turn ast::ident into a struct

This will help with the auto_serialize2 migration. We have to change
ident from a type alias to uint into a unique type. We need to use
a struct instead of a "enum ident = token::str_num" because structs
support constants, but newtypes do not.

Author: erickt

src/fuzzer/fuzzer.rs

@@ -243,7 +243,7 @@ fn check_variants_T<T: Copy>(
   filename: &Path,
   thing_label: ~str,
   things: ~[T],
-  stringifier: fn@(@T, syntax::parse::token::ident_interner) -> ~str,
+  stringifier: fn@(@T, @syntax::parse::token::ident_interner) -> ~str,
   replacer: fn@(ast::crate, uint, T, test_mode) -> ast::crate,
   cx: context
   ) {

src/libsyntax/ast.rs

@@ -38,6 +38,10 @@ macro_rules! interner_key (
         (-3 as uint, 0u)))
 )
 
+// FIXME(#3534): Replace with the struct-based newtype when it's been
+// implemented.
+struct ident { repr: uint }
+
 fn serialize_ident<S: Serializer>(s: S, i: ident) {
     let intr = match unsafe{
         task::local_data::local_data_get(interner_key!())
@@ -59,7 +63,16 @@ fn deserialize_ident<D: Deserializer>(d: D) -> ident  {
     (*intr).intern(@d.read_str())
 }
 
-type ident = token::str_num;
+impl ident: cmp::Eq {
+    pure fn eq(other: &ident) -> bool { self.repr == other.repr }
+    pure fn ne(other: &ident) -> bool { !self.eq(other) }
+}
+
+impl ident: to_bytes::IterBytes {
+    pure fn iter_bytes(lsb0: bool, f: to_bytes::Cb) {
+        self.repr.iter_bytes(lsb0, f)
+    }
+}
 
 // Functions may or may not have names.
 #[auto_serialize]

src/libsyntax/ast_map.rs

@@ -34,7 +34,7 @@ impl path_elt : cmp::Eq {
 type path = ~[path_elt];
 
 /* FIXMEs that say "bad" are as per #2543 */
-fn path_to_str_with_sep(p: path, sep: ~str, itr: ident_interner) -> ~str {
+fn path_to_str_with_sep(p: path, sep: ~str, itr: @ident_interner) -> ~str {
     let strs = do vec::map(p) |e| {
         match *e {
           path_mod(s) => *itr.get(s),
@@ -44,7 +44,7 @@ fn path_to_str_with_sep(p: path, sep: ~str, itr: ident_interner) -> ~str {
     str::connect(strs, sep)
 }
 
-fn path_ident_to_str(p: path, i: ident, itr: ident_interner) -> ~str {
+fn path_ident_to_str(p: path, i: ident, itr: @ident_interner) -> ~str {
     if vec::is_empty(p) {
         //FIXME /* FIXME (#2543) */ copy *i
         *itr.get(i)
@@ -53,7 +53,7 @@ fn path_ident_to_str(p: path, i: ident, itr: ident_interner) -> ~str {
     }
 }
 
-fn path_to_str(p: path, itr: ident_interner) -> ~str {
+fn path_to_str(p: path, itr: @ident_interner) -> ~str {
     path_to_str_with_sep(p, ~"::", itr)
 }
 
@@ -326,7 +326,7 @@ fn map_stmt(stmt: @stmt, cx: ctx, v: vt) {
     visit::visit_stmt(stmt, cx, v);
 }
 
-fn node_id_to_str(map: map, id: node_id, itr: ident_interner) -> ~str {
+fn node_id_to_str(map: map, id: node_id, itr: @ident_interner) -> ~str {
     match map.find(id) {
       None => {
         fmt!("unknown node (id=%d)", id)

src/libsyntax/ast_util.rs

@@ -23,7 +23,7 @@ pure fn dummy_sp() -> span { return mk_sp(0u, 0u); }
 
 
 
-pure fn path_name_i(idents: ~[ident], intr: token::ident_interner) -> ~str {
+pure fn path_name_i(idents: ~[ident], intr: @token::ident_interner) -> ~str {
     // FIXME: Bad copies (#2543 -- same for everything else that says "bad")
     str::connect(idents.map(|i| *intr.get(*i)), ~"::")
 }

src/libsyntax/ext/tt/transcribe.rs

@@ -25,7 +25,7 @@ type tt_frame = @{
 
 type tt_reader = @{
     sp_diag: span_handler,
-    interner: ident_interner,
+    interner: @ident_interner,
     mut cur: tt_frame,
     /* for MBE-style macro transcription */
     interpolations: std::map::HashMap<ident, @named_match>,
@@ -39,7 +39,7 @@ type tt_reader = @{
 /** This can do Macro-By-Example transcription. On the other hand, if
  *  `src` contains no `tt_seq`s and `tt_nonterminal`s, `interp` can (and
  *  should) be none. */
-fn new_tt_reader(sp_diag: span_handler, itr: ident_interner,
+fn new_tt_reader(sp_diag: span_handler, itr: @ident_interner,
                  interp: Option<std::map::HashMap<ident,@named_match>>,
                  src: ~[ast::token_tree])
     -> tt_reader {

src/libsyntax/parse.rs

@@ -25,7 +25,7 @@ type parse_sess = @{
     cm: codemap::codemap,
     mut next_id: node_id,
     span_diagnostic: span_handler,
-    interner: ident_interner,
+    interner: @ident_interner,
     // these two must be kept up to date
     mut chpos: uint,
     mut byte_pos: uint

src/libsyntax/parse/lexer.rs

@@ -13,7 +13,7 @@ trait reader {
     fn next_token() -> {tok: token::token, sp: span};
     fn fatal(~str) -> !;
     fn span_diag() -> span_handler;
-    pure fn interner() -> token::ident_interner;
+    pure fn interner() -> @token::ident_interner;
     fn peek() -> {tok: token::token, sp: span};
     fn dup() -> reader;
 }
@@ -26,15 +26,15 @@ type string_reader = @{
     mut curr: char,
     mut chpos: uint,
     filemap: codemap::filemap,
-    interner: token::ident_interner,
+    interner: @token::ident_interner,
     /* cached: */
     mut peek_tok: token::token,
     mut peek_span: span
 };
 
 fn new_string_reader(span_diagnostic: span_handler,
                      filemap: codemap::filemap,
-                     itr: token::ident_interner) -> string_reader {
+                     itr: @token::ident_interner) -> string_reader {
     let r = new_low_level_string_reader(span_diagnostic, filemap, itr);
     string_advance_token(r); /* fill in peek_* */
     return r;
@@ -43,7 +43,7 @@ fn new_string_reader(span_diagnostic: span_handler,
 /* For comments.rs, which hackily pokes into 'pos' and 'curr' */
 fn new_low_level_string_reader(span_diagnostic: span_handler,
                                filemap: codemap::filemap,
-                               itr: token::ident_interner)
+                               itr: @token::ident_interner)
     -> string_reader {
     let r = @{span_diagnostic: span_diagnostic, src: filemap.src,
               mut col: 0u, mut pos: 0u, mut curr: -1 as char,
@@ -78,7 +78,7 @@ impl string_reader: reader {
         self.span_diagnostic.span_fatal(copy self.peek_span, m)
     }
     fn span_diag() -> span_handler { self.span_diagnostic }
-    pure fn interner() -> token::ident_interner { self.interner }
+    pure fn interner() -> @token::ident_interner { self.interner }
     fn peek() -> {tok: token::token, sp: span} {
         {tok: self.peek_tok, sp: self.peek_span}
     }
@@ -100,7 +100,7 @@ impl tt_reader: reader {
         self.sp_diag.span_fatal(copy self.cur_span, m);
     }
     fn span_diag() -> span_handler { self.sp_diag }
-    pure fn interner() -> token::ident_interner { self.interner }
+    pure fn interner() -> @token::ident_interner { self.interner }
     fn peek() -> {tok: token::token, sp: span} {
         { tok: self.cur_tok, sp: self.cur_span }
     }

src/libsyntax/parse/parser.rs

@@ -237,7 +237,7 @@ struct parser {
     mut restriction: restriction,
     mut quote_depth: uint, // not (yet) related to the quasiquoter
     reader: reader,
-    interner: interner<@~str>,
+    interner: @token::ident_interner,
     keywords: HashMap<~str, ()>,
     strict_keywords: HashMap<~str, ()>,
     reserved_keywords: HashMap<~str, ()>,

src/libsyntax/parse/token.rs

@@ -12,9 +12,6 @@ use std::serialization::{Serializer,
                             serialize_bool,
                             deserialize_bool};
 
-#[auto_serialize]
-type str_num = uint;
-
 #[auto_serialize]
 enum binop {
     PLUS,
@@ -72,17 +69,17 @@ enum token {
     LIT_INT(i64, ast::int_ty),
     LIT_UINT(u64, ast::uint_ty),
     LIT_INT_UNSUFFIXED(i64),
-    LIT_FLOAT(str_num, ast::float_ty),
-    LIT_STR(str_num),
+    LIT_FLOAT(ast::ident, ast::float_ty),
+    LIT_STR(ast::ident),
 
     /* Name components */
-    IDENT(str_num, bool),
+    IDENT(ast::ident, bool),
     UNDERSCORE,
 
     /* For interpolation */
     INTERPOLATED(nonterminal),
 
-    DOC_COMMENT(str_num),
+    DOC_COMMENT(ast::ident),
     EOF,
 }
 
@@ -95,7 +92,7 @@ enum nonterminal {
     nt_pat( @ast::pat),
     nt_expr(@ast::expr),
     nt_ty(  @ast::ty),
-    nt_ident(str_num, bool),
+    nt_ident(ast::ident, bool),
     nt_path(@ast::path),
     nt_tt(  @ast::token_tree), //needs @ed to break a circularity
     nt_matchers(~[ast::matcher])
@@ -116,7 +113,7 @@ fn binop_to_str(o: binop) -> ~str {
     }
 }
 
-fn to_str(in: interner<@~str>, t: token) -> ~str {
+fn to_str(in: @ident_interner, t: token) -> ~str {
     match t {
       EQ => ~"=",
       LT => ~"<",
@@ -174,7 +171,7 @@ fn to_str(in: interner<@~str>, t: token) -> ~str {
         }
         body + ast_util::float_ty_to_str(t)
       }
-      LIT_STR(s) => { ~"\"" + str::escape_default( *in.get(s)) + ~"\"" }
+      LIT_STR(s) => { ~"\"" + str::escape_default(*in.get(s)) + ~"\"" }
 
       /* Name components */
       IDENT(s, _) => *in.get(s),
@@ -281,49 +278,66 @@ pure fn is_bar(t: token) -> bool {
 mod special_idents {
     #[legacy_exports];
     use ast::ident;
-    const underscore : ident = 0u;
-    const anon : ident = 1u;
-    const dtor : ident = 2u; // 'drop', but that's reserved
-    const invalid : ident = 3u; // ''
-    const unary : ident = 4u;
-    const not_fn : ident = 5u;
-    const idx_fn : ident = 6u;
-    const unary_minus_fn : ident = 7u;
-    const clownshoes_extensions : ident = 8u;
-
-    const self_ : ident = 9u; // 'self'
+    const underscore : ident = ident { repr: 0u };
+    const anon : ident = ident { repr: 1u };
+    const dtor : ident = ident { repr: 2u }; // 'drop', but that's reserved
+    const invalid : ident = ident { repr: 3u }; // ''
+    const unary : ident = ident { repr: 4u };
+    const not_fn : ident = ident { repr: 5u };
+    const idx_fn : ident = ident { repr: 6u };
+    const unary_minus_fn : ident = ident { repr: 7u };
+    const clownshoes_extensions : ident = ident { repr: 8u };
+
+    const self_ : ident = ident { repr: 9u }; // 'self'
 
     /* for matcher NTs */
-    const item : ident = 10u;
-    const block : ident = 11u;
-    const stmt : ident = 12u;
-    const pat : ident = 13u;
-    const expr : ident = 14u;
-    const ty : ident = 15u;
-    const ident : ident = 16u;
-    const path : ident = 17u;
-    const tt : ident = 18u;
-    const matchers : ident = 19u;
-
-    const str : ident = 20u; // for the type
+    const item : ident = ident { repr: 10u };
+    const block : ident = ident { repr: 11u };
+    const stmt : ident = ident { repr: 12u };
+    const pat : ident = ident { repr: 13u };
+    const expr : ident = ident { repr: 14u };
+    const ty : ident = ident { repr: 15u };
+    const ident : ident = ident { repr: 16u };
+    const path : ident = ident { repr: 17u };
+    const tt : ident = ident { repr: 18u };
+    const matchers : ident = ident { repr: 19u };
+
+    const str : ident = ident { repr: 20u }; // for the type
 
     /* outside of libsyntax */
-    const ty_visitor : ident = 21u;
-    const arg : ident = 22u;
-    const descrim : ident = 23u;
-    const clownshoe_abi : ident = 24u;
-    const clownshoe_stack_shim : ident = 25u;
-    const tydesc : ident = 26u;
-    const literally_dtor : ident = 27u;
-    const main : ident = 28u;
-    const opaque : ident = 29u;
-    const blk : ident = 30u;
-    const static : ident = 31u;
-    const intrinsic : ident = 32u;
-    const clownshoes_foreign_mod: ident = 33;
+    const ty_visitor : ident = ident { repr: 21u };
+    const arg : ident = ident { repr: 22u };
+    const descrim : ident = ident { repr: 23u };
+    const clownshoe_abi : ident = ident { repr: 24u };
+    const clownshoe_stack_shim : ident = ident { repr: 25u };
+    const tydesc : ident = ident { repr: 26u };
+    const literally_dtor : ident = ident { repr: 27u };
+    const main : ident = ident { repr: 28u };
+    const opaque : ident = ident { repr: 29u };
+    const blk : ident = ident { repr: 30u };
+    const static : ident = ident { repr: 31u };
+    const intrinsic : ident = ident { repr: 32u };
+    const clownshoes_foreign_mod: ident = ident { repr: 33 };
 }
 
-type ident_interner = util::interner::interner<@~str>;
+struct ident_interner {
+    priv interner: util::interner::interner<@~str>,
+}
+
+impl ident_interner {
+    fn intern(val: @~str) -> ast::ident {
+        ast::ident { repr: self.interner.intern(val) }
+    }
+    fn gensym(val: @~str) -> ast::ident {
+        ast::ident { repr: self.interner.gensym(val) }
+    }
+    pure fn get(idx: ast::ident) -> @~str {
+        self.interner.get(idx.repr)
+    }
+    fn len() -> uint {
+        self.interner.len()
+    }
+}
 
 /** Key for thread-local data for sneaking interner information to the
  * serializer/deserializer. It sounds like a hack because it is one.
@@ -335,7 +349,7 @@ macro_rules! interner_key (
         (-3 as uint, 0u)))
 )
 
-fn mk_ident_interner() -> ident_interner {
+fn mk_ident_interner() -> @ident_interner {
     /* the indices here must correspond to the numbers in special_idents */
     let init_vec = ~[@~"_", @~"anon", @~"drop", @~"", @~"unary", @~"!",
                      @~"[]", @~"unary-", @~"__extensions__", @~"self",
@@ -346,7 +360,9 @@ fn mk_ident_interner() -> ident_interner {
                      @~"dtor", @~"main", @~"<opaque>", @~"blk", @~"static",
                      @~"intrinsic", @~"__foreign_mod__"];
 
-    let rv = interner::mk_prefill::<@~str>(init_vec);
+    let rv = @ident_interner {
+        interner: interner::mk_prefill::<@~str>(init_vec)
+    };
 
     /* having multiple interners will just confuse the serializer */
     unsafe {
@@ -360,8 +376,8 @@ fn mk_ident_interner() -> ident_interner {
 
 /* for when we don't care about the contents; doesn't interact with TLD or
    serialization */
-fn mk_fake_ident_interner() -> ident_interner {
-    interner::mk::<@~str>()
+fn mk_fake_ident_interner() -> @ident_interner {
+    @ident_interner { interner: interner::mk::<@~str>() }
 }
 
 /**