Get vtable class name

tools/isledecomp/isledecomp/parser/parser.py

@@ -7,6 +7,7 @@
     is_blank_or_comment,
     match_marker,
     is_marker_exact,
+    get_class_name,
     get_synthetic_name,
     remove_trailing_comment,
 )
@@ -195,14 +196,18 @@ def _vtable_marker(self, marker: DecompMarker):
             self._syntax_warning(ParserError.DUPLICATE_MODULE)
         self.state = ReaderState.IN_VTABLE
 
-    def _vtable_done(self):
+    def _vtable_done(self, class_name: str = None):
+        if class_name is None:
+            # Best we can do
+            class_name = self.last_line.strip()
+
         for marker in self.tbl_markers.iter():
             self.vtables.append(
                 ParserVtable(
                     line_number=self.line_number,
                     module=marker.module,
                     offset=marker.offset,
-                    class_name=self.last_line.strip(),
+                    class_name=class_name,
                 )
             )
 
@@ -380,8 +385,9 @@ def read_line(self, line: str):
                 self._variable_done()
 
         elif self.state == ReaderState.IN_VTABLE:
-            if not is_blank_or_comment(line):
-                self._vtable_done()
+            vtable_class = get_class_name(line)
+            if vtable_class is not None:
+                self._vtable_done(class_name=vtable_class)
 
     def read_lines(self, lines: Iterable):
         for line in lines:

tools/isledecomp/isledecomp/parser/util.py

@@ -63,3 +63,33 @@ def match_marker(line: str) -> DecompMarker | None:
 
 def is_marker_exact(line: str) -> bool:
     return markerExactRegex.match(line) is not None
+
+
+template_class_decl_regex = re.compile(
+    r"\s*(?:\/\/)?\s*class (\w+)<([\w]+)\s*(\*+)?\s*>"
+)
+
+
+class_decl_regex = re.compile(r"\s*(?:\/\/)?\s*class (\w+)")
+
+
+def get_class_name(line: str) -> str | None:
+    """For VTABLE markers, extract the class name from the code line or comment
+    where it appears."""
+
+    match = template_class_decl_regex.match(line)
+    if match is not None:
+        # For template classes, we should reformat the class name so it matches
+        # the output from cvdump: one space between the template type and any asterisks
+        # if it is a pointer type.
+        (class_name, template_type, asterisks) = match.groups()
+        if asterisks is not None:
+            return f"{class_name}<{template_type} {asterisks}>"
+
+        return f"{class_name}<{template_type}>"
+
+    match = class_decl_regex.match(line)
+    if match is not None:
+        return match.group(1)
+
+    return None

tools/isledecomp/tests/test_parser.py

@@ -193,6 +193,7 @@ def test_multiple_vtables(parser):
     )
     assert len(parser.alerts) == 0
     assert len(parser.vtables) == 2
+    assert parser.vtables[0].class_name == "MxString"
 
 
 def test_multiple_vtables_same_module(parser):

tools/isledecomp/tests/test_parser_util.py

@@ -5,6 +5,7 @@
     is_blank_or_comment,
     match_marker,
     is_marker_exact,
+    get_class_name,
 )
 
 
@@ -96,3 +97,36 @@ def test_marker_dict_type_replace():
     markers = list(d.iter())
     assert len(markers) == 1
     assert markers[0].type == "FUNCTION"
+
+
+class_name_match_cases = [
+    ("class MxString {", "MxString"),
+    ("// class MxString", "MxString"),
+    ("class MxString : public MxCore {", "MxString"),
+    ("class MxPtrList<MxPresenter>", "MxPtrList<MxPresenter>"),
+    # If it is possible to match the symbol MxList<LegoPathController *>::`vftable'
+    # we should get the correct class name if possible. If the template type is a pointer,
+    # the asterisk and class name are separated by one space.
+    ("// class MxList<LegoPathController *>", "MxList<LegoPathController *>"),
+    ("// class MxList<LegoPathController*>", "MxList<LegoPathController *>"),
+    ("// class MxList<LegoPathController* >", "MxList<LegoPathController *>"),
+    # I don't know if this would ever come up, but sure, why not?
+    ("// class MxList<LegoPathController**>", "MxList<LegoPathController **>"),
+]
+
+
+@pytest.mark.parametrize("line, class_name", class_name_match_cases)
+def test_get_class_name(line: str, class_name: str):
+    assert get_class_name(line) == class_name
+
+
+class_name_no_match_cases = [
+    "MxString { ",
+    "clas MxString",
+    "// MxPtrList<MxPresenter>::`scalar deleting destructor'",
+]
+
+
+@pytest.mark.parametrize("line", class_name_no_match_cases)
+def test_get_class_name_none(line: str):
+    assert get_class_name(line) is None