Get #include working, provisionally

hera/main.py

@@ -96,10 +96,13 @@ def execute_program(
         vm = VirtualMachine()
 
     try:
-        program = parse(program)
+        program = parse(program, expand_includes=True)
     except HERAError as e:
         emit_error(str(e), line=e.line, column=e.column, exit=True)
 
+    # Filter out #include statements for now.
+    program = [op for op in program if op.name != "#include"]
+
     symtab = get_symtab(program)
 
     typecheck(program, symtab)

hera/parser.py

@@ -18,13 +18,31 @@
 class TreeToOplist(Transformer):
     """Transform Lark's parse tree into a list of HERA ops."""
 
+    def start(self, matches):
+        # TODO: Figure out why all this is necessary.
+        if len(matches) == 2:
+            if isinstance(matches[0], tuple):
+                return [matches[0]] + matches[1]
+            else:
+                return matches[0] + matches[1]
+        elif len(matches) > 2:
+            return matches[:-1] + matches[-1]
+        else:
+            return matches[0]
+
     def cpp_program(self, matches):
         emit_warning("void HERA_main() { ... } is not necessary")
         return matches
 
+    def hera_program(self, matches):
+        return matches
+
     def op(self, matches):
         return Op(matches[0], matches[1:])
 
+    def include(self, matches):
+        return Op("#include", [matches[0]])
+
     def value(self, matches):
         line = matches[0].line
         column = matches[0].column
@@ -70,17 +88,18 @@ def value(self, matches):
 
 _parser = Lark(
     r"""
-    ?start: cpp_program | _hera_program
+    start: include* (cpp_program | hera_program)
 
-    cpp_program: _INCLUDE* _cpp_open op* _CPP_CLOSE
-    _hera_program: op*
+    cpp_program: _cpp_open (op | include)* _CPP_CLOSE
+    hera_program: op (op | include)* |
 
     // Underscores before everything so that no tokens end up in the tree.
-    _INCLUDE: /#include.*/
     _cpp_open: "void" _SYMBOL "(" ")" "{"
     _CPP_CLOSE: "}"
     _SYMBOL: SYMBOL
 
+    include: "#include" ( STRING | /<[^>]+>/ )
+
     op: SYMBOL "(" _arglist? ")" ";"?
 
     _arglist: ( value "," )* value
@@ -106,7 +125,7 @@ def value(self, matches):
 )
 
 
-def parse(text):
+def parse(text, *, expand_includes=False):
     """Parse a HERA program into a list of Op objects."""
     try:
         tree = _parser.parse(text)
@@ -121,11 +140,25 @@ def parse(text):
         raise HERAError("invalid syntax", e.line, e.column) from None
 
     if isinstance(tree, Tree):
-        return tree.children
+        ops = tree.children
     elif isinstance(tree, Op):
-        return [tree]
+        ops = [tree]
+    else:
+        ops = tree
+
+    if expand_includes:
+        expanded_ops = []
+        for op in ops:
+            if op.name == "#include" and len(op.args) == 1 and not op.args[0].startswith("<"):
+                # Strip off the leading and trailing quote.
+                fpath = op.args[0][1:-1]
+                with open(fpath) as f:
+                    expanded_ops.extend(parse(f.read()))
+            else:
+                expanded_ops.append(op)
+        return expanded_ops
     else:
-        return tree
+        return ops
 
 
 def replace_escapes(s):

test/assets/lib/add.hera

@@ -0,0 +1,7 @@
+BR(end_of_add)
+
+LABEL(add)
+  ADD(R3, R1, R2)
+  RETURN(R12, R13)
+
+LABEL(end_of_add)

test/assets/simple_include.hera

@@ -0,0 +1,6 @@
+// TODO: This should be just "lib/add.hera"
+#include "test/assets/lib/add.hera"
+
+SET(R1, 20)
+SET(R2, 22)
+CALL(R12, add)

test/test_integration.py

@@ -338,6 +338,24 @@ def test_manual_strings_dot_hera():
     assert not vm.flag_sign
 
 
+def test_simple_include_dot_hera():
+    vm = VirtualMachine()
+    main(["test/assets/simple_include.hera"], vm)
+
+    assert vm.registers[1] == 20
+    assert vm.registers[2] == 22
+    assert vm.registers[3] == 42
+
+    for r in vm.registers[4:11]:
+        assert r == 0
+
+    assert not vm.flag_sign
+    assert not vm.flag_zero
+    assert not vm.flag_overflow
+    assert not vm.flag_carry
+    assert not vm.flag_carry_block
+
+
 def test_error_message_for_missing_comma(capsys):
     line = "SETLO(R1 40)"
     with pytest.raises(SystemExit):

test/test_parser.py

@@ -94,12 +94,13 @@ def test_parse_single_line_comment():
 
 def test_parse_hera_boilerplate():
     assert parse("#include <HERA.h>\nvoid HERA_main() {SETLO(R1, 42)}") == [
-        Op("SETLO", ["R1", 42])
+        Op("#include", ["<HERA.h>"]),
+        Op("SETLO", ["R1", 42]),
     ]
 
 
 def test_parse_hera_boilerplate_weird_whitespace_and_spelling():
-    assert parse("#include <HERA.h>\nvoid   HeRA_mAin( \t)\n {\n\n}") == []
+    assert parse("#include <HERA.h>\nvoid   HeRA_mAin( \t)\n {\n\n}") == [Op("#include", ["<HERA.h>"])]
 
 
 def test_parse_hera_boilerplate_no_includes():
@@ -132,6 +133,11 @@ def test_parse_multiline_comment():
     assert parse(program) == [Op("SETLO", ["R1", 1])]
 
 
+def test_parse_include_amidst_instructions():
+    program = 'SETLO(R1, 42)\n#include "whatever"\n'
+    assert parse(program) == [Op("SETLO", ["R1", 42]), Op("#include", ['"whatever"'])]
+
+
 def test_parse_missing_comma():
     with pytest.raises(HERAError):
         parse("ADD(R1, R2 R3)")
@@ -155,3 +161,14 @@ def test_parse_exception_has_line_number():
         assert e.line == 3
     else:
         assert False, "expected excepton"
+
+
+def test_parse_expands_include():
+    program = '#include "test/assets/lib/add.hera"'
+    assert parse(program, expand_includes=True) == [
+        Op("BR", ["end_of_add"]),
+        Op("LABEL", ["add"]),
+        Op("ADD", ["R3", "R1", "R2"]),
+        Op("RETURN", ["R12", "R13"]),
+        Op("LABEL", ["end_of_add"]),
+    ]