feat: parse colon continuations and header stems

emulators/beetle_saturn.yml

@@ -5,7 +5,7 @@ core: mednafen_saturn_libretro
 source: "https://github.com/libretro/beetle-saturn-libretro"
 upstream: "https://mednafen.github.io/"
 profiled_date: "2026-03-24"
-source_commit: "b4df47a9f0f30d09eb95b07a4435d0f435a2e95d"
+source_commit: "84461434f249c1b5cd10c83ae922feae84e1acf8"
 core_version: "v0.9.45.1"
 display_name: "Sega - Saturn (Beetle Saturn)"
 cores:
@@ -36,7 +36,7 @@ files:
     size: 524288
     validation: [size]
     note: "Saturn BIOS v1.01 (Japan). Used when region is JP or Asia NTSC."
-    source_ref: "mednafen/ss/ss.cpp:928-929,937-949"
+    source_ref: "mednafen/ss/ss.c:1761 (region filename), mednafen/ss/ss.c:1784-1899 (open, size check, read, sha256)"
 
   - name: "mpr-17933.bin"
     system: sega-saturn
@@ -45,25 +45,25 @@ files:
     size: 524288
     validation: [size]
     note: "Saturn BIOS (NA/EU). Used for all non-JP/Asia-NTSC regions."
-    source_ref: "mednafen/ss/ss.cpp:930-931,937-949"
+    source_ref: "mednafen/ss/ss.c:1763 (region filename), mednafen/ss/ss.c:1784-1899 (open, size check, read, sha256)"
 
   - name: "mpr-18811-mx.ic1"
     system: sega-saturn
     required: false
     size: 2097152
     note: "KOF95 expansion ROM cart. Loaded when cart type set to KOF95."
-    source_ref: "mednafen/ss/cart.cpp:157-172, mednafen/ss/cart/rom.cpp:32-43"
+    source_ref: "mednafen/ss/cart.c:200-215 (firmware path + open), mednafen/ss/cart/rom.c:51 (CART_ROM_Init)"
 
   - name: "mpr-19367-mx.ic1"
     system: sega-saturn
     required: false
     size: 2097152
     note: "Ultraman expansion ROM cart. Loaded when cart type set to Ultraman."
-    source_ref: "mednafen/ss/cart.cpp:157-172, mednafen/ss/cart/rom.cpp:32-43"
+    source_ref: "mednafen/ss/cart.c:200-215 (firmware path + open), mednafen/ss/cart/rom.c:51 (CART_ROM_Init)"
 
   - name: "satar4mp.bin"
     system: sega-saturn
     required: false
     size: 262144
     note: "Action Replay 4M Plus firmware. Unfinished in source (ar4mp.cpp:22-24), needs CPU UBC emulation. Cart type not exposed in core options."
-    source_ref: "mednafen/ss/cart.cpp:174-188, mednafen/ss/cart/ar4mp.cpp:120-159"
+    source_ref: "mednafen/ss/cart.c:218 (CART_AR4MP case), mednafen/ss/cart/ar4mp.c:150 (CART_AR4MP_Init)"

scripts/profile_sync.py

@@ -46,7 +46,8 @@ PIN = "pin"
 HEAD = "head"
 
 _REF_RE = re.compile(r"^(?P<path>[^:]+?)(?::(?P<start>\d+)(?:-(?P<end>\d+))?)?$")
-_BARE_RANGE_RE = re.compile(r"^\d+(?:-\d+)?$")
+_BARE_RANGE_RE = re.compile(r"^:?(\d+(?:-\d+)?)$")
+HEADER_SUFFIXES = (".h", ".hpp", ".hh", ".hxx", ".inc", ".inl")
 _SPLIT_RE = re.compile(r"[,;]")
 _ANNOTATION_RE = re.compile(r"\s*\([^)]*\)?")
 _LOCATED_RE = re.compile(r"^[^\s:]+:\d+(?:-\d+)?$")
@@ -128,8 +129,9 @@ def split_source_ref(ref: str) -> list[RefPart]:
         chunk = _trim_prose(_ANNOTATION_RE.sub("", raw).strip())
         if not chunk:
             continue
-        if _BARE_RANGE_RE.match(chunk) and parts:
-            start, _, end = chunk.partition("-")
+        bare = _BARE_RANGE_RE.match(chunk)
+        if bare and parts:
+            start, _, end = bare.group(1).partition("-")
             parts.append(
                 RefPart(parts[-1].path, int(start), int(end or start), raw)
             )
@@ -370,13 +372,25 @@ def resolve_rename(
     return None, matches
 
 
+def _is_header(path: str) -> bool:
+    return path.lower().endswith(HEADER_SUFFIXES)
+
+
 def _narrow(matches: list[str], path: str) -> list[str]:
-    """Prefer candidates sitting in the directory the file came from."""
+    """Prefer candidates of the same kind, then of the same directory.
+
+    A stem search on `ss.cpp` matches both `ss.c` and `ss.h`, and only one of
+    them is the implementation the ref meant.
+    """
     if len(matches) < 2:
         return matches
+    kind = [p for p in matches if _is_header(p) == _is_header(path)]
+    if len(kind) == 1:
+        return kind
+    pool = kind or matches
     directory = posixpath.dirname(path)
-    same = [p for p in matches if posixpath.dirname(p) == directory]
-    return same if len(same) == 1 else matches
+    same = [p for p in pool if posixpath.dirname(p) == directory]
+    return same if len(same) == 1 else pool
 
 
 def locate_by_declared_value(

tests/test_profile_sync.py

@@ -167,6 +167,23 @@ class TestSplitSourceRef(unittest.TestCase):
         parts = split_source_ref("munt ROMInfo.cpp")
         self.assertEqual(parts[0].path, "munt ROMInfo.cpp")
 
+    def test_leading_colon_continues_the_previous_file(self):
+        parts = split_source_ref("sms_mapper.c:12, :61, :162-170")
+        self.assertEqual(
+            [(p.path, p.start, p.end) for p in parts],
+            [("sms_mapper.c", 12, 12), ("sms_mapper.c", 61, 61),
+             ("sms_mapper.c", 162, 170)],
+        )
+
+    def test_leading_colon_with_annotations(self):
+        parts = split_source_ref(
+            "genesis.rs:41-51 (read), :256-262 (bios_path_for_region)"
+        )
+        self.assertEqual(
+            [(p.path, p.start, p.end) for p in parts],
+            [("genesis.rs", 41, 51), ("genesis.rs", 256, 262)],
+        )
+
     def test_leading_bare_range_stays_a_path(self):
         parts = split_source_ref("273-285")
         self.assertEqual(parts[0].path, "273-285")
@@ -531,10 +548,22 @@ class TestResolveRename(unittest.TestCase):
         )
         self.assertEqual(found, "libretro.c")
 
-    def test_same_directory_tie_break_needs_a_single_winner(self):
-        found, candidates = resolve_rename(
+    def test_an_implementation_wins_over_a_header(self):
+        found, _ = resolve_rename(
             CompareResult([], True), "src/a.cpp", ["src/a.c", "src/a.h"]
         )
+        self.assertEqual(found, "src/a.c")
+
+    def test_a_header_ref_picks_the_header(self):
+        found, _ = resolve_rename(
+            CompareResult([], True), "src/a.hpp", ["src/a.c", "src/a.h"]
+        )
+        self.assertEqual(found, "src/a.h")
+
+    def test_two_implementations_stay_ambiguous(self):
+        found, candidates = resolve_rename(
+            CompareResult([], True), "src/a.cpp", ["one/a.c", "two/a.cc"]
+        )
         self.assertIsNone(found)
         self.assertEqual(len(candidates), 2)