Require a bang pattern when unlifted types are where/let bound; #3182
[ghc-hetmet.git] / compiler / parser / Lexer.x
index c813e36..edfbecd 100644 (file)
@@ -32,7 +32,7 @@
 --       qualified varids.
 
 {
-{-# OPTIONS -w #-}
+{-# OPTIONS -Wwarn #-}
 -- The above warning supression flag is a temporary kludge.
 -- While working on this module you are encouraged to remove it and fix
 -- any warnings in the module. See
@@ -70,11 +70,9 @@ import Util          ( maybePrefixMatch, readRational )
 
 import Control.Monad
 import Data.Bits
-import Data.Char       ( chr, ord, isSpace )
+import Data.Char
 import Data.Ratio
 import Debug.Trace
-
-import Unicode ( GeneralCategory(..), generalCategory, isPrint, isUpper )
 }
 
 $unispace    = \x05 -- Trick Alex into handling Unicode. See alexGetChar.
@@ -223,7 +221,8 @@ $tab+         { warn Opt_WarnTabs (text "Tab character") }
 
 <0,option_prags> \n                            { begin bol }
 
-"{-#" $whitechar* (line|LINE)          { begin line_prag2 }
+"{-#" $whitechar* (line|LINE) / { notFollowedByPragmaChar }
+                            { begin line_prag2 }
 
 -- single-line line pragmas, of the form
 --    # <line> "<file>" <extra-stuff> \n
@@ -240,31 +239,45 @@ $tab+         { warn Opt_WarnTabs (text "Tab character") }
    -- with older versions of GHC which generated these.
 
 <0,option_prags> {
-  "{-#" $whitechar* (RULES|rules)       { rulePrag }
-  "{-#" $whitechar* (INLINE|inline)    { token (ITinline_prag True) }
-  "{-#" $whitechar* (NO(T?)INLINE|no(t?)inline)
+  "{-#" $whitechar* (RULES|rules)  / { notFollowedByPragmaChar } { rulePrag }
+  "{-#" $whitechar* (INLINE|inline)     / { notFollowedByPragmaChar }
+                    { token (ITinline_prag True) }
+  "{-#" $whitechar* (NO(T?)INLINE|no(t?)inline) / { notFollowedByPragmaChar }
                                        { token (ITinline_prag False) }
-  "{-#" $whitechar* (SPECIALI[SZ]E|speciali[sz]e)
+  "{-#" $whitechar* (INLINE|inline)
+        $whitechar+ (CONLIKE|conlike) / { notFollowedByPragmaChar }
+                                        { token (ITinline_conlike_prag True) }
+  "{-#" $whitechar* (NO(T)?INLINE|no(t?)inline)
+        $whitechar+ (CONLIKE|constructorlike) / { notFollowedByPragmaChar }
+                                        { token (ITinline_conlike_prag False) }
+  "{-#" $whitechar* (SPECIALI[SZ]E|speciali[sz]e) / { notFollowedByPragmaChar }
                                        { token ITspec_prag }
   "{-#" $whitechar* (SPECIALI[SZ]E|speciali[sz]e)
-       $whitechar* (INLINE|inline)     { token (ITspec_inline_prag True) }
+       $whitechar+ (INLINE|inline) / { notFollowedByPragmaChar }
+                    { token (ITspec_inline_prag True) }
   "{-#" $whitechar* (SPECIALI[SZ]E|speciali[sz]e)
-       $whitechar* (NO(T?)INLINE|no(t?)inline)
+       $whitechar+ (NO(T?)INLINE|no(t?)inline) / { notFollowedByPragmaChar }
                                        { token (ITspec_inline_prag False) }
-  "{-#" $whitechar* (SOURCE|source)    { token ITsource_prag }
-  "{-#" $whitechar* (WARNING|warning)
+  "{-#" $whitechar* (SOURCE|source) / { notFollowedByPragmaChar }
+                    { token ITsource_prag }
+  "{-#" $whitechar* (WARNING|warning) / { notFollowedByPragmaChar }
                                        { token ITwarning_prag }
-  "{-#" $whitechar* (DEPRECATED|deprecated)
+  "{-#" $whitechar* (DEPRECATED|deprecated) / { notFollowedByPragmaChar }
                                        { token ITdeprecated_prag }
-  "{-#" $whitechar* (SCC|scc)          { token ITscc_prag }
-  "{-#" $whitechar* (GENERATED|generated)
+  "{-#" $whitechar* (SCC|scc)  / { notFollowedByPragmaChar }
+                    { token ITscc_prag }
+  "{-#" $whitechar* (GENERATED|generated) / { notFollowedByPragmaChar }
                                        { token ITgenerated_prag }
-  "{-#" $whitechar* (CORE|core)                { token ITcore_prag }
-  "{-#" $whitechar* (UNPACK|unpack)    { token ITunpack_prag }
+  "{-#" $whitechar* (CORE|core) / { notFollowedByPragmaChar }
+                    { token ITcore_prag }
+  "{-#" $whitechar* (UNPACK|unpack) / { notFollowedByPragmaChar }
+                    { token ITunpack_prag }
+  "{-#" $whitechar* (ANN|ann) / { notFollowedByPragmaChar }
+                    { token ITann_prag }
 
   -- We ignore all these pragmas, but don't generate a warning for them
   -- CFILES is a hugs-only thing.
-  "{-#" $whitechar* (OPTIONS_HUGS|options_hugs|OPTIONS_NHC98|options_nhc98|OPTIONS_JHC|options_jhc|CFILES|cfiles)
+  "{-#" $whitechar* (OPTIONS_(HUGS|hugs|NHC98|nhc98|JHC|jhc|YHC|yhc|CATCH|catch|DERIVE|derive)|CFILES|cfiles|CONTRACT|contract) / { notFollowedByPragmaChar }
                     { nested_comment lexToken }
 
   -- ToDo: should only be valid inside a pragma:
@@ -272,19 +285,23 @@ $tab+         { warn Opt_WarnTabs (text "Tab character") }
 }
 
 <option_prags> {
-  "{-#"  $whitechar* (OPTIONS|options)   { lex_string_prag IToptions_prag }
-  "{-#"  $whitechar* (OPTIONS_GHC|options_ghc)
+  "{-#"  $whitechar* (OPTIONS|options) / { notFollowedByPragmaChar }
+                                        { lex_string_prag IToptions_prag }
+  "{-#"  $whitechar* (OPTIONS_GHC|options_ghc) / { notFollowedByPragmaChar }
                                         { lex_string_prag IToptions_prag }
   "{-#"  $whitechar* (OPTIONS_HADDOCK|options_haddock)
+                   / { notFollowedByPragmaChar }
                                          { lex_string_prag ITdocOptions }
   "-- #"                                 { multiline_doc_comment }
-  "{-#"  $whitechar* (LANGUAGE|language) { token ITlanguage_prag }
-  "{-#"  $whitechar* (INCLUDE|include)   { lex_string_prag ITinclude_prag }
+  "{-#"  $whitechar* (LANGUAGE|language) / { notFollowedByPragmaChar }
+                                         { token ITlanguage_prag }
+  "{-#"  $whitechar* (INCLUDE|include) / { notFollowedByPragmaChar }
+                                         { lex_string_prag ITinclude_prag }
 }
 
 <0> {
   -- In the "0" mode we ignore these pragmas
-  "{-#"  $whitechar* (OPTIONS|options|OPTIONS_GHC|options_ghc|OPTIONS_HADDOCK|options_haddock|LANGUAGE|language|INCLUDE|include)
+  "{-#"  $whitechar* (OPTIONS|options|OPTIONS_GHC|options_ghc|OPTIONS_HADDOCK|options_haddock|LANGUAGE|language|INCLUDE|include) / { notFollowedByPragmaChar }
                      { nested_comment lexToken }
 }
 
@@ -479,6 +496,7 @@ data Token
 
        -- Pragmas
   | ITinline_prag Bool         -- True <=> INLINE, False <=> NOINLINE
+  | ITinline_conlike_prag Bool  -- same
   | ITspec_prag                        -- SPECIALISE   
   | ITspec_inline_prag Bool    -- SPECIALISE INLINE (or NOINLINE)
   | ITsource_prag
@@ -490,6 +508,7 @@ data Token
   | ITgenerated_prag
   | ITcore_prag                 -- hdaume: core annotations
   | ITunpack_prag
+  | ITann_prag
   | ITclose_prag
   | IToptions_prag String
   | ITinclude_prag String
@@ -700,16 +719,15 @@ reservedSymsFM = listToUFM $
        ,("!",   ITbang,     always)
 
         -- For data T (a::*) = MkT
-       ,("*", ITstar, \i -> kindSigsEnabled i || tyFamEnabled i)
+       ,("*", ITstar, always) -- \i -> kindSigsEnabled i || tyFamEnabled i)
         -- For 'forall a . t'
-       ,(".", ITdot, \i -> explicitForallEnabled i || inRulePrag i)
+       ,(".", ITdot,  always) -- \i -> explicitForallEnabled i || inRulePrag i)
 
        ,("-<",  ITlarrowtail, arrowsEnabled)
        ,(">-",  ITrarrowtail, arrowsEnabled)
        ,("-<<", ITLarrowtail, arrowsEnabled)
        ,(">>-", ITRarrowtail, arrowsEnabled)
 
-#if __GLASGOW_HASKELL__ >= 605
        ,("∷",   ITdcolon, unicodeSyntaxEnabled)
        ,("⇒",   ITdarrow, unicodeSyntaxEnabled)
        ,("∀",   ITforall, \i -> unicodeSyntaxEnabled i &&
@@ -720,7 +738,6 @@ reservedSymsFM = listToUFM $
         -- ToDo: ideally, → and ∷ should be "specials", so that they cannot
         -- form part of a large operator.  This would let us have a better
         -- syntax for kinds: ɑ∷*→* would be a legal kind signature. (maybe).
-#endif
        ]
 
 -- -----------------------------------------------------------------------------
@@ -769,6 +786,9 @@ notFollowedBy char _ _ _ (AI _ _ buf)
 notFollowedBySymbol _ _ _ (AI _ _ buf)
   = nextCharIs buf (`notElem` "!#$%&*+./<=>?@\\^|-~")
 
+notFollowedByPragmaChar _ _ _ (AI _ _ buf)
+  = nextCharIs buf (\c -> not (isAlphaNum c || c == '_'))
+
 -- We must reject doc comments as being ordinary comments everywhere.
 -- In some cases the doc comment will be selected as the lexeme due to
 -- maximal munch, but not always, because the nested comment rule is
@@ -1230,11 +1250,11 @@ lex_char_tok span _buf _len = do        -- We've seen '
                -- We've seen 'x, where x is a valid character
                --  (i.e. not newline etc) but not a quote or backslash
           case alexGetChar' i2 of      -- Look ahead one more character
-               Nothing -> lit_error
                Just ('\'', i3) -> do   -- We've seen 'x'
                        setInput i3 
                        finish_char_tok loc c
                _other -> do            -- We've seen 'x not followed by quote
+                                       -- (including the possibility of EOF)
                                        -- If TH is on, just parse the quote only
                        th_exts <- extension thEnabled  
                        let (AI end _ _) = i1