From 7e2ecc6e1f9e49439696adc72dcbefd699168e44 Mon Sep 17 00:00:00 2001
From: uvok
Date: Mon, 12 Jan 2026 19:16:17 +0100
Subject: Add first version of assembler

---
 nandgame/assembler/parser.py | 115 +++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 115 insertions(+)
 create mode 100644 nandgame/assembler/parser.py

(limited to 'nandgame/assembler/parser.py')

diff --git a/nandgame/assembler/parser.py b/nandgame/assembler/parser.py
new file mode 100644
index 0000000..b2cc3f9
--- /dev/null
+++ b/nandgame/assembler/parser.py
@@ -0,0 +1,115 @@
+import sys
+
+import ply.yacc as yacc
+from ply.lex import LexToken
+
+# Get the token map from the lexer.  This is required.
+from lexer import tokens
+
+# Error rule for syntax errors
+
+def p_error(p: LexToken):
+    if p:
+        print(f"Unexpected {p.value} on line {p.lineno}")
+    else:
+        print("Unexpected end of file.")
+
+def p_program(p):
+    '''program : instruction_list
+               | empty'''
+    pass
+
+def p_empty(p):
+    '''empty :'''
+    pass
+
+def p_inss(p):
+    '''instruction_list : instruction_list instruction
+                        | instruction
+    '''
+    if len(p) == 2:
+        p[0] = [p[1]]
+    else:
+        p[0] = p[1] + [p[2]]
+
+    # print(f"INSes: {len(p)}")
+    # print(f"    {p[1]}")
+    # if len(p) > 2:
+    #     print(f"    {p[2]}")
+    pass
+
+def p_inss2(p):
+    '''instruction_list2 : instruction instruction_list2
+                        | instruction
+    '''
+    if len(p) == 2:
+        p[0] = [p[1]]
+    else:
+        p[0] = [p[1]] + p[2]
+
+    print(f"INSes2: {len(p)}")
+    print(f"    {p[1]}")
+    if len(p) > 2:
+        print(f"    {p[2]}")
+    pass
+
+def p_instruction(p):
+    '''instruction : noarg
+                   | onearg
+                   | twoarg
+                   | jumpdest
+    '''
+                #    | invalid_arg
+    print(f"INS: {p[1]}")
+    p[0] = p[1]
+    pass
+
+def p_jumpdest(p):
+    '''jumpdest : SYMBOL COLON'''
+    p[0] = (p[1], )
+
+def p_twoarg(p):
+    '''twoarg : op REG COMMA argument COMMA argument'''
+    p[0] = (*p[1], p[2], p[4], p[6])
+    pass
+
+def p_argument(p):
+    '''argument : number
+                | REG
+                | SYMBOL
+    '''
+    pass
+
+def p_onearg(p):
+    '''onearg : op REG COMMA argument'''
+    p[0] = (*p[1], p[2], p[4], None)
+    pass
+
+def p_noarg(p):
+    '''noarg : op'''
+    p[0] = (*p[1], None, None, None)
+    pass
+
+def p_invalid_arg(p):
+    '''invalid_arg : op argument'''
+    print(f"Invalid opcode, destination missing: {p[1:]}")
+    sys.exit(1)
+    pass
+
+def p_op(p):
+    '''op : OP DOT JUMP
+          | OP
+    '''
+    if len(p) == 2:
+        p[0] = (p[1], None)
+    else:
+        p[0] = (p[1], p[3])
+
+
+def p_number(p):
+    '''number : NUMBER
+              | HEXNUMBER
+    '''
+    p[0] = p[1]
+
+parser = yacc.yacc()
-- 
cgit v1.2.3