lexer: insert semicolon after end of line for less erroneous expressi…

…on parsing

examples/declaration.luna

@@ -1,7 +1,6 @@
 # declaration
 
-let a = 2
-    b = 5
+let a = 2, b = "foo"
 
 type foo
   a, b: int
@@ -24,7 +23,7 @@ def repeat(a: string, n = 0)
   let res: string
   while n > 0
     res += a
-    n -= 1
+    --n
   end
   return res
 end

src/lexer.c

@@ -48,6 +48,14 @@
 
 #define error(msg) (self->error = msg, token(ILLEGAL))
 
+/*
+ * True if the lexer should insert a semicolon after `t`.
+ */
+
+#define need_semi(t) \
+  (t == LUNA_TOKEN_ID || t == LUNA_TOKEN_FLOAT || t == LUNA_TOKEN_INT || \
+   t == LUNA_TOKEN_STRING || t == LUNA_TOKEN_RETURN)
+
 /*
  * Initialize lexer with the given `source` and `filename`.
  */
@@ -267,7 +275,6 @@ scan_number(luna_lexer_t *self, int c) {
 int
 luna_scan(luna_lexer_t *self) {
   int c;
-  token(ILLEGAL);
 
   // scan
   scan:
@@ -350,8 +357,13 @@ luna_scan(luna_lexer_t *self) {
     case '#':
       while ((c = next) != '\n' && c) ; undo;
       goto scan;
+    case ';':
+      return token(SEMICOLON);
     case '\n':
     case '\r':
+      if (need_semi(self->tok.type)) {
+        return undo, token(SEMICOLON);
+      }
       ++self->lineno;
       goto scan;
     case '"':
@@ -363,6 +375,7 @@ luna_scan(luna_lexer_t *self) {
     default:
       if (isalpha(c) || '_' == c) return scan_ident(self, c);
       if (isdigit(c) || '.' == c) return scan_number(self, c);
+      token(ILLEGAL);
       error("illegal character");
       return 0;
   }

src/parser.c

@@ -337,9 +337,9 @@ unary_expr(luna_parser_t *self) {
     || is(OP_PLUS)
     || is(OP_MINUS)
     || is(OP_NOT)) {
-    luna_token_t *op = self->tok;
+    int op = self->tok->type;
     next;
-    return (luna_node_t *) luna_unary_op_node_new(op->type, unary_expr(self), 0);
+    return (luna_node_t *) luna_unary_op_node_new(op, unary_expr(self), 0);
   }
   return postfix_expr(self);
 }
@@ -931,11 +931,17 @@ type_stmt(luna_parser_t *self) {
   type = luna_type_node_new(name);
   next;
 
+  // semicolon might have been inserted here
+  accept(SEMICOLON);
+
   // type fields
   do {
     luna_node_t *decl = decl_expr(self, true);
     if (!decl) return error("expecting field");
 
+    // semicolon might have been inserted here
+    accept(SEMICOLON);
+
     luna_vec_push(type->fields, luna_node(decl));
   } while (!accept(END));
 
@@ -1013,6 +1019,9 @@ if_stmt(luna_parser_t *self) {
     return NULL;
   }
 
+  // semicolon might have been inserted here
+  accept(SEMICOLON);
+
   // block
   context("if statement");
   if (!(body = block(self))) {
@@ -1030,6 +1039,10 @@ if_stmt(luna_parser_t *self) {
     if (accept(IF)) {
       context("else if statement condition");
       if (!(cond = expr(self))) return NULL;
+
+      // semicolon might have been inserted here
+      accept(SEMICOLON);
+
       context("else if statement");
       if (!(body = block(self))) return NULL;
       luna_vec_push(node->else_ifs, luna_node((luna_node_t *) luna_if_node_new(0, cond, body)));
@@ -1066,6 +1079,9 @@ while_stmt(luna_parser_t *self) {
   if (!(cond = expr(self))) return NULL;
   context("while statement");
 
+  // semicolon might have been inserted here
+  accept(SEMICOLON);
+
   // block
   if (!(body = block(self))) return NULL;
 
@@ -1126,6 +1142,8 @@ block(luna_parser_t *self) {
 
   do {
     if (!(node = stmt(self))) return NULL;
+    accept(SEMICOLON);
+
     luna_vec_push(block->stmts, luna_node(node));
   } while (!accept(END) && !is(ELSE));
 
@@ -1145,6 +1163,7 @@ program(luna_parser_t *self) {
   next;
   while (!is(EOS)) {
     if (node = stmt(self)) {
+      accept(SEMICOLON);
       luna_vec_push(block->stmts, luna_node(node));
     } else {
       return NULL;

src/prettyprint.c

@@ -212,9 +212,16 @@ visit_string(luna_visitor_t *self, luna_string_node_t *node) {
 
 static void
 visit_unary_op(luna_visitor_t *self, luna_unary_op_node_t *node) {
-  int c = node->postfix ? '@' : 0;
-  print_func("(%c%s ", c, luna_token_type_string(node->op));
-  visit(node->expr);
+  print_func("(");
+
+  if (node->postfix) {
+    visit(node->expr);
+    print_func(" %s", luna_token_type_string(node->op));
+  } else {
+    print_func("%s ", luna_token_type_string(node->op));
+    visit(node->expr);
+  }
+
   print_func(")");
 }
 

src/utils.c

@@ -41,7 +41,7 @@ file_read(const char *filename) {
   size_t read = fread(buf, sizeof(char), len, fh);
 
   fclose(fh);
-  buf[len] = 0;
+  buf[read] = 0;
   return buf;
 }