Add file name and line number info to exceptions

Author: dgelessus

rezparser/common.py

@@ -12,7 +12,22 @@
 
 
 class RezParserError(Exception):
-	__slots__ = ()
+	__slots__ = ("message", "filename", "lineno")
+	
+	def __init__(self, message, *, filename=None, lineno=None):
+		full_message = str(message)
+		
+		if filename is not None:
+			full_message += f", in file {filename!r}"
+		
+		if lineno is not None:
+			full_message += f", on line {lineno}"
+		
+		super().__init__(full_message)
+		
+		self.message = message
+		self.filename = filename
+		self.lineno = lineno
 
 
 class Token(ply.lex.LexToken):

rezparser/lexer.py

@@ -41,6 +41,20 @@ def lexpos(self, lexpos):
 		else:
 			self.lexer.lexpos = lexpos
 
+	@property
+	def filename(self):
+		if self.lexer is None:
+			return self._filename
+		else:
+			return self.lexer.filename
+	
+	@filename.setter
+	def filename(self, filename):
+		if self.lexer is None:
+			self._filename = filename
+		else:
+			self.lexer.filename = filename
+	
 	def __init__(self, lexer=None):
 		super().__init__()
 
@@ -66,7 +80,12 @@ def token(self):
 				return self.lexer.token()
 
 	def clone(self):
-		cloned = NoOpLexer(self.lexer.clone())
+		cloned = NoOpLexer(None if self.lexer is None else self.lexer.clone())
+		for attr in ("_lineno", "_lexpos", "_filename"):
+			try:
+				setattr(cloned, attr, getattr(self, attr))
+			except AttributeError:
+				pass
 		cloned.input(self.tokens)
 		return cloned
 
@@ -233,7 +252,12 @@ class RezLexer(object):
 	tokens += tuple("FUN_" + fun[2:].upper() for fun in rez_functions)
 
 	def t_error(self, t):
-		raise LexError(t)
+		try:
+			newline_pos = t.value.index("\n", 1)
+		except ValueError:
+			newline_pos = len(t.value)
+		
+		raise LexError(repr(t.value[:newline_pos]), filename=self.filename, lineno=self.lineno)
 
 	_pp = r"(?m:^)[ \t]*\#[ \t]*"
 	_id = r"[A-Za-z_][A-Za-z0-9_]*"
@@ -360,7 +384,7 @@ def t_IDENTIFIER(self, t):
 	@ply.lex.TOKEN(r"\$\$"+_id)
 	def t_REZ_FUNCTION(self, t):
 		if t.value.lower() not in RezLexer.rez_functions:
-			raise LexError("Unknown Rez function: " + t.value)
+			raise LexError("Unknown Rez function: " + t.value, filename=self.filename, lineno=self.lineno)
 
 		t.type = "FUN_" + t.value[2:].upper()
 
@@ -443,11 +467,23 @@ def lexpos(self):
 	def lexpos(self, lexpos):
 		self.lexer.lexpos = lexpos
 
-	def __init__(self, *, _lexer=None, **kwargs):
+	@property
+	def filename(self):
+		try:
+			return self.lexer.filename
+		except AttributeError:
+			return None
+	
+	@filename.setter
+	def filename(self, filename):
+		self.lexer.filename = filename
+	
+	def __init__(self, *, filename="<input>", _lexer=None, **kwargs):
 		super().__init__()
 
 		if _lexer is None:
 			self.lexer = ply.lex.lex(module=self, lextab="_table_lexer", **kwargs)
+			self.filename = filename
 		else:
 			self.lexer = _lexer
 
@@ -462,4 +498,6 @@ def token(self):
 		return self.lexer.token()
 
 	def clone(self):
-		return RezLexer(_lexer=self.lexer.clone())
+		new_lexer = RezLexer(_lexer=self.lexer.clone())
+		new_lexer.filename = self.filename
+		return new_lexer

rezparser/parser.py

@@ -98,8 +98,8 @@ class RezParser(object):
 
 	start = "Start symbol must be set manually"
 
-	def p_error(self, p):
-		raise ParseError(p)
+	def p_error(self, t):
+		raise ParseError(t, filename=t.lexer.filename, lineno=t.lineno)
 
 	def p_empty(self, p):
 		"""empty : """
@@ -132,7 +132,11 @@ def p_intlit(self, p):
 		"""
 
 		if p[1].startswith("'"):
-			value = int.from_bytes(_unescape_string(p[1][1:-1]), "big")
+			try:
+				unescaped = _unescape_string(p[1][1:-1])
+			except ValueError as e:
+				raise ParseError(str(e), filename=p[-1].lexer.filename, lineno=p[-1].lineno)
+			value = int.from_bytes(unescaped, "big")
 		elif p[1].startswith("$"):
 			value = int(p[1][1:], 16)
 		elif p[1].startswith("0X") or p[1].startswith("0x"):
@@ -514,7 +518,11 @@ def p_single_string(self, p):
 			if p[1].startswith("$"):
 				p[0] = ast.StringLiteral(value=bytes.fromhex(p[1][2:-1]))
 			else:
-				p[0] = ast.StringLiteral(value=_unescape_string(p[1][1:-1]))
+				try:
+					unescaped = _unescape_string(p[1][1:-1])
+				except ValueError as e:
+					raise ParseError(str(e), filename=p[-1].lexer.filename, lineno=p[-1].lineno)
+				p[0] = ast.StringLiteral(value=unescaped)
 		else:
 			p[0] = p[1]
 
@@ -972,13 +980,13 @@ def p_array_field(self, p):
 		for mod in p[1]:
 			mod = mod.lower()
 			if mod in seen:
-				raise ParseError(f"Duplicate attribute {mod!r}")
+				raise ParseError(f"Duplicate attribute {mod!r}", filename=p[-1].lexer.filename, lineno=p[-1].lineno)
 			seen.add(mod)
 
 			if mod == "wide":
 				wide = True
 			else:
-				raise ParseError(f"Unsupported modifier {mod!r} for type array")
+				raise ParseError(f"Unsupported modifier {mod!r} for type array", filename=p[-1].lexer.filename, lineno=p[-1].lineno)
 
 		if len(p) > 9:
 			p[0] = ast.ArrayField(wide=wide, label=None, count=p[4], fields=p[7])
@@ -1038,29 +1046,29 @@ def p_field(self, p):
 				for mod in modifiers:
 					mod = mod.lower()
 					if mod in seen:
-						raise ParseError(f"Duplicate attribute {mod!r}")
+						raise ParseError(f"Duplicate attribute {mod!r}", filename=p[-1].lexer.filename, lineno=p[-1].lineno)
 					seen.add(mod)
 
 					if mod == "key":
 						is_key = True
 					elif mod == "unsigned":
 						if typename not in ("bitstring", "byte", "integer", "longint"):
-							raise ParseError(f"Unsupported modifier {mod!r} for type {typename!r}")
+							raise ParseError(f"Unsupported modifier {mod!r} for type {typename!r}", filename=p[-1].lexer.filename, lineno=p[-1].lineno)
 
 						signed = False
 					elif mod in ("binary", "octal", "decimal", "hex", "literal"):
 						if mod == "hex" and typename == "string":
 							# Special case: hex is allowed on string
 							pass
 						elif typename not in ("bitstring", "byte", "integer", "longint"):
-							raise ParseError(f"Unsupported modifier {mod!r} for type {typename!r}")
+							raise ParseError(f"Unsupported modifier {mod!r} for type {typename!r}", filename=p[-1].lexer.filename, lineno=p[-1].lineno)
 
 						if base is None:
 							base = mod
 						else:
-							raise ParseError(f"Duplicate base modifier {mod!r} (base was previously set to {base!r}")
+							raise ParseError(f"Duplicate base modifier {mod!r} (base was previously set to {base!r}", filename=p[-1].lexer.filename, lineno=p[-1].lineno)
 					else:
-						raise ParseError(f"Invalid modifier: {mod!r}")
+						raise ParseError(f"Invalid modifier: {mod!r}", filename=p[-1].lexer.filename, lineno=p[-1].lineno)
 
 				if typename == "int":
 					typename = "integer"
@@ -1087,7 +1095,7 @@ def p_field(self, p):
 				elif typename == "rect":
 					fieldtype = ast.RectFieldType()
 				else:
-					raise ParseError(f"Unknown field type {typename!r}")
+					raise ParseError(f"Unknown field type {typename!r}", filename=p[-1].lexer.filename, lineno=p[-1].lineno)
 
 				if value_or_symconsts is None:
 					value = None