Make token trees parseable.

Author: paulstansifer

src/libsyntax/ast.rs

@@ -17,6 +17,8 @@ import std::serialization::{serializer,
                             deserialize_str,
                             serialize_bool,
                             deserialize_bool};
+import parse::token;
+
 
 /* Note #1972 -- spans are serialized but not deserialized */
 fn serialize_span<S>(_s: S, _v: span) {
@@ -371,6 +373,16 @@ enum blk_sort {
 }
 */
 
+#[auto_serialize]
+type token_tree = spanned<token_tree_>;
+
+#[auto_serialize]
+enum token_tree_ {
+    /* for macro invocations; parsing is the macro's job */
+    tt_delim(token::token, [token_tree]),
+    tt_flat(token::token)
+}
+
 #[auto_serialize]
 type mac = spanned<mac_>;
 
@@ -386,6 +398,7 @@ type mac_body = option<mac_body_>;
 #[auto_serialize]
 enum mac_ {
     mac_invoc(@path, mac_arg, mac_body),
+    mac_invoc_tt(@path, token_tree), //will kill mac_invoc and steal its name
     mac_embed_type(@ty),
     mac_embed_block(blk),
     mac_ellipsis,

src/libsyntax/ext/simplext.rs

@@ -549,6 +549,7 @@ fn p_t_s_r_mac(cx: ext_ctxt, mac: ast::mac, s: selector, b: binders) {
     alt mac.node {
       ast::mac_ellipsis { cx.span_fatal(mac.span, "misused `...`"); }
       ast::mac_invoc(_, _, _) { no_des(cx, mac.span, "macro calls"); }
+      ast::mac_invoc_tt(_, _) { no_des(cx, mac.span, "macro calls"); }
       ast::mac_embed_type(ty) {
         alt ty.node {
           ast::ty_path(pth, _) {

src/libsyntax/fold.rs

@@ -119,6 +119,7 @@ fn fold_mac_(m: mac, fld: ast_fold) -> mac {
                  mac_invoc(fld.fold_path(pth),
                            option::map(arg, fld.fold_expr), body)
                }
+               mac_invoc_tt(pth, tt) { m.node }
                mac_embed_type(ty) { mac_embed_type(fld.fold_ty(ty)) }
                mac_embed_block(blk) { mac_embed_block(fld.fold_block(blk)) }
                mac_ellipsis { mac_ellipsis }

src/libsyntax/parse/attr.rs

@@ -24,6 +24,7 @@ impl parser_attr for parser {
                 ret some(left([first_attr] + self.parse_outer_attributes()));
             } else if !(self.look_ahead(1u) == token::LT
                         || self.look_ahead(1u) == token::LBRACKET
+                        || self.look_ahead(1u) == token::POUND
                         || expect_item_next) {
                 self.bump();
                 ret some(right(self.parse_syntax_ext_naked(lo)));

src/libsyntax/parse/parser.rs

@@ -4,7 +4,7 @@ import std::map::{hashmap, str_hash};
 import token::{can_begin_expr, is_ident, is_plain_ident};
 import codemap::{span,fss_none};
 import util::interner;
-import ast_util::{spanned, mk_sp, ident_to_path, operator_prec};
+import ast_util::{spanned, respan, mk_sp, ident_to_path, operator_prec};
 import ast::*;
 import lexer::reader;
 import prec::{as_prec, token_to_binop};
@@ -801,10 +801,16 @@ class parser {
                 {|p| p.parse_expr()});
             hi = self.span.hi;
             ex = expr_vec(es, mutbl);
+        } else if self.token == token::POUND
+            && self.look_ahead(1u) == token::POUND {
+            self.bump(); self.bump();
+            let macname = self.parse_path_without_tps();
+            let macbody = self.parse_token_tree();
+            ret pexpr(self.mk_mac_expr(lo, self.span.hi,
+                                       mac_invoc_tt(macname, macbody)));
         } else if self.token == token::POUND
             && self.look_ahead(1u) == token::LT {
-            self.bump();
-            self.bump();
+            self.bump(); self.bump();
             let ty = self.parse_ty(false);
             self.expect(token::GT);
 
@@ -813,8 +819,7 @@ class parser {
                                        mac_embed_type(ty)));
         } else if self.token == token::POUND
             && self.look_ahead(1u) == token::LBRACE {
-            self.bump();
-            self.bump();
+            self.bump(); self.bump();
             let blk = mac_embed_block(
                 self.parse_block_tail(lo, default_blk));
             ret pexpr(self.mk_mac_expr(lo, self.span.hi, blk));
@@ -1053,6 +1058,47 @@ class parser {
     ret e;
 }
 
+    fn parse_token_tree() -> token_tree {
+        #[doc="what's the opposite delimiter?"]
+        fn flip(t: token::token) -> token::token {
+            alt t {
+              token::LPAREN { token::RPAREN }
+              token::LBRACE { token::RBRACE }
+              token::LBRACKET { token::RBRACKET }
+              _ { fail }
+            }
+        }
+
+        fn parse_tt_flat(p: parser, delim_ok: bool) -> token_tree {
+            alt p.token {
+              token::RPAREN | token::RBRACE | token::RBRACKET
+              if !delim_ok {
+                p.fatal("incorrect close delimiter: `"
+                           + token_to_str(p.reader, p.token) + "`");
+              }
+              token::EOF {
+                p.fatal("file ended in the middle of a macro invocation");
+              }
+              _ { /* ok */ }
+            }
+            let res = tt_flat(p.span.lo, p.token);
+            p.bump();
+            ret res;
+        }
+
+        ret alt self.token {
+          token::LPAREN | token::LBRACE | token::LBRACKET {
+            let ket = flip(self.token);
+            tt_delim([parse_tt_flat(self, true)] +
+                     self.parse_seq_to_before_end(ket, seq_sep_none(),
+                                                  {|p| p.parse_token_tree()})
+                     + [parse_tt_flat(self, true)])
+          }
+          _ { parse_tt_flat(self, false) }
+        };
+    }
+
+
     fn parse_prefix_expr() -> pexpr {
         let lo = self.span.lo;
         let mut hi;

src/libsyntax/parse/token.rs

@@ -1,9 +1,21 @@
 import util::interner;
 import util::interner::interner;
 import std::map::{hashmap, str_hash};
+import std::serialization::{serializer,
+                            deserializer,
+                            serialize_uint,
+                            deserialize_uint,
+                            serialize_i64,
+                            deserialize_i64,
+                            serialize_u64,
+                            deserialize_u64,
+                            serialize_bool,
+                            deserialize_bool};
 
+#[auto_serialize]
 type str_num = uint;
 
+#[auto_serialize]
 enum binop {
     PLUS,
     MINUS,
@@ -17,6 +29,7 @@ enum binop {
     SHR,
 }
 
+#[auto_serialize]
 enum token {
     /* Expression-operator symbols. */
     EQ,

src/libsyntax/syntax.rc

@@ -30,6 +30,7 @@ mod util {
 mod parse {
     export parser;
     export lexer;
+    export token;
     export comments;
     export prec;
     export classify;

src/libsyntax/visit.rs

@@ -351,6 +351,7 @@ fn visit_mac<E>(m: mac, e: E, v: vt<E>) {
     alt m.node {
       ast::mac_invoc(pth, arg, body) {
         option::map(arg) {|arg| v.visit_expr(arg, e, v)}; }
+      ast::mac_invoc_tt(pth, tt) { /* no user-serviceable parts inside */ }
       ast::mac_embed_type(ty) { v.visit_ty(ty, e, v); }
       ast::mac_embed_block(blk) { v.visit_block(blk, e, v); }
       ast::mac_ellipsis { }