Parse and store suffixes on literals.

This adds an optional suffix at the end of a literal token:
`"foo"bar`. An actual use of a suffix in a expression (or other literal
that the compiler reads) is rejected in the parser.

This doesn't switch the handling of numbers to this system, and doesn't
outlaw illegal suffixes for them yet.

3 files changed, 179 insertions, 61 deletions

diff --git a/src/libsyntax/parse/lexer/mod.rs b/src/libsyntax/parse/lexer/mod.rs
index b7598c7c428..55c4335941d 100644
--- a/src/libsyntax/parse/lexer/mod.rs
+++ b/src/libsyntax/parse/lexer/mod.rs
@@ -369,6 +369,25 @@ impl<'a> StringReader<'a> {
         self.nextnextch() == Some(c)
     }
 
+    /// Eats <XID_start><XID_continue>*, if possible.
+    fn scan_optional_raw_name(&mut self) -> Option<ast::Name> {
+        if !ident_start(self.curr) {
+            return None
+        }
+        let start = self.last_pos;
+        while ident_continue(self.curr) {
+            self.bump();
+        }
+
+        self.with_str_from(start, |string| {
+            if string == "_" {
+                None
+            } else {
+                Some(token::intern(string))
+            }
+        })
+    }
+
     /// PRECONDITION: self.curr is not whitespace
     /// Eats any kind of comment.
     fn scan_comment(&mut self) -> Option<TokenAndSpan> {
@@ -638,7 +657,7 @@ impl<'a> StringReader<'a> {
     }
 
     /// Lex a LIT_INTEGER or a LIT_FLOAT
-    fn scan_number(&mut self, c: char) -> token::Token {
+    fn scan_number(&mut self, c: char) -> token::Lit {
         let mut num_digits;
         let mut base = 10;
         let start_bpos = self.last_pos;
@@ -655,17 +674,17 @@ impl<'a> StringReader<'a> {
                 }
                 'u' | 'i' => {
                     self.scan_int_suffix();
-                    return token::Literal(token::Integer(self.name_from(start_bpos)));
+                    return token::Integer(self.name_from(start_bpos));
                 },
                 'f' => {
                     let last_pos = self.last_pos;
                     self.scan_float_suffix();
                     self.check_float_base(start_bpos, last_pos, base);
-                    return token::Literal(token::Float(self.name_from(start_bpos)));
+                    return token::Float(self.name_from(start_bpos));
                 }
                 _ => {
                     // just a 0
-                    return token::Literal(token::Integer(self.name_from(start_bpos)));
+                    return token::Integer(self.name_from(start_bpos));
                 }
             }
         } else if c.is_digit_radix(10) {
@@ -678,7 +697,7 @@ impl<'a> StringReader<'a> {
             self.err_span_(start_bpos, self.last_pos, "no valid digits found for number");
             // eat any suffix
             self.scan_int_suffix();
-            return token::Literal(token::Integer(token::intern("0")));
+            return token::Integer(token::intern("0"));
         }
 
         // might be a float, but don't be greedy if this is actually an
@@ -696,13 +715,13 @@ impl<'a> StringReader<'a> {
             }
             let last_pos = self.last_pos;
             self.check_float_base(start_bpos, last_pos, base);
-            return token::Literal(token::Float(self.name_from(start_bpos)));
+            return token::Float(self.name_from(start_bpos));
         } else if self.curr_is('f') {
             // or it might be an integer literal suffixed as a float
             self.scan_float_suffix();
             let last_pos = self.last_pos;
             self.check_float_base(start_bpos, last_pos, base);
-            return token::Literal(token::Float(self.name_from(start_bpos)));
+            return token::Float(self.name_from(start_bpos));
         } else {
             // it might be a float if it has an exponent
             if self.curr_is('e') || self.curr_is('E') {
@@ -710,11 +729,11 @@ impl<'a> StringReader<'a> {
                 self.scan_float_suffix();
                 let last_pos = self.last_pos;
                 self.check_float_base(start_bpos, last_pos, base);
-                return token::Literal(token::Float(self.name_from(start_bpos)));
+                return token::Float(self.name_from(start_bpos));
             }
             // but we certainly have an integer!
             self.scan_int_suffix();
-            return token::Literal(token::Integer(self.name_from(start_bpos)));
+            return token::Integer(self.name_from(start_bpos));
         }
     }
 
@@ -967,7 +986,9 @@ impl<'a> StringReader<'a> {
         }
 
         if is_dec_digit(c) {
-            return self.scan_number(c.unwrap());
+            let num = self.scan_number(c.unwrap());
+            let suffix = self.scan_optional_raw_name();
+            return token::Literal(num, suffix)
         }
 
         if self.read_embedded_ident {
@@ -1126,17 +1147,19 @@ impl<'a> StringReader<'a> {
             }
             let id = if valid { self.name_from(start) } else { token::intern("0") };
             self.bump(); // advance curr past token
-            return token::Literal(token::Char(id));
+            let suffix = self.scan_optional_raw_name();
+            return token::Literal(token::Char(id), suffix);
           }
           'b' => {
             self.bump();
-            return match self.curr {
+            let lit = match self.curr {
                 Some('\'') => self.scan_byte(),
                 Some('"') => self.scan_byte_string(),
                 Some('r') => self.scan_raw_byte_string(),
                 _ => unreachable!()  // Should have been a token::Ident above.
             };
-
+            let suffix = self.scan_optional_raw_name();
+            return token::Literal(lit, suffix);
           }
           '"' => {
             let start_bpos = self.last_pos;
@@ -1157,7 +1180,8 @@ impl<'a> StringReader<'a> {
             let id = if valid { self.name_from(start_bpos + BytePos(1)) }
                      else { token::intern("??") };
             self.bump();
-            return token::Literal(token::Str_(id));
+            let suffix = self.scan_optional_raw_name();
+            return token::Literal(token::Str_(id), suffix);
           }
           'r' => {
             let start_bpos = self.last_pos;
@@ -1224,7 +1248,8 @@ impl<'a> StringReader<'a> {
             } else {
                 token::intern("??")
             };
-            return token::Literal(token::StrRaw(id, hash_count));
+            let suffix = self.scan_optional_raw_name();
+            return token::Literal(token::StrRaw(id, hash_count), suffix);
           }
           '-' => {
             if self.nextch_is('>') {
@@ -1293,7 +1318,7 @@ impl<'a> StringReader<'a> {
      || (self.curr_is('#') && self.nextch_is('!') && !self.nextnextch_is('['))
     }
 
-    fn scan_byte(&mut self) -> token::Token {
+    fn scan_byte(&mut self) -> token::Lit {
         self.bump();
         let start = self.last_pos;
 
@@ -1314,10 +1339,10 @@ impl<'a> StringReader<'a> {
 
         let id = if valid { self.name_from(start) } else { token::intern("??") };
         self.bump(); // advance curr past token
-        return token::Literal(token::Byte(id));
+        return token::Byte(id);
     }
 
-    fn scan_byte_string(&mut self) -> token::Token {
+    fn scan_byte_string(&mut self) -> token::Lit {
         self.bump();
         let start = self.last_pos;
         let mut valid = true;
@@ -1336,10 +1361,10 @@ impl<'a> StringReader<'a> {
         }
         let id = if valid { self.name_from(start) } else { token::intern("??") };
         self.bump();
-        return token::Literal(token::Binary(id));
+        return token::Binary(id);
     }
 
-    fn scan_raw_byte_string(&mut self) -> token::Token {
+    fn scan_raw_byte_string(&mut self) -> token::Lit {
         let start_bpos = self.last_pos;
         self.bump();
         let mut hash_count = 0u;
@@ -1387,9 +1412,9 @@ impl<'a> StringReader<'a> {
             self.bump();
         }
         self.bump();
-        return token::Literal(token::BinaryRaw(self.name_from_to(content_start_bpos,
-                                                                 content_end_bpos),
-                                               hash_count));
+        return token::BinaryRaw(self.name_from_to(content_start_bpos,
+                                                  content_end_bpos),
+                                hash_count);
     }
 }
 
@@ -1536,17 +1561,17 @@ mod test {
 
     #[test] fn character_a() {
         assert_eq!(setup(&mk_sh(), "'a'".to_string()).next_token().tok,
-                   token::Literal(token::Char(token::intern("a"))));
+                   token::Literal(token::Char(token::intern("a")), None));
     }
 
     #[test] fn character_space() {
         assert_eq!(setup(&mk_sh(), "' '".to_string()).next_token().tok,
-                   token::Literal(token::Char(token::intern(" "))));
+                   token::Literal(token::Char(token::intern(" ")), None));
     }
 
     #[test] fn character_escaped() {
         assert_eq!(setup(&mk_sh(), "'\\n'".to_string()).next_token().tok,
-                   token::Literal(token::Char(token::intern("\\n"))));
+                   token::Literal(token::Char(token::intern("\\n")), None));
     }
 
     #[test] fn lifetime_name() {
@@ -1558,7 +1583,38 @@ mod test {
         assert_eq!(setup(&mk_sh(),
                          "r###\"\"#a\\b\x00c\"\"###".to_string()).next_token()
                                                                  .tok,
-                   token::Literal(token::StrRaw(token::intern("\"#a\\b\x00c\""), 3)));
+                   token::Literal(token::StrRaw(token::intern("\"#a\\b\x00c\""), 3), None));
+    }
+
+    #[test] fn literal_suffixes() {
+        macro_rules! test {
+            ($input: expr, $tok_type: ident, $tok_contents: expr) => {{
+                assert_eq!(setup(&mk_sh(), format!("{}suffix", $input)).next_token().tok,
+                           token::Literal(token::$tok_type(token::intern($tok_contents)),
+                                          Some(token::intern("suffix"))));
+                // with a whitespace separator:
+                assert_eq!(setup(&mk_sh(), format!("{} suffix", $input)).next_token().tok,
+                           token::Literal(token::$tok_type(token::intern($tok_contents)),
+                                          None));
+            }}
+        }
+
+        test!("'a'", Char, "a");
+        test!("b'a'", Byte, "a");
+        test!("\"a\"", Str_, "a");
+        test!("b\"a\"", Binary, "a");
+        test!("1234", Integer, "1234");
+        test!("0b101", Integer, "0b101");
+        test!("0xABC", Integer, "0xABC");
+        test!("1.0", Float, "1.0");
+        test!("1.0e10", Float, "1.0e10");
+
+        assert_eq!(setup(&mk_sh(), "r###\"raw\"###suffix".to_string()).next_token().tok,
+                   token::Literal(token::StrRaw(token::intern("raw"), 3),
+                                  Some(token::intern("suffix"))));
+        assert_eq!(setup(&mk_sh(), "br###\"raw\"###suffix".to_string()).next_token().tok,
+                   token::Literal(token::BinaryRaw(token::intern("raw"), 3),
+                                  Some(token::intern("suffix"))));
     }
 
     #[test] fn line_doc_comments() {
@@ -1574,7 +1630,7 @@ mod test {
             token::Comment => { },
             _ => panic!("expected a comment!")
         }
-        assert_eq!(lexer.next_token().tok, token::Literal(token::Char(token::intern("a"))));
+        assert_eq!(lexer.next_token().tok, token::Literal(token::Char(token::intern("a")), None));
     }
 
 }
diff --git a/src/libsyntax/parse/parser.rs b/src/libsyntax/parse/parser.rs
index 4edcb182e53..b9e1fe07e71 100644
--- a/src/libsyntax/parse/parser.rs
+++ b/src/libsyntax/parse/parser.rs
@@ -646,6 +646,20 @@ impl<'a> Parser<'a> {
         }
     }
 
+    pub fn expect_no_suffix(&mut self, sp: Span, kind: &str, suffix: Option<ast::Name>) {
+        match suffix {
+            None => {/* everything ok */}
+            Some(suf) => {
+                let text = suf.as_str();
+                if text.is_empty() {
+                    self.span_bug(sp, "found empty non-None literal suffix")
+                }
+                self.span_err(sp, &*format!("a {} with a suffix is illegal", kind));
+            }
+        }
+    }
+
+
     /// Attempt to consume a `<`. If `<<` is seen, replace it with a single
     /// `<` and continue. If a `<` is not seen, return false.
     ///
@@ -968,6 +982,9 @@ impl<'a> Parser<'a> {
     pub fn span_err(&mut self, sp: Span, m: &str) {
         self.sess.span_diagnostic.span_err(sp, m)
     }
+    pub fn span_bug(&mut self, sp: Span, m: &str) -> ! {
+        self.sess.span_diagnostic.span_bug(sp, m)
+    }
     pub fn abort_if_errors(&mut self) {
         self.sess.span_diagnostic.handler().abort_if_errors();
     }
@@ -1640,24 +1657,40 @@ impl<'a> Parser<'a> {
     /// Matches token_lit = LIT_INTEGER | ...
     pub fn lit_from_token(&mut self, tok: &token::Token) -> Lit_ {
         match *tok {
-            token::Literal(token::Byte(i)) => LitByte(parse::byte_lit(i.as_str()).val0()),
-            token::Literal(token::Char(i)) => LitChar(parse::char_lit(i.as_str()).val0()),
-            token::Literal(token::Integer(s)) => parse::integer_lit(s.as_str(),
-                                                        &self.sess.span_diagnostic,
-                                                       self.last_span),
-            token::Literal(token::Float(s)) => parse::float_lit(s.as_str()),
-            token::Literal(token::Str_(s)) => {
-                LitStr(token::intern_and_get_ident(parse::str_lit(s.as_str()).as_slice()),
-                       ast::CookedStr)
-            }
-            token::Literal(token::StrRaw(s, n)) => {
-                LitStr(token::intern_and_get_ident(parse::raw_str_lit(s.as_str()).as_slice()),
-                       ast::RawStr(n))
+            token::Literal(lit, suf) => {
+                let (suffix_illegal, out) = match lit {
+                    token::Byte(i) => (true, LitByte(parse::byte_lit(i.as_str()).val0())),
+                    token::Char(i) => (true, LitChar(parse::char_lit(i.as_str()).val0())),
+                    token::Integer(s) => (false, parse::integer_lit(s.as_str(),
+                                                            &self.sess.span_diagnostic,
+                                                            self.last_span)),
+                    token::Float(s) => (false, parse::float_lit(s.as_str())),
+                    token::Str_(s) => {
+                        (true,
+                         LitStr(token::intern_and_get_ident(parse::str_lit(s.as_str()).as_slice()),
+                                ast::CookedStr))
+                    }
+                    token::StrRaw(s, n) => {
+                        (true,
+                         LitStr(
+                            token::intern_and_get_ident(
+                                parse::raw_str_lit(s.as_str()).as_slice()),
+                            ast::RawStr(n)))
+                    }
+                    token::Binary(i) =>
+                        (true, LitBinary(parse::binary_lit(i.as_str()))),
+                    token::BinaryRaw(i, _) =>
+                        (true,
+                         LitBinary(Rc::new(i.as_str().as_bytes().iter().map(|&x| x).collect()))),
+                };
+
+                if suffix_illegal {
+                    let sp = self.last_span;
+                    self.expect_no_suffix(sp, &*format!("{} literal", lit.short_name()), suf)
+                }
+
+                out
             }
-            token::Literal(token::Binary(i)) =>
-                LitBinary(parse::binary_lit(i.as_str())),
-            token::Literal(token::BinaryRaw(i, _)) =>
-                LitBinary(Rc::new(i.as_str().as_bytes().iter().map(|&x| x).collect())),
             _ => { self.unexpected_last(tok); }
         }
     }
@@ -2424,7 +2457,10 @@ impl<'a> Parser<'a> {
                         }
                     }
                   }
-                  token::Literal(token::Integer(n)) => {
+                  token::Literal(token::Integer(n), suf) => {
+                    let sp = self.span;
+                    self.expect_no_suffix(sp, "tuple index", suf);
+
                     let index = n.as_str();
                     let dot = self.last_span.hi;
                     hi = self.span.hi;
@@ -2449,7 +2485,7 @@ impl<'a> Parser<'a> {
                         }
                     }
                   }
-                  token::Literal(token::Float(n)) => {
+                  token::Literal(token::Float(n), _suf) => {
                     self.bump();
                     let last_span = self.last_span;
                     let fstr = n.as_str();
@@ -5085,12 +5121,17 @@ impl<'a> Parser<'a> {
                 self.expect(&token::Semi);
                 (path, the_ident)
             },
-            token::Literal(token::Str_(..)) | token::Literal(token::StrRaw(..)) => {
-                let path = self.parse_str();
+            token::Literal(token::Str_(..), suf) | token::Literal(token::StrRaw(..), suf) => {
+                let sp = self.span;
+                self.expect_no_suffix(sp, "extern crate name", suf);
+                // forgo the internal suffix check of `parse_str` to
+                // avoid repeats (this unwrap will always succeed due
+                // to the restriction of the `match`)
+                let (s, style, _) = self.parse_optional_str().unwrap();
                 self.expect_keyword(keywords::As);
                 let the_ident = self.parse_ident();
                 self.expect(&token::Semi);
-                (Some(path), the_ident)
+                (Some((s, style)), the_ident)
             },
             _ => {
                 let span = self.span;
@@ -5267,7 +5308,9 @@ impl<'a> Parser<'a> {
     /// the `extern` keyword, if one is found.
     fn parse_opt_abi(&mut self) -> Option<abi::Abi> {
         match self.token {
-            token::Literal(token::Str_(s)) | token::Literal(token::StrRaw(s, _)) => {
+            token::Literal(token::Str_(s), suf) | token::Literal(token::StrRaw(s, _), suf) => {
+                let sp = self.span;
+                self.expect_no_suffix(sp, "ABI spec", suf);
                 self.bump();
                 let the_string = s.as_str();
                 match abi::lookup(the_string) {
@@ -5902,21 +5945,27 @@ impl<'a> Parser<'a> {
     }
 
     pub fn parse_optional_str(&mut self)
-                              -> Option<(InternedString, ast::StrStyle)> {
-        let (s, style) = match self.token {
-            token::Literal(token::Str_(s)) => (self.id_to_interned_str(s.ident()), ast::CookedStr),
-            token::Literal(token::StrRaw(s, n)) => {
-                (self.id_to_interned_str(s.ident()), ast::RawStr(n))
+                              -> Option<(InternedString, ast::StrStyle, Option<ast::Name>)> {
+        let ret = match self.token {
+            token::Literal(token::Str_(s), suf) => {
+                (self.id_to_interned_str(s.ident()), ast::CookedStr, suf)
+            }
+            token::Literal(token::StrRaw(s, n), suf) => {
+                (self.id_to_interned_str(s.ident()), ast::RawStr(n), suf)
             }
             _ => return None
         };
         self.bump();
-        Some((s, style))
+        Some(ret)
     }
 
     pub fn parse_str(&mut self) -> (InternedString, StrStyle) {
         match self.parse_optional_str() {
-            Some(s) => { s }
+            Some((s, style, suf)) => {
+                let sp = self.last_span;
+                self.expect_no_suffix(sp, "str literal", suf);
+                (s, style)
+            }
             _ =>  self.fatal("expected string literal")
         }
     }
diff --git a/src/libsyntax/parse/token.rs b/src/libsyntax/parse/token.rs
index bfa6ca798b2..4272b57a4dc 100644
--- a/src/libsyntax/parse/token.rs
+++ b/src/libsyntax/parse/token.rs
@@ -72,6 +72,19 @@ pub enum Lit {
     BinaryRaw(ast::Name, uint), /* raw binary str delimited by n hash symbols */
 }
 
+impl Lit {
+    pub fn short_name(&self) -> &'static str {
+        match *self {
+            Byte(_) => "byte",
+            Char(_) => "char",
+            Integer(_) => "integer",
+            Float(_) => "float",
+            Str_(_) | StrRaw(..) => "str",
+            Binary(_) | BinaryRaw(..) => "binary str"
+        }
+    }
+}
+
 #[allow(non_camel_case_types)]
 #[deriving(Clone, Encodable, Decodable, PartialEq, Eq, Hash, Show)]
 pub enum Token {
@@ -111,7 +124,7 @@ pub enum Token {
     CloseDelim(DelimToken),
 
     /* Literals */
-    Literal(Lit),
+    Literal(Lit, Option<ast::Name>),
 
     /* Name components */
     Ident(ast::Ident, IdentStyle),
@@ -151,7 +164,7 @@ impl Token {
             Ident(_, _)                 => true,
             Underscore                  => true,
             Tilde                       => true,
-            Literal(_)                  => true,
+            Literal(_, _)               => true,
             Pound                       => true,
             At                          => true,
             Not                         => true,
@@ -172,7 +185,7 @@ impl Token {
     /// Returns `true` if the token is any literal
     pub fn is_lit(&self) -> bool {
         match *self {
-            Literal(_) => true,
+            Literal(_, _) => true,
             _          => false,
         }
     }