From 5ef7a9fabf66d54b61f8e3089352f5bd1842288a Mon Sep 17 00:00:00 2001
From: Abdessamad Derraz <3028866+Abdess@users.noreply.github.com>
Date: Thu, 5 Nov 2026 11:57:08 +0100
Subject: [PATCH] refactor: hash each profile entry once

---
 scripts/exporter/candidates.py | 27 +++++++++++++++++++++------
 tests/test_entry_categories.py | 23 +++++++++++++++++++++++
 2 files changed, 44 insertions(+), 6 deletions(-)

diff --git a/scripts/exporter/candidates.py b/scripts/exporter/candidates.py
index 59df9a4ea..5ceb69f2d 100644
--- a/scripts/exporter/candidates.py
+++ b/scripts/exporter/candidates.py
@@ -55,11 +55,16 @@ def _same(declared: str, accepted: str) -> bool:
     )
 
 
-def _accepts(profile_entry: dict, declared: dict[str, set[str]]) -> bool:
-    """True when the profile entry, or a member of its archive, accepts a
-    hash the platform entry declares."""
-    accepted = _entry_hashes(profile_entry)
-    return any(_same(d, a) for values in declared.values() for d in values for a in accepted)
+def _accepts(accepted: set[str], declared: dict[str, set[str]]) -> bool:
+    """True when the hashes a profile entry accepts, the members of its
+    archive included, hold one the platform entry declares."""
+    for values in declared.values():
+        if values & accepted:
+            return True
+        cut = [d for d in values if _CUT_MD5 <= len(d) < _FULL_MD5]
+        if cut and any(a.startswith(d) for d in cut for a in accepted):
+            return True
+    return False
 
 
 def _contradicts(profile_entry: dict, declared: dict[str, set[str]]) -> bool:
@@ -366,6 +371,10 @@ class EntryCategories:
         self._guessed = not config.get("cores")
         self._romsets = sorted(key for key in examined if _loads_romsets(profiles[key]))
         self._profiles = profiles
+        # The hashes each profile entry accepts, read once: every platform
+        # entry with a proof walks its reader's whole profile, and a MAME
+        # profile's archive members were hashed again for each.
+        self._accepted: dict[int, set[str]] = {}
         # One index per core and per kind, so a proof names its reader.
         self._readers: list[tuple[str, str, ReadEvidence]] = []
         for key in sorted(examined):
@@ -403,6 +412,12 @@ class EntryCategories:
                     sorted(unresolved_platform_cores(own, profiles)),
                 )
 
+    def _accepted_by(self, profile_entry: dict) -> set[str]:
+        accepted = self._accepted.get(id(profile_entry))
+        if accepted is None:
+            accepted = self._accepted[id(profile_entry)] = _entry_hashes(profile_entry)
+        return accepted
+
     def _matching(self, key: str, kind: str, entry: dict) -> list[dict]:
         """The profile entries of `key` that read the platform entry.
 
@@ -425,7 +440,7 @@ class EntryCategories:
             if not isinstance(f, dict) or (f.get("category") == GAME_DATA) != (kind == GAME_DATA):
                 continue
             cited = {k: f[k] for k in ("name", "source_ref") if f.get(k)}
-            if declared and _accepts(f, declared):
+            if declared and _accepts(self._accepted_by(f), declared):
                 by_hash.append(cited)
                 continue
             own = {str(n).casefold() for n in (f.get("name"), *(f.get("aliases") or [])) if n}
diff --git a/tests/test_entry_categories.py b/tests/test_entry_categories.py
index e62a3570a..d3c68861c 100644
--- a/tests/test_entry_categories.py
+++ b/tests/test_entry_categories.py
@@ -42,6 +42,29 @@ def _config(cores: list[str], files: list[dict], system: str = "s") -> dict:
             "systems": {system: {"files": files}}}
 
 
+
+class EachProfileEntryIsHashedOnce(unittest.TestCase):
+    """verify --all hashed 9,058 profile entries 2.9 million times: each
+    platform entry with a proof walked its reader's profile from scratch."""
+
+    def test_two_platform_entries_read_one_profile_entry_once(self):
+        from unittest import mock
+
+        import exporter.candidates as candidates
+
+        profiles = {"core": {"emulator": "Core", "type": "libretro", "cores": ["core"],
+                             "systems": ["s"],
+                             "files": [{"name": "a.bin", "md5": "a" * 32,
+                                        "contents": [{"name": "m", "crc32": "12345678"}]}]}}
+        cats = EntryCategories(_config(["core"], []), profiles)
+        real = candidates._entry_hashes
+        with mock.patch.object(candidates, "_entry_hashes", side_effect=real) as counted:
+            for md5 in ("a" * 32, "a" * 29):
+                self.assertEqual(cats.category({"name": "a.bin", "md5": md5}, "a.bin", "s"), BIOS)
+        self.assertEqual(counted.call_count, 1)
+        self.assertTrue(candidates._accepts({"a" * 32}, {"md5": {"a" * 29}}))
+        self.assertFalse(candidates._accepts({"a" * 32}, {"md5": {"b" * 32}}))
+
 class Category(unittest.TestCase):
     def test_each_category(self):
         cats = EntryCategories(_config(["bluemsx", "devilutionx"], []), PROFILES)