fix: never pair files whose contents or dirs differ

scripts/exporter/baseline.py

@@ -323,13 +323,21 @@ def build_native_model(
             ]
 
             def by_destination(candidate: NativeFile) -> bool:
+                # Dolphin writes dolphin-emu/Sys/GC/USA/IPL.bin for the truth's
+                # GC/USA/IPL.bin: a destination ending in the path is the file.
                 theirs = _match_key(candidate.platform or {})[0]
-                return bool(t_dest) and theirs == t_dest
+                return bool(t_dest) and (theirs == t_dest or theirs.endswith("/" + t_dest))
 
             def by_name(candidate: NativeFile) -> bool:
-                theirs = _match_key(candidate.platform or {})[1]
+                theirs_dest, theirs = _match_key(candidate.platform or {})
                 if not t_name or theirs != t_name:
                     return False
+                # Two declared directories that differ name two files: the
+                # truth's BeebFile/BIOS.rom is not Recalbox's np2kai/bios.rom.
+                if "/" in t_dest and "/" in theirs_dest and not theirs_dest.endswith(
+                    "/" + t_dest.rsplit("/", 1)[0].rsplit("/", 1)[-1] + "/" + t_name
+                ):
+                    return False
                 # A shared name with two declared sizes is two files: fbneo's
                 # 480-byte boot.bin is not RomM's 2 MB Dreamcast boot.bin.
                 t_size = truth_entry.get("size")

scripts/truth.py

@@ -165,12 +165,30 @@ def _same_file(files: list[dict], file_entry: dict, emu_name: str) -> dict | Non
     same_name = [f for f in files if f["name"].lower() == name_lower]
     return (
         next((f for f in same_name if path and other_path(f) == path), None)
-        or next((f for f in same_name if emu_name not in f.get("_cores", ())), None)
+        # Another core's file of the same name is the same file unless their
+        # contents disagree: beebem's 16 KB BBC BIOS.rom and np2kai's PC-98
+        # bios.rom merged, and the export wrote the BBC md5 over Recalbox's.
+        or next(
+            (f for f in same_name
+             if emu_name not in f.get("_cores", ()) and not _contents_disagree(f, file_entry)),
+            None,
+        )
         # Revisions accepted under one name and no path fill one slot.
         or next((f for f in same_name if not path or not other_path(f)), None)
     )
 
 
+def _contents_disagree(a: dict, b: dict) -> bool:
+    """A hash both declare without a value in common, or two declared sizes."""
+    for field in ("sha1", "md5", "sha256", "crc32"):
+        ours = {str(v).lower() for v in (a.get(field) if isinstance(a.get(field), list) else [a.get(field)]) if v}
+        theirs = {str(v).lower() for v in (b.get(field) if isinstance(b.get(field), list) else [b.get(field)]) if v}
+        if ours and theirs and not ours & theirs:
+            return True
+    size_a, size_b = a.get("size"), b.get("size")
+    return isinstance(size_a, int) and isinstance(size_b, int) and size_a != size_b
+
+
 def _fold_into(existing: dict, file_entry: dict, emu_name: str) -> None:
     """Add one core's declaration to the entry that already stands for it."""
     existing["_cores"] = existing.get("_cores", set()) | {emu_name}

tests/test_export_counts.py

@@ -195,6 +195,16 @@ class MisterCountsWhatItWrites(unittest.TestCase):
         self.assertFalse(exporter.states(absent, "md5"))
 
 
+class OneMd5FormatsAddWhole(unittest.TestCase):
+    def test_an_addition_with_several_md5_is_refused(self):
+        from exporter.batocera_exporter import Exporter as Batocera  # noqa: PLC0415
+
+        several = NativeFile("boot.bin", "boot.bin", "dc", truth={"md5": [A, B, C]})
+        single = NativeFile("boot2.bin", "boot2.bin", "dc", truth={"md5": A})
+        self.assertFalse(Batocera.writable(several))
+        self.assertTrue(Batocera.writable(single))
+
+
 class ModelKeepsOneFileOneEntry(unittest.TestCase):
     def test_a_list_of_sizes_is_not_one_size(self):
         """A profile may accept several revisions; int() on the list crashed."""
@@ -275,6 +285,31 @@ class RommKeepsItsOwnKeys(unittest.TestCase):
         self.assertEqual(written["dc:boot.bin"]["size"], "2097152")
         self.assertFalse(exporter.writable(ours))
 
+class NamesInAnotherDirectory(unittest.TestCase):
+    def test_a_truth_path_in_another_directory_does_not_match(self):
+        scraped = {"systems": {"pc98": {"files": [
+            {"name": "bios.rom", "destination": "np2kai/bios.rom", "md5": A},
+        ]}}}
+        truth = {"systems": {"pc98": {"files": [
+            {"name": "bios.rom", "path": "BeebFile/BIOS.rom", "md5": B},
+        ]}}}
+        systems, _report = build_native_model(truth, scraped)
+        declared = next(f for f in systems["pc98"].files if f.platform is not None)
+        self.assertIsNone(declared.truth)
+
+    def test_a_destination_ending_in_the_path_matches(self):
+        scraped = {"systems": {"gc": {"files": [
+            {"name": "IPL.bin", "destination": "dolphin-emu/Sys/GC/USA/IPL.bin", "md5": A},
+            {"name": "IPL.bin", "destination": "dolphin-emu/Sys/GC/EUR/IPL.bin", "md5": B},
+        ]}}}
+        truth = {"systems": {"gc": {"files": [
+            {"name": "IPL.bin", "path": "GC/EUR/IPL.bin", "md5": B},
+            {"name": "IPL.bin", "path": "GC/USA/IPL.bin", "md5": A},
+        ]}}}
+        systems, report = build_native_model(truth, scraped)
+        self.assertEqual(report.hashes_corrected, [])
+
+
 class RecalboxKeepsItsOwnNotes(unittest.TestCase):
     def test_no_profile_prose_reaches_a_note(self):
         fe = NativeFile("bios.bin", "bios.bin", "psx", platform={"md5": A},

tests/test_truth_diff.py

@@ -115,6 +115,17 @@ class OneNameSeveralFiles(unittest.TestCase):
         self.assertEqual(_diff_system(self._system(), scraped), {})
 
 
+class DifferentContentsStayApart(unittest.TestCase):
+    def test_two_cores_same_name_different_contents(self):
+        """beebem's BBC BIOS.rom and np2kai's PC-98 bios.rom merged into one."""
+        system: dict = {}
+        _merge_file_into_system(system, {"name": "bios.rom", "path": "np2kai/bios.rom",
+                                         "md5": "a" * 32}, "np2kai", None)
+        _merge_file_into_system(system, {"name": "bios.rom", "path": "BeebFile/BIOS.rom",
+                                         "md5": "b" * 32, "size": 16384}, "beebem", None)
+        self.assertEqual(len(system["files"]), 2)
+
+
 class RenameMatching(unittest.TestCase):
     def test_a_shared_sha1_pairs_the_two_names(self):
         truth = [_entry("bios_CD_U.bin", sha1="a" * 40)]