dissect.cstruct 5.0.dev14__py3-none-any.whl → 5.0.dev15__py3-none-any.whl

dissect/cstruct/lexer.py

@@ -125,6 +125,7 @@ _SINGLE_CHARS = {
 
 _RE_IDENTIFIER = re.compile(r"[a-zA-Z_.][a-zA-Z0-9_.]*")
 _RE_WHITESPACE = re.compile(r"[ \t\r\n]+")
+_RE_DEFINE_LITERAL = re.compile(r"""(["'])(?:\\.|(?!\1)[^\\\n])*\1""")
 
 IDENTIFIER_TYPES = set(_C_KEYWORDS.values()) | {TokenType.IDENTIFIER}
 
@@ -292,10 +293,10 @@ class Lexer:
             if peek == "*":
                 self._take(2)  # Consume /*
                 end = self.data.find("*/", self._pos)
-                if end != -1:
-                    self._take(end - self._pos + 2)
-                else:
-                    self._take(len(self.data) - self._pos)
+                if end == -1:
+                    raise self._error("unterminated comment")
+
+                self._take(end - self._pos + 2)
 
             elif peek == "/":
                 self._take(2)  # Consume //
@@ -390,6 +391,23 @@ class Lexer:
 
         return result
 
+    def _read_define_value(self) -> str:
+        """Read the value of a ``#define`` up to the end of the line."""
+        value = []
+
+        while not self.eof and (ch := self._current()) != "\n":
+            if ch == "/" and self._peek() in ("/", "*"):
+                self._skip_comment()
+                value.append(" ")
+            elif ch in ('"', "'"):
+                if not (match := _RE_DEFINE_LITERAL.match(self.data, self._pos)):
+                    raise self._error("unterminated string literal")
+                value.append(self._take(match.end() - self._pos))
+            else:
+                value.append(self._read_until("\n\"'/") or self._take())
+
+        return "".join(value).strip()
+
     def _read_preprocessor(self) -> None:
         """Read a preprocessor directive starting with ``#``."""
         line = self._line
@@ -423,12 +441,7 @@ class Lexer:
                 raise self._error("expected identifier after '#define'", lineno=line)
             self._emit(TokenType.IDENTIFIER, name, line)
 
-            self._skip_whitespace_and_comments()
-            if self.eof or self._line != line:
-                # No value, just a simple macro definition
-                return
-
-            if (value := self._read_until("\n")).strip():
+            if value := self._read_define_value():
                 self._emit(TokenType.STRING, value, line)
 
         elif token_type == TokenType.PP_INCLUDE:

dissect/cstruct/parser.py

@@ -20,6 +20,10 @@ if TYPE_CHECKING:
     from dissect.cstruct.types import BaseType, Enum, Flag
 
 
+# A string or the name of a constant in the value of a define
+_RE_DEFINE_STRING = re.compile(r"""\s*(?:([bB]?)(["'])((?:\\.|(?!\2)[^\\\n])*)\2|([a-zA-Z_.][a-zA-Z0-9_.]*))""")
+
+
 class Parser(TokenCursor):
     """Base class for definition parsers.
 
@@ -42,7 +46,7 @@ class Parser(TokenCursor):
 
 def _join_line_continuations(string: str) -> str:
     # Join lines ending with backslash
-    return re.sub(r"\\\n", "", string)
+    return re.sub(r"\\\r?\n", "", string)
 
 
 class CStyleParser(Parser):
@@ -138,11 +142,16 @@ class CStyleParser(Parser):
             elif token.type == TokenType.TYPEDEF:
                 self._parse_typedef()
             elif token.type in (TokenType.STRUCT, TokenType.UNION):
-                self._parse_struct_or_union()
+                type_ = self._parse_struct_or_union()
 
                 # Skip variable declarations after struct/union definitions
-                while not self._at(TokenType.SEMICOLON, TokenType.EOF):
-                    self._pos += 1
+                has_variable = self._at(TokenType.IDENTIFIER, TokenType.STAR)
+                while has_variable:
+                    _, _, bits = self._parse_field_name(type_)
+                    if bits is not None:
+                        raise self._error("variables cannot have bitfields")
+
+                    has_variable = self._match(TokenType.COMMA)
 
                 self._expect(TokenType.SEMICOLON)
             elif token.type in (TokenType.ENUM, TokenType.FLAG):
@@ -174,21 +183,28 @@ class CStyleParser(Parser):
             value = self._take().value
 
         if value:
-            if value[0] in ('"', "'"):
-                quote = value[0]
-                if value[-1] != quote:
-                    raise self._error("unterminated string literal", token=token)
-
-                # Remove the surrounding quotes and flatten escape sequences
-                value = value[1:-1].encode().decode("unicode_escape")
-            elif value[:2].lower() in ("b'", 'b"'):
-                quote = value[1]
-                if value[-1] != quote:
-                    raise self._error("unterminated bytes literal", token=token)
-
-                # Remove the leading b and surrounding quotes and flatten escape sequences
-                value = value[2:-1].encode().decode("unicode_escape")
-                value = ast.literal_eval(f"b{value!r}")
+            # Adjacent strings and constants that are strings are concatenated like in C
+            strings = []
+            pos = 0
+            if '"' in value or "'" in value:
+                while pos < len(value) and (match := _RE_DEFINE_STRING.match(value, pos)):
+                    is_bytes, _, string, name = match.groups()
+                    if name is None:
+                        # Flatten escape sequences
+                        string = string.encode().decode("unicode_escape")
+                        strings.append(string.encode("latin-1") if is_bytes else string)
+                    elif isinstance(constant := self.cs.consts.get(name), (str, bytes)):
+                        strings.append(constant)
+                    else:
+                        break
+
+                    pos = match.end()
+
+            if strings and pos == len(value):
+                try:
+                    value = strings[0][:0].join(strings)
+                except TypeError:
+                    raise self._error("cannot concatenate strings and bytes", token=token) from None
             else:
                 try:
                     # Lazy mode, try to evaluate as a Python literal first (for simple constants)

{dissect_cstruct-5.0.dev14.dist-info → dissect_cstruct-5.0.dev15.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: dissect.cstruct
-Version: 5.0.dev14
+Version: 5.0.dev15
 Summary: A Dissect module implementing a parser for C-like structures: structure parsing in Python made easy
 Author-email: Dissect Team <dissect@fox-it.com>
 License-Expression: Apache-2.0

{dissect_cstruct-5.0.dev14.dist-info → dissect_cstruct-5.0.dev15.dist-info}/RECORD

@@ -4,8 +4,8 @@ dissect/cstruct/compiler.py,sha256=0u58J5pB4izosBj8G4wiBI1JSAAgHFMLM63IaSG4YO4,5
 dissect/cstruct/cstruct.py,sha256=mJerzPouWV29RdK5LqjGitDDMvNKcadoT2hj1zpExnM,26551
 dissect/cstruct/exception.py,sha256=RqD08LuBkbMqa3iQ1A05DNN8XgEmQR_SUnguwe9LGo8,316
 dissect/cstruct/expression.py,sha256=Y27-N48BomBoRFozbWhGDF_DS2lfklRwWVUbCzGnxHE,10873
-dissect/cstruct/lexer.py,sha256=NDUfAWgyxgRNRuVO8ExXQ51LqQXS2mv9bywVa_M7tFQ,19665
-dissect/cstruct/parser.py,sha256=5vz1frbenevKyn-CobKyp9_22dnEbFej_nUXQeL5W8U,21368
+dissect/cstruct/lexer.py,sha256=FbAoUVGAoH4tvnLGBwVoEf6FAxfpz9QQXpJGzykAoc4,20237
+dissect/cstruct/parser.py,sha256=92SCg1lldmFy4CUZjf3gWz9UJaGdYeprL1PV-JXAmY4,22122
 dissect/cstruct/util.py,sha256=vAf-kdOkk3zprW_gD6QmQeN3VIRgQqhPS4B_ha1jEpQ,18040
 dissect/cstruct/tools/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 dissect/cstruct/tools/stubgen.py,sha256=MeZkLmSeNSLNcAEwlAdIGv1lKvlvK3UQ4WN85ovsxTM,7672
@@ -21,10 +21,10 @@ dissect/cstruct/types/pointer.py,sha256=FNuB9bfENWRebbKX-R6Vs6xWPMHepjvOhhomeUhr
 dissect/cstruct/types/structure.py,sha256=mxDvurQ0oAjI2flykhAPlT-03280cDpwGTz0-bqbr-0,39558
 dissect/cstruct/types/void.py,sha256=fV4hKk50uop_EKbR1nfy1cdtAzvUYrSymXsdakCyFk8,1222
 dissect/cstruct/types/wchar.py,sha256=VrnPXsP_lsyA-1xr4D2aSIVijumMhUuZWNlDHcn12-s,3964
-dissect_cstruct-5.0.dev14.dist-info/licenses/COPYRIGHT,sha256=H-18RXfshdH9AdHwW2eO1Xa-5s6tY5eipHh5c0whDu4,316
-dissect_cstruct-5.0.dev14.dist-info/licenses/LICENSE,sha256=PhUqiw6jAh2KbBdVRPBq_hfAvfcTBin7nZ3CK7NQbTM,11341
-dissect_cstruct-5.0.dev14.dist-info/METADATA,sha256=0EhDz7fNITZQWzSYoy46-j0nObu4B1Am3-nEipozmxc,9256
-dissect_cstruct-5.0.dev14.dist-info/WHEEL,sha256=mOdXR9C3wMVov7cSY8PY-XCzo3SmmWCHvGQr5csJoao,91
-dissect_cstruct-5.0.dev14.dist-info/entry_points.txt,sha256=z53zqZqwD2OLrAkRwrP4wTeiU9CQe7xrMly0T2c0_wQ,71
-dissect_cstruct-5.0.dev14.dist-info/top_level.txt,sha256=Mn-CQzEYsAbkxrUI0TnplHuXnGVKzxpDw_po_sXpvv4,8
-dissect_cstruct-5.0.dev14.dist-info/RECORD,,
+dissect_cstruct-5.0.dev15.dist-info/licenses/COPYRIGHT,sha256=H-18RXfshdH9AdHwW2eO1Xa-5s6tY5eipHh5c0whDu4,316
+dissect_cstruct-5.0.dev15.dist-info/licenses/LICENSE,sha256=PhUqiw6jAh2KbBdVRPBq_hfAvfcTBin7nZ3CK7NQbTM,11341
+dissect_cstruct-5.0.dev15.dist-info/METADATA,sha256=9aR35qikIsQ5Y746lVzVxt7NV1pYQ7GyzCiHbtB0BXE,9256
+dissect_cstruct-5.0.dev15.dist-info/WHEEL,sha256=mOdXR9C3wMVov7cSY8PY-XCzo3SmmWCHvGQr5csJoao,91
+dissect_cstruct-5.0.dev15.dist-info/entry_points.txt,sha256=z53zqZqwD2OLrAkRwrP4wTeiU9CQe7xrMly0T2c0_wQ,71
+dissect_cstruct-5.0.dev15.dist-info/top_level.txt,sha256=Mn-CQzEYsAbkxrUI0TnplHuXnGVKzxpDw_po_sXpvv4,8
+dissect_cstruct-5.0.dev15.dist-info/RECORD,,