refactor compilation of builtin forms

2 files changed, 159 insertions, 102 deletions

diff --git a/com.c b/com.c
index 45b1f80..c68e663 100644
--- a/com.c
+++ b/com.c
@@ -20,16 +20,144 @@ static void patch(State *S, Chunk *ch, size_t addr, uint16_t val) {
 }
 
 
+
+static void compile_node(State *S, Chunk *ch, AstNode a);
+
+typedef void (*form_compiler)(State *S, Chunk *ch, AstVec l, Op op);
+typedef struct {
+	char *name;
+	int min_params;
+	bool ellipsis;
+	form_compiler action;
+	Op op;
+} BuiltinForm;
+
+
+void single_form(State *S, Chunk *ch, AstVec l, Op op) {
+	compile_node(S, ch, l.vals[1]);
+	chunk_wbc(S, ch, op);
+}
+
+void set_form(State *S, Chunk *ch, AstVec l, Op op) {
+	AstNode ident = l.vals[1];
+	if (ident.ty != AST_IDENT) {
+		fprintf(stderr, "set's first argument must be identifier");
+		exit(1);
+	}
+	ObjString *o = objstring_copy_cstr(S, ident.as.str);
+	compile_node(S, ch, l.vals[2]);
+	chunk_wbc(S, ch, OP_SETGLOBAL);
+	chunk_wbc(S, ch, chunk_wconst(S, ch, VAL_OBJ(o)));
+}
+
+void do_form(State *S, Chunk *ch, AstVec l, Op op) {
+	for (int i = 1; i < l.len - 1; i++) {
+		compile_node(S, ch, l.vals[i]);
+		chunk_wbc(S, ch, OP_DROP);
+	}
+	compile_node(S, ch, l.vals[l.len - 1]);
+}
+
+void if_form(State *S, Chunk *ch, AstVec l, Op op) {
+	// (if cond if-true if-false)
+	// cond
+	// 0branch ->A
+	// if-true
+	// skip ->B
+	// A: if-false
+	// B:
+	compile_node(S, ch, l.vals[1]);
+	chunk_wbc(S, ch, OP_0BRANCH);
+	size_t ph_a = placeholder(S, ch);
+	compile_node(S, ch, l.vals[2]);
+	chunk_wbc(S, ch, OP_SKIP);
+	size_t ph_b = placeholder(S, ch);
+	size_t dest_a = ch->bc.len;
+	compile_node(S, ch, l.vals[3]);
+	size_t dest_b = ch->bc.len;
+
+	patch(S, ch, ph_a, dest_a - ph_a - 2);
+	patch(S, ch, ph_b, dest_b - ph_b - 2);
+}
+
+void while_form(State *S, Chunk *ch, AstVec l, Op op) {
+	// (while cond body ...)
+	// A:
+	// cond
+	// 0branch ->B
+	// body ....
+	// redo ->A
+	// B:
+	// nil   (while loop always returns nil)
+	size_t dest_a = ch->bc.len;
+	compile_node(S, ch, l.vals[1]);
+	chunk_wbc(S, ch, OP_0BRANCH);
+	size_t ph_b = placeholder(S, ch);
+	for (int i = 2; i < l.len; i++)  {
+		compile_node(S, ch, l.vals[i]);
+		chunk_wbc(S, ch, OP_DROP);
+	}
+	chunk_wbc(S, ch, OP_REDO);
+	size_t ph_a = placeholder(S, ch);
+	size_t dest_b = ch->bc.len;
+	chunk_wbc(S, ch, OP_NIL);
+
+	patch(S, ch, ph_a, ph_a - dest_a + 2);
+	patch(S, ch, ph_b, dest_b - ph_b - 2);
+}
+
+void arith_form(State *S, Chunk *ch, AstVec l, Op op) {
+	compile_node(S, ch, l.vals[1]);
+	compile_node(S, ch, l.vals[2]);
+	chunk_wbc(S, ch, op);
+}
+
+static BuiltinForm builtin_forms[] = {
+	{ "puts",  1, false, single_form, OP_PUTS },
+	{ "print", 1, false, single_form, OP_PRINT },
+	{ "set",   2, false, set_form, 0 },
+	{ "do",    1,  true, do_form, 0 },
+	{ "if",    3, false, if_form, 0 },
+	{ "while", 2,  true, while_form, 0 },
+#define ARITH_OP(str, op) \
+	{ str, 2, false, arith_form, op },
+	ARITH_OP("+", OP_ADD)
+	ARITH_OP("-", OP_SUB)
+	ARITH_OP("*", OP_MUL)
+	ARITH_OP("/", OP_DIV)
+	ARITH_OP("=", OP_EQU)
+	ARITH_OP("<", OP_CMP)
+	ARITH_OP("%", OP_MOD)
+#undef ARITH_OP
+	{ 0 },
+};
+
+typedef struct {
+	char *name;
+	Op op;
+} BuiltinIdent;
+static BuiltinIdent builtin_idents[] = {
+	{ "true", OP_TRUE },
+	{ "false", OP_FALSE },
+	{ "nil", OP_NIL },
+	{ 0 },
+};
+
 static void compile_node(State *S, Chunk *ch, AstNode a) {
 	switch (a.ty) {
 	case AST_IDENT:;
 		char *ident = a.as.str;
-
-		if (0 == strcmp(ident, "true")) chunk_wbc(S, ch, OP_TRUE);
-		else if (0 == strcmp(ident, "false")) chunk_wbc(S, ch, OP_FALSE);
-		else if (0 == strcmp(ident, "nil")) chunk_wbc(S, ch, OP_NIL);
-		else {
-			// global read
+		BuiltinIdent *found_builtin = NULL;
+		for (BuiltinIdent *b = builtin_idents; b->name != NULL; b++) {
+			if (0 == strcmp(b->name, ident)) {
+				found_builtin = b;
+				break;
+			}
+		}
+		if (found_builtin != NULL) {
+			chunk_wbc(S, ch, found_builtin->op);
+		} else {
+			// read global variable
 			ObjString *o = objstring_copy_cstr(S, a.as.str);
 			chunk_wbc(S, ch, OP_GETGLOBAL);
 			chunk_wbc(S, ch, chunk_wconst(S, ch, VAL_OBJ(o)));
@@ -47,113 +175,43 @@ static void compile_node(State *S, Chunk *ch, AstNode a) {
 	}
 	case AST_LIST: {
 		AstVec l = a.as.list;
+
 		#define CK(cond, msg) if (!(cond)) { fputs(msg "\n", stderr); exit(1); }
 		CK(l.len > 0, "can't handle empty list");
 		CK(l.vals[0].ty == AST_IDENT, "can only call ops");
 
-		char *name = l.vals[0].as.str;
-
-		if (0 == strcmp(name, "puts")) {
-			CK(l.len == 2, "puts requires exactly 1 argument");
-			compile_node(S, ch, l.vals[1]);
-			chunk_wbc(S, ch, OP_PUTS);
-		} else if (0 == strcmp(name, "print")) {
-			CK(l.len == 2, "print requires exactly 1 argument");
-			compile_node(S, ch, l.vals[1]);
-			chunk_wbc(S, ch, OP_PRINT);
-		} else if (0 == strcmp(name, "set")) {
-			CK(l.len == 3, "set requires exactly 2 arguments");
-			AstNode ident = l.vals[1];
-			CK(ident.ty == AST_IDENT, "set's first argument must be identifier");
-
-			ObjString *o = objstring_copy_cstr(S, ident.as.str);
-			compile_node(S, ch, l.vals[2]);
-			chunk_wbc(S, ch, OP_SETGLOBAL);
-			chunk_wbc(S, ch, chunk_wconst(S, ch, VAL_OBJ(o)));
-		} else if (0 == strcmp(name, "do")) {
-			for (int i = 1; i < l.len - 1; i++) {
-				compile_node(S, ch, l.vals[i]);
-				chunk_wbc(S, ch, OP_DROP);
-			}
-			compile_node(S, ch, l.vals[l.len - 1]);
-		} else if (0 == strcmp(name, "if")) {
-			CK(l.len == 4, "if requires exactly 3 arguments");
-			// (if cond if-true if-false)
-			// cond
-			// 0branch ->A
-			// if-true
-			// skip ->B
-			// A: if-false
-			// B:
-			compile_node(S, ch, l.vals[1]);
-			chunk_wbc(S, ch, OP_0BRANCH);
-			size_t ph_a = placeholder(S, ch);
-			compile_node(S, ch, l.vals[2]);
-			chunk_wbc(S, ch, OP_SKIP);
-			size_t ph_b = placeholder(S, ch);
-			size_t dest_a = ch->bc.len;
-			compile_node(S, ch, l.vals[3]);
-			size_t dest_b = ch->bc.len;
-
-			patch(S, ch, ph_a, dest_a - ph_a - 2);
-			patch(S, ch, ph_b, dest_b - ph_b - 2);
-
-		} else if (0 == strcmp(name, "while")) { 
-			CK(l.len >= 3, "while requires at least 2 arguments");
-			// (while cond body ...)
-			// A:
-			// cond
-			// 0branch ->B
-			// body ....
-			// redo ->A
-			// B:
-			// nil   (while loop always returns nil)
-			size_t dest_a = ch->bc.len;
-			compile_node(S, ch, l.vals[1]);
-			chunk_wbc(S, ch, OP_0BRANCH);
-			size_t ph_b = placeholder(S, ch);
-			for (int i = 2; i < l.len; i++)  {
-				compile_node(S, ch, l.vals[i]);
-				chunk_wbc(S, ch, OP_DROP);
+		char *head = l.vals[0].as.str;
+
+		BuiltinForm *form = NULL;
+		for (BuiltinForm *b = builtin_forms; b->name != NULL; b++) {
+			if (0 == strcmp(b->name, head)) {
+				form = b;
+				break;
 			}
-			chunk_wbc(S, ch, OP_REDO);
-			size_t ph_a = placeholder(S, ch);
-			size_t dest_b = ch->bc.len;
-			chunk_wbc(S, ch, OP_NIL);
+		}
 
-			patch(S, ch, ph_a, ph_a - dest_a + 2);
-			patch(S, ch, ph_b, dest_b - ph_b - 2);
+		if (form != NULL) {
+			if (form->ellipsis && l.len < form->min_params + 1) {
+				fprintf(stderr, "%s requires at least %d parameters\n",
+					form->name, form->min_params);
+				exit(1);
+			} else if (!form->ellipsis && l.len != form->min_params + 1) {
+				fprintf(stderr, "%s requires exactly %d parameters\n",
+					form->name, form->min_params);
+				exit(1);
+			}
 
+			form->action(S, ch, l, form->op);
 		} else {
-				
-			CK(l.len == 3, "can only compile binary ops");
-			char opchar = l.vals[0].as.str[0];
-			Op op;
-			switch (opchar) {
-			#define OP(char, code) case char: op = code; break;
-				OP('+', OP_ADD)
-				OP('-', OP_SUB)
-				OP('*', OP_MUL)
-				OP('/', OP_DIV)
-				OP('=', OP_EQU)
-				OP('<', OP_CMP)
-				OP('%', OP_MOD)
-			#undef OP
-				default:
-					printf("unkown op %s\n",l.vals[0].as.str);
-					exit(1);
-					break;
-			}
-			compile_node(S, ch, l.vals[1]);
-			compile_node(S, ch, l.vals[2]);
-			chunk_wbc(S, ch, op);
+			fprintf(stderr, "unknown form %s\n", head);
+			exit(1);
 		}
-		#undef CK
+
+		break;
 	}
 	}
 }
 
-
 int main(int argc, char **argv) {
 	State st = state_new();
 	State *S = &st;
@@ -183,4 +241,3 @@ int main(int argc, char **argv) {
 
 	return runvm(S);
 }
-
diff --git a/tests/compile_error.out b/tests/compile_error.out
index b7b8184..2bf3ee6 100644
--- a/tests/compile_error.out
+++ b/tests/compile_error.out
@@ -1 +1 @@
-while requires at least 2 arguments
+while requires at least 2 parameters