tools/PascalPreprocessor.hs
author unc0rr
Tue, 22 May 2012 00:29:32 +0400
changeset 7103 1c1afb5a1565
parent 7067 f98ec3aecf4e
child 7315 59b5b19e6604
permissions -rw-r--r--
Feel free to add real checksum function
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
6412
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
     1
module PascalPreprocessor where
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
     2
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
     3
import Text.Parsec
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
     4
import Control.Monad.IO.Class
6414
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
     5
import Control.Monad
6412
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
     6
import System.IO
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
     7
import qualified Data.Map as Map
6414
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
     8
import Data.Char
6412
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
     9
6413
6714531e7bd2 Preprocessor strips comments
unc0rr
parents: 6412
diff changeset
    10
6714531e7bd2 Preprocessor strips comments
unc0rr
parents: 6412
diff changeset
    11
-- comments are removed
6714531e7bd2 Preprocessor strips comments
unc0rr
parents: 6412
diff changeset
    12
comment = choice [
6714531e7bd2 Preprocessor strips comments
unc0rr
parents: 6412
diff changeset
    13
        char '{' >> notFollowedBy (char '$') >> manyTill anyChar (try $ char '}') >> return ""
6714531e7bd2 Preprocessor strips comments
unc0rr
parents: 6412
diff changeset
    14
        , (try $ string "(*") >> manyTill anyChar (try $ string "*)") >> return ""
6714531e7bd2 Preprocessor strips comments
unc0rr
parents: 6412
diff changeset
    15
        , (try $ string "//") >> manyTill anyChar (try newline) >> return "\n"
6714531e7bd2 Preprocessor strips comments
unc0rr
parents: 6412
diff changeset
    16
        ]
6714531e7bd2 Preprocessor strips comments
unc0rr
parents: 6412
diff changeset
    17
7038
d853e4385241 Some more definitions and slight fixes
unc0rr
parents: 6964
diff changeset
    18
initDefines = Map.fromList [
d853e4385241 Some more definitions and slight fixes
unc0rr
parents: 6964
diff changeset
    19
    ("FPC", "")
d853e4385241 Some more definitions and slight fixes
unc0rr
parents: 6964
diff changeset
    20
    , ("PAS2C", "")
d853e4385241 Some more definitions and slight fixes
unc0rr
parents: 6964
diff changeset
    21
    ]
6425
1ef4192aa80d - Parse unions, sets, function type, packed arrays and some more imporvements to the parser. Now it parses uVariable, uConsts and even SDLh.pas
unc0rr
parents: 6414
diff changeset
    22
        
6412
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    23
preprocess :: String -> IO String
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    24
preprocess fn = do
6425
1ef4192aa80d - Parse unions, sets, function type, packed arrays and some more imporvements to the parser. Now it parses uVariable, uConsts and even SDLh.pas
unc0rr
parents: 6414
diff changeset
    25
    r <- runParserT (preprocessFile fn) (initDefines, [True]) "" ""
6412
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    26
    case r of
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    27
         (Left a) -> do
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    28
             hPutStrLn stderr (show a)
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    29
             return ""
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    30
         (Right a) -> return a
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    31
    
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    32
    where
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    33
    preprocessFile fn = do
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    34
        f <- liftIO (readFile fn)
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    35
        setInput f
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    36
        preprocessor
6413
6714531e7bd2 Preprocessor strips comments
unc0rr
parents: 6412
diff changeset
    37
        
6414
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    38
    preprocessor, codeBlock, switch :: ParsecT String (Map.Map String String, [Bool]) IO String
6413
6714531e7bd2 Preprocessor strips comments
unc0rr
parents: 6412
diff changeset
    39
    
6414
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    40
    preprocessor = chainr codeBlock (return (++)) ""
6413
6714531e7bd2 Preprocessor strips comments
unc0rr
parents: 6412
diff changeset
    41
    
6414
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    42
    codeBlock = do
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    43
        s <- choice [
6412
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    44
            switch
6413
6714531e7bd2 Preprocessor strips comments
unc0rr
parents: 6412
diff changeset
    45
            , comment
6414
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    46
            , char '\'' >> many (noneOf "'\n") >>= \s -> char '\'' >> return ('\'' : s ++ "'")
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    47
            , identifier >>= replace
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    48
            , noneOf "{" >>= \a -> return [a]
6412
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    49
            ]
6414
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    50
        (_, ok) <- getState
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    51
        return $ if and ok then s else ""
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    52
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    53
    --otherChar c = c `notElem` "{/('_" && not (isAlphaNum c)
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    54
    identifier = do
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    55
        c <- letter <|> oneOf "_"
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    56
        s <- many (alphaNum <|> oneOf "_")
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    57
        return $ c:s
6413
6714531e7bd2 Preprocessor strips comments
unc0rr
parents: 6412
diff changeset
    58
            
6412
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    59
    switch = do
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    60
        try $ string "{$"
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    61
        s <- choice [
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    62
            include
6414
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    63
            , ifdef
6453
11c578d30bd3 Countless imporvements to the parser and countless help to the parser in sources.
unc0rr
parents: 6425
diff changeset
    64
            , if'
6414
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    65
            , elseSwitch
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    66
            , endIf
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    67
            , define
6412
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    68
            , unknown
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    69
            ]
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    70
        return s
6413
6714531e7bd2 Preprocessor strips comments
unc0rr
parents: 6412
diff changeset
    71
        
6412
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    72
    include = do
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    73
        try $ string "INCLUDE"
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    74
        spaces
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    75
        (char '"')
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    76
        fn <- many1 $ noneOf "\"\n"
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    77
        char '"'
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    78
        spaces
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    79
        char '}'
6964
6dde80ae7049 Raise exception when .inc file isn't found
unc0rr
parents: 6891
diff changeset
    80
        f <- liftIO (readFile fn `catch` error ("File not found: " ++ fn))
6412
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    81
        c <- getInput
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    82
        setInput $ f ++ c
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    83
        return ""
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
    84
6414
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    85
    ifdef = do
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    86
        s <- try (string "IFDEF") <|> try (string "IFNDEF")
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    87
        let f = if s == "IFNDEF" then not else id
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    88
        
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    89
        spaces
6425
1ef4192aa80d - Parse unions, sets, function type, packed arrays and some more imporvements to the parser. Now it parses uVariable, uConsts and even SDLh.pas
unc0rr
parents: 6414
diff changeset
    90
        d <- identifier
6414
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    91
        spaces
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    92
        char '}'
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    93
        
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    94
        updateState $ \(m, b) ->
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    95
            (m, (f $ d `Map.member` m) : b)
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    96
      
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
    97
        return ""
6453
11c578d30bd3 Countless imporvements to the parser and countless help to the parser in sources.
unc0rr
parents: 6425
diff changeset
    98
11c578d30bd3 Countless imporvements to the parser and countless help to the parser in sources.
unc0rr
parents: 6425
diff changeset
    99
    if' = do
11c578d30bd3 Countless imporvements to the parser and countless help to the parser in sources.
unc0rr
parents: 6425
diff changeset
   100
        s <- try (string "IF" >> notFollowedBy alphaNum)
6414
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   101
        
6453
11c578d30bd3 Countless imporvements to the parser and countless help to the parser in sources.
unc0rr
parents: 6425
diff changeset
   102
        manyTill anyChar (char '}')
11c578d30bd3 Countless imporvements to the parser and countless help to the parser in sources.
unc0rr
parents: 6425
diff changeset
   103
        --char '}'
11c578d30bd3 Countless imporvements to the parser and countless help to the parser in sources.
unc0rr
parents: 6425
diff changeset
   104
        
11c578d30bd3 Countless imporvements to the parser and countless help to the parser in sources.
unc0rr
parents: 6425
diff changeset
   105
        updateState $ \(m, b) ->
11c578d30bd3 Countless imporvements to the parser and countless help to the parser in sources.
unc0rr
parents: 6425
diff changeset
   106
            (m, False : b)
11c578d30bd3 Countless imporvements to the parser and countless help to the parser in sources.
unc0rr
parents: 6425
diff changeset
   107
      
11c578d30bd3 Countless imporvements to the parser and countless help to the parser in sources.
unc0rr
parents: 6425
diff changeset
   108
        return ""
11c578d30bd3 Countless imporvements to the parser and countless help to the parser in sources.
unc0rr
parents: 6425
diff changeset
   109
6414
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   110
    elseSwitch = do
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   111
        try $ string "ELSE}"
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   112
        updateState $ \(m, b:bs) -> (m, (not b):bs)
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   113
        return ""
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   114
    endIf = do
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   115
        try $ string "ENDIF}"
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   116
        updateState $ \(m, b:bs) -> (m, bs)
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   117
        return ""
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   118
    define = do
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   119
        try $ string "DEFINE"
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   120
        spaces
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   121
        i <- identifier        
6425
1ef4192aa80d - Parse unions, sets, function type, packed arrays and some more imporvements to the parser. Now it parses uVariable, uConsts and even SDLh.pas
unc0rr
parents: 6414
diff changeset
   122
        d <- ((string ":=" >> return ())<|> spaces) >> many (noneOf "}")
6414
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   123
        char '}'
7067
f98ec3aecf4e A solution to char vs string problem: mark single-letter strings with _S macro
unc0rr
parents: 7059
diff changeset
   124
        updateState $ \(m, b) -> (if (and b) && (head i /= '_') then Map.insert i d m else m, b)
6414
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   125
        return ""
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   126
    replace s = do
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   127
        (m, _) <- getState
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   128
        return $ Map.findWithDefault s s m
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   129
        
6412
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
   130
    unknown = do
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
   131
        fn <- many1 $ noneOf "}\n"
4b9a59116535 - Split PascalParser into modules
unc0rr
parents:
diff changeset
   132
        char '}'
6414
8474b7fa84d6 Finish preprocessor. Now it correctly handles $IFDEF, $IFNDEF, $ELSE, $ENDIF and $DEFINE.
unc0rr
parents: 6413
diff changeset
   133
        return $ "{$" ++ fn ++ "}"