fix: read gaps by normalized system, sha256, file names

tests/test_cross_reference_sources.py

@@ -0,0 +1,59 @@
+"""The gap report reads the collection the way the resolver does.
+
+Three ways it called held files missing, or a missing file held: system IDs
+compared verbatim (sega-megacd against sega-mega-cd), no sha256 in the hash
+fallback (rust_dos names its SC-55 ROMs by sha256 alone), and directory names
+indexed as files (BasiliskII's required ROM read as held because cpcemu keeps
+a directory called ROM).
+"""
+
+from __future__ import annotations
+
+import sys
+import tempfile
+import unittest
+from pathlib import Path
+
+REPO_ROOT = Path(__file__).resolve().parent.parent
+sys.path.insert(0, str(REPO_ROOT / "scripts"))
+
+from cross_reference import (  # noqa: E402
+    _build_supplemental_index,
+    cross_reference,
+    entry_source,
+)
+
+
+def _index(**over) -> dict:
+    base = {
+        "by_name": {}, "by_name_lower": {}, "by_path_suffix": {}, "by_md5": {},
+        "by_sha256": {}, "by_crc32": {}, "db_files": {}, "data_names": set(),
+    }
+    base.update(over)
+    return base
+
+
+class GapReportSources(unittest.TestCase):
+    def test_a_sha256_alone_finds_the_file(self):
+        entry = {"name": "wave1.bin", "sha256": "AB" * 32}
+        self.assertEqual(entry_source(entry, _index(by_sha256={"ab" * 32: "s"})), "bios")
+
+    def test_a_directory_does_not_stand_for_a_file(self):
+        with tempfile.TemporaryDirectory(dir=REPO_ROOT / "tmp") as tmp:
+            (Path(tmp) / "bios" / "cpcemu" / "ROM").mkdir(parents=True)
+            names = _build_supplemental_index(str(Path(tmp) / "data"), str(Path(tmp) / "bios"))
+        self.assertIsNone(entry_source({"name": "ROM"}, _index(data_names=names)))
+        directory = {"name": "ROM", "type": "directory"}
+        self.assertEqual(entry_source(directory, _index(data_names=names)), "data")
+
+    def test_a_system_spelled_otherwise_is_still_declared(self):
+        profiles = {"gpgx": {"emulator": "gpgx", "systems": ["sega-megacd"],
+                             "files": [{"name": "bios_CD_U.bin"}]}}
+        declared = {"megacd": {"bios_CD_U.bin"}}  # sega-mega-cd, normalized
+        db = {"files": {}, "indexes": {}}
+        report = cross_reference(profiles, declared, db)
+        self.assertEqual(report["gpgx"]["gaps"], 0)
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_e2e.py

@@ -5562,11 +5562,13 @@ struct BurnDriver BurnDrvneogeo = {
                 with open(os.path.join(plat_dir, f"{name}.yml"), "w") as f:
                     yaml.dump(cfg, f)
 
+            # Keyed by the normalized system ID, as profiles spell systems
+            # differently from platforms.
             everything, _ = load_platform_files(plat_dir)
-            self.assertEqual(everything["shared-system"], {"a.bin", "b.bin"})
+            self.assertEqual(everything["sharedsystem"], {"a.bin", "b.bin"})
 
             only_alpha, _ = load_platform_files(plat_dir, ["alpha"])
-            self.assertEqual(only_alpha["shared-system"], {"a.bin"})
+            self.assertEqual(only_alpha["sharedsystem"], {"a.bin"})
 
 
     def test_227_manifest_injected_compressed(self):