Some rearranging in perparation for MBE-style TT transcription.

Author: paulstansifer

src/libsyntax/ast.rs

@@ -374,12 +374,17 @@ enum blk_sort {
 */
 
 #[auto_serialize]
+#[doc="For macro invocations; parsing is delegated to the macro"]
 enum token_tree {
-    /* for macro invocations; parsing is the macro's job */
     tt_delim(~[token_tree]),
-    tt_flat(span, token::token)
+    tt_flat(span, token::token),
+    /* These only make sense for right-hand-sides of MBE macros*/
+    tt_dotdotdot(~[token_tree]),
+    tt_interpolate(ident)
 }
 
+
+
 #[auto_serialize]
 type matcher = spanned<matcher_>;
 

src/libsyntax/ext/earley_parser.rs

@@ -1,12 +1,15 @@
 // Earley-like parser for macros.
 import parse::token;
 import parse::token::{token, EOF, to_str, whole_nt};
-import parse::lexer::{reader, tt_reader, tt_reader_as_reader};
+import parse::lexer::*; //resolve bug?
+//import parse::lexer::{reader, tt_reader, tt_reader_as_reader};
 import parse::parser::{parser,SOURCE_FILE};
-import parse::common::parser_common;
+//import parse::common::parser_common;
+import parse::common::*; //resolve bug?
 import parse::parse_sess;
 import dvec::{dvec, extensions};
-import ast::{matcher, mtc_tok, mtc_rep, mtc_bb};
+import ast::{matcher, mtc_tok, mtc_rep, mtc_bb, ident};
+import std::map::{hashmap, box_str_hash};
 
 /* This is an Earley-like parser, without support for nonterminals.  This
 means that there are no completer or predictor rules, and therefore no need to
@@ -66,8 +69,31 @@ enum arb_depth { leaf(whole_nt), seq(~[@arb_depth]) }
 type earley_item = matcher_pos;
 
 
+fn nameize(&&p_s: parse_sess, ms: ~[matcher], &&res: ~[@arb_depth])
+    -> hashmap<ident,@arb_depth> {
+    fn n_rec(&&p_s: parse_sess, &&m: matcher, &&res: ~[@arb_depth],
+             &&ret_val: hashmap<ident, @arb_depth>) {
+        alt m {
+          {node: mtc_tok(_), span: _} { }
+          {node: mtc_rep(more_ms, _, _), span: _} {
+            for more_ms.each() |next_m| { n_rec(p_s, next_m, res, ret_val) };
+          }
+          {node: mtc_bb(bind_name, _, idx), span: sp} {
+            if ret_val.contains_key(bind_name) {
+                p_s.span_diagnostic.span_fatal(sp, "Duplicated bind name: "
+                                               + *bind_name)
+            }
+            ret_val.insert(bind_name, res[idx]);
+          }
+        }
+    }
+    let ret_val = box_str_hash::<@arb_depth>();
+    for ms.each() |m| { n_rec(p_s, m, res, ret_val) };
+    ret ret_val;
+}
+
 fn parse(sess: parse_sess, cfg: ast::crate_cfg, rdr: reader, ms: ~[matcher])
-    -> ~[@arb_depth] {
+    -> hashmap<ident,@arb_depth> {
     let mut cur_eis = ~[];
     vec::push(cur_eis, new_matcher_pos(ms, none));
 
@@ -164,9 +190,9 @@ fn parse(sess: parse_sess, cfg: ast::crate_cfg, rdr: reader, ms: ~[matcher])
 
         /* error messages here could be improved with links to orig. rules */
         if tok == EOF {
-            if eof_eis.len() == 1u {
-                let ret_val = vec::map(eof_eis[0u].matches, |dv| dv.pop());
-                ret ret_val; /* success */
+            if eof_eis.len() == 1u { /* success */
+                ret nameize(sess, ms,
+                            vec::map(eof_eis[0u].matches, |dv| dv.pop()));
             } else if eof_eis.len() > 1u {
                 rdr.fatal("Ambiguity: multiple successful parses");
             } else {

src/libsyntax/ext/tt/transcribe.rs

@@ -0,0 +1,114 @@
+import util::interner::interner;
+import diagnostic::span_handler;
+import ast::{tt_delim,tt_flat,tt_dotdotdot,tt_interpolate,ident};
+import ext::earley_parser::arb_depth;
+import codemap::span;
+import parse::token::{EOF,token};
+
+export tt_reader,  new_tt_reader, dup_tt_reader, tt_next_token;
+
+enum tt_frame_up { /* to break a circularity */
+    tt_frame_up(option<tt_frame>)
+}
+
+/* TODO: figure out how to have a uniquely linked stack, and change to `~` */
+///an unzipping of `token_tree`s
+type tt_frame = @{
+    readme: [ast::token_tree]/~,
+    mut idx: uint,
+    up: tt_frame_up
+};
+
+type tt_reader = @{
+    span_diagnostic: span_handler,
+    interner: @interner<@str>,
+    mut cur: tt_frame,
+    /* for MBE-style macro transcription */
+    interpolations: std::map::hashmap<ident, @arb_depth>,
+    /* cached: */
+    mut cur_tok: token,
+    mut cur_span: span
+};
+
+/** This can do Macro-By-Example transcription. On the other hand, if
+ *  `doc` contains no `tt_dotdotdot`s and `tt_interpolate`s, `interp` can (and
+ *  should) be none. */
+fn new_tt_reader(span_diagnostic: span_handler, itr: @interner<@str>,
+                 interp: option<std::map::hashmap<ident,@arb_depth>>,
+                 src: [ast::token_tree]/~)
+    -> tt_reader {
+    let r = @{span_diagnostic: span_diagnostic, interner: itr,
+              mut cur: @{readme: src, mut idx: 0u,
+                         up: tt_frame_up(option::none)},
+              interpolations: alt interp { /* just a convienience */
+                none { std::map::box_str_hash::<@arb_depth>() }
+                some(x) { x }
+              },
+              /* dummy values, never read: */
+              mut cur_tok: EOF,
+              mut cur_span: ast_util::mk_sp(0u,0u)
+             };
+    tt_next_token(r); /* get cur_tok and cur_span set up */
+    ret r;
+}
+
+pure fn dup_tt_frame(&&f: tt_frame) -> tt_frame {
+    @{readme: f.readme, mut idx: f.idx,
+      up: alt f.up {
+        tt_frame_up(some(up_frame)) {
+          tt_frame_up(some(dup_tt_frame(up_frame)))
+        }
+        tt_frame_up(none) { tt_frame_up(none) }
+      }
+     }
+}
+
+pure fn dup_tt_reader(&&r: tt_reader) -> tt_reader {
+    @{span_diagnostic: r.span_diagnostic, interner: r.interner,
+      mut cur: dup_tt_frame(r.cur),
+      interpolations: r.interpolations,
+      mut cur_tok: r.cur_tok, mut cur_span: r.cur_span}
+}
+
+
+fn tt_next_token(&&r: tt_reader) -> {tok: token, sp: span} {
+    let ret_val = { tok: r.cur_tok, sp: r.cur_span };
+    if r.cur.idx >= vec::len(r.cur.readme) {
+        /* done with this set; pop */
+        alt r.cur.up {
+          tt_frame_up(none) {
+            r.cur_tok = EOF;
+            ret ret_val;
+          }
+          tt_frame_up(some(tt_f)) {
+            r.cur = tt_f;
+            /* the above `if` would need to be a `while` if we didn't know
+            that the last thing in a `tt_delim` is always a `tt_flat` */
+            r.cur.idx += 1u;
+          }
+        }
+    }
+    /* if `tt_delim`s could be 0-length, we'd need to be able to switch
+    between popping and pushing until we got to an actual `tt_flat` */
+    loop { /* because it's easiest, this handles `tt_delim` not starting
+    with a `tt_flat`, even though it won't happen */
+        alt copy r.cur.readme[r.cur.idx] {
+          tt_delim(tts) {
+            r.cur = @{readme: tts, mut idx: 0u,
+                      up: tt_frame_up(option::some(r.cur)) };
+          }
+          tt_flat(sp, tok) {
+            r.cur_span = sp; r.cur_tok = tok;
+            r.cur.idx += 1u;
+            ret ret_val;
+          }
+          tt_dotdotdot(tts) {
+            fail;
+          }
+          tt_interpolate(ident) {
+            fail;
+          }
+        }
+    }
+
+}

src/libsyntax/parse.rs

@@ -13,8 +13,10 @@ export parse_expr_from_source_str, parse_item_from_source_str;
 export parse_from_source_str;
 
 import parser::parser;
-import attr::parser_attr;
-import common::parser_common;
+//import attr::parser_attr;
+import attr::*; //resolve bug?
+//import common::parser_common;
+import common::*; //resolve bug?
 import ast::node_id;
 import util::interner;
 // FIXME (#1935): resolve badness
@@ -199,6 +201,7 @@ fn new_parser_from_file(sess: parse_sess, cfg: ast::crate_cfg, +path: str,
 
 fn new_parser_from_tt(sess: parse_sess, cfg: ast::crate_cfg,
                       tt: ~[ast::token_tree]) -> parser {
-    let trdr = lexer::new_tt_reader(sess.span_diagnostic, sess.interner, tt);
+    let trdr = lexer::new_tt_reader(sess.span_diagnostic, sess.interner,
+                                    none, tt);
     ret parser(sess, cfg, trdr as reader, parser::SOURCE_FILE)
 }

src/libsyntax/parse/attr.rs

@@ -1,6 +1,7 @@
 import either::{either, left, right};
 import ast_util::spanned;
-import common::{parser_common, seq_sep_trailing_disallowed};
+import common::*; //resolve bug?
+//import common::{parser_common, seq_sep_trailing_disallowed};
 
 export attr_or_ext;
 export parser_attr;