nqasm: Initial stab at a lexer and parser

So far, this only understands instructions, but no operands or labels.  It
doesn't handle bad input gracefully.  It's just enough to test the basics.

5 files changed, 278 insertions, 0 deletions

diff --git a/Makefile b/Makefile
new file mode 100644
index 0000000..554c540
--- /dev/null
+++ b/Makefile
@@ -0,0 +1,30 @@
+OPTFLAGS = -Os
+CFLAGS   = -std=c99 -D_POSIX_C_SOURCE=200809L $(OPTFLAGS)
+
+GENSRCS  = lexer.c parser.c
+GENHDRS  = lexer.h parser.h
+SRCS     = $(sort $(wildcard *.c) $(GENSRCS))
+OBJS     = $(SRCS:.c=.o)
+TARGET   = nqasm
+
+all: $(TARGET)
+
+clean:
+	rm -f $(GENSRCS) $(GENHDRS) $(OBJS) $(TARGET)
+
+lexer.c lexer.h: lexer.l
+	flex $<
+
+parser.c parser.h: parser.y
+	bison $<
+
+lexer.o: parser.h
+parser.o: lexer.h
+
+%.o: %.c
+	$(CC) $(CFLAGS) -c $< -o $@
+
+$(TARGET): $(OBJS)
+	$(CC) $(CFLAGS) $^ -o $@
+
+.PHONY: all clean
diff --git a/lexer.l b/lexer.l
new file mode 100644
index 0000000..9a34d66
--- /dev/null
+++ b/lexer.l
@@ -0,0 +1,61 @@
+%{
+
+#include "parser.h"
+
+%}
+
+%option outfile="lexer.c" header-file="lexer.h"
+%option noyywrap never-interactive
+
+%s inst
+%s args
+
+COMMENT    #[^\n]*
+SP         [ \t]
+EOI        [ \t\n#]
+
+%%
+
+<INITIAL>{SP}+           { BEGIN(inst);    return T_SPACE; }
+
+<inst>add/{EOI}          { BEGIN(args);    return T_ADD;    }
+<inst>sub/{EOI}          { BEGIN(args);    return T_SUB;    }
+<inst>mul/{EOI}          { BEGIN(args);    return T_MUL;    }
+<inst>div/{EOI}          { BEGIN(args);    return T_DIV;    }
+<inst>and/{EOI}          { BEGIN(args);    return T_AND;    }
+<inst>or/{EOI}           { BEGIN(args);    return T_OR;     }
+<inst>xor/{EOI}          { BEGIN(args);    return T_XOR;    }
+<inst>lsl/{EOI}          { BEGIN(args);    return T_LSL;    }
+<inst>lsr/{EOI}          { BEGIN(args);    return T_LSR;    }
+<inst>asl/{EOI}          { BEGIN(args);    return T_ASL;    }
+<inst>asr/{EOI}          { BEGIN(args);    return T_ASR;    }
+<inst>rol/{EOI}          { BEGIN(args);    return T_ROL;    }
+<inst>ror/{EOI}          { BEGIN(args);    return T_ROR;    }
+<inst>not/{EOI}          { BEGIN(args);    return T_NOT;    }
+<inst>neg/{EOI}          { BEGIN(args);    return T_NEG;    }
+<inst>btc/{EOI}          { BEGIN(args);    return T_BTC;    }
+<inst>bts/{EOI}          { BEGIN(args);    return T_BTS;    }
+<inst>mov/{EOI}          { BEGIN(args);    return T_MOV;    }
+<inst>st\.b/{EOI}        { BEGIN(args);    return T_ST_B;   }
+<inst>st\.w/{EOI}        { BEGIN(args);    return T_ST_W;   }
+<inst>ld\.bl/{EOI}       { BEGIN(args);    return T_LD_BL;  }
+<inst>ld\.bh/{EOI}       { BEGIN(args);    return T_LD_BH;  }
+<inst>ld\.w/{EOI}        { BEGIN(args);    return T_LD_W;   }
+<inst>ldi\.bl/{EOI}      { BEGIN(args);    return T_LDI_BL; }
+<inst>ldi\.bh/{EOI}      { BEGIN(args);    return T_LDI_BH; }
+<inst>beq/{EOI}          { BEGIN(args);    return T_BEQ;    }
+<inst>bne/{EOI}          { BEGIN(args);    return T_BNE;    }
+<inst>bgt/{EOI}          { BEGIN(args);    return T_BGT;    }
+<inst>bge/{EOI}          { BEGIN(args);    return T_BGE;    }
+<inst>blt/{EOI}          { BEGIN(args);    return T_BLT;    }
+<inst>ble/{EOI}          { BEGIN(args);    return T_BLE;    }
+<inst>bra/{EOI}          { BEGIN(args);    return T_BRA;    }
+<inst>jmp/{EOI}          { BEGIN(args);    return T_JMP;    }
+<inst>addpc/{EOI}        { BEGIN(args);    return T_ADDPC;  }
+<inst>nop/{EOI}          { BEGIN(args);    return T_NOP;    }
+
+{SP}*{COMMENT}?\n        { BEGIN(INITIAL); return T_EOL;    }
+{SP}*{COMMENT}           { BEGIN(INITIAL); return T_SPACE;  }
+{SP}+                    {                 return T_SPACE;  }
+
+%%
diff --git a/nqasm.c b/nqasm.c
new file mode 100644
index 0000000..a42f8f4
--- /dev/null
+++ b/nqasm.c
@@ -0,0 +1,20 @@
+#include <stdio.h>
+
+#include "mnemonics.h"
+#include "nqasm.h"
+
+#include "lexer.h"
+#include "parser.h"
+
+void add_instruction(const struct instruction *i)
+{
+	const struct mnemonic *m = &mnemonics[i->mnem];
+	printf("%04x\n", m->bits);
+}
+
+int main(int argc, char **argv)
+{
+	yyparse();
+	return 0;
+}
+
diff --git a/nqasm.h b/nqasm.h
new file mode 100644
index 0000000..3712849
--- /dev/null
+++ b/nqasm.h
@@ -0,0 +1,11 @@
+#ifndef NQ_NQASM_H
+#define NQ_NQASM_H
+
+struct instruction {
+	int mnem;
+};
+
+void add_instruction(const struct instruction *);
+
+#endif
+
diff --git a/parser.y b/parser.y
new file mode 100644
index 0000000..78bf123
--- /dev/null
+++ b/parser.y
@@ -0,0 +1,156 @@
+%{
+
+#include <stdio.h>
+
+#include "lexer.h"
+#include "mnemonics.h"
+
+void yyerror(const char *msg)
+{
+	fprintf(stderr, "%s\n", msg);
+}
+
+#define INSTRUCTION(mnem) (struct instruction) { MNEM_ ## mnem }
+
+%}
+
+%output  "parser.c"
+%defines "parser.h"
+
+%code requires {
+	#include "nqasm.h"
+}
+
+%union {
+	struct instruction inst;
+}
+
+%token END 0 "end of file"
+
+%token T_SPACE
+%token T_EOL
+
+%token T_ADD
+%token T_SUB
+%token T_MUL
+%token T_DIV
+%token T_AND
+%token T_OR
+%token T_XOR
+%token T_LSL
+%token T_LSR
+%token T_ASL
+%token T_ASR
+%token T_ROL
+%token T_ROR
+%token T_NOT
+%token T_NEG
+%token T_BTC
+%token T_BTS
+%token T_MOV
+%token T_ST_B
+%token T_ST_W
+%token T_LD_BL
+%token T_LD_BH
+%token T_LD_W
+%token T_LDI_BL
+%token T_LDI_BH
+%token T_BEQ
+%token T_BNE
+%token T_BGT
+%token T_BGE
+%token T_BLT
+%token T_BLE
+%token T_BRA
+%token T_JMP
+%token T_ADDPC
+%token T_NOP
+
+%type <inst> inst
+%%
+
+input:	%empty
+	|	input line
+	;
+
+line:	T_SPACE inst eol
+		{ add_instruction(&$2); }
+	|	eol
+	;
+
+inst:	T_ADD
+		{ $$ = INSTRUCTION(ADD); }
+	|	T_SUB
+		{ $$ = INSTRUCTION(SUB); }
+	|	T_MUL
+		{ $$ = INSTRUCTION(MUL); }
+	|	T_DIV
+		{ $$ = INSTRUCTION(DIV); }
+	|	T_AND
+		{ $$ = INSTRUCTION(AND); }
+	|	T_OR
+		{ $$ = INSTRUCTION(OR); }
+	|	T_XOR
+		{ $$ = INSTRUCTION(XOR); }
+	|	T_LSL
+		{ $$ = INSTRUCTION(LSL); }
+	|	T_LSR
+		{ $$ = INSTRUCTION(LSR); }
+	|	T_ASL
+		{ $$ = INSTRUCTION(ASL); }
+	|	T_ASR
+		{ $$ = INSTRUCTION(ASR); }
+	|	T_ROL
+		{ $$ = INSTRUCTION(ROL); }
+	|	T_ROR
+		{ $$ = INSTRUCTION(ROR); }
+	|	T_NOT
+		{ $$ = INSTRUCTION(NOT); }
+	|	T_NEG
+		{ $$ = INSTRUCTION(NEG); }
+	|	T_BTC
+		{ $$ = INSTRUCTION(BTC); }
+	|	T_BTS
+		{ $$ = INSTRUCTION(BTS); }
+	|	T_MOV
+		{ $$ = INSTRUCTION(MOV); }
+	|	T_ST_B
+		{ $$ = INSTRUCTION(ST_B); }
+	|	T_ST_W
+		{ $$ = INSTRUCTION(ST_W); }
+	|	T_LD_BL
+		{ $$ = INSTRUCTION(LD_BL); }
+	|	T_LD_BH
+		{ $$ = INSTRUCTION(LD_BH); }
+	|	T_LD_W
+		{ $$ = INSTRUCTION(LD_W); }
+	|	T_LDI_BL
+		{ $$ = INSTRUCTION(LDI_BL); }
+	|	T_LDI_BH
+		{ $$ = INSTRUCTION(LDI_BH); }
+	|	T_BEQ
+		{ $$ = INSTRUCTION(BEQ); }
+	|	T_BNE
+		{ $$ = INSTRUCTION(BNE); }
+	|	T_BGT
+		{ $$ = INSTRUCTION(BGT); }
+	|	T_BGE
+		{ $$ = INSTRUCTION(BGE); }
+	|	T_BLT
+		{ $$ = INSTRUCTION(BLT); }
+	|	T_BLE
+		{ $$ = INSTRUCTION(BLE); }
+	|	T_BRA
+		{ $$ = INSTRUCTION(BRA); }
+	|	T_JMP
+		{ $$ = INSTRUCTION(JMP); }
+	|	T_ADDPC
+		{ $$ = INSTRUCTION(ADDPC); }
+	|	T_NOP
+		{ $$ = INSTRUCTION(NOP); }
+	;
+
+eol:	T_EOL
+	|	T_SPACE END
+	|	END
+	;