aboutsummaryrefslogtreecommitdiff
path: root/Parser.hs
blob: f525ba61b96d4f2cae81031aada7c1285df67d01 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
{-# LANGUAGE DeriveFunctor #-}
{-# LANGUAGE LambdaCase #-}
module Parser where

import Control.Applicative
import Data.Char
import Control.Monad.Chronicle
import Control.Monad.Reader
import Control.Monad.State.Strict
import Data.Foldable (asum)
import Data.These

import AST


-- Positions are zero-based in both dimensions
data PS = PS
    { psRefCol :: Int
    , psLine :: Int
    , psCol :: Int
    , psRest :: String }
  deriving (Show)

data Context = Context { ctxFile :: FilePath }
  deriving (Show)

type Parser = ReaderT Context (ChronicleT [ErrMsg] (State PS))

-- Positions are zero-based in both dimensions
data ErrMsg = ErrMsg { errFile :: FilePath
                     , errLine :: Int
                     , errCol :: Int
                     , errMsg :: String }
  deriving (Show)

printErrMsg :: ErrMsg -> String
printErrMsg (ErrMsg fp y x s) = fp ++ ":" ++ show (y + 1) ++ ":" ++ show (x + 1) ++ ": " ++ s

parse :: FilePath -> String -> These [ErrMsg] (Program ())
parse = runParser pProgram

runParser :: Parser a -> FilePath -> String -> These [ErrMsg] a
runParser p fp source = 
    flip evalState (PS 0 0 0 source)
    . runChronicleT
    . flip runReaderT (Context fp)
    $ p
                            
pProgram :: Parser (Program ())
pProgram = do
    prog <- Program <$> many pFunDef
    skipWhiteComment
    assertEOF Error
    return prog

pFunDef :: Parser (FunDef ())
pFunDef = do
    skipWhiteComment
    mtypesig <- optional pStandaloneTypesig0
    _

pStandaloneTypesig0 :: Parser (Name, Type)
pStandaloneTypesig0 = do
    assertAtCol 0 Fatal "Expected top-level definition, found indented stuff"
    withRefCol 0 $ do
        name <- pIdentifier0 Lowercase
        inlineWhite
        string "::"
        ty <- pType
        return (name, ty)

pType :: Parser Type
pType = do
    ty1 <- pTypeApp
    asum [do inlineWhite
             string "->"
             ty2 <- pType
             return (TFun ty1 ty2)
         ,return ty1]

pTypeApp :: Parser Type
pTypeApp = many pTypeAtom >>= \case
    [] -> raise Error "Expected type" >> return (TTup [])
    [t] -> return t
    t:ts -> return (TApp t ts)

pTypeAtom :: Parser Type
pTypeAtom = pTypeParens <|> pTypeList <|> pTypeCon <|> pTypeVar
  where
    pTypeParens = do
        inlineWhite
        string "("
        asum [do inlineWhite
                 string ")"
                 return (TTup [])
             ,do ty1 <- pType
                 ty2s <- many $ do
                     inlineWhite
                     string ","
                     pType
                 inlineWhite
                 string ")"
                 case ty2s of
                   [] -> return ty1
                   _ -> return (TTup (ty1 : ty2s))]

    pTypeList = do
        inlineWhite
        string "["
        ty <- pType
        string "]"
        return (TList ty)

    pTypeCon = inlineWhite >> TCon <$> pIdentifier0 Uppercase
    pTypeVar = inlineWhite >> TVar <$> pIdentifier0 Lowercase

data Case = Uppercase | Lowercase
  deriving (Show)

-- | Consumes an identifier (word or parenthesised operator) at the current
-- position. The `var` production in Haskell2010.
-- var -> varid | "(" varsym ")"
pIdentifier0 :: Case -> Parser Name
pIdentifier0 cs = pAlphaName0 cs <|> pParens0 (pSymbol0 cs)

-- | Consumes a word-like name at the current position with the given case. The
-- `varid` production in Haskell2010 for 'Lowercase', `conid' for 'Uppercase'.
--
-- > varid -> (small {small | large | digit | "'"}) without reservedid
pAlphaName0 :: Case -> Parser Name
pAlphaName0 cs = do
    (_, s) <- readInline (\case True -> \case Just c | isInitNameChar c -> Just (Right False)
                                              _ -> Nothing
                                False -> \case Just c | isNameContChar c -> Just (Right False)
                                               _ -> Just (Left ()))
                         True
    name <- case cs of
      Uppercase | isLower (head s) -> do
          raise Error "Unexpected uppercase word at this position, assuming typo"
          return (toUpper (head s) : tail s)
      Lowercase | isUpper (head s) -> do
          raise Error "Unexpected lowercase word at this position, assuming typo"
          return (toLower (head s) : tail s)
      _ -> return s
    guard (name `notElem` ["case", "class", "data", "default", "deriving", "do", "else"
                          ,"foreign", "if", "import", "in", "infix", "infixl"
                          ,"infixr", "instance", "let", "module", "newtype", "of"
                          ,"then", "type", "where", "_"])
    return (Name name)
  where
    isInitNameChar, isNameContChar :: Char -> Bool
    isInitNameChar c = isLetter c || c == '_'
    isNameContChar c = isInitNameChar c || isDigit c || c == '\''

-- | Consumes a symbol at the current position. The `varsym` production in
-- Haskell2010 for 'Lowercase', `consym` otherwise.
--
-- > varsym -> ((symbol without ":") {symbol}) without (reservedop | dashes)
-- > consym -> (":" {symbol}) without reservedop
-- > symbol -> ascSymbol | uniSymbol without (special | "_" | "\"" | "'")
-- > ascSymbol -> ```!#$%&⋆+./<=>?@^|-~:```
-- > uniSymbol -> unicode symbol or punctuation
-- > dashes -> "--" {"-"}
-- > special -> ```(),;[]`{}```
-- > reservedop -> ".." | ":" | "::" | "=" | "\" | "|" | "<-" | "->" | "@" | "~" | "=>"
pSymbol0 :: Case -> Parser Name
pSymbol0 cs = do
    let isSpecialExt c = c `elem` "(),;[]`{}_\"'"
        isAscSymbol c = c `elem` "!#$%&⋆+./<=>?@^|-~:"
        isUniSymbol c = isSymbol c || isPunctuation c
        isSymbolChar c = (isAscSymbol c || isUniSymbol c) && not (isSpecialExt c)
    name <- (:) <$> (case cs of Lowercase -> satisfy (\c -> isSymbolChar c && c /= ':')
                                Uppercase -> satisfy (== ':'))
                <*> many (satisfy isSymbolChar)
    guard (name `notElem` ["..", ":", "::", "=", "\\", "|", "<-", "->", "@", "~", "=>"])
    guard (take 2 name /= "--")
    return (Name name)

-- | Parser between parens, with the opening paren at the current position.
-- Enforces that all components are within the current indented block.
pParens0 :: Parser a -> Parser a
pParens0 p = do
    string "("
    skipWhiteComment
    assertWithinBlock Error "Unexpected dedent after opening parenthesis"
    res <- p
    assertWithinBlock Error "Unexpected dedent in parenthesised expression"
    skipWhiteComment
    assertWithinBlock Error "Unexpected dedent in parenthesised expression"
    string ")"
    return res


-- | Run a parser under a modified psRefCol. The current psRefCol is reinstated
-- after completion of this parser.
withRefCol :: Int -> Parser a -> Parser a
withRefCol refcol p = do
    old <- gets psRefCol
    modify (\ps -> ps { psRefCol = refcol })
    res <- p
    modify (\ps -> ps { psRefCol = old })
    return res

data Fatality = Error | Fatal
  deriving (Show)

-- | Raise an error with the given fatality and description.
raise :: Fatality -> String -> Parser ()
raise fat msg = do
    fp <- asks ctxFile
    ps <- get
    let fun = case fat of
                Error -> dictate . pure
                Fatal -> confess . pure
    fun (ErrMsg fp (psLine ps) (psCol ps) msg)

-- | Raises an error if we're not currently at the given column.
assertAtCol :: Int -> Fatality -> String -> Parser ()
assertAtCol col fat msg = gets psCol >>= \col' ->
    when (col' /= col) $ raise fat msg

-- | Raises an error if psCol is not greater than psRefCol.
assertWithinBlock :: Fatality -> String -> Parser ()
assertWithinBlock fat msg = get >>= \ps ->
    when (psCol ps <= psRefCol ps) $ raise fat msg

-- | Raises an error if we're not currently at EOF.
assertEOF :: Fatality -> Parser ()
assertEOF fat = gets psRest >>= \case
    [] -> return ()
    _ -> raise fat "Unexpected stuff"

-- | Consumes an inline token at the current position, asserting that psCol >
-- psRefCol at the start. The token is defined by a pure stateful parser.
-- If encountering a newline or EOF, the parser is run on this character
-- ('Nothing' for EOF); if this produces a result, the result is returned;
-- otherwise, the parser fails.
readInline :: (s -> Maybe Char -> Maybe (Either r s)) -> s -> Parser (r, String)
readInline f s0 = do
    ps0 <- get
    when (psCol ps0 <= psRefCol ps0) $
        raise Fatal "Expected stuff, but found end of indented expression"
    let loop :: (s -> Maybe Char -> Maybe (Either r s)) -> s -> Parser (r, String)
        loop f' st = do
            ps <- get
            case psRest ps of
              []       | Just (Left res) <- f' st Nothing     -> return (res, "")
                       | otherwise -> empty
              '\n' : _ | Just (Left res) <- f' st (Just '\n') -> return (res, "")
              c : cs -> case f' st (Just c) of
                          Nothing -> empty
                          Just (Left res) -> return (res, "")
                          Just (Right st') -> do
                              put (ps { psCol = psCol ps + 1, psRest = cs })
                              fmap (c :) <$> loop f' st'
    loop f s0

-- | Consumes all whitespace and comments (including newlines), but only if
-- this then leaves psCol > psRefCol. If not, this fails.
inlineWhite :: Parser ()
inlineWhite = do
    skipWhiteComment
    ps <- get
    when (psCol ps <= psRefCol ps) empty

-- | Consumes all whitespace and comments (including newlines). Note: this may
-- leave psCol <= psRefCol.
skipWhiteComment :: Parser ()
skipWhiteComment = do
    inlineSpaces
    _ <- many (inlineComment >> inlineSpaces)
    _ <- optional lineComment
    (consumeNewline >> skipWhiteComment) <|> return ()

-- | Consumes some inline whitespace.
inlineSpaces :: Parser ()
inlineSpaces = readWhileInline isSpace

-- | Consumes an inline comment including both end markers. Note: this may
-- leave psCol < psRefCol.
inlineComment :: Parser ()
inlineComment = do
    string "{-"
    let loop = do
            readWhileInline (`notElem` "{-")
            asum [string "-}"
                 ,inlineComment >> loop
                 ,consumeNewline >> loop]
    loop

-- | Consumes a line comment marker and the rest of the line, excluding
-- newline.
lineComment :: Parser ()
lineComment = string "--" >> readWhileInline (const True)

-- | Consumes characters while the predicate holds or until (and excluding)
-- a newline, whichever comes first.
readWhileInline :: (Char -> Bool) -> Parser ()
readWhileInline p = do
    (taken, rest) <- span (\c -> p c && c /= '\n') <$> gets psRest
    modify (\ps -> ps { psCol = psCol ps + length taken
                      , psRest = rest })

-- | Consumes exactly one newline at the current position.
consumeNewline :: Parser ()
consumeNewline = gets psRest >>= \case
    '\n' : rest -> modify (\ps -> ps { psLine = psLine ps + 1
                                     , psCol = 0
                                     , psRest = rest })
    _ -> empty

-- | Consumes exactly one character, unequal to newline, at the current position.
satisfy :: (Char -> Bool) -> Parser Char
satisfy p = gets psRest >>= \case
    c : rest | c /= '\n', p c -> do
        modify (\ps -> ps { psCol = psCol ps + 1
                          , psRest = rest })
        return c
    _ -> empty

-- | Consumes exactly this string at the current position. The string must not
-- contain a newline.
string :: String -> Parser ()
string s | any (== '\n') s = error "Newline in 'string' argument"
string s = do
    ps <- get
    if take (length s) (psRest ps) == s
        then put (ps { psCol = psCol ps + length s
                     , psRest = drop (length s) (psRest ps) })
        else empty