X-Git-Url: http://git.megacz.com/?a=blobdiff_plain;f=ghc%2Fcompiler%2FbasicTypes%2FOccName.lhs;h=94b50bb38ebf293f5dd0ce44cbef5a40193ea6de;hb=34c2d268ed257a0ae063f8867adfb4144409dd38;hp=0735434200c75ac9ccbb5981e9b16afa957c23a3;hpb=0554dc08d9e05e812d264a682679b798fce1ff78;p=ghc-hetmet.git diff --git a/ghc/compiler/basicTypes/OccName.lhs b/ghc/compiler/basicTypes/OccName.lhs index 0735434..94b50bb 100644 --- a/ghc/compiler/basicTypes/OccName.lhs +++ b/ghc/compiler/basicTypes/OccName.lhs @@ -7,19 +7,19 @@ \begin{code} module OccName ( -- The NameSpace type; abstact - NameSpace, tcName, clsName, tcClsName, dataName, varName, tvName, - nameSpaceString, + NameSpace, tcName, clsName, tcClsName, dataName, varName, ipName, + tvName, uvName, nameSpaceString, -- The OccName type OccName, -- Abstract, instance of Outputable pprOccName, - mkSrcOccFS, mkSysOcc, mkSysOccFS, mkSrcVarOcc, mkKindOccFS, + mkSrcOccFS, mkSysOcc, mkSysOccFS, mkCCallOcc, mkSrcVarOcc, mkKindOccFS, mkSuperDictSelOcc, mkDFunOcc, mkForeignExportOcc, - mkDictOcc, mkWorkerOcc, mkMethodOcc, mkDefaultMethodOcc, + mkDictOcc, mkIPOcc, mkWorkerOcc, mkMethodOcc, mkDefaultMethodOcc, mkDerivedTyConOcc, mkClassTyConOcc, mkClassDataConOcc, mkSpecOcc, - isTvOcc, isDataOcc, isDataSymOcc, isSymOcc, + isSysOcc, isTvOcc, isUvOcc, isDataOcc, isDataSymOcc, isSymOcc, isIPOcc, isValOcc, occNameFS, occNameString, occNameUserString, occNameSpace, occNameFlavour, setOccNameSpace, @@ -58,7 +58,7 @@ code the encoding operation is not performed on each occurrence. These type synonyms help documentation. \begin{code} -type UserFS = FAST_STRING -- As the user typed it +type UserFS = FAST_STRING -- As the user typed it type EncodedFS = FAST_STRING -- Encoded form type UserString = String -- As the user typed it @@ -68,10 +68,10 @@ type EncodedString = String -- Encoded form pprEncodedFS :: EncodedFS -> SDoc pprEncodedFS fs = getPprStyle $ \ sty -> - if userStyle sty then - text (decode (_UNPK_ fs)) - else - ptext fs + if userStyle sty + -- ptext (decodeFS fs) would needlessly pack the string again + then text (decode (_UNPK_ fs)) + else ptext fs \end{code} %************************************************************************ @@ -82,10 +82,12 @@ pprEncodedFS fs \begin{code} data NameSpace = VarName -- Variables + | IPName -- Implicit Parameters | DataName -- Data constructors | TvName -- Type variables + | UvName -- Usage variables | TcClsName -- Type constructors and classes; Haskell has them - -- in the same name space for now. + -- in the same name space for now. deriving( Eq, Ord ) -- Though type constructors and classes are in the same name space now, @@ -96,13 +98,17 @@ tcClsName = TcClsName -- Not sure which! dataName = DataName tvName = TvName +uvName = UvName varName = VarName +ipName = IPName nameSpaceString :: NameSpace -> String nameSpaceString DataName = "Data constructor" nameSpaceString VarName = "Variable" +nameSpaceString IPName = "Implicit Param" nameSpaceString TvName = "Type variable" +nameSpaceString UvName = "Usage variable" nameSpaceString TcClsName = "Type constructor or class" \end{code} @@ -156,13 +162,20 @@ already encoded \begin{code} mkSysOcc :: NameSpace -> EncodedString -> OccName -mkSysOcc occ_sp str = ASSERT( alreadyEncoded str ) +mkSysOcc occ_sp str = ASSERT2( alreadyEncoded str, text str ) OccName occ_sp (_PK_ str) mkSysOccFS :: NameSpace -> EncodedFS -> OccName mkSysOccFS occ_sp fs = ASSERT2( alreadyEncodedFS fs, ppr fs ) OccName occ_sp fs +mkCCallOcc :: EncodedString -> OccName +-- This version of mkSysOcc doesn't check that the string is already encoded, +-- because it will be something like "{__ccall f dyn Int# -> Int#}" +-- This encodes a lot into something that then parses like an Id. +-- But then alreadyEncoded complains about the braces! +mkCCallOcc str = OccName varName (_PK_ str) + -- Kind constructors get a speical function. Uniquely, they are not encoded, -- so that they have names like '*'. This means that *even in interface files* -- we'll get kinds like (* -> (* -> *)). We can't use mkSysOcc because it @@ -211,23 +224,33 @@ occNameFlavour (OccName sp _) = nameSpaceString sp \end{code} \begin{code} -isTvOcc, isDataSymOcc, isSymOcc :: OccName -> Bool +isTvOcc, isDataSymOcc, isSymOcc, isUvOcc :: OccName -> Bool isTvOcc (OccName TvName _) = True isTvOcc other = False +isUvOcc (OccName UvName _) = True +isUvOcc other = False + +isValOcc (OccName VarName _) = True +isValOcc (OccName DataName _) = True +isValOcc other = False + -- Data constructor operator (starts with ':', or '[]') -- Pretty inefficient! isDataSymOcc (OccName DataName s) = isLexConSym (decodeFS s) isDataSymOcc other = False isDataOcc (OccName DataName _) = True -isDataOcc oter = False +isDataOcc other = False -- Any operator (data constructor or variable) -- Pretty inefficient! isSymOcc (OccName DataName s) = isLexConSym (decodeFS s) isSymOcc (OccName VarName s) = isLexSym (decodeFS s) + +isIPOcc (OccName IPName _) = True +isIPOcc _ = False \end{code} @@ -271,7 +294,7 @@ mk_deriv occ_sp sys_prefix str = mkSysOcc occ_sp (encode sys_prefix ++ str) \end{code} \begin{code} -mkDictOcc, mkWorkerOcc, mkDefaultMethodOcc, +mkDictOcc, mkIPOcc, mkWorkerOcc, mkDefaultMethodOcc, mkClassTyConOcc, mkClassDataConOcc, mkSpecOcc :: OccName -> OccName @@ -282,10 +305,18 @@ mkDerivedTyConOcc = mk_simple_deriv tcName ":" -- The : prefix makes sure it mkClassTyConOcc = mk_simple_deriv tcName ":T" -- as a tycon/datacon mkClassDataConOcc = mk_simple_deriv dataName ":D" -- mkDictOcc = mk_simple_deriv varName "$d" +mkIPOcc = mk_simple_deriv varName "$i" mkSpecOcc = mk_simple_deriv varName "$s" mkForeignExportOcc = mk_simple_deriv varName "$f" mk_simple_deriv sp px occ = mk_deriv sp px (occNameString occ) + + +isSysOcc :: OccName -> Bool -- True for all these '$' things +isSysOcc occ = case occNameUserString occ of + ('$' : _ ) -> True + other -> False -- We don't care about the ':' ones + -- isSysOcc is only called for Ids anyway \end{code} \begin{code} @@ -298,18 +329,16 @@ mkSuperDictSelOcc index cls_occ \begin{code} -mkDFunOcc :: OccName -- class, eg "Ord" - -> OccName -- tycon (or something convenient from the instance type) - -- eg "Maybe" - -> Int -- Unique to distinguish dfuns which share the previous two - -- eg 3 - -> OccName -- "dOrdMaybe3" - -mkDFunOcc cls_occ tycon_occ index - = mk_deriv VarName "$f" (show_index ++ cls_str ++ tycon_str) +mkDFunOcc :: EncodedString -- Typically the class and type glommed together e.g. "OrdMaybe" + -> Int -- Unique to distinguish dfuns which share the previous two + -- eg 3 + -- The requirement is that the (string,index) pair be unique in this module + + -> OccName -- "$fOrdMaybe3" + +mkDFunOcc string index + = mk_deriv VarName "$f" (show_index ++ string) where - cls_str = occNameString cls_occ - tycon_str = occNameString tycon_occ show_index | index == 0 = "" | otherwise = show index \end{code} @@ -400,29 +429,29 @@ The basic encoding scheme is this. * Tuples (,,,) are coded as Z3T -* Alphabetic characters (upper and lower), digits, and '_' +* Alphabetic characters (upper and lower) and digits all translate to themselves; except 'Z', which translates to 'ZZ' and 'z', which translates to 'zz' We need both so that we can preserve the variable/tycon distinction -* Most other printable characters translate to 'Zx' for some +* Most other printable characters translate to 'zx' or 'Zx' for some alphabetic character x -* The others translate as 'Zxdd' where 'dd' is exactly two hexadecimal +* The others translate as 'zxdd' where 'dd' is exactly two hexadecimal digits for the ord of the character Before After -------------------------- Trak Trak - foo_wib foo_wib - > Zg - >1 Zg1 - foo# fooZh - foo## fooZhZh - foo##1 fooZhXh1 + foo_wib foozuwib + > zg + >1 zg1 + foo# foozh + foo## foozhzh + foo##1 foozhzh1 fooZ fooZZ - :+ ZcZp + :+ Zczp () Z0T (,,,,) Z4T @@ -434,7 +463,10 @@ The basic encoding scheme is this. alreadyEncoded :: String -> Bool alreadyEncoded s = all ok s where - ok '_' = True + ok ' ' = True -- This is a bit of a lie; if we really wanted spaces + -- in names we'd have to encode them. But we do put + -- spaces in ccall "occurrences", and we don't want to + -- reject them here ok ch = ISALPHANUM ch alreadyEncodedFS :: FAST_STRING -> Bool @@ -464,7 +496,6 @@ encodeFS fast_str | all unencodedChar str = fast_str str = _UNPK_ fast_str unencodedChar :: Char -> Bool -- True for chars that don't need encoding -unencodedChar '_' = True unencodedChar 'Z' = False unencodedChar 'z' = False unencodedChar c = ISALPHANUM c @@ -484,6 +515,7 @@ encode_ch 'Z' = "ZZ" encode_ch 'z' = "zz" encode_ch '&' = "za" encode_ch '|' = "zb" +encode_ch '^' = "zc" encode_ch '$' = "zd" encode_ch '=' = "ze" encode_ch '>' = "zg" @@ -497,7 +529,7 @@ encode_ch '\'' = "zq" encode_ch '\\' = "zr" encode_ch '/' = "zs" encode_ch '*' = "zt" -encode_ch '^' = "zu" +encode_ch '_' = "zu" encode_ch '%' = "zv" encode_ch c = ['z', 'x', intToDigit hi, intToDigit lo] where @@ -518,16 +550,17 @@ decode (c : rest) = c : decode rest decode_escape :: EncodedString -> UserString -decode_escape ('Z' : rest) = 'Z' : decode rest -decode_escape ('C' : rest) = ':' : decode rest decode_escape ('L' : rest) = '(' : decode rest decode_escape ('R' : rest) = ')' : decode rest decode_escape ('M' : rest) = '[' : decode rest decode_escape ('N' : rest) = ']' : decode rest +decode_escape ('C' : rest) = ':' : decode rest +decode_escape ('Z' : rest) = 'Z' : decode rest decode_escape ('z' : rest) = 'z' : decode rest decode_escape ('a' : rest) = '&' : decode rest decode_escape ('b' : rest) = '|' : decode rest +decode_escape ('c' : rest) = '^' : decode rest decode_escape ('d' : rest) = '$' : decode rest decode_escape ('e' : rest) = '=' : decode rest decode_escape ('g' : rest) = '>' : decode rest @@ -541,7 +574,7 @@ decode_escape ('q' : rest) = '\'' : decode rest decode_escape ('r' : rest) = '\\' : decode rest decode_escape ('s' : rest) = '/' : decode rest decode_escape ('t' : rest) = '*' : decode rest -decode_escape ('u' : rest) = '^' : decode rest +decode_escape ('u' : rest) = '_' : decode rest decode_escape ('v' : rest) = '%' : decode rest decode_escape ('x' : d1 : d2 : rest) = chr (digitToInt d1 * 16 + digitToInt d2) : decode rest @@ -581,32 +614,29 @@ isLexSym cs = isLexConSym cs || isLexVarSym cs isLexConId cs -- Prefix type or data constructors | _NULL_ cs = False -- e.g. "Foo", "[]", "(,)" | cs == SLIT("[]") = True - | c == '(' = True -- (), (,), (,,), ... - | otherwise = isUpper c || isUpperISO c - where - c = _HEAD_ cs + | otherwise = startsConId (_HEAD_ cs) isLexVarId cs -- Ordinary prefix identifiers | _NULL_ cs = False -- e.g. "x", "_x" - | otherwise = isLower c || isLowerISO c || c == '_' - where - c = _HEAD_ cs + | otherwise = startsVarId (_HEAD_ cs) isLexConSym cs -- Infix type or data constructors | _NULL_ cs = False -- e.g. ":-:", ":", "->" - | otherwise = c == ':' - || cs == SLIT("->") - where - c = _HEAD_ cs + | cs == SLIT("->") = True + | otherwise = startsConSym (_HEAD_ cs) isLexVarSym cs -- Infix identifiers | _NULL_ cs = False -- e.g. "+" - | otherwise = isSymbolASCII c - || isSymbolISO c - where - c = _HEAD_ cs + | otherwise = startsVarSym (_HEAD_ cs) ------------- +startsVarSym, startsVarId, startsConSym, startsConId :: Char -> Bool +startsVarSym c = isSymbolASCII c || isSymbolISO c -- Infix Ids +startsConSym c = c == ':' -- Infix data constructors +startsVarId c = isLower c || isLowerISO c || c == '_' -- Ordinary Ids +startsConId c = isUpper c || isUpperISO c || c == '(' -- Ordinary type constructors and data constructors + + isSymbolASCII c = c `elem` "!#$%&*+./<=>?@\\^|~-" isSymbolISO c = ord c `elem` (0xd7 : 0xf7 : [0xa1 .. 0xbf]) isUpperISO (C# c#) = c# `geChar#` '\xc0'# && c# `leChar#` '\xde'# && c# `neChar#` '\xd7'#