Skeleton for the assembler tool

dasm.py

@@ -0,0 +1,111 @@
+#!/usr/bin/env python3
+
+from rply import LexerGenerator, ParserGenerator
+
+
+_lexer = LexerGenerator()
+_lexer.add("DIRECTIVE", r"\.org|\.entry|\.dw|\.ds|\.dsz|\.dsp")
+_lexer.add("OPCODE2", r"SET|ADD|SUB|MUL|MLI|DIV|MOD|MDI|AND|BOR|XOR|SHR|ASR|SHL|IFB|IFC|IFE|IFN|IFG|IFA|IFL|IFU|ADX|SBX|STI|STD")
+_lexer.add("OPCODE1", r"JSR|INT|IAG|IAS|RFI|IAQ|HWN|HWQ|HWI")
+_lexer.add("OPCODE0", r"HLT")
+_lexer.add("GREG", r"A|B|C|X|Y|Z|I|J")
+_lexer.add("SREG", r"SP|PC|EX")
+_lexer.add("STACK", r"PUSH|POP|PEEK|PICK")
+_lexer.add("COMMA", r",")
+_lexer.add("COLON", r":")
+_lexer.add("OPENSB", r"\[")
+_lexer.add("CLOSESB", r"\]")
+_lexer.add("BINOP", r"[+*/-]|>>|<<")
+_lexer.add("OPENRB", r"\(")
+_lexer.add("CLOSERB", r"\)")
+_lexer.add("NUMBER", r"0x[0-9a-fA-F]+|[0-9]+")
+_lexer.add("SYMBOL", r"[a-zA-Z._@][a-zA-Z._@0-9]*")
+#_lexer.add("EOL", r"([;#][^\n]*)?(\r?\n)+")
+#_lexer.ignore(r"[ \t]+")
+_lexer.ignore(r"\s+|([;#][^\n]*)?\n")
+lexer = _lexer.build()
+
+
+_parser = ParserGenerator([x.name for x in lexer.rules])
+
+@_parser.production("expression : NUMBER")
+def expression_number(p):
+    n = p[0].getstr()
+    if n.startswith('0x'):
+        return int(n, 16)
+    else:
+        return int(n, 10)
+
+@_parser.production("expression : SYMBOL")
+def expression_symbol(p):
+    return ('SYMBOL', p[0].getstr())
+
+@_parser.production("expression : OPENRB expression CLOSERB")
+def expression_parens(p):
+    return p[1]
+
+@_parser.production("expression : expression BINOP expression")
+def expression_binop(p):
+    a, op, b = tuple(p)
+    print(op,a,b)
+    return (op, a, b)
+
+@_parser.production("value : GREG")
+@_parser.production("value : SREG")
+def value_reg(p):
+    return p[1]
+
+@_parser.production("value : OPENSB GREG CLOSESB")
+def value_greg_indirect(p):
+    pass
+
+@_parser.production("value : OPENSB GREG BINOP expression CLOSESB")
+def value_greg_indirect_offset(p):
+    pass
+
+@_parser.production("code : OPCODE0")
+def opcode0(p):
+    pass
+
+@_parser.production("code : OPCODE1 value")
+def opcode1(p):
+    pass
+
+@_parser.production("code : OPCODE2 value COMMA value")
+def opcode2(p):
+    pass
+
+@_parser.production("directive : DIRECTIVE expression")
+def directive1(p):
+    pass
+
+@_parser.production("source : LABEL")
+def source_label(p):
+    pass
+
+@_parser.production("source : code")
+def source_code(p):
+    pass
+
+@_parser.production("source : directive")
+def source_directive(p):
+    pass
+
+parser = _parser.build()
+
+
+def assemble(filename):
+    with open(filename, "r") as f:
+        text = f.read()
+        print(len(text))
+        for token in lexer.lex(text):
+            print(token.source_pos, token)
+
+
+if __name__ == '__main__':
+    import sys
+    
+    for f in sys.argv[1:]:
+        assemble(f)
+
+# vim:et:sw=4:ts=4:sts=4: