Detect type errors even after parse errors

CHANGELOG.md

@@ -10,6 +10,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 - Execution times of long-running programs has been cut down by roughly 50%.
 - Use of `SWI` and `RTI` operations now result in parse-time errors instead of run-time warnings.
 - HERA files may no longer contain non-ASCII bytes.
+- Type errors are detected even in files with parse errors.
 
 ## [0.5.2] - 2019-02-02
 ### Added

hera/checker.py

@@ -23,10 +23,10 @@
 
 def check(
     oplist: List[AbstractOperation], settings: Settings
-) -> Tuple[Optional[Program], Messages]:
+) -> Tuple[Program, Messages]:
     symbol_table, messages = typecheck(oplist, settings=settings)
     if messages.errors:
-        return (None, messages)
+        return (Program([], [], {}), messages)
 
     oplist, preprocess_messages = convert_ops(oplist, symbol_table)
     messages.extend(preprocess_messages)

hera/loader.py

@@ -18,8 +18,10 @@ def load_program(text: str, settings=Settings()) -> Program:
 
     The return value of this function is valid input to the VirtualMachine.run method.
     """
-    oplist = handle_messages(settings, parse(text, settings=settings))
-    return handle_messages(settings, check(oplist, settings))
+    oplist, parse_messages = parse(text, settings=settings)
+    program, check_messages = check(oplist, settings=settings)
+    handle_messages(settings, parse_messages.extend(check_messages))
+    return program
 
 
 def load_program_from_file(path: str, settings=Settings()) -> Program:
@@ -48,5 +50,7 @@ def load_program_from_file(path: str, settings=Settings()) -> Program:
         except HERAError as e:
             handle_messages(settings, Messages(str(e) + "."))
 
-    oplist = handle_messages(settings, parse(text, path=path, settings=settings))
-    return handle_messages(settings, check(oplist, settings))
+    oplist, parse_messages = parse(text, path=path, settings=settings)
+    program, check_messages = check(oplist, settings=settings)
+    handle_messages(settings, parse_messages.extend(check_messages))
+    return program

hera/parser.py

@@ -100,6 +100,9 @@ def match_op(self, name_tkn: Token) -> Optional[AbstractOperation]:
         self.lexer.next_token()
         args = self.match_optional_arglist()
         self.lexer.next_token()
+        if args is None:
+            return None
+
         try:
             cls = name_to_class[name_tkn.value]
         except KeyError:
@@ -110,13 +113,20 @@ def match_op(self, name_tkn: Token) -> Optional[AbstractOperation]:
 
     VALUE_TOKENS = {Token.INT, Token.REGISTER, Token.SYMBOL, Token.STRING, Token.CHAR}
 
-    def match_optional_arglist(self) -> List[Token]:
+    def match_optional_arglist(self) -> Optional[List[Token]]:
+        """Match zero or more comma-separated values. Exits with the right parenthesis
+        as the current token. Make sure to distinguish between a None return value (the
+        arglist could not be parsed) and a [] return value (an empty arglist was parsed
+        successfully).
+        """
         if self.lexer.tkn.type == Token.RPAREN:
             return []
 
         args = []
+        hit_error = False
         while True:
             if not self.expect(self.VALUE_TOKENS, "expected value"):
+                hit_error = True
                 self.skip_until({Token.COMMA, Token.RPAREN})
                 if self.lexer.tkn.type == Token.COMMA:
                     self.lexer.next_token()
@@ -131,6 +141,7 @@ def match_optional_arglist(self) -> List[Token]:
             if self.lexer.tkn.type == Token.RPAREN:
                 break
             elif self.lexer.tkn.type != Token.COMMA:
+                hit_error = True
                 self.err("expected comma or right parenthesis")
                 self.skip_until({Token.COMMA, Token.RPAREN})
                 if (
@@ -140,7 +151,8 @@ def match_optional_arglist(self) -> List[Token]:
                     break
             else:
                 self.lexer.next_token()
-        return args
+
+        return args if not hit_error else None
 
     def match_value(self) -> Token:
         if self.lexer.tkn.type == Token.INT:

test/assets/error/mega_error.hera

@@ -0,0 +1,20 @@
+// A parse error (missing comma)
+SET(R1 40)
+
+// A type error
+FOFF(R1)
+
+// Multiple type errors on one line
+ADD('c', "abc")
+
+// Warning for zero-prefixed octal literal
+SET(R2, 01)
+
+// Error for over-long character literal
+SETLO(R7, 'ab')
+
+// Warning for invalid backslash escape, and error for data after code
+LP_STRING("\y")
+
+// Finish off with an unclosed string literal
+SET(R1, "

test/test_error.py

@@ -415,3 +415,65 @@ def test_error_for_interrupt_instructions(capsys):
 
 """
     )
+
+
+def test_mega_error(capsys):
+    with pytest.raises(SystemExit):
+        main(["test/assets/error/mega_error.hera"])
+
+    captured = capsys.readouterr().err
+    assert (
+        captured
+        == """\
+Warning: consider using "0o" prefix for octal numbers, line 11 col 9 of test/assets/error/mega_error.hera
+
+  SET(R2, 01)
+          ^
+
+Warning: unrecognized backslash escape, line 17 col 13 of test/assets/error/mega_error.hera
+
+  LP_STRING("\\y")
+              ^
+
+Error: expected comma or right parenthesis, line 2 col 8 of test/assets/error/mega_error.hera
+
+  SET(R1 40)
+         ^
+
+Error: over-long character literal, line 14 col 11 of test/assets/error/mega_error.hera
+
+  SETLO(R7, 'ab')
+            ^
+
+Error: unclosed string literal, line 20 col 9 of test/assets/error/mega_error.hera
+
+  SET(R1, "
+          ^
+
+Error: expected integer, line 5 col 6 of test/assets/error/mega_error.hera
+
+  FOFF(R1)
+       ^
+
+Error: too few args to ADD (expected 3), line 8 col 1 of test/assets/error/mega_error.hera
+
+  ADD('c', "abc")
+  ^
+
+Error: expected register, line 8 col 6 of test/assets/error/mega_error.hera
+
+  ADD('c', "abc")
+       ^
+
+Error: expected register, line 8 col 10 of test/assets/error/mega_error.hera
+
+  ADD('c', "abc")
+           ^
+
+Error: data statement after code, line 17 col 1 of test/assets/error/mega_error.hera
+
+  LP_STRING("\\y")
+  ^
+
+"""
+    )