From 3a6f2c8770499e021466151c4a453e3594410b76 Mon Sep 17 00:00:00 2001
From: Abdessamad Derraz <3028866+Abdess@users.noreply.github.com>
Date: Fri, 16 Oct 2026 14:29:05 +0200
Subject: [PATCH] fix: key the opcode cache by content

---
 scripts/profile_sync.py    | 16 +++++++++++-----
 tests/test_profile_sync.py | 17 +++++++++++++++++
 2 files changed, 28 insertions(+), 5 deletions(-)

diff --git a/scripts/profile_sync.py b/scripts/profile_sync.py
index 12d6be4f..c31b36d7 100644
--- a/scripts/profile_sync.py
+++ b/scripts/profile_sync.py
@@ -723,22 +723,28 @@ def _map_index(opcodes, index: int) -> int | None:
     return None
 
 
-_OPCODE_CACHE: dict[tuple[int, int], list] = {}
+_OPCODE_CACHE: dict[tuple[tuple[str, ...], tuple[str, ...]], list] = {}
+_OPCODE_CACHE_SIZE = 8
 
 
 def _opcodes(pin: list[str], head: list[str]) -> list:
     """Opcodes for one pair of revisions, computed once per run.
 
     Every ref citing the same file would otherwise redo the same diff, and a
-    driver of thirty thousand lines takes seconds each time.
+    driver of thirty thousand lines takes seconds each time. Keyed by the
+    lines themselves: anchor_block hands over lists it builds and frees, and
+    an id() key served one file's opcodes to the next pair at the same
+    addresses.
     """
-    key = (id(pin), id(head))
-    cached = _OPCODE_CACHE.get(key)
+    key = (tuple(pin), tuple(head))
+    cached = _OPCODE_CACHE.pop(key, None)
     if cached is None:
         cached = difflib.SequenceMatcher(
             None, pin, head, autojunk=False
         ).get_opcodes()
-        _OPCODE_CACHE[key] = cached
+    _OPCODE_CACHE[key] = cached
+    while len(_OPCODE_CACHE) > _OPCODE_CACHE_SIZE:
+        del _OPCODE_CACHE[next(iter(_OPCODE_CACHE))]
     return cached
 
 
diff --git a/tests/test_profile_sync.py b/tests/test_profile_sync.py
index d31adf31..145a688a 100644
--- a/tests/test_profile_sync.py
+++ b/tests/test_profile_sync.py
@@ -3397,6 +3397,23 @@ class TestWrappedProse(unittest.TestCase):
         self.assertEqual(self.path.read_text(), self.TEXT)
 
 
+class TestOpcodeCache(unittest.TestCase):
+    """The diff cache was keyed by id() of lists anchor_block builds and frees
+    on return: a later pair at the same addresses read another file's
+    opcodes, and a CHANGED ref could be mapped onto the wrong line."""
+
+    def test_each_pair_gets_its_own_diff(self):
+        import difflib
+
+        for n in range(200):
+            pin = [f"line {n} {i}" for i in range(n % 7 + 2)]
+            head = [f"line {n} {i}" for i in range(n % 5 + 1)] + [f"new {n}"]
+            expected = difflib.SequenceMatcher(
+                None, pin, head, autojunk=False
+            ).get_opcodes()
+            self.assertEqual(profile_sync._opcodes(list(pin), list(head)), expected)
+
+
 class TestCommentCitations(unittest.TestCase):
     """The loader drops comments: zesarux's `# rom_load switch,
     src/cpu.c:3212-3451` stayed on the old lines under a new pin."""