From 0157e235e45eb95ba2cb37577ee206652c3b3323 Mon Sep 17 00:00:00 2001
From: Paul Mathieu <paulmathieu@google.com>
Date: Thu, 18 Feb 2021 21:20:35 -0800
Subject: [PATCH] Add support for char litterals and <<

Also streamline cc.py to do it all:
- cpp
- cc
- as

LD is still a separate step
---
 tools/cc.ebnf |   3 +-
 tools/cc.py   | 109 +++++++++++++++++++++++++++++++++++---------------
 2 files changed, 78 insertions(+), 34 deletions(-)

diff --git a/tools/cc.ebnf b/tools/cc.ebnf
index a3ba3d6..52f29d9 100644
--- a/tools/cc.ebnf
+++ b/tools/cc.ebnf
@@ -123,7 +123,7 @@ initializer_list: "{" [init_list_field ("," init_list_field)* ","? ] "}"
 
 
 label: IDENTIFIER
-litteral: SIGNED_NUMBER | ESCAPED_STRING | HEX_LITTERAL
+litteral: SIGNED_NUMBER | ESCAPED_STRING | HEX_LITTERAL | CHARACTER
 field: IDENTIFIER
 identifier: IDENTIFIER
 ?symbol: IDENTIFIER
@@ -145,6 +145,7 @@ struct_field: type IDENTIFIER sized_array* ";"
 IDENTIFIER: /[_a-zA-Z]\w*/
 COMMENT: /\/\/.*/
 HEX_LITTERAL: /0x[a-fA-F0-9]+/
+CHARACTER: /'[^']'/
 
 
 %import common.WS
diff --git a/tools/cc.py b/tools/cc.py
index b315b1f..dc3b3a1 100644
--- a/tools/cc.py
+++ b/tools/cc.py
@@ -1,11 +1,18 @@
+import argparse
 import collections
 import contextlib
+import importlib
+import io
 import re
+import subprocess
 import sys
 
 import lark
 
+asmod = importlib.import_module("as")
+
 GRAMMAR_FILE = '/home/paulmathieu/vhdl/bin/cc.ebnf'
+CPP = ('cpp', '-P')
 
 
 class Scope:
@@ -91,7 +98,7 @@ class RegBank:
         if not self.available:
             assert self.vars, "nothing to clean, no more regs :/"
             # storing one random var
-            var = self.vars.keys()[0]
+            var = list(self.vars.keys())[0]
             self.evict(var)
         return self.available.pop(0)
 
@@ -267,6 +274,7 @@ class CcTransform(lark.visitors.Transformer):
     mul = _binary_op(lambda a, b: a*b)
     shl = _binary_op(lambda a, b: a<<b)
 
+    CHARACTER = lambda _, x: ord(x[1])
     IDENTIFIER = str
     SIGNED_NUMBER = int
     HEX_LITTERAL = lambda _, x: int(x[2:], 16)
@@ -321,26 +329,19 @@ AndOp = make_cpu_bin_op('and')
 orOp = make_cpu_bin_op('or')
 XorOp = make_cpu_bin_op('xor')
 
-
-def combo(UnOp2, Op1):
-    """Apply UnOp2 on the output of Op1."""
-
-    class _C(AsmOp):
-        scratch_need = max(Op1.scratch_need, UnOp2.scratch_need)
-
-        @property
-        def out(self):
-            return self.op2.out
-
-        def __init__(self, fun, ops):
-            self.op1 = Op1(fun, ops)
-            self.op2 = Op2(fun, self.op1.out)
-
-        def synth(self, scratches):
-            self.op1.synth(scratches)
-            self.op2.synth(scratches)
-
-    return _C
+class ShlOp(BinOp):
+    scratch_need = 3
+    def synth(self, scratches):
+        sc0, sc1, sc2 = scratches
+        return [f'or {sc0}, {self.right}, {self.right}',
+                f'or {self.dest}, {self.left}, {self.left}',
+                f'set {sc1}, 1',
+                f'set {sc2}, 0',
+                f'cmp {sc0}, {sc2}',
+                f'beq [pc, 6]',
+                f'add {self.dest}, {self.dest}, {self.dest}',
+                f'sub {sc0}, {sc0}, {sc1}',
+                f'bneq [pc, -6]']
 
 
 class LtOp(BinOp):
@@ -394,11 +395,6 @@ class NeqOp(BinOp):
     def synth(self, scratches):
         return [f'sub {self.dest}, {self.left}, {self.right}']
 
-EqOp = combo(BoolNot, NeqOp)
-LeOp = combo(BoolNot, GtOp)
-GeOp = combo(BoolNot, LtOp)
-
-
 
 class FnCall(AsmOp):
     scratch_need = 1
@@ -862,6 +858,13 @@ class CcInterp(lark.visitors.Interpreter):
             self.cur_fun.regs.give(right)
         return _f
 
+    def _combo(uop, bop):
+        def _f(self, tree):
+            bop.__get__(self)(tree)
+            uop.__get__(self)(tree)
+        return _f
+
+    shl = _binary_op(ShlOp)
     add = _binary_op(AddOp)
     sub = _binary_op(SubOp)
     mul = _binary_op(MulOp)
@@ -870,6 +873,7 @@ class CcInterp(lark.visitors.Interpreter):
     gt = _binary_op(GtOp)
     lt = _binary_op(LtOp)
     neq = _binary_op(NeqOp)
+    eq = _combo(bool_not, neq)
 
     def _forward_op(self, tree):
         self.visit_children(tree)
@@ -968,11 +972,12 @@ def filter_dupes(ops):
             continue
         yield op
 
-def parse_tree(tree):
+def parse_tree(tree, debug=False):
     tr = CcTransform()
     tree = tr.transform(tree)
 
-#    print(tree.pretty())
+    if debug:
+        print(tree.pretty())
 
     inte = CcInterp()
     inte.visit(tree)
@@ -985,13 +990,51 @@ def parse_tree(tree):
     return '\n'.join(filter_dupes(out))
 
 
-def larkparse(f):
+def larkparse(f, debug=False):
     with open(GRAMMAR_FILE) as g:
         asparser = lark.Lark(g.read())
-    tree = asparser.parse(f.read())
-    return parse_tree(tree)
+    data = f.read()
+    if isinstance(data, bytes):
+        data = data.decode()
+    tree = asparser.parse(data)
+    return parse_tree(tree, debug=debug)
+
+
+def parse_args():
+    parser = argparse.ArgumentParser(description='Compile.')
+    group = parser.add_mutually_exclusive_group()
+    parser.add_argument('--debug', action='store_true',
+                        help='print the AST')
+    parser.add_argument('--assembly', '-S', action='store_true',
+                        help='output assembly')
+    parser.add_argument('--compile', '-c', action='store_true',
+                        help='compile a single file')
+    parser.add_argument('input', type=argparse.FileType('r'),
+                        default=sys.stdin, help='input file (default: stdin)')
+    parser.add_argument('--output', '-o', type=argparse.FileType('wb'),
+                        default=sys.stdout.buffer, help='output file')
+    return parser.parse_args()
+
+
+def preprocess(fin):
+    p = subprocess.Popen(CPP, stdin=fin, stdout=subprocess.PIPE)
+    return p.stdout
+
+
+def assemble(text, fout):
+    fin = io.StringIO(text)
+    asmod.write_obj(fout, *asmod.larkparse(fin))
+
+
+def main():
+    args = parse_args()
+
+    assy = larkparse(preprocess(args.input), debug=args.debug)
+    if args.assembly:
+        args.output.write(assy.encode() + b'\n')
+    else:
+        assemble(assy, args.output)
 
 
 if __name__ == "__main__":
-#    print(',\n'.join(parse(sys.stdin)))
-    print(larkparse(sys.stdin))
+    main()