Add support for newlines inside "NAME" token

dissect/cstruct/expression.py

@@ -142,7 +142,7 @@ def tokenize(self) -> list[str]:
                 self.tokens.append(">>")
             elif self.match(expected="<", append=False) and self.match(expected="<", append=False):
                 self.tokens.append("<<")
-            elif self.match(expected={" ", "\t"}, append=False):
+            elif self.match(expected={" ", "\n", "\t"}, append=False):
                 continue
             else:
                 raise ExpressionTokenizerError(

dissect/cstruct/parser.py

@@ -63,7 +63,7 @@ def _tokencollection() -> TokenCollection:
             "ENUM",
         )
         TOK.add(r"(?<=})\s*(?P<defs>(?:[a-zA-Z0-9_]+\s*,\s*)+[a-zA-Z0-9_]+)\s*(?=;)", "DEFS")
-        TOK.add(r"(?P<name>\**?\s*[a-zA-Z0-9_]+)(?:\s*:\s*(?P<bits>\d+))?(?:\[(?P<count>[^;\n]*)\])?\s*(?=;)", "NAME")
+        TOK.add(r"(?P<name>\**?\s*[a-zA-Z0-9_]+)(?:\s*:\s*(?P<bits>\d+))?(?:\[(?P<count>[^;]*)\])?\s*(?=;)", "NAME")
         TOK.add(r"[a-zA-Z_][a-zA-Z0-9_]*", "IDENTIFIER")
         TOK.add(r"[{}]", "BLOCK")
         TOK.add(r"\$(?P<name>[^\s]+) = (?P<value>{[^}]+})\w*[\r\n]+", "LOOKUP")

tests/test_types_structure.py

@@ -761,3 +761,37 @@ def __init__(self, _0 = None, _1 = None, _2 = None, _3 = None, _4 = None):
     cached = structure._make_structure__init__(5)
     assert structure._make_structure__init__.cache_info() == (1, 1, 128, 1)
     assert result is cached
+
+
+def test_structure_definition_newline(cs: cstruct, compiled: bool) -> None:
+    cdef = """
+    struct test {
+        char    magic[4
+        ];
+
+        wchar   wmagic[4
+        ];
+        uint8   a;
+        uint16  b;
+        uint32  c;
+        char    string[];
+        wchar   wstring[];
+    };
+    """
+    cs.endian = ">"
+    cs.load(cdef, compiled=compiled)
+
+    assert verify_compiled(cs.test, compiled)
+
+    buf = b"test\x00t\x00e\x00s\x00t\x01\x02\x03\x04\x05\x06\x07lalala\x00\x00t\x00e\x00s\x00t\x00\x00"
+
+    obj = cs.test()
+    obj.magic = "test"
+    obj.wmagic = "test"
+    obj.a = 0x01
+    obj.b = 0x0203
+    obj.c = 0x04050607
+    obj.string = b"lalala"
+    obj.wstring = "test"
+
+    assert obj.dumps() == buf