refactor: require a hash before relocating a ref

emulators/lime3ds.yml

@@ -101,7 +101,7 @@ files:
     size: 273
     validation: [size]
     note: "console region and serial number; loaded on a size check alone"
-    source_ref: "src/core/hle/service/cfg/cfg.cpp:689-709, src/core/hle/service/cfg/cfg.h:180-186"
+    source_ref: "src/core/hle/service/cfg/cfg.cpp:608-610, 689-709, src/core/hle/service/cfg/cfg.h:180-186"
 
   - name: LocalFriendCodeSeed_B
     path: nand/rw/sys/LocalFriendCodeSeed_B
@@ -110,4 +110,4 @@ files:
     size: 272
     validation: [size]
     note: "friend code generation seed; loaded on a size check alone"
-    source_ref: "src/core/hle/service/cfg/cfg.cpp:711-732, src/core/hle/service/cfg/cfg.h:188-193"
+    source_ref: "src/core/hle/service/cfg/cfg.cpp:604-606, 711-732, src/core/hle/service/cfg/cfg.h:188-193"

emulators/mandarine.yml

@@ -111,7 +111,7 @@ files:
     size: 273
     validation: [size]
     note: "console region and serial number; loaded on a size check alone, and paired with LocalFriendCodeSeed_B"
-    source_ref: "src/core/hle/service/cfg/cfg.cpp:689-709, src/core/hle/service/cfg/cfg.h:180-186"
+    source_ref: "src/core/hle/service/cfg/cfg.cpp:608-610, 689-709, src/core/hle/service/cfg/cfg.h:180-186"
 
   - name: LocalFriendCodeSeed_B
     path: nand/rw/sys/LocalFriendCodeSeed_B
@@ -120,4 +120,4 @@ files:
     size: 272
     validation: [size]
     note: "friend code generation seed; loaded on a size check alone, and paired with SecureInfo_A"
-    source_ref: "src/core/hle/service/cfg/cfg.cpp:711-732, src/core/hle/service/cfg/cfg.h:188-193"
+    source_ref: "src/core/hle/service/cfg/cfg.cpp:604-606, 711-732, src/core/hle/service/cfg/cfg.h:188-193"

emulators/opentyrian.yml

@@ -240,40 +240,40 @@ files:
   # Level tileset shapes (loaded per-level from .lvl data)
   # The level file specifies a character via fgetc() at tyrian2.c:3039,
   # which maps to shapes{char}.dat at tyrian2.c:3079.
-  # Only 5 characters are used by actual levels: ), W, X, Y, Z.
+  # Episode 4 uses Z, W, X and } (lvlmast.c:128-131).
   # Note: shapeFile[34] in lvlmast.c is a DIFFERENT mechanism —
   # it maps indices to characters for newsh?.shp (enemy sprites),
   # NOT for shapes?.dat (level tilesets).
   # -----------------------------------------------------------
-  - name: "shapes).dat"
+  - name: "shapes}.dat"
     category: game_data
     system: tyrian
     required: true
-    source_ref: "src/tyrian2.c:3039-3079"
+    source_ref: "src/tyrian2.c:3039, 3079, src/lvlmast.c:128-131"
 
   - name: "shapesw.dat"
     category: game_data
     system: tyrian
     required: true
-    source_ref: "src/tyrian2.c:3039-3079"
+    source_ref: "src/tyrian2.c:3039, 3079, src/lvlmast.c:128-131"
 
   - name: "shapesx.dat"
     category: game_data
     system: tyrian
     required: true
-    source_ref: "src/tyrian2.c:3039-3079"
+    source_ref: "src/tyrian2.c:3039, 3079, src/lvlmast.c:128-131"
 
   - name: "shapesy.dat"
     category: game_data
     system: tyrian
     required: true
-    source_ref: "src/tyrian2.c:3039-3079"
+    source_ref: "src/tyrian2.c:3039, 3079, src/lvlmast.c:128-131"
 
   - name: "shapesz.dat"
     category: game_data
     system: tyrian
     required: true
-    source_ref: "src/tyrian2.c:3039-3079"
+    source_ref: "src/tyrian2.c:3039, 3079, src/lvlmast.c:128-131"
 
   # -----------------------------------------------------------
   # Compressed shape files (newshX.shp, loaded per-level)

emulators/same_cdi.yml

@@ -42,7 +42,7 @@ files:
     system: cdi
     required: true
     note: "Forced system. Main CPU BIOS (3 region variants via MAME BIOS switch) + servo/slave MCU dumps."
-    source_ref: "src/mame/drivers/cdi.cpp:428-443"
+    source_ref: "src/mame/drivers/cdi.cpp:829-443"
     contents:
       - name: cdi200.rom
         description: Magnavox CD-i 200 BIOS (default, US)
@@ -100,7 +100,7 @@ files:
     system: cdi
     required: false
     note: "CD-i BIOS for Quizard arcade. MACHINE_IS_BIOS_ROOT. Subset of cdimono1 + slave MCU."
-    source_ref: "src/mame/drivers/cdi.cpp:630-640"
+    source_ref: "src/mame/drivers/cdi.cpp:836-640"
     contents:
       - name: cdi200.rom
         description: Magnavox CD-i 200 BIOS (same as cdimono1)
@@ -124,7 +124,7 @@ files:
     system: cdi
     required: false
     note: "CD-i 910-17P Mini-MMC (PAL). MACHINE_NOT_WORKING. Not in .info firmware list."
-    source_ref: "src/mame/drivers/cdi.cpp:319-335"
+    source_ref: "src/mame/drivers/cdi.cpp:831-335"
     contents:
       - name: "philips__cd-i_2.1__mb834200b-15__26b_aa__9224_z01.tc574200.7211"
         description: CD-i 910-17P Mini-MMC BIOS (default)

scripts/profile_sync.py

@@ -746,7 +746,7 @@ def branch_in_url(url: str) -> str | None:
 SELF_CHECK_CONTEXT = 2
 
 
-def verify_at_pin(part: RefPart, pin_lines, tokens) -> PartResult:
+def verify_at_pin(part: RefPart, pin_lines, tokens, hash_tokens=()) -> PartResult:
     """Check a ref against its own revision instead of against HEAD.
 
     A profile pinned to a superseded tag documents a program HEAD no longer
@@ -780,7 +780,12 @@ def verify_at_pin(part: RefPart, pin_lines, tokens) -> PartResult:
         return PartResult(part, "ANCHORED", None, None, None, [])
     # A ref that cites loading logic never spells the value out, so its absence
     # here proves nothing. Only finding the value somewhere else in the file
-    # shows the ref points at the wrong place.
+    # shows the ref points at the wrong place -- and only a hash carries that
+    # weight. A filename fragment matches by coincidence: `tyrian2` hits every
+    # line naming `tyrian2.c`, and a name built at runtime from a format
+    # string appears nowhere at all.
+    if not hash_tokens:
+        return PartResult(part, "ANCHORED", None, None, None, [])
     elsewhere = sorted(
         {
             index
@@ -877,6 +882,7 @@ def build_report(
             entry.get("name", "") + (f" [{label}]" if label else ""),
             value,
             _anchor_tokens(entry),
+            entry_hashes(entry),
         )
         for entry in (profile.get("files") or [])
         if isinstance(entry, dict) and entry.get("source_ref")
@@ -1028,10 +1034,12 @@ def build_report(
     if self_check:
         staged = [
             (entry_name, ref, reconcile_self_check([
-                verify_at_pin(part, fetch(PIN, part.path, part.start), tokens)
+                verify_at_pin(
+                    part, fetch(PIN, part.path, part.start), tokens, hashes
+                )
                 for part in split_source_ref(ref)
             ]))
-            for entry_name, ref, tokens in refs
+            for entry_name, ref, tokens, hashes in refs
         ]
     else:
         staged = [
@@ -1039,7 +1047,7 @@ def build_report(
                 anchor_part(part, fetch, rename_getter, describe, tokens)
                 for part in split_source_ref(ref)
             ])
-            for entry_name, ref, tokens in refs
+            for entry_name, ref, tokens, hashes in refs
         ]
 
     shifts = dominant_shifts([p for _, _, parts in staged for p in parts])

tests/test_profile_sync.py

@@ -337,14 +337,18 @@ class TestVerifyAtPin(unittest.TestCase):
     def test_declared_value_carried_by_another_line(self):
         lines = ["load()", "", "", "", "", "", 'rom("cafebabe")']
         part = RefPart("a.c", 1, 1, "a.c:1")
-        result = profile_sync.verify_at_pin(part, lines, ["cafebabe"])
+        result = profile_sync.verify_at_pin(
+            part, lines, ["cafebabe"], ["cafebabe"]
+        )
         self.assertEqual(result.status, "MOVED")
         self.assertEqual(result.start, 7)
 
     def test_declared_value_on_several_lines_is_ambiguous(self):
         lines = ["load()", "", "", "", 'a("cafebabe")', "", 'b("cafebabe")']
         part = RefPart("a.c", 1, 1, "a.c:1")
-        result = profile_sync.verify_at_pin(part, lines, ["cafebabe"])
+        result = profile_sync.verify_at_pin(
+            part, lines, ["cafebabe"], ["cafebabe"]
+        )
         self.assertEqual(result.status, "AMBIGUOUS")
 
     def test_one_anchored_part_settles_the_others(self):
@@ -386,6 +390,13 @@ class TestVerifyAtPin(unittest.TestCase):
         result = profile_sync.verify_at_pin(part, None, ["cafebabe"])
         self.assertEqual(result.status, "EXTERNAL")
 
+    def test_a_name_fragment_elsewhere_is_not_evidence(self):
+        # Only a hash is unique enough to prove a ref points elsewhere.
+        lines = ["load()", "", "", "", "", "", "tyrian2.c line"]
+        part = RefPart("a.c", 1, 1, "a.c:1")
+        result = profile_sync.verify_at_pin(part, lines, ["tyrian2"], [])
+        self.assertEqual(result.status, "ANCHORED")
+
     def test_missing_file(self):
         part = RefPart("a.c", 1, 1, "a.c:1")
         self.assertEqual(