Make symbol table an explicit parameter of typechecker and preprocessor

hera/main.py

@@ -21,6 +21,7 @@
 from . import config
 from .parser import parse
 from .preprocessor import preprocess
+from .symtab import get_symtab
 from .typechecker import typecheck
 from .utils import emit_error, print_register_debug, HERAError
 from .vm import VirtualMachine
@@ -96,11 +97,13 @@ def execute_program(program, *, lines_to_exec=None, no_dump_state=False, vm=None
     except HERAError as e:
         emit_error(str(e), line=e.line, column=e.column, exit=True)
 
-    typecheck(program)
+    symtab = get_symtab(program)
+
+    typecheck(program, symtab)
     if config.SEEN_ERROR:
         sys.exit(3)
 
-    program = preprocess(program)
+    program = preprocess(program, symtab)
     if config.SEEN_ERROR:
         sys.exit(3)
 

hera/preprocessor.py

@@ -6,11 +6,10 @@
 from lark import Token
 
 from .parser import Op
-from .symtab import get_symtab
 from .utils import copy_token, emit_error, is_symbol, to_u16
 
 
-def preprocess(program):
+def preprocess(program, symtab):
     """Preprocess the program (a list of Op objects) into valid input for the
     exec_many method on the VirtualMachine class.
 
@@ -19,22 +18,18 @@ def preprocess(program):
         - Resolves labels into their line numbers.
     """
     program = [op for old_op in program for op in convert(old_op)]
-
-    labels = get_symtab(program)
-
-    program = [substitute_label(op, labels) for op in program]
+    program = [substitute_label(op, symtab) for op in program]
     program = [op for op in program if op.name not in ("LABEL", "DLABEL", "CONSTANT")]
-
     return program
 
 
-def substitute_label(op, labels):
+def substitute_label(op, symtab):
     """Substitute any label in the instruction with its concrete value."""
     if op.name == "SETLO" and is_symbol(op.args[1]):
         d, v = op.args
         name = copy_token("SETLO", op.name)
         try:
-            label = labels[v]
+            label = symtab[v]
         except KeyError:
             emit_error(
                 "undefined symbol `{}`".format(v), line=op.name.line, column=v.column
@@ -44,7 +39,7 @@ def substitute_label(op, labels):
     elif op.name == "SETHI" and is_symbol(op.args[1]):
         d, v = op.args
         name = copy_token("SETHI", op.name)
-        return Op(name, [d, labels[v] >> 8])
+        return Op(name, [d, symtab[v] >> 8])
     else:
         return op
 

hera/symtab.py

@@ -1,3 +1,4 @@
+from .preprocessor import convert
 from .utils import emit_error
 
 
@@ -25,7 +26,7 @@ def get_symtab(program):
         elif op.name == "LP_STRING":
             dc += len(op.args[0]) + 1
         else:
-            pc += 1
+            pc += len(convert(op))
 
         if dc >= 0xFFFF and odc < 0xFFFF:
             emit_error("past the end of available memory", line=op.name.line)

hera/typechecker.py

@@ -11,7 +11,7 @@
 DATA_STATEMENTS = set(["CONSTANT", "DLABEL", "INTEGER", "LP_STRING", "DSKIP"])
 
 
-def typecheck(program):
+def typecheck(program, symtab):
     """Type-check the program and emit errors as appropriate."""
     end_of_data = False
     for op in program:
@@ -21,14 +21,14 @@ def typecheck(program):
         else:
             if op.name in DATA_STATEMENTS:
                 emit_error("data statement after instruction", line=op.name.line)
-        typecheck_one(op)
+        typecheck_one(op, symtab)
 
 
-def typecheck_one(op):
+def typecheck_one(op, symtab):
     """Type-check a single HERA op and emit errors as appropriate."""
     params = _types_map.get(op.name)
     if params is not None:
-        check_types(op.name, params, op.args)
+        check_types(op.name, params, op.args, symtab)
     else:
         emit_error("unknown instruction `{}`".format(op.name), line=op.name.line)
 
@@ -131,7 +131,7 @@ def typecheck_one(op):
 }
 
 
-def check_types(name, expected, got):
+def check_types(name, expected, got, symtab):
     """Verify that the given args match the expected ones and emit errors as
     appropriate. `name` is the name of the HERA op, as a Token object. `expected` is a
     tuple or list of constants (REGISTER, U16, etc., defined above) representing the

test/test_preprocessor.py

@@ -4,12 +4,7 @@
 from lark import Token
 
 from hera.parser import Op
-from hera.preprocessor import (
-    convert,
-    convert_set,
-    preprocess,
-    substitute_label,
-)
+from hera.preprocessor import convert, convert_set, preprocess, substitute_label
 from hera.utils import HERAError, IntToken
 
 
@@ -167,8 +162,8 @@ def test_convert_not():
 
 
 def test_preprocess_constant():
-    program = [
-        Op(Token("SYMBOL", "CONSTANT"), [Token("SYMBOL", "n"), IntToken(100)]),
-        Op(Token("SYMBOL", "SET"), [R("R1"), Token("SYMBOL", "n")]),
+    program = [Op(Token("SYMBOL", "SET"), [R("R1"), Token("SYMBOL", "n")])]
+    assert preprocess(program, {"n": 100}) == [
+        Op("SETLO", ["R1", 100]),
+        Op("SETHI", ["R1", 0]),
     ]
-    assert preprocess(program) == [Op("SETLO", ["R1", 100]), Op("SETHI", ["R1", 0])]

test/test_typechecker.py

@@ -1,3 +1,4 @@
+import functools
 import pytest
 from unittest.mock import patch
 
@@ -17,6 +18,12 @@
 from hera.utils import HERAError, IntToken
 
 
+# TODO: Get rid of these and explicitly pass the empty symbol table in.
+typecheck = functools.partial(typecheck, symtab={})
+typecheck_one = functools.partial(typecheck_one, symtab={})
+check_types = functools.partial(check_types, symtab={})
+
+
 def R(s):
     return Token("REGISTER", s)