if resolve_status == "hash_mismatch": # No md5 to compare does not mean nothing was compared: the entry # declares a sha1 or a crc32 the local file contradicts, and the # builder drops it for exactly that. Reporting OK here said the # collection holds bytes it does not. return { **base, "status": Status.UNTESTED, "path": local_path, "reason": "declared hash contradicted by the local file", } return {**base, "status": Status.OK, "path": local_path} if resolve_status == "md5_exact": return {**base, "status": Status.OK, "path": local_path} # The builder's comparison, so the two count the same files as held. if frontend_digest_matches(file_entry, local_path, "md5"): return {**base, "status": Status.OK, "path": local_path} actual_md5 = md5sum(local_path) return { **base, "status": Status.UNTESTED, "path": local_path, "reason": f"expected {md5_list[0][:12]}... got {actual_md5[:12]}...", } def verify_entry_sha1( file_entry: dict, local_path: str | None, ) -> dict: """SHA1 verification -BizHawk firmware hash check.""" name = file_entry.get("name", "") expected_sha1 = file_entry.get("sha1", "") required = file_entry.get("required", True) base = {"name": name, "required": required} if not local_path: return {**base, "status": Status.MISSING} if not expected_sha1: return {**base, "status": Status.OK, "path": local_path} hashes = compute_hashes(local_path) actual_sha1 = hashes["sha1"].lower() if actual_sha1 == expected_sha1.lower(): return {**base, "status": Status.OK, "path": local_path} return { **base, "status": Status.UNTESTED, "path": local_path, "reason": f"expected {expected_sha1[:12]}... got {actual_sha1[:12]}...", } # Severity mapping per platform def _emulator_severity(status: str, required: bool, hle_fallback: bool) -> str: """Severity when the emulator's own code is the check. A file present but refused by the core's validation, or a different dump than the one the profile declares, is not a file the core runs with: read in existence mode it counted OK, and `verify --emulator` announced 1/1 OK above the line that said the core would refuse it. """ if status == Status.UNTESTED: return Severity.WARNING return compute_severity(status, required, "existence", hle_fallback) def compute_severity( status: str, required: bool, mode: str, hle_fallback: bool = False, ) -> str: """Map (status, required, verification_mode, hle_fallback) -> severity. Based on native platform behavior + emulator HLE capability: - RetroArch (existence): required+missing = warning, optional+missing = info - Batocera/Recalbox/RetroBat/EmuDeck (md5): hash-based verification - BizHawk (sha1): same severity rules as md5 - hle_fallback: core works without this file via HLE -> always INFO when missing """ if status == Status.OK: return Severity.OK # HLE fallback: core works without this file regardless of platform requirement if hle_fallback and status == Status.MISSING: return Severity.INFO if not reads_file_contents(mode): if status == Status.MISSING: return Severity.WARNING if required else Severity.INFO return Severity.OK # md5 mode (Batocera, Recalbox, RetroBat, EmuDeck) if status == Status.MISSING: return Severity.CRITICAL if required else Severity.WARNING if status == Status.UNTESTED: return Severity.WARNING return Severity.OK # Cross-reference: undeclared files used by cores def _build_expected(file_entry: dict, checks: list[str]) -> dict: """Extract expected validation values from an emulator profile file entry.""" expected: dict = {} if not checks: return expected if "size" in checks: for key in ("size", "min_size", "max_size"): if file_entry.get(key) is not None: expected[key] = file_entry[key] for hash_type in ("crc32", "md5", "sha1", "sha256"): if hash_type in checks and file_entry.get(hash_type): expected[hash_type] = file_entry[hash_type] adler_val = file_entry.get("known_hash_adler32") or file_entry.get("adler32") if adler_val: expected["adler32"] = adler_val return expected def _name_in_index( name: str, by_name: dict, by_path_suffix: dict | None = None, data_names: set[str] | None = None, by_name_lower: dict[str, str] | None = None, ) -> bool: """Check if a name is resolvable in the database indexes or data directories.""" # Strip trailing slash for directory-type entries (e.g. nestopia/samples/foo/) name = name.rstrip("/") if name in by_name: return True basename = name.rsplit("/", 1)[-1] if "/" in name else name if basename != name and basename in by_name: return True if by_name_lower: key = name.lower() if key in by_name_lower: return True if basename != name and basename.lower() in by_name_lower: return True if by_path_suffix and name in by_path_suffix: return True if data_names: if name in data_names or name.lower() in data_names: return True if basename != name and ( basename in data_names or basename.lower() in data_names ): return True return False def _candidate_verdict( file_entry: dict, fname: str, is_standalone: bool, include_all: bool, declared_names: Mapping[str, list[tuple[int | None, str]]] | set[str], dest: str = "", folded_slots: frozenset[str] = frozenset(), ) -> str: """Whether a profile entry can be a gap, and whether it is settled. Three answers. "keep" means it is a candidate. "settled" means something has answered for it (the platform declares it, or the profile documents it as unsourceable), so the same requirement reached from another profile must not be reconsidered. "skip" means it does not apply in this context, which leaves it open for a profile where it does: the same file can be libretro-only here and standalone-only there. """ if file_entry.get("unsourceable"): return "settled" if outside_gap_scope(file_entry, is_standalone): return "skip" if not include_all and declaration_settles( file_entry, fname, declared_names, dest, folded_slots ): return "settled" return "keep" def _identity_of(entry: dict) -> dict: """The fields that tell a profile's file apart from a same-named one.""" return { field: entry.get(field) for field in PROFILE_IDENTITY_FIELDS if entry.get(field) not in (None, "", []) } def _held_record(report: dict, db: dict) -> dict | None: """The database record of the file the collection holds for a report entry.""" local, status = resolve_local_file(report, db, dest_hint=report.get("path") or "") if local is None or status in ("not_found", "hash_mismatch"): return None for record in db.get("files", {}).values(): if record.get("path") == local: return record return None def find_undeclared_files( config: dict, emulators_dir: str, db: dict, emu_profiles: dict | None = None, target_cores: set[str] | None = None, data_names: set[str] | None = None, include_all: bool = False, declared_names: Mapping[str, list[tuple[int | None, str]]] | None = None, ) -> list[dict]: """Find files needed by cores but not declared in platform config. declared_names overrides the default enriched declarations from platform_declarations. Pass the strict ones (YAML names only) when building packs so alias-only names still get packed. """ if declared_names is None: declared_names = platform_declarations(config, db) folded_slots = case_folded_slots(config, declared_names) # Whether the builder drops a file whose local copy contradicts its # declared hash, which decides if such a copy counts as held here. shipped_on_mismatch = hash_mismatch_excludes_file( config.get("verification_mode") ) declared_dd: set[str] = set() for sys_id, system in config.get("systems", {}).items(): for dd in system.get("data_directories", []): ref = dd.get("ref", "") if ref: declared_dd.add(ref) by_name = db.get("indexes", {}).get("by_name", {}) by_name_lower = {k.lower(): k for k in by_name} by_path_suffix = db.get("indexes", {}).get("by_path_suffix", {}) profiles = ( emu_profiles if emu_profiles is not None else load_emulator_profiles(emulators_dir) ) relevant = resolve_platform_cores(config, profiles, target_cores=target_cores) standalone_set = platform_standalone_set(config) # What the platform's own launchers open, matched to the cores it runs: # the one decision the builder and this report both read. launcher_layer = LauncherLayer( config, profiles, relevant, db, declared_destinations(config) ) launcher_candidates: list[tuple[str, bool, dict, dict]] = [] undeclared = [] seen_files: set[tuple] = set() # Key -> the entry reported for it. A later profile reading the same # requirement is kept on that entry rather than counted again: a pack # built without the first profile still has to name the file. reported: dict[tuple, dict] = {} # Track archives: archive_name -> {in_repo, emulator, files: [...], ...} archive_entries: dict[tuple, dict] = {} for emu_name, profile in sorted(profiles.items()): if profile.get("type") in ("launcher", "alias"): continue if emu_name not in relevant: continue # Skip agnostic profiles entirely (filename-agnostic BIOS detection) if profile.get("bios_mode") == "agnostic": continue for profile, is_standalone in platform_layouts(emu_name, profile, standalone_set): for f in expand_directory_entries( profile.get("files", []), db, is_standalone ): fname = f.get("name", "") # One destination rule for the dedup key and for the entry: a # standalone build without standalone_path falls back on path. # The key used to skip that fallback, so clk's Acorn/basic.rom and # Electron/basic.rom both keyed as "basic.rom" under Batocera's # standalone clk and the second was dropped. if is_standalone: dest = f.get("standalone_path") or f.get("path") or fname else: dest = system_dir_path(f, f.get("path") or fname) raw_regions = f.get("region") or [] region_key = tuple( str(value) for value in ( raw_regions if isinstance(raw_regions, list) else [raw_regions] ) ) # Same-name requirements at different paths, for different systems, # or in distinct variant groups are not interchangeable. seen_key = ( fname, f.get("archive"), dest, f.get("system"), f.get("variant_group"), region_key, ) if not fname or unread_on(f, config.get("_key", "")): continue first = reported.get(seen_key) duplicate = seen_key in seen_files if duplicate and (first is None or first["profile"] == emu_name): continue verdict = _candidate_verdict( f, fname, is_standalone, include_all, declared_names, dest, folded_slots, ) if verdict == "settled": seen_files.add(seen_key) continue if verdict == "skip": continue archive = f.get("archive") seen_files.add(seen_key) if archive: archive_key = ( archive, f.get("system_dir"), f.get("system"), f.get("variant_group"), region_key, is_standalone, ) if archive_key not in archive_entries: in_repo = _name_in_index( archive, by_name, by_path_suffix, data_names, by_name_lower, ) archive_entries[archive_key] = { "profile": emu_name, "emulator": profile.get("emulator", emu_name), "systems": entry_system_ids(f, profile), "system": f.get("system"), "region": f.get("region"), "variant_group": f.get("variant_group"), "name": archive, "archive": archive, "path": system_dir_path(f, archive), "required": False, "hle_fallback": False, "category": f.get("category", "bios"), "in_repo": in_repo, "note": "", "checks": [], "source_ref": None, "expected": {}, "archive_file_count": 0, "archive_required_count": 0, "members": archive_members(archive, profile.get("files", [])), } if f.get("system_dir"): archive_entries[archive_key]["system_dir"] = f["system_dir"] entry = archive_entries[archive_key] entry["archive_file_count"] += 1 if f.get("required", False): entry["archive_required_count"] += 1 entry["required"] = True continue # Resolution: name, then path basename # storage: release names where the bytes are served from, not # that the collection holds them: resolved like any file. if shipped_on_mismatch and (f.get("md5") or f.get("sha1")): # The entry states what its content should be and the builder # drops a copy that contradicts it, so content decides here # too. A name match would not do: generic names collide across # systems, and answering yes on one describes a pack that will # not contain the file. _lp, _st = resolve_local_file(f, db, dest_hint=dest) in_repo = _lp is not None and _st != "hash_mismatch" else: in_repo = _name_in_index( fname, by_name, by_path_suffix, data_names, by_name_lower, ) if not in_repo and dest != fname: path_base = dest.rsplit("/", 1)[-1] in_repo = _name_in_index( path_base, by_name, by_path_suffix, data_names, by_name_lower, ) if in_repo and not any( f.get(h) for h in ("sha1", "md5", "crc32", "sha256") ): # A name alone answers nothing when every copy of that # name fails the size the code checks: the resolver # rejects them all. NanoBoyAdvance's 16 KB bios.bin met # only PCem's and QEMU's PC BIOS images of that name. # The fallback below then decides as for any mismatch. _lp, _st = resolve_local_file(f, db, dest_hint=dest) if _st == "hash_mismatch": in_repo = False if not in_repo: # Hash fallback: the repo may hold the content under a # different filename (exos21.rom vs exos21.bin). # # A copy contradicting the declared hash counts as held # only where the builder would ship it. Under existence # the frontend never opens the file, so the pack carries # it and reports the divergence. Under a digest mode the # frontend would reject it, the builder leaves it out, and # calling it available here would describe a pack that # does not contain it. _lp, _st = resolve_local_file(f, db, dest_hint=dest) in_repo = _st != "not_found" and _lp is not None if in_repo and _st == "hash_mismatch" and shipped_on_mismatch: in_repo = False if in_repo and "size" in _parse_validation(f.get("validation")): # On any platform the core measures what it opens: a # copy it turns away by its own size check is not # held for it, the rule the builder's lookup applies # (packextras.ExtraResolver). The Apple IIGS' 128 KB # ROM is not Basilisk II's ROM. _lp, _st = resolve_local_file(f, db, dest_hint=dest) if core_size_refusal( f, _lp, same_slot_entries(profile.get("files"), fname, dest) ): in_repo = False checks = _parse_validation(f.get("validation")) entry = { "profile": emu_name, "emulator": profile.get("emulator", emu_name), "systems": entry_system_ids(f, profile), "system": f.get("system"), "region": f.get("region"), "variant_group": f.get("variant_group"), "name": fname, "path": dest, "required": f.get("required", False), "hle_fallback": f.get("hle_fallback", False), "bundled": installed_with_emulator(f, config.get("_key", "")), "category": f.get("category", "bios"), "in_repo": in_repo, "note": f.get("note", ""), "checks": sorted(checks) if checks else [], "source_ref": f.get("source_ref"), "expected": _build_expected(f, checks), **_identity_of(f), "sha1": f.get("sha1"), "md5": f.get("md5"), } if f.get("contents"): entry["contents"] = f["contents"] if f.get("system_dir"): entry["system_dir"] = f["system_dir"] members = with_archive_members(f, profile.get("files", [])).get("members") if members: entry["members"] = members if launcher_layer: launcher_candidates.append((emu_name, is_standalone, f, entry)) if duplicate: first.setdefault("other_claims", []).append(entry) continue reported[seen_key] = entry undeclared.append(entry) if launcher_layer: added, settled = launcher_layer.settle( launcher_candidates, lambda report: _held_record(report, db) ) undeclared = [u for u in undeclared if id(u) not in settled] undeclared.extend(added) for entry in sorted(archive_entries.values(), key=lambda e: e["name"]): undeclared.append(entry) return undeclared def find_exclusion_notes( config: dict, emulators_dir: str, emu_profiles: dict | None = None, target_cores: set[str] | None = None, ) -> list[dict]: """Document why certain emulator files are intentionally excluded. Reports: - Launchers (BIOS managed by standalone emulator) - Standalone-only files (not needed in libretro mode) - Frozen snapshots with files: [] (code doesn't load .info firmware) - Files covered by data_directories """ profiles = ( emu_profiles if emu_profiles is not None else load_emulator_profiles(emulators_dir) ) platform_systems = set() for sys_id in config.get("systems", {}): platform_systems.add(sys_id) relevant = resolve_platform_cores(config, profiles, target_cores=target_cores) notes = [] for emu_name, profile in sorted(profiles.items()): emu_systems = set(profile.get("systems", [])) # Match by core resolution OR system intersection (documents all potential emulators) if emu_name not in relevant and not (emu_systems & platform_systems): continue emu_display = profile.get("emulator", emu_name) if profile.get("type") == "launcher": notes.append( { "emulator": emu_display, "reason": "launcher", "detail": profile.get( "exclusion_note", "BIOS managed by standalone emulator" ), } ) continue # Profile-level exclusion note (frozen snapshots, etc.) exclusion_note = profile.get("exclusion_note") if exclusion_note: notes.append( { "emulator": emu_display, "reason": "exclusion_note", "detail": exclusion_note, } ) continue # Count standalone-only files, but only report as excluded if the # platform does NOT use this emulator in standalone mode standalone_set = platform_standalone_set(config) is_standalone = runs_standalone(emu_name, profile, standalone_set) if not is_standalone: standalone_files = [ f for f in profile.get("files", []) if f.get("mode") == "standalone" ] if standalone_files: names = [f["name"] for f in standalone_files[:3]] more = ( f" +{len(standalone_files) - 3}" if len(standalone_files) > 3 else "" ) notes.append( { "emulator": emu_display, "reason": "standalone_only", "detail": f"{len(standalone_files)} files for standalone mode only ({', '.join(names)}{more})", } ) return notes # Platform verification _UNPLACED_REASONS = { "no_platform_slug": "no platform slug for its system", "below_platform_folder": "below the folder the platform lists", } def _mark_unplaced(config: dict, undeclared: list[dict], profiles: dict) -> None: """What the builder cannot place is not in the pack, held or not.""" from packextras import unplaceable_extras unplaced = unplaceable_extras(config, undeclared, profiles) for u in undeclared: if u.get("bundled"): # Installed with the emulator, so already on the user's machine: # the pack leaves it out, held or not. u["in_pack"] = False u["omitted"] = "installed with its emulator" continue reason = unplaced.get( (u.get("emulator", ""), u.get("name", ""), u.get("path") or "") ) if reason: u["in_pack"] = False u["omitted"] = _UNPLACED_REASONS[reason] def _storage_verdict(file_entry: dict, emulator_pack: bool) -> dict | None: """The result a storage tier fixes before any lookup, as the builder sees it. A user_provided entry is never taken from the collection: a platform pack writes its instructions in the slot and counts it filled, an emulator pack lists it to provide. An external one is fetched from its URL when the pack is built, never read from bios/. Resolving either by name credited a same-named file the pack does not carry. """ storage = file_entry.get("storage", "embedded") base = {"name": file_entry.get("name", ""), "required": file_entry.get("required", True)} if storage == "user_provided": if emulator_pack: stated = file_entry.get("unsourceable") why = ( stated if isinstance(stated, str) and stated else file_entry.get("instructions") or "the user provides it" ) return {**base, "status": Status.MISSING, "unsourceable": why} return {**base, "status": Status.OK, "user_provided": True} if storage == "external": return {**base, "status": Status.MISSING, "reason": "fetched from its URL when the pack is built"} return None _HIDDEN_REASONS = { "case_twin": "a packed file of the same name in another case takes its place", "path_conflict": "a packed file occupies its directory", } def _mark_hidden( config: dict, undeclared: list[dict], emulators_dir: str, db: dict, emu_profiles: dict | None, shipped: dict[str, str | None], target_cores: set[str] | None, region_drops: set[str] | None = None, report: list[dict] | None = None, resolver: object | None = None, placed_kind: dict[str, str] | None = None, system_ids: list[str] | None = None, ) -> set[str]: """What the builder holds but cannot place beside the files it packed. Returns the full destinations the builder fills with a core's file. ``report`` is the gap report the caller already computed with the builder's arguments. Without it the builder's pass computes its own. """ from packextras import hidden_extras placed: set[str] = set() hidden = hidden_extras( config, emulators_dir, db, emu_profiles if emu_profiles is not None else load_emulator_profiles(emulators_dir), shipped, target_cores, region_drops, placed, report, resolver, placed_kind, system_ids=system_ids, ) for u in undeclared: reason = hidden.get( (u.get("emulator", ""), u.get("name", ""), u.get("path") or "") ) if reason and u.get("in_repo") and u.get("in_pack", True): u["in_pack"] = False u["omitted"] = _HIDDEN_REASONS[reason] return placed def _builder_placement( packed_order: list[tuple[str, str, str | None]], case_insensitive: bool, ) -> tuple[dict[str, str], dict[str, str | None]]: """The platform files the builder writes, and every path it keeps. The builder walks the systems in name order and skips a path that a written file blocks as a directory, or the reverse: RetroArch declares SGB1.sfc and SGB1.sfc/program.rom, and only the first is in the pack. An entry without a destination (EmuDeck's system:md5 whitelist) is judged but takes no path. Returns (path -> local copy written there, path -> copy or None for each slot kept, a user_provided or external one included). """ from packpaths import _has_path_conflict, _register_path shipped: dict[str, str] = {} occupied: dict[str, str | None] = {} seen: set[str] = set() folded: set[str] = set() parents: set[str] = set() for _sys_id, full, local in sorted(packed_order, key=lambda item: item[0]): # EmuDeck's system:md5 entries name no file: the builder writes none. if not full or full.endswith("/"): continue if ( full in occupied or (case_insensitive and full.lower() in folded) or _has_path_conflict(full, seen, parents) ): continue _register_path(full, seen, parents) folded.add(full.lower()) occupied[full] = local if local is not None: shipped[full] = local return shipped, occupied def _dest_key(file_entry: dict, sys_id: str) -> str: """The path a declaration occupies, as the builder keys its pack.""" dest = sanitize_pack_path(file_entry.get("destination", file_entry.get("name", ""))) return dest or f"{sys_id}/{file_entry.get('name', '')}" def _twin_index( verify_systems: dict, db: dict, base_dest: str, zip_contents: dict, ) -> tuple[dict[str, int], dict[int, tuple[str, dict]]]: """Which declaration the pack ships at each destination, and every declaration by id.""" # From the module that defines the rule, not the builder's CLI: importing # generate_pack made every verify depend on its argparse and manifest code. from packresolve import preferred_entries as _preferred_entries preferred_entries = _preferred_entries( verify_systems, db, DEFAULT_BIOS_DIR, base_dest, False, zip_contents, True, ) winners = { id(fe): (sid, fe) for sid, system in verify_systems.items() for fe in system.get("files", []) } return preferred_entries, winners def _resolve_declaration( file_entry: dict, sys_id: str, twins: tuple, db: dict, zip_contents: dict, data_dir_registry: dict | None, slot_overrides: dict, mode: str, platform_profiles: dict, ) -> tuple[str | None, str, str | None]: """Resolve a declaration to the file the pack ships at its destination. When another declaration holds the destination, that one's file is returned and the evidence reset: the winner's hash match is not this declaration's, which is hashed against the shipped file. """ preferred_entries, winners, base_dest = twins bare = sanitize_pack_path(file_entry.get("destination", file_entry.get("name", ""))) preferred = preferred_entries.get(f"{base_dest}/{bare}" if base_dest else bare) held_by, winner = (None, file_entry) if preferred is not None and preferred != id(file_entry): held_by, winner = winners[preferred] local_path, resolve_status = _resolve_for_platform( winner, held_by or sys_id, db, zip_contents, data_dir_registry, slot_overrides, mode, platform_profiles, ) return local_path, "" if held_by else resolve_status, held_by def _twin_unmet( result: dict, held_by: str | None, file_entry: dict | None = None, sys_id: str = "", destination: str = "", ) -> bool: """Mark a declaration unmet at a path another declaration holds. The path is answered by the winner's file, which the pack counts once. This declaration's failure is reported, never folded into the destination's status. The message names the path, both systems and what this declaration expects: "is not met at the path coco holds" named neither the path nor the file wanted there. """ if held_by is None or result["status"] == Status.OK: return False entry = file_entry or {} expects = next( (f" ({h} {str(entry[h]).split(',')[0]})" for h in ("sha1", "md5", "crc32") if entry.get(h)), "", ) owner = sys_id or "this" result["discrepancy"] = ( f"{destination or result.get('name', '')} is declared for {owner}" f" and for {held_by}. The {owner} declaration{expects} is not met at" f" the path, which holds the {held_by} file" ) # Read by reports that list entries: the destination itself is held, and # a count by destination does not include this one. result["settled_by"] = held_by return True def _resolve_for_platform( file_entry: dict, sys_id: str, db: dict, zip_contents: dict, data_dir_registry: dict | None, slot_overrides: dict[str, str], mode: str, platform_profiles: dict, ) -> tuple[str | None, str]: """Resolve one platform entry the way the pack builder does. The builder's own resolution (packresolve.resolve_declaration), read from disk and the large-file cache: verify never downloads. A frontend that never reads the bytes takes the filename-free core's own image. """ from packresolve import pending_release_asset, resolve_declaration, run_is_offline override = slot_overrides.get( sanitize_pack_path(file_entry.get("destination", file_entry.get("name", ""))) ) local_path, status = resolve_declaration( file_entry, db, DEFAULT_BIOS_DIR, zip_contents, data_dir_registry, override, offline=True, ) if status in ("not_found", "hash_mismatch") and not run_is_offline(): # Not in the cache, and the release holds a copy that answers the # declaration: an online build fetches and packs it (resolve_file). asset = pending_release_asset(file_entry, db) if asset: return asset, RELEASE_ASSET_PENDING if status == "not_found" and not reads_file_contents(mode): found = agnostic_substitute(file_entry, sys_id, db, platform_profiles) if found: return found[0], "agnostic_fallback" return local_path, status def verify_platform( config: dict, db: dict, emulators_dir: str = DEFAULT_EMULATORS_DIR, emu_profiles: dict | None = None, target_cores: set[str] | None = None, data_dir_registry: dict | None = None, regions: list[str] | None = None, system_filter: list[str] | None = None, ) -> dict: """Verify all BIOS files for a platform, including cross-reference gaps. A region priority list narrows the report to the files a pack built with the same list would carry, using the same selection function as the builder. ``system_filter`` does the same for a pack built with --system: the systems it keeps and the core extras those systems own. """ # Normalized once: a typo in a platform YAML must not verify with one # mode and be scored with another. mode = normalize_mode(config.get("verification_mode")) platform = config.get("platform", "unknown") # The builder settles a destination claimed by both layers, and this must read # the same decision, or the two tools describe different packs. has_zipped = any( fe.get("zipped_file") for sys in config.get("systems", {}).values() for fe in sys.get("files", []) ) zip_contents = build_zip_contents_index(db) if has_zipped else {} # Loaded before the slot overrides: a caller passing no profiles (the # README's coverage) got no override and reported a discrepancy at a slot # the pack settles. profiles = ( emu_profiles if emu_profiles is not None else load_emulator_profiles(emulators_dir) ) base_dest = config.get("base_destination", "") slot_overrides = { (key[len(base_dest) + 1:] if base_dest and key.startswith(f"{base_dest}/") else key): path for key, path in slots.pack_overrides( config, profiles, db, zip_contents, data_dir_registry, target_cores=target_cores, ).items() } # Build HLE + validation indexes from emulator profiles # Ground truth comes from the emulators the platform runs. A standalone # profile that loads a same-named file of its own (ZEsarUX's 48K # cpc6128.rom against cap32's 32K one) has nothing to say about a # RetroArch pack, nor does its HLE: dosbox-x's FONT.ROM fallback turned a # missing required Batocera BIOS into INFO. plat_cores = resolve_platform_cores(config, profiles) platform_profiles = {name: profiles[name] for name in plat_cores} hle_index = { f.get("name", ""): True for profile in platform_profiles.values() for f in profile.get("files", []) if f.get("hle_fallback") } validation_index = _build_validation_index(platform_profiles) validation_owners = destination_owners(platform_profiles) if not target_cores: plat_cores = None verify_systems = filter_systems_by_target( config.get("systems", {}), profiles, target_cores, platform_cores=plat_cores, ) if system_filter: verify_systems = select_systems(verify_systems, system_filter) details = [] file_status: dict[str, str] = {} file_required: dict[str, bool] = {} file_severity: dict[str, str] = {} # What each entry is: firmware, a game's data, or a name no core the # platform runs opens. Only firmware gaps carry a severity. file_category: dict[str, str] = {} categories = EntryCategories(config, profiles, data_dir_registry) region_drops: set[str] = set() region_fallbacks: list[str] = [] region_extra_dests: dict[tuple[str, str, str], str] = {} if regions: import region as region_mod # The builder owns pack composition, so it owns the grouping the region # pass reads. Grouping the platform files here and the core extras # there let the two answer differently on one request: the report kept # every core extra a region run withdraws from the pack. from packextras import platform_region_groups held: set[str] = set() region_groups, region_extra_dests = platform_region_groups( config, verify_systems, emulators_dir, db, config.get("base_destination", ""), profiles, target_cores=target_cores, held_out=held, ) region_index = region_mod.platform_region_index(config, profiles, target_cores) region_drops = region_mod.resolve_region_drops( region_groups, region_index, regions, held, ) # The pack README names these: no file answers the request, so every # candidate stays. Reported here too, or verify reads all OK for a # system that has no BIOS for the region asked. region_fallbacks = region_mod.fallback_groups( region_groups, region_index, regions ) # A destination two systems declare with two hashes ships one file, the # one the builder prefers. The other declaration is scored against that # file, as the frontend will score it after installation: RetroDECK's # xroar component hashes bios/disk.rom and finds the PC-88 ROM. base_dest = config.get("base_destination", "") preferred_entries, winners = _twin_index( verify_systems, db, base_dest, zip_contents ) # (system, full destination, local copy) of each declaration the builder # writes, and None for one whose slot it keeps without a file from the # collection (user_provided, external). Placed below in the builder's order. packed_order: list[tuple[str, str, str | None]] = [] shipped_keys: set[str] = set() results_at: dict[str, list[dict]] = {} for sys_id, system in verify_systems.items(): for file_entry in system.get("files", []): if region_drops and ( sanitize_pack_path( file_entry.get("destination", file_entry.get("name", "")) ) in region_drops ): continue local_path, resolve_status, held_by = _resolve_declaration( file_entry, sys_id, (preferred_entries, winners, base_dest), db, zip_contents, data_dir_registry, slot_overrides, mode, platform_profiles, ) destination = sanitize_pack_path( file_entry.get("destination", file_entry.get("name", "")) ) stored = _storage_verdict(file_entry, emulator_pack=False) if resolve_status == RELEASE_ASSET_PENDING: # Counted as an online build counts it, which fetches the # asset. Under --offline the builder cannot, and neither # does this report (_resolve_for_platform). result = { "name": file_entry.get("name", ""), "required": file_entry.get("required", True), "status": Status.OK, "release_asset": local_path, "reason": "release asset, fetched when the pack is built", } packed_order.append(( sys_id, f"{base_dest}/{destination}" if base_dest else destination, None, )) shipped_keys.add(_dest_key(file_entry, sys_id)) local_path = None elif stored is not None: result = stored local_path = None # The builder keeps the slot: instructions for a user_provided # file, the download for an external one, whether it arrives # or not. No core file goes there. packed_order.append(( sys_id, f"{base_dest}/{destination}" if base_dest else destination, None, )) elif not reads_file_contents(mode): result = verify_entry_existence( file_entry, local_path, validation_index, db, destination, validation_owners, resolve_status, config.get("platform", ""), ) elif digest_algorithm(mode) == "sha1": result = verify_entry_sha1(file_entry, local_path) else: result = verify_entry_md5( file_entry, local_path, resolve_status, config.get("hash_rule", "") ) # Emulator-level validation: informational for platform packs. # Platform verification (MD5) is the authority. Emulator # mismatches are reported as discrepancies, not failures. if result["status"] == Status.OK: _chosen, disagreement = validated_choice( file_entry, local_path, db, validation_index, "bios", digest_algorithm(mode), destination, validation_owners, ) if disagreement: result["discrepancy"] = f"{platform} says OK but {disagreement}" result["system"] = sys_id result["hle_fallback"] = hle_index.get(file_entry.get("name", ""), False) result["ground_truth"] = build_ground_truth( file_entry.get("name", ""), validation_index, ) details.append(result) # Occupied only where the builder ships the declaration: a hash # the frontend refuses is left out, and the destination stays free # for a core's file the pack does carry. if ( local_path and result["status"] != Status.MISSING and baseline_verdict( file_entry, local_path, resolve_status, mode, config.get("hash_rule", "") )[1] is None ): packed_order.append(( sys_id, f"{base_dest}/{destination}" if base_dest else destination, local_path, )) shipped_keys.add(_dest_key(file_entry, sys_id)) if _twin_unmet( result, held_by, file_entry, sys_id, f"{base_dest}/{destination}" if base_dest else destination, ): continue # Aggregate by destination, keyed as the builder keys its pack. # A declaration that finds nothing leaves the path to one that # holds a file: the builder packs that file there, whatever the # order the two are read in. dest = _dest_key(file_entry, sys_id) results_at.setdefault(dest, []).append(result) required = file_entry.get("required", True) cur = result["status"] prev = file_status.get(dest) hle = hle_index.get(file_entry.get("name", ""), False) sev = compute_severity(cur, required, mode, hle) explained = categories.explain(file_entry, destination, sys_id) category = explained.pop("category") result["category"] = category result["category_evidence"] = explained if cur != Status.MISSING: # A held entry is firmware for the counts: its category only # documents it. category = BIOS if file_category.get(dest, category) == BIOS: category = BIOS file_category[dest] = category if prev is not None and (cur == Status.MISSING) != (prev == Status.MISSING): if cur == Status.MISSING: continue del file_status[dest], file_severity[dest] prev = None if prev is None or _STATUS_ORDER.get(cur, 0) > _STATUS_ORDER.get(prev, 0): file_status[dest] = cur file_required[dest] = required prev_sev = file_severity.get(dest) if prev_sev is None or _SEVERITY_ORDER.get(sev, 0) > _SEVERITY_ORDER.get( prev_sev, 0 ): file_severity[dest] = sev shipped, occupied = _builder_placement( packed_order, bool(config.get("case_insensitive_fs", False)) ) # Cross-reference undeclared files, called exactly as the builder calls # it: its declarations, not the enriched ones (a core file the platform # also declares under another name is packed at the core's own path, 21 # such files in the RetroArch pack), and no names from inside archives # or data trees (a loose file whose name exists only as a member of # naomi2.zip was counted "in pack" while the builder shipped nothing). undeclared = find_undeclared_files( config, emulators_dir, db, emu_profiles, target_cores=target_cores, declared_names=platform_declarations(config, db, enrich=False), ) # The builder's own pass reads the same report: computed once here and # handed over as it stands now, before the markings below. report_for_builder = [dict(u) for u in undeclared] if system_filter: # A system pack carries the extras its systems own. from packextras import report_owned_by_systems undeclared = report_owned_by_systems(undeclared, system_filter) if region_drops: undeclared = [ u for u in undeclared if region_extra_dests.get( (u.get("emulator", ""), u.get("name", ""), u.get("path") or "") ) not in region_drops ] _mark_unplaced(config, undeclared, profiles) from packextras import ExtraResolver placed_kind: dict[str, str] = {} placed = _mark_hidden( config, undeclared, emulators_dir, db, emu_profiles, occupied, target_cores, region_drops, report_for_builder, ExtraResolver( db, DEFAULT_BIOS_DIR, zip_contents, data_dir_registry, validation_index, validation_owners, profiles=profiles, ), placed_kind, system_filter, ) # What the builder places from the cores, one per pack path, by category: # several entries claim one path (rom.key from three Amiga cores) and the # builder ships one file there. The category travels on the extra: a copy # the second pass makes has no gap report entry to look it up by. packed_extras = {"bios": 0, "game_data": 0} for category in placed_kind.values(): packed_extras["game_data" if category == "game_data" else "bios"] += 1 if not reads_file_contents(mode): # An existence frontend is satisfied by any file at the path, and the # builder fills a free platform destination with a core's file. for dest, status in file_status.items(): full = f"{base_dest}/{dest}" if base_dest else dest if status == Status.MISSING and full in placed: file_status[dest] = Status.OK file_severity[dest] = Severity.OK for result in results_at.get(dest, []): result["status"] = Status.OK counts = { Severity.OK: 0, Severity.INFO: 0, Severity.WARNING: 0, Severity.CRITICAL: 0, } not_firmware: dict[str, list[str]] = {} for dest, s in file_severity.items(): category = file_category.get(dest, BIOS) if category != BIOS and file_status.get(dest) == Status.MISSING: not_firmware.setdefault(category, []).append(dest) continue counts[s] = counts.get(s, 0) + 1 # Held but refused: on a hash platform the builder leaves out a file whose # declared hash the local copy contradicts. Counted apart, so a coverage # figure never calls present what the pack does not carry. excluded = sorted( dest for dest, status in file_status.items() if status == Status.UNTESTED and dest not in shipped_keys ) # Count by file status (ok/untested/missing) status_counts: dict[str, int] = {} for s in file_status.values(): status_counts[s] = status_counts.get(s, 0) + 1 exclusions = find_exclusion_notes( config, emulators_dir, emu_profiles, target_cores=target_cores ) gt_filenames = set(validation_index) profiled_names: set[str] = set() for profile in profiles.values(): if profile.get("type") in ("launcher", "alias"): continue for f in profile.get("files", []): name = f.get("name", "") if name: profiled_names.add(name) profiled_names.update(f.get("aliases") or []) dest_to_name: dict[str, str] = {} for sys_id, system in verify_systems.items(): for fe in system.get("files", []): dest_to_name.setdefault(_dest_key(fe, sys_id), fe.get("name", "")) with_validation = sum( 1 for dest in file_status if dest_to_name.get(dest, "") in gt_filenames ) with_profile = sum( 1 for dest in file_status if dest_to_name.get(dest, "") in profiled_names ) total = len(file_status) # The paths a pack built from this report holds, as written in the ZIP # (base_destination stripped, as the builder flattens it). The pipeline # compares this set with the archive: equal counts can hide one file # credited here and another written there. from packpaths import _flat pack_destinations = sorted( {_flat(dest, base_dest, True) for dest in shipped} | { _flat(dest, base_dest, True) for dest in placed } ) return { "platform": platform, **({"systems": sorted(verify_systems)} if system_filter else {}), "verification_mode": mode, "total_files": total, "pack_destinations": pack_destinations, "severity_counts": counts, "status_counts": status_counts, "missing_not_firmware": {k: sorted(v) for k, v in sorted(not_firmware.items())}, "excluded_files": excluded, "packed_extras": packed_extras, "undeclared_files": undeclared, "exclusion_notes": exclusions, "region_fallbacks": sorted(region_fallbacks), "details": details, "ground_truth_coverage": { "with_validation": with_validation, "with_profile": with_profile, "platform_only": total - with_validation, "total": total, "applicable": bool(resolve_platform_cores(config, profiles)), }, } # Output def _format_ground_truth_aggregate(ground_truth: list[dict]) -> str: """Format ground truth as a single aggregated line. Example: beetle_psx [md5], pcsx_rearmed [existence] """ parts = [] for gt in ground_truth: checks_label = "+".join(gt["checks"]) if gt["checks"] else "existence" parts.append(f"{gt['emulator']} [{checks_label}]") return ", ".join(parts) def _format_ground_truth_verbose(ground_truth: list[dict]) -> list[str]: """Format ground truth as one line per core with expected values and source ref. Example: handy validates crc32=0d973c9d,size=512 [lynx/rom.h:48-49] """ lines = [] for gt in ground_truth: checks_label = "+".join(gt["checks"]) if gt["checks"] else "existence" expected = gt.get("expected", {}) if expected: vals = ",".join(f"{k}={v}" for k, v in sorted(expected.items())) part = f"{gt['emulator']} validates {vals}" else: part = f"{gt['emulator']} validates {checks_label}" if gt.get("source_ref"): part += f" [{gt['source_ref']}]" lines.append(part) return lines def _print_ground_truth(gt: list[dict], verbose: bool) -> None: """Print ground truth lines for a file entry.""" if not gt: return if verbose: for line in _format_ground_truth_verbose(gt): print(f" {line}") else: print(f" {_format_ground_truth_aggregate(gt)}") def _discrepancy_keys(details: list[dict]) -> set[str]: """The DISCREPANCY lines _print_detail_entries prints for these details. A file already reported MISSING or UNTESTED on its own ground is not printed again as a discrepancy, and is not counted as one either. """ shown = { f"{d['system']}/{d['name']}" for d in details if d["status"] == Status.MISSING or (d["status"] == Status.UNTESTED and not d.get("discrepancy")) } return { key for d in details if d.get("discrepancy") and (key := f"{d['system']}/{d['name']}") not in shown } # A platform entry that is not firmware, as the report names it. _NOT_FIRMWARE = { "game_data": "game data", "unread": "no core the platform runs opens it", } def category_evidence_text(evidence: dict) -> str: """One line saying what establishes a missing entry's category.""" if evidence.get("read_by"): readers = [] for reader in evidence["read_by"]: refs = [ f"{e['name']} ({e['source_ref']})" if e.get("source_ref") else e["name"] for e in reader.get("entries", []) ] readers.append( f"{reader['emulator']}: {', '.join(refs)}" if refs else f"{reader['emulator']} (by {reader['proof']})" ) return "read by " + "; ".join(readers) examined = evidence.get("cores_examined") or [] cores = f"cores examined: {', '.join(examined)}" if examined else "no core examined" if evidence.get("unsettled"): return f"no reader found, not settled: {evidence['unsettled']} ({cores})" return f"no reader found ({cores})" def _print_category_evidence(detail: dict) -> None: evidence = detail.get("category_evidence") if evidence: print(f" {category_evidence_text(evidence)}") def _print_detail_entries(details: list[dict], seen: set[str], verbose: bool) -> None: """Print UNTESTED, MISSING, and DISCREPANCY entries from verification details.""" for d in details: if d["status"] == Status.UNTESTED: if d.get("discrepancy"): # Reported below with the ground for it: a declaration # unmet at a path another declaration holds. continue key = f"{d['system']}/{d['name']}" if key in seen: continue seen.add(key) req = "required" if d.get("required", True) else "optional" hle = ", HLE available" if d.get("hle_fallback") else "" reason = d.get("reason", "") print(f" UNTESTED ({req}{hle}): {key}: {reason}") _print_ground_truth(d.get("ground_truth", []), verbose) for d in details: if d["status"] == Status.MISSING: key = f"{d['system']}/{d['name']}" if key in seen: continue label = _NOT_FIRMWARE.get(d.get("category", BIOS)) if label: # Counted on their own line; named on request. if verbose: seen.add(key) print(f" MISSING ({label}): {key}") _print_category_evidence(d) continue seen.add(key) req = "required" if d.get("required", True) else "optional" hle = ", HLE available" if d.get("hle_fallback") else "" print(f" MISSING ({req}{hle}): {key}") if verbose: _print_category_evidence(d) _print_ground_truth(d.get("ground_truth", []), verbose) for d in details: disc = d.get("discrepancy") if disc: key = f"{d['system']}/{d['name']}" if key in seen: continue seen.add(key) print(f" DISCREPANCY: {key}: {disc}") _print_ground_truth(d.get("ground_truth", []), verbose) if verbose: for d in details: if d["status"] == Status.OK: key = f"{d['system']}/{d['name']}" if key in seen: continue seen.add(key) gt = d.get("ground_truth", []) if gt: req = "required" if d.get("required", True) else "optional" print(f" OK ({req}): {key}") for line in _format_ground_truth_verbose(gt): print(f" {line}") def _print_undeclared_entry(u: dict, prefix: str, verbose: bool) -> None: """Print a single undeclared file entry with its validation checks.""" arc_count = u.get("archive_file_count") if arc_count: name_label = f"{u['name']} ({arc_count} file{'s' if arc_count != 1 else ''})" else: name_label = u["name"] print(f" {prefix}: {u['emulator']} needs {name_label}") checks = u.get("checks", []) if checks: if verbose: expected = u.get("expected", {}) if expected: vals = ",".join(f"{k}={v}" for k, v in sorted(expected.items())) ref_part = f" [{u['source_ref']}]" if u.get("source_ref") else "" print(f" validates {vals}{ref_part}") else: checks_label = "+".join(checks) ref_part = f" [{u['source_ref']}]" if u.get("source_ref") else "" print(f" validates {checks_label}{ref_part}") else: print(f" [{'+'.join(checks)}]") def _split_undeclared(undeclared: list[dict]) -> tuple[list[dict], list[dict], list[dict]]: """Game data, firmware the pack places, firmware it holds but cannot place. Everything that is not game data is firmware the core loads, archives included: a bios_zip sat in neither list, so a required one that was missing was never printed. """ game_data = [u for u in undeclared if u.get("category", "bios") == "game_data"] firmware = [u for u in undeclared if u.get("category", "bios") != "game_data"] unplaced = [u for u in firmware if u["in_repo"] and not u.get("in_pack", True)] placed = [u for u in firmware if u.get("in_pack", True)] return game_data, placed, unplaced def _core_required_missing(bios_files: list[dict]) -> list[dict]: """Core files a core cannot run without, absent from the collection. A file bundled with the emulator (DaedalusX64's n64.psh, the shader library its renderer installs beside the binary) arrives with it: the collection does not hold it by design, and listing it as missing sent readers after a file every install already has. """ return [ u for u in bios_files if u["required"] and not u["in_repo"] and not u.get("hle_fallback") and not u.get("bundled") ] def _print_undeclared_section(result: dict, verbose: bool) -> None: """Print cross-reference section for undeclared files used by cores.""" undeclared = result.get("undeclared_files", []) if not undeclared: return game_data, bios_files, unplaced = _split_undeclared(undeclared) req_not_in_repo = _core_required_missing(bios_files) req_hle_not_in_repo = [ u for u in bios_files if u["required"] and not u["in_repo"] and u.get("hle_fallback") ] req_in_repo = [u for u in bios_files if u["required"] and u["in_repo"]] opt_in_repo = [u for u in bios_files if not u["required"] and u["in_repo"]] opt_not_in_repo = [u for u in bios_files if not u["required"] and not u["in_repo"]] bundled = [ u for u in bios_files if u["required"] and not u["in_repo"] and u.get("bundled") and not u.get("hle_fallback") ] # The builder's count, one file per pack path, when the platform pass # replayed it. Counting entries said 1130 for RetroArch, 48 more than the # pack holds: rom.key alone was claimed by three Amiga cores. packed = result.get("packed_extras") core_in_pack = ( packed["bios"] if packed is not None else len(req_in_repo) + len(opt_in_repo) ) core_missing_req = len(req_not_in_repo) + len(req_hle_not_in_repo) core_missing_opt = len(opt_not_in_repo) # Required gaps are always named below, the optional ones on request. listed = "" if verbose or not core_missing_opt else " (--verbose lists them)" print( f" Core files: {core_in_pack} in pack, {core_missing_req} required missing, " f"{core_missing_opt} optional missing{listed}" ) if bundled: names = ", ".join(sorted({u["name"] for u in bundled})) print(f" Bundled with their emulator, not collected: {names}") if unplaced: print(f" Core files held but not packed: {len(unplaced)} ({unplaced[0]['omitted']})") for u in req_not_in_repo: _print_undeclared_entry(u, "MISSING (required)", verbose) for u in req_hle_not_in_repo: _print_undeclared_entry(u, "MISSING (required, HLE fallback)", verbose) if verbose: for u in opt_not_in_repo: _print_undeclared_entry(u, "MISSING (optional)", verbose) if game_data: gd_missing = [u for u in game_data if not u["in_repo"]] gd_packed = ( packed["game_data"] if packed is not None else sum(1 for u in game_data if u["in_repo"] and u.get("in_pack", True)) ) gd_unplaced = [u for u in game_data if u["in_repo"] and not u.get("in_pack", True)] if gd_missing or gd_packed or gd_unplaced: listed = "" if verbose or not gd_missing else " (--verbose lists them)" print(f" Game data: {gd_packed} in pack, {len(gd_missing)} missing{listed}") if verbose: for u in gd_missing: _print_undeclared_entry(u, "MISSING (game data)", verbose) if gd_unplaced: print( f" Game data held but not packed: {len(gd_unplaced)} " f"({gd_unplaced[0].get('omitted', 'not placed')})" ) def print_platform_result( result: dict, group: list[str], verbose: bool = False ) -> None: mode = result["verification_mode"] total = result["total_files"] c = result["severity_counts"] label = " / ".join(group) if result.get("systems"): label += f" ({', '.join(result['systems'])})" ok_count = c[Severity.OK] problems = total - ok_count if not reads_file_contents(mode): if problems: missing = c.get(Severity.WARNING, 0) + c.get(Severity.CRITICAL, 0) optional_missing = c.get(Severity.INFO, 0) parts = [f"{ok_count}/{total} present"] if missing: parts.append(f"{missing} missing") if optional_missing: parts.append(f"{optional_missing} optional missing") else: parts = [f"{ok_count}/{total} present"] else: sc = result.get("status_counts", {}) untested = sc.get(Status.UNTESTED, 0) missing = sc.get(Status.MISSING, 0) parts = [f"{ok_count}/{total} OK"] if untested: parts.append(f"{untested} untested") if missing: parts.append(f"{missing} missing") print(f"{label}: {', '.join(parts)} [{mode}]") fetched = sorted({d["name"] for d in result["details"] if d.get("release_asset")}) if fetched: # Counted present: an online build downloads them. --offline counts # them missing, as an offline build leaves them out. shown = ", ".join(fetched[:10]) + (f" (+{len(fetched) - 10})" if len(fetched) > 10 else "") print(f" Release assets fetched when the pack is built ({len(fetched)}): {shown}") fallbacks = result.get("region_fallbacks") or [] if fallbacks: from packreadme import _group_label print(f" No BIOS for the requested regions, all kept ({len(fallbacks)}): " + ", ".join(_group_label(g) for g in fallbacks)) not_firmware = result.get("missing_not_firmware") or {} if not_firmware: counts = ", ".join( f"{len(dests)} {_NOT_FIRMWARE[category]}" for category, dests in not_firmware.items() ) listed = "" if verbose else " (--verbose lists them)" print(f" Missing, not firmware: {counts}{listed}") seen_details: set[str] = set() _print_detail_entries(result["details"], seen_details, verbose) _print_undeclared_section(result, verbose) # The same notes come back for every platform running these cores and # filled 985 of the 1114 lines of verify --all: counted, listed on ask. exclusions = result.get("exclusion_notes", []) if exclusions and verbose: print(f" No external files ({len(exclusions)}):") for ex in exclusions: print(f" {ex['emulator']}: {ex['detail']} [{ex['reason']}]") elif exclusions: print(f" No external files: {len(exclusions)} cores (--verbose lists them)") gt_cov = result.get("ground_truth_coverage") if gt_cov and not gt_cov.get("applicable", True): # MiSTer declares no core by design: its BiosDB is the answer, and # the site shows "-" for it rather than a 0% gap. print(" Ground truth: not applicable, the platform runs no profiled core") elif gt_cov and gt_cov["total"] > 0: # Rounded, as the site rounds: truncated, RetroArch read 20% here # and 21% there for one measure. pct = round(gt_cov["with_validation"] * 100 / gt_cov["total"]) print( f" Ground truth: {gt_cov['with_validation']}/{gt_cov['total']} files have emulator validation ({pct}%)" ) no_profile = gt_cov["total"] - gt_cov.get("with_profile", gt_cov["with_validation"]) if no_profile: print(f" {no_profile} platform-only (no emulator profile)") # Emulator/system mode verification def _effective_validation_label(details: list[dict], validation_index: dict) -> str: """Determine the bracket label for the report. Returns the union of all check types used, e.g. [crc32+existence+size]. """ all_checks: set[str] = set() has_files = False for d in details: fname = d.get("name", "") if d.get("note"): continue # skip informational entries (empty profiles) has_files = True entry = validation_index.get(fname) if entry: all_checks.update(entry["checks"]) else: all_checks.add("existence") if not has_files: return "existence" return "+".join(sorted(all_checks)) def _select_profiles( profile_names: list[str], all_profiles: dict, standalone: bool, ) -> list[tuple[str, dict]]: """Resolve the named profiles, stopping the run on a bad name.""" try: return select_emulator_profiles(profile_names, all_profiles, standalone) except ProfileSelectionError as exc: print(f"Error: {exc}", file=sys.stderr) sys.exit(1) def verify_emulator( profile_names: list[str], emulators_dir: str, db: dict, standalone: bool = False, regions: list[str] | None = None, platforms_dir: str = _REPO_PLATFORMS, system_ids: list[str] | None = None, ) -> dict: """Verify files for specific emulator profiles. A region priority list narrows the report the same way a pack built with the same list would be narrowed, through the emulator pack's own drop set. `system_ids` keeps the entries of those systems, as the system pack does. """ from packextras import expand_agnostic_entries from packresolve import resolve_file from packpaths import ( _archive_destination, _register_path, _resolve_destination, placement_block, ) # Also load aliases for redirect messages all_profiles = load_emulator_profiles(emulators_dir, skip_aliases=False) selected = _select_profiles(profile_names, all_profiles, standalone) # The member index only answers a zipped_file entry, and decompressing # every archive took 23 of the 24 seconds of a run that had none. zip_contents = ( build_zip_contents_index(db) if any(f.get("zipped_file") for _name, profile in selected for f in profile.get("files") or []) else {} ) # Build validation index from selected profiles only selected_profiles = {n: p for n, p in selected} validation_index = _build_validation_index(selected_profiles) data_registry = load_data_dir_registry(platforms_dir) details = [] file_status: dict[str, str] = {} file_severity: dict[str, str] = {} dest_to_name: dict[str, str] = {} data_dir_notices: list[str] = [] # The paths the emulator pack writes a profile file to. A copy whose hash # the profile contradicts is written too (UNTESTED here, a DISCREPANCY in # the pack), so the status alone does not say what the pack holds. packed_dests: set[str] = set() # The emulator pack withdraws these, and the report withdraws the same. region_drops: set[str] = set() if regions: from packextras import emulator_region_drops region_drops = emulator_region_drops(selected, standalone, regions, db=db) # The emulator pack walks profiles in name order and keeps the first # file at a path, folding case: a later profile's different file there # (zesarux pentagon.rom after speccy's PENTAGON.ROM) is not in the pack. sha1_by_path = { entry.get("path"): sha1 for sha1, entry in db.get("files", {}).items() } # Destination -> (file packed there, its profile, the path as written). # Archives and data trees take paths too, before and after the loose # files of their profile, as the pack writes them. placed: dict[str, tuple[str | None, str, str]] = {} placed_files: set[str] = set() placed_parents: set[str] = set() content_ids: dict[str, str | None] = {} def content_id(path: str | None) -> str | None: if not path: return None if path not in content_ids: content_ids[path] = sha1_by_path.get(path) or ( compute_hashes(path)["sha1"] if os.path.isfile(path) else None ) return content_ids[path] def occupy(dest: str, path: str | None, owner: str) -> None: if placement_block(dest, placed_files, set(placed), placed_parents, True): return placed[dest.lower()] = (path, owner, dest) _register_path(dest, placed_files, placed_parents) for emu_name, profile in sorted(selected): files = expand_agnostic_entries( expand_directory_entries( deliverable(profile.get("files", []), standalone), db, standalone, ), emu_name, profile, db, standalone, all_profiles, ) if system_ids: files = [fe for fe in files if entry_serves_systems(fe, profile, system_ids)] if region_drops: # Loose files only, as the emulator pack: an archive ships whole, # and dropping kronos' regional stvbios.zip members rebuilt the # archive here from 4 of its 14 members and changed its required. files = [ fe for fe in files if fe.get("archive") or _resolve_destination(fe, standalone) not in region_drops ] # Check data directories (only notice if not cached) for dd in deliverable( profile.get("data_directories", []), standalone ): ref = dd.get("ref", "") if not ref: continue if data_registry and ref in data_registry: cache_path = data_registry[ref].get("local_cache", "") if cache_path and os.path.isdir(cache_path): continue # cached, no notice needed data_dir_notices.append(ref) if not files: details.append( { "name": f"({emu_name})", "status": Status.OK, "required": False, "system": "", "note": f"No files needed for {profile.get('emulator', emu_name)}", "ground_truth": [], } ) continue # Verify archives as units (e.g., neogeo.zip, aes.zip) seen_archives: set[str] = set() for file_entry in files: archive = file_entry.get("archive") if archive and archive not in seen_archives: seen_archives.add(archive) local_path, resolve_status = resolve_file( archive_entry(archive, files), db, DEFAULT_BIOS_DIR, zip_contents, data_dir_registry=data_registry, offline=True, ) required = any( f.get("archive") == archive and f.get("required", True) for f in files ) if local_path and resolve_status == "hash_mismatch": # Same policy as the loose-file branch below: the name # matched while the bytes contradict the hash the profile # declares on the container. Reporting it covered says the # collection holds an archive it does not. result = { "name": archive, "status": Status.UNTESTED, "required": required, "path": local_path, "reason": "declared hash contradicted by the local file", } elif local_path: result = { "name": archive, "status": Status.OK, "required": required, "path": local_path, } else: result = { "name": archive, "status": Status.MISSING, "required": required, } result["system"] = file_entry.get("system", "") result["hle_fallback"] = False result["ground_truth"] = build_ground_truth(archive, validation_index) details.append(result) # Where the pack puts it: an archive_prefix or pack_structure # moves it, and a loose same-named entry is another file. dest = _archive_destination( archive, profile.get("archive_prefix", ""), profile.get("pack_structure"), standalone, ) result["destination"] = dest block = placement_block( dest, placed_files, set(placed), placed_parents, True ) if block and local_path and result["status"] != Status.MISSING: other_path, owner, other_dest = placed.get(dest.lower(), (None, "", "")) if block == "path_conflict" or content_id(local_path) != content_id(other_path): # The pack serves the other profile's file here: this # archive is not in it, whatever this profile resolves. result["status"] = Status.MISSING result["reason"] = ( "a packed file occupies its directory" if block == "path_conflict" else f"the pack holds {owner}'s {other_dest} at this path" ) elif local_path: occupy(dest, local_path, emu_name) packed_dests.add(dest) dest_to_name[dest] = archive cur = result["status"] prev = file_status.get(dest) if prev is None or _STATUS_ORDER.get(cur, 0) > _STATUS_ORDER.get( prev, 0 ): file_status[dest] = cur sev = _emulator_severity(cur, required, False) prev_sev = file_severity.get(dest) if prev_sev is None or _SEVERITY_ORDER.get( sev, 0 ) > _SEVERITY_ORDER.get(prev_sev, 0): file_severity[dest] = sev # Results of this emulator, by destination. Several entries of one # profile on one destination are the releases the code accepts # there (a shareware and a retail HOG, each GRP in a size table): # holding one of them fills the slot, and the others are not gaps. emu_results: dict[str, list[tuple[dict, bool, bool]]] = {} for file_entry in files: # Skip archived files (verified as archive units above) if file_entry.get("archive"): continue dest_hint = file_entry.get("path", "") # The emulator pack's resolution, release assets included. local_path, resolve_status = resolve_file( {**with_archive_members(file_entry, files), "source_profile": emu_name}, db, DEFAULT_BIOS_DIR, zip_contents, dest_hint=dest_hint, data_dir_registry=data_registry, offline=True, ) if resolve_status in ("not_found", "user_provided", "external"): local_path = None name = file_entry.get("name", "") required = file_entry.get("required", True) hle = file_entry.get("hle_fallback", False) stored = _storage_verdict(file_entry, emulator_pack=True) if stored is not None: result = stored elif not local_path: result = {"name": name, "status": Status.MISSING, "required": required} # An entry the profile documents as unsourceable is absent by # design: a per-user key, a slot the user fills, a dump nobody # has made. Counting it beside a file somebody could still # find invites the wrong repair: dropping the flag, or # chasing a vendor's whole install tree. unsourceable = file_entry.get("unsourceable") if unsourceable: result["unsourceable"] = ( unsourceable if isinstance(unsourceable, str) else "" ) else: # Apply emulator validation # Judged by this emulator's own entries, as it runs its check, # through the selection the emulator pack makes. chosen, disagreement = validated_choice( file_entry, local_path, db, validation_index, DEFAULT_BIOS_DIR, None, emulators={emu_name}, ) if disagreement: result = { "name": name, "status": Status.UNTESTED, "required": required, "path": local_path, "reason": disagreement.split(" says ", 1)[-1], } elif chosen != local_path: result = { "name": name, "status": Status.OK, "required": required, "path": chosen, } elif resolve_status == "hash_mismatch": # Nothing in validation: caught it, but the name matched # while the bytes contradict the hash the profile # declares. That is a different file wearing the right # name (config.ini and ROM collide across systems), so # calling it covered reports the collection as holding # something it does not. Validation runs first because it # names the specific field that disagrees. result = { "name": name, "status": Status.UNTESTED, "required": required, "path": local_path, "reason": "declared hash contradicted by the local file", } else: result = { "name": name, "status": Status.OK, "required": required, "path": local_path, } result["system"] = file_entry.get("system", "") result["hle_fallback"] = hle result["ground_truth"] = build_ground_truth(name, validation_index) # The slot the emulator pack places this entry in: standalone_path # under --standalone, the sanitised path. dest = _resolve_destination(file_entry, standalone) or name emu_results.setdefault(dest, []).append((result, required, hle)) for dest, alternatives in emu_results.items(): # What the emulator pack places here: an accepted file, else a # file whose hash the profile contradicts. A refused copy is held # but never packed. fallback_reason = "declared hash contradicted by the local file" if len(alternatives) > 1: # Reduced in the order generate_emulator_pack chooses, so the # destination's status is the one the pack reports: a slot it # fills with a contradicted copy, or counts as refused by the # core, is not missing because another alternative is absent. for keep in ( lambda a: a[0]["status"] == Status.OK, lambda a: a[0]["status"] == Status.UNTESTED and a[0].get("reason") == fallback_reason, lambda a: a[0]["status"] != Status.MISSING, ): chosen = [a for a in alternatives if keep(a)] if chosen: alternatives = chosen[:1] break packed = next( (a[0].get("path") for a in alternatives if a[0]["status"] == Status.OK and a[0].get("path")), None, ) or next( (a[0].get("path") for a in alternatives if a[0]["status"] == Status.UNTESTED and a[0].get("reason") == fallback_reason and a[0].get("path")), None, ) held = packed or next( (a[0].get("path") for a in alternatives if a[0]["status"] != Status.MISSING and a[0].get("path")), None, ) block = placement_block( dest, placed_files, set(placed), placed_parents, True ) block_reason = "" if block and held is not None: other_path, owner, other_dest = placed.get(dest.lower(), (None, "", "")) same = content_id(held) is not None and content_id(held) == content_id( other_path ) if block == "path_conflict": block_reason = "a packed file occupies its directory" elif block == "case_twin": block_reason = ( f"packed only as {other_dest}, from {owner}" if same else f"the pack holds {owner}'s {other_dest} at this path" ) elif not same: block_reason = f"the pack holds {owner}'s {other_dest} at this path" elif not block and packed is not None: occupy(dest, packed, emu_name) packed_dests.add(dest) for result, required, hle in alternatives: if block_reason and result["status"] != Status.MISSING: result["status"] = Status.MISSING result["reason"] = block_reason result["destination"] = dest details.append(result) # Aggregate by destination (path if available, else name) dest_to_name[dest] = result["name"] cur = result["status"] prev = file_status.get(dest) if prev is None or _STATUS_ORDER.get(cur, 0) > _STATUS_ORDER.get( prev, 0 ): file_status[dest] = cur sev = _emulator_severity(cur, required, hle) prev_sev = file_severity.get(dest) if prev_sev is None or _SEVERITY_ORDER.get( sev, 0 ) > _SEVERITY_ORDER.get(prev_sev, 0): file_severity[dest] = sev # The profile's data trees follow its files in the pack. for dd in deliverable(profile.get("data_directories", []), standalone): cache = (data_registry or {}).get(dd.get("ref", ""), {}).get("local_cache", "") if not cache or not os.path.isdir(cache): continue dd_dest = dd.get("destination", "") structure = profile.get("pack_structure") prefix = ( structure.get("standalone" if standalone else "libretro", "") if structure else "" ) if prefix: dd_dest = f"{prefix}/{dd_dest}" if dd_dest else prefix for root, _dirs, names in os.walk(cache): for fname in names: src = os.path.join(root, fname) rel = os.path.relpath(src, cache).replace(os.sep, "/") occupy(f"{dd_dest}/{rel}" if dd_dest else rel, src, emu_name) counts = { Severity.OK: 0, Severity.INFO: 0, Severity.WARNING: 0, Severity.CRITICAL: 0, } for s in file_severity.values(): counts[s] = counts.get(s, 0) + 1 status_counts: dict[str, int] = {} for s in file_status.values(): status_counts[s] = status_counts.get(s, 0) + 1 label = _effective_validation_label(details, validation_index) gt_filenames = set(validation_index) total = len(file_status) with_validation = sum( 1 for dest in file_status if dest_to_name.get(dest, "") in gt_filenames ) return { "emulators": [n for n, _ in selected], "verification_mode": label, "total_files": total, # The paths the emulator pack fills with the profile's files, for the # pipeline's comparison with the pack built in memory. "pack_destinations": sorted(packed_dests), "severity_counts": counts, "status_counts": status_counts, "details": details, "data_dir_notices": sorted(set(data_dir_notices)), "ground_truth_coverage": { "with_validation": with_validation, "platform_only": total - with_validation, "total": total, }, } def verify_system( system_ids: list[str], emulators_dir: str, db: dict, standalone: bool = False, regions: list[str] | None = None, platforms_dir: str = _REPO_PLATFORMS, ) -> dict: """Verify files for all emulators supporting given system IDs.""" profiles = load_emulator_profiles(emulators_dir) matching = profiles_for_systems(profiles, system_ids, standalone) if not matching: all_systems: set[str] = set() for p in profiles.values(): all_systems.update(p.get("systems", [])) if standalone: print( f"No standalone emulators found for system(s): {', '.join(system_ids)}", file=sys.stderr, ) else: print( f"No emulators found for system(s): {', '.join(system_ids)}", file=sys.stderr, ) print(suggest_names(system_ids, sorted(all_systems), "--list-systems"), file=sys.stderr) sys.exit(1) return verify_emulator( matching, emulators_dir, db, standalone, regions=regions, platforms_dir=platforms_dir, system_ids=system_ids, ) def print_emulator_result(result: dict, verbose: bool = False) -> None: """Print verification result for emulator/system mode.""" label = " + ".join(result["emulators"]) mode = result["verification_mode"] total = result["total_files"] c = result["severity_counts"] ok_count = c[Severity.OK] sc = result.get("status_counts", {}) untested = sc.get(Status.UNTESTED, 0) missing = sc.get(Status.MISSING, 0) # Counted by slot, as missing is: xash3d's thirteen unsourceable # pak0.pak and liblist.gam were three names and ten "missing". missing_slots: dict[str, bool] = {} for d in result["details"]: if d["status"] == Status.MISSING: slot = d.get("destination") or d["name"] missing_slots[slot] = missing_slots.get(slot, True) and "unsourceable" in d unsourceable = sum(missing_slots.values()) parts = [f"{ok_count}/{total} OK"] if untested: parts.append(f"{untested} untested") if missing - unsourceable > 0: parts.append(f"{missing - unsourceable} missing") if unsourceable: parts.append(f"{unsourceable} unsourceable") print(f"{label}: {', '.join(parts)} [{mode}]") seen = set() for d in result["details"]: if d["status"] == Status.UNTESTED: if d["name"] in seen: continue seen.add(d["name"]) req = "required" if d.get("required", True) else "optional" hle = ", HLE available" if d.get("hle_fallback") else "" reason = d.get("reason", "") print(f" UNTESTED ({req}{hle}): {d['name']}: {reason}") gt = d.get("ground_truth", []) if gt: if verbose: for line in _format_ground_truth_verbose(gt): print(f" {line}") else: print(f" {_format_ground_truth_aggregate(gt)}") for d in result["details"]: if d["status"] == Status.MISSING and "unsourceable" not in d: if d["name"] in seen: continue seen.add(d["name"]) req = "required" if d.get("required", True) else "optional" hle = ", HLE available" if d.get("hle_fallback") else "" print(f" MISSING ({req}{hle}): {d['name']}") gt = d.get("ground_truth", []) if gt: if verbose: for line in _format_ground_truth_verbose(gt): print(f" {line}") else: print(f" {_format_ground_truth_aggregate(gt)}") for d in result["details"]: if d["status"] == Status.MISSING and "unsourceable" in d: if d["name"] in seen: continue seen.add(d["name"]) why = d.get("unsourceable") or "documented as unobtainable" print(f" UNSOURCEABLE: {d['name']}: {why}") for d in result["details"]: if d.get("note"): print(f" {d['note']}") if verbose: for d in result["details"]: if d["status"] == Status.OK: if d["name"] in seen: continue seen.add(d["name"]) gt = d.get("ground_truth", []) if gt: req = "required" if d.get("required", True) else "optional" print(f" OK ({req}): {d['name']}") for line in _format_ground_truth_verbose(gt): print(f" {line}") for ref in result.get("data_dir_notices", []): print( f" Note: data directory '{ref}' required but not included (use refresh_data_dirs.py)" ) gt_cov = result.get("ground_truth_coverage") if gt_cov and gt_cov["total"] > 0: pct = round(gt_cov["with_validation"] * 100 / gt_cov["total"]) print( f" Ground truth: {gt_cov['with_validation']}/{gt_cov['total']} files have emulator validation ({pct}%)" ) if gt_cov["platform_only"]: print(f" {gt_cov['platform_only']} without declared validation") def _refuse_listing_narrowings( args: argparse.Namespace, parser: argparse.ArgumentParser ) -> None: """A listing mode reads --platform at most, anything else is refused. Printing past a narrowing flag lets the user believe it applied (generate_pack applies the same table). """ listing = next( (flag for flag, on in ( ("--list-emulators", args.list_emulators), ("--list-systems", args.list_systems), ("--list-targets", args.list_targets), ) if on), None, ) if not listing: return reads_platform = listing != "--list-emulators" for flag, on in ( ("--platform", args.platform and not reads_platform), ("--all", args.all), ("--emulator", args.emulator), ("--system", args.system), ("--region", getattr(args, "region", None)), ("--target", getattr(args, "target", None)), ("--standalone", args.standalone), ("--include-archived", args.include_archived), ("--json", args.json), ("--strict", args.strict), ("--verbose", args.verbose), ("--destinations", args.destinations), ("--offline", args.offline), ): if on: parser.error(f"{flag} is incompatible with {listing}") def _refuse_mode_flags(args: argparse.Namespace, parser: argparse.ArgumentParser) -> None: """A flag the chosen mode does not read is refused, never ignored.""" for refused, message in ( (args.standalone and not (args.emulator or args.system), "--standalone requires --emulator or --system"), (args.standalone and args.platform, "--standalone is incompatible with --platform"), (args.include_archived and not args.all, "--include-archived requires --all"), (args.target and not (args.platform or args.all), "--target requires --platform or --all"), (args.target and (args.emulator or (args.system and not args.platform)), "--target is incompatible with --emulator and --system"), (args.destinations and args.system, "--destinations describes the full pack and is incompatible with --system"), (args.destinations and not (args.platform or args.all), "--destinations requires --platform or --all"), (args.offline and not (args.platform or args.all), "--offline requires --platform or --all"), (args.destinations and getattr(args, "region", None), "--destinations describes the full pack and is incompatible with --region"), ): if refused: parser.error(message) def _run_listing(args: argparse.Namespace, parser: argparse.ArgumentParser) -> None: if args.list_emulators: list_emulator_profiles(args.emulators_dir) elif args.list_systems and args.platform: list_platform_system_ids(args.platform, args.platforms_dir) elif args.list_systems: list_system_ids(args.emulators_dir) else: if not args.platform: parser.error("--list-targets requires --platform") targets = list_available_targets(args.platform, args.platforms_dir) if not targets: print(f"No targets configured for platform '{args.platform}'") return for t in targets: print(target_listing_line(t)) def main() -> None: with contextlib.ExitStack() as held: _main(held) def _main(held: contextlib.ExitStack) -> None: parser = argparse.ArgumentParser(allow_abbrev=False, description="Platform-native BIOS verification") parser.add_argument("--platform", "-p", type=platform_arg, help="Platform name") parser.add_argument( "--all", action="store_true", help="Verify all active platforms" ) parser.add_argument( "--emulator", "-e", help="Emulator profile name(s), comma-separated" ) parser.add_argument("--system", "-s", help="System ID(s), comma-separated") parser.add_argument("--standalone", action="store_true", help="Use standalone mode") parser.add_argument( "--list-emulators", action="store_true", help="List available emulators" ) parser.add_argument( "--list-systems", action="store_true", help="List available systems" ) parser.add_argument( "--include-archived", action="store_true", help="With --all, also verify the archived platforms (RetroPie)", ) parser.add_argument("--target", "-t", help="Hardware target (e.g., switch, rpi4)") parser.add_argument( "--region", help="Region priority list, best first (e.g. us,eu,jp)" ) parser.add_argument( "--list-targets", action="store_true", help="List available targets for the platform", ) parser.add_argument( "--db", default=DEFAULT_DB, help="database file to read (default: %(default)s)", ) parser.add_argument( "--platforms-dir", default=DEFAULT_PLATFORMS_DIR, help="platform YAML directory (default: %(default)s)", ) parser.add_argument( "--emulators-dir", default=DEFAULT_EMULATORS_DIR, help="emulator profile directory (default: %(default)s)", ) parser.add_argument( "--verbose", "-v", action="store_true", help="Show emulator ground truth details", ) parser.add_argument("--json", action="store_true", help="JSON output") parser.add_argument( "--offline", action="store_true", help="Report as an offline build packs: a release asset the cache " "lacks is missing. Without it, it counts as fetched at build time", ) parser.add_argument( "--destinations", metavar="FILE", help="With --platform or --all, also write the paths each pack holds " "according to this report to FILE (JSON), for comparison with the archive", ) parser.add_argument( "--strict", action="store_true", help="Exit 1 on a required file missing: where the frontend checks " "hashes for a platform, without HLE for --emulator and --system. A file " "present with an unconfirmed hash is a warning, not a failure", ) args = parser.parse_args() work_from_repo_root(args, parser, ["db", "platforms_dir", "emulators_dir"]) if args.platform: require_platform(args.platform, args.platforms_dir) if args.offline: import packresolve packresolve.set_offline(True) requested_regions: list[str] = [] if getattr(args, "region", None): import region as region_mod try: requested_regions = region_mod.parse_requested(args.region) except ValueError as exc: parser.error(str(exc)) _refuse_listing_narrowings(args, parser) if args.list_emulators or args.list_systems or args.list_targets: _run_listing(args, parser) return # --system beside --platform narrows that platform, as in generate_pack. system_mode = bool(args.system) and not args.platform modes = sum(1 for x in (args.platform, args.all, args.emulator, system_mode) if x) if modes == 0: parser.error("Specify --platform, --all, --emulator, or --system") if modes > 1: parser.error( "--platform, --all, --emulator, and --system are mutually exclusive" ) _refuse_mode_flags(args, parser) platform_systems: list[str] | None = None if args.platform and args.system: config = load_platform_config(args.platform, args.platforms_dir) try: platform_systems = resolve_system_arg( args.system, list(config.get("systems", {})), config, f"--platform {args.platform} --list-systems", ) except ValueError as exc: parser.error(f"--system {args.system}: {exc}") # The trees stay put while names resolve against them and files are read. from refresh_data_dirs import hold_data_dirs hold_data_dirs(held, load_data_dir_registry(args.platforms_dir)) db = load_database(args.db) if args.emulator: names = [n.strip() for n in args.emulator.split(",") if n.strip()] result = verify_emulator( names, args.emulators_dir, db, args.standalone, regions=requested_regions, platforms_dir=args.platforms_dir, ) if args.json: result["details"] = _json_details(result["details"], args.verbose) print(json.dumps(result, indent=2)) else: print_emulator_result(result, verbose=args.verbose) _exit_if_strict_emulator(args.strict, result) return if system_mode: system_ids = [s.strip() for s in args.system.split(",") if s.strip()] result = verify_system( system_ids, args.emulators_dir, db, args.standalone, regions=requested_regions, platforms_dir=args.platforms_dir, ) if args.json: result["details"] = _json_details(result["details"], args.verbose) print(json.dumps(result, indent=2)) else: print_emulator_result(result, verbose=args.verbose) _exit_if_strict_emulator(args.strict, result) return # Platform mode (existing) if args.all: platforms = list_registered_platforms( args.platforms_dir, include_archived=args.include_archived ) elif args.platform: platforms = [args.platform] else: parser.error("Specify --platform or --all") return # Load emulator profiles once for cross-reference (not per-platform) emu_profiles = load_emulator_profiles(args.emulators_dir) data_registry = load_data_dir_registry(args.platforms_dir) target_cores_cache: dict[str, set[str] | None] = {} if args.target: try: target_cores_cache, platforms = build_target_cores_cache( platforms, args.target, args.platforms_dir, is_all=args.all, ) except (FileNotFoundError, ValueError) as e: print(f"ERROR: {e}", file=sys.stderr) sys.exit(1) # Group identical platforms (same function as generate_pack) groups = group_identical_platforms( platforms, args.platforms_dir, target_cores_cache if args.target else None ) all_results = {} group_results: list[tuple[dict, list[str]]] = [] for group_platforms, representative in groups: config = load_platform_config(representative, args.platforms_dir) tc = target_cores_cache.get(representative) if args.target else None result = verify_platform( config, db, args.emulators_dir, emu_profiles, target_cores=tc, data_dir_registry=data_registry, regions=requested_regions, system_filter=platform_systems, ) names = [ load_platform_config(p, args.platforms_dir).get("platform", p) for p in group_platforms ] group_results.append((result, names)) for p in group_platforms: all_results[p] = result if not args.json: for result, group in group_results: print_platform_result(result, group, verbose=args.verbose) print() if len(group_results) > 1: print(_overall_summary(group_results)) if args.destinations: held = {} for (result, names), (group, _rep) in zip(group_results, groups): for name, key in zip(names, group): held[name] = { "platform": key, "pack_destinations": result["pack_destinations"], } Path(args.destinations).write_text(json.dumps(held, indent=1, sort_keys=True)) if args.json: for r in all_results.values(): r["details"] = _json_details(r["details"], args.verbose) print(json.dumps(all_results, indent=2)) _exit_if_strict(args.strict, [r for r, _ in group_results]) def _json_details(details: list[dict], verbose: bool) -> list[dict]: """The entries a JSON report lists: what needs attention, or every one. --verbose shows each entry with its ground truth in the text report. It used to be accepted beside --json and dropped, the output byte for byte the same, so a script asking for the passing entries got nothing. """ return list(details) if verbose else [ d for d in details if d["status"] != Status.OK or d.get("release_asset") ] def _exit_if_strict(strict: bool, results: list[dict]) -> None: """Exit 1 under --strict when a platform result counts a critical file: a required file missing where the frontend reads hashes. A file present whose hash the frontend does not confirm is UNTESTED, a warning (Batocera's BiosStatus calls it so): reported, not failed. """ critical = sum(r["severity_counts"].get(Severity.CRITICAL, 0) for r in results) if strict and critical: print(f"ERROR: {critical} required file(s) missing where the frontend " "checks hashes", file=sys.stderr) sys.exit(1) def _exit_if_strict_emulator(strict: bool, result: dict) -> None: """Exit 1 under --strict when an emulator lacks a file it cannot run without. The emulator, not a frontend, is the judge here: a required file with no HLE fallback that the collection does not hold stops the core. """ missing = sorted({ d["name"] for d in result["details"] if d["status"] == Status.MISSING and d.get("required", True) and not d.get("hle_fallback") }) if strict and missing: print(f"ERROR: {len(missing)} required file(s) missing: " f"{', '.join(missing)}", file=sys.stderr) sys.exit(1) def _overall_summary(group_results: list[tuple[dict, list[str]]]) -> str: """The closing lines of a multi-platform run, worst severity first. Each count names the platforms carrying it, and the core files a core cannot run without are counted too: a reader of the summary alone used to conclude nothing required was missing, and could not tell where the one warning was without reading the whole output. """ platforms = sum(len(group) for _, group in group_results) def where(counts: dict[str, int]) -> str: named = [f"{name} {count}" if count > 1 else name for name, count in counts.items() if count] return f" ({', '.join(named)})" if named else "" parts = [] for severity, label in ((Severity.CRITICAL, "critical"), (Severity.WARNING, "warning"), (Severity.INFO, "info")): counts = { " / ".join(group): r["severity_counts"].get(severity, 0) for r, group in group_results } parts.append(f"{sum(counts.values())} {label}{where(counts)}") # Shipped files whose bytes contradict a declaration or a core's own # check: the body printed 19 of them under a last line that read clean. counts = { " / ".join(group): len(_discrepancy_keys(r.get("details", []))) for r, group in group_results } total = sum(counts.values()) parts.append(f"{total} {'discrepancy' if total == 1 else 'discrepancies'}{where(counts)}") lines = [f"{platforms} platforms: {', '.join(parts)}"] missing = { " / ".join(group): sorted({ u["name"] for u in _core_required_missing(_split_undeclared( r.get("undeclared_files", []))[1]) }) for r, group in group_results } total = sum(len(names) for names in missing.values()) if total: detail = "; ".join(f"{group}: {', '.join(names)}" for group, names in missing.items() if names) lines.append(f"{total} core file(s) a core cannot run without are missing ({detail})") for category, label in _NOT_FIRMWARE.items(): counts = { " / ".join(group): len((r.get("missing_not_firmware") or {}).get(category, [])) for r, group in group_results } total = sum(counts.values()) if total: lines.append(f"{total} platform entries missing, not firmware, {label}{where(counts)}") return "\n".join(lines) if __name__ == "__main__": from common import run_cli run_cli(main)