8906d99 · elk · git.iko.soy

git.iko.soy/elk

on master help sign in

Move string parsing to Parse.String and use the new Elm.String type

8906d99Evan Czaplicki7 years ago

compiler/src/Parse/Utf8.hs → compiler/src/Parse/String.hsrenamed+34−33

compare with master

unifiedsplit

@@ -1,6 +1,6 @@

1{-# OPTIONS_GHC -Wall -fno-warn-unused-do-bind #-}

2{-# LANGUAGE BangPatterns, MagicHash, OverloadedStrings, UnboxedTuples #-}

3−module Parse.Utf8

3+module Parse.String

4 ( string

5 , character

6 )

@@ -11,6 +11,7 @@ import qualified Data.Utf8 as Utf8

11import Data.Word (Word8, Word16)

12import Foreign.Ptr (Ptr, plusPtr, minusPtr)

14+import qualified Elm.String as ES

14import Parse.Primitives (Parser, Row, Col)

15import Parse.Primitives (Parser, Row, Col)

15import qualified Parse.Number as Number

16import qualified Parse.Number as Number

16import qualified Parse.Primitives as P

17import qualified Parse.Primitives as P

@@ -21,9 +22,9 @@ import qualified Reporting.Error.Syntax as E

21-- CHARACTER

22-- CHARACTER

24−character :: (Row -> Col -> x) -> (E.Char -> Row -> Col -> x) -> Parser x Utf8.String

25+character :: (Row -> Col -> x) -> (E.Char -> Row -> Col -> x) -> Parser x ES.String

25character toExpectation toError =

26character toExpectation toError =

26− P.Parser $ \(P.State pos end indent row col) cok _ cerr eerr ->

27+ P.Parser $ \(P.State src pos end indent row col) cok _ cerr eerr ->

27 if pos >= end || P.unsafeIndex pos /= 0x27 {- ' -} then

28 if pos >= end || P.unsafeIndex pos /= 0x27 {- ' -} then

28 eerr row col toExpectation

29 eerr row col toExpectation

@@ -34,8 +35,8 @@ character toExpectation toError =

34 cerr row col (toError (E.CharNotString (fromIntegral (newCol - col))))

35 cerr row col (toError (E.CharNotString (fromIntegral (newCol - col))))

35 else

36 else

36 let

37 let

37− !newState = P.State newPos end indent row newCol

38+ !newState = P.State src newPos end indent row newCol

38− !char = Utf8.fromChunks [mostRecent]

39+ !char = ES.fromChunks [mostRecent]

39 in

40 in

40 cok char newState

41 cok char newState

@@ -47,12 +48,12 @@ character toExpectation toError =

49data CharResult

50data CharResult

50− = Good (Ptr Word8) Col Word16 Utf8.Chunk

51+ = Good (Ptr Word8) Col Word16 ES.Chunk

51 | CharEndless Col

52 | CharEndless Col

52 | CharEscape Row Col E.Escape

53 | CharEscape Row Col E.Escape

55−chompChar :: Ptr Word8 -> Ptr Word8 -> Row -> Col -> Word16 -> Utf8.Chunk -> CharResult

56+chompChar :: Ptr Word8 -> Ptr Word8 -> Row -> Col -> Word16 -> ES.Chunk -> CharResult

56chompChar pos end row col numChars mostRecent =

57chompChar pos end row col numChars mostRecent =

57 if pos >= end then

58 if pos >= end then

58 CharEndless col

59 CharEndless col

@@ -73,10 +74,10 @@ chompChar pos end row col numChars mostRecent =

73 else if word == 0x5C {- \ -} then

74 else if word == 0x5C {- \ -} then

74 case eatEscape (plusPtr pos 1) end row col of

75 case eatEscape (plusPtr pos 1) end row col of

75 EscapeNormal ->

76 EscapeNormal ->

76− chompChar (plusPtr pos 2) end row (col + 2) (numChars + 1) (Utf8.Slice pos 2)

77+ chompChar (plusPtr pos 2) end row (col + 2) (numChars + 1) (ES.Slice pos 2)

78 EscapeUnicode delta code ->

79 EscapeUnicode delta code ->

79− chompChar (plusPtr pos delta) end row (col + fromIntegral delta) (numChars + 1) (Utf8.CodePoint code)

80+ chompChar (plusPtr pos delta) end row (col + fromIntegral delta) (numChars + 1) (ES.CodePoint code)

81 EscapeProblem r c badEscape ->

82 EscapeProblem r c badEscape ->

82 CharEscape r c badEscape

83 CharEscape r c badEscape

@@ -89,16 +90,16 @@ chompChar pos end row col numChars mostRecent =

89 !width = P.getCharWidth pos end word

90 !width = P.getCharWidth pos end word

90 !newPos = plusPtr pos width

91 !newPos = plusPtr pos width

91 in

92 in

92− chompChar newPos end row (col + 1) (numChars + 1) (Utf8.Slice pos width)

93+ chompChar newPos end row (col + 1) (numChars + 1) (ES.Slice pos width)

96-- STRINGS

97-- STRINGS

99−string :: (Row -> Col -> x) -> (E.String -> Row -> Col -> x) -> Parser x Utf8.String

100+string :: (Row -> Col -> x) -> (E.String -> Row -> Col -> x) -> Parser x ES.String

100string toExpectation toError =

101string toExpectation toError =

101− P.Parser $ \(P.State pos end indent row col) cok _ cerr eerr ->

102+ P.Parser $ \(P.State src pos end indent row col) cok _ cerr eerr ->

102 if isDoubleQuote pos end then

103 if isDoubleQuote pos end then

103

104

104 let

105 let

@@ -121,7 +122,7 @@ string toExpectation toError =

121 Ok newPos newRow newCol utf8 ->

122 Ok newPos newRow newCol utf8 ->

122 let

123 let

123 !newState =

124 !newState =

124− P.State newPos end indent newRow newCol

125+ P.State src newPos end indent newRow newCol

125 in

126 in

126 cok utf8 newState

127 cok utf8 newState

127

128

@@ -139,32 +140,32 @@ isDoubleQuote pos end =

139

140

141

141data StringResult

142data StringResult

142− = Ok (Ptr Word8) Row Col !Utf8.String

143+ = Ok (Ptr Word8) Row Col !ES.String

143 | Err Row Col E.String

144 | Err Row Col E.String

144

145

146

146−finalize :: Ptr Word8 -> Ptr Word8 -> [Utf8.Chunk] -> Utf8.String

147+finalize :: Ptr Word8 -> Ptr Word8 -> [ES.Chunk] -> ES.String

147finalize start end revChunks =

148finalize start end revChunks =

148− Utf8.fromChunks $ reverse $

149+ ES.fromChunks $ reverse $

149 if start == end then

150 if start == end then

150 revChunks

151 revChunks

151 else

152 else

152− Utf8.Slice start (minusPtr end start) : revChunks

153+ ES.Slice start (minusPtr end start) : revChunks

153

154

155

155−addEscape :: Utf8.Chunk -> Ptr Word8 -> Ptr Word8 -> [Utf8.Chunk] -> [Utf8.Chunk]

156+addEscape :: ES.Chunk -> Ptr Word8 -> Ptr Word8 -> [ES.Chunk] -> [ES.Chunk]

156addEscape chunk start end revChunks =

157addEscape chunk start end revChunks =

157 if start == end then

158 if start == end then

158 chunk : revChunks

159 chunk : revChunks

159 else

160 else

160− chunk : Utf8.Slice start (minusPtr end start) : revChunks

161+ chunk : ES.Slice start (minusPtr end start) : revChunks

161

162

163

164

164-- SINGLE STRINGS

165-- SINGLE STRINGS

165

166

167

167−singleString :: Ptr Word8 -> Ptr Word8 -> Word16 -> Word16 -> Ptr Word8 -> [Utf8.Chunk] -> StringResult

168+singleString :: Ptr Word8 -> Ptr Word8 -> Word16 -> Word16 -> Ptr Word8 -> [ES.Chunk] -> StringResult

168singleString pos end row col initialPos revChunks =

169singleString pos end row col initialPos revChunks =

169 if pos >= end then

170 if pos >= end then

170 Err row col E.StringEndless_Single

171 Err row col E.StringEndless_Single

@@ -193,7 +194,7 @@ singleString pos end row col initialPos revChunks =

193 EscapeUnicode delta code ->

194 EscapeUnicode delta code ->

194 let !newPos = plusPtr pos delta in

195 let !newPos = plusPtr pos delta in

195 singleString newPos end row (col + fromIntegral delta) newPos $

196 singleString newPos end row (col + fromIntegral delta) newPos $

196− addEscape (Utf8.CodePoint code) initialPos pos revChunks

197+ addEscape (ES.CodePoint code) initialPos pos revChunks

197

198

198 EscapeProblem r c x ->

199 EscapeProblem r c x ->

199 Err r c (E.StringEscape x)

200 Err r c (E.StringEscape x)

@@ -210,7 +211,7 @@ singleString pos end row col initialPos revChunks =

210-- MULTI STRINGS

211-- MULTI STRINGS

211

212

213

213−multiString :: Ptr Word8 -> Ptr Word8 -> Word16 -> Word16 -> Ptr Word8 -> Word16 -> Word16 -> [Utf8.Chunk] -> StringResult

214+multiString :: Ptr Word8 -> Ptr Word8 -> Word16 -> Word16 -> Ptr Word8 -> Word16 -> Word16 -> [ES.Chunk] -> StringResult

214multiString pos end row col initialPos sr sc revChunks =

215multiString pos end row col initialPos sr sc revChunks =

215 if pos >= end then

216 if pos >= end then

216 Err sr sc E.StringEndless_Multi

217 Err sr sc E.StringEndless_Multi

@@ -244,7 +245,7 @@ multiString pos end row col initialPos sr sc revChunks =

244 EscapeUnicode delta code ->

245 EscapeUnicode delta code ->

245 let !newPos = plusPtr pos delta in

246 let !newPos = plusPtr pos delta in

246 multiString newPos end row (col + fromIntegral delta) newPos sr sc $

247 multiString newPos end row (col + fromIntegral delta) newPos sr sc $

247− addEscape (Utf8.CodePoint code) initialPos pos revChunks

248+ addEscape (ES.CodePoint code) initialPos pos revChunks

248

249

249 EscapeProblem r c x ->

250 EscapeProblem r c x ->

250 Err r c (E.StringEscape x)

251 Err r c (E.StringEscape x)

@@ -313,30 +314,30 @@ eatUnicode pos end row col =

313

314

315

315{-# NOINLINE singleQuote #-}

316{-# NOINLINE singleQuote #-}

316−singleQuote :: Utf8.Chunk

317+singleQuote :: ES.Chunk

317singleQuote =

318singleQuote =

318− Utf8.Escape 0x27 {- ' -}

319+ ES.Escape 0x27 {-'-}

319

320

321

321{-# NOINLINE doubleQuote #-}

322{-# NOINLINE doubleQuote #-}

322−doubleQuote :: Utf8.Chunk

323+doubleQuote :: ES.Chunk

323doubleQuote =

324doubleQuote =

324− Utf8.Escape 0x22 {- " -}

325+ ES.Escape 0x22 {-"-}

325

326

327

327{-# NOINLINE newline #-}

328{-# NOINLINE newline #-}

328−newline :: Utf8.Chunk

329+newline :: ES.Chunk

329newline =

330newline =

330− Utf8.Escape 0x6E {- n -}

331+ ES.Escape 0x6E {-n-}

331

332

333

333{-# NOINLINE carriageReturn #-}

334{-# NOINLINE carriageReturn #-}

334−carriageReturn :: Utf8.Chunk

335+carriageReturn :: ES.Chunk

335carriageReturn =

336carriageReturn =

336− Utf8.Escape 0x72 {- \r -}

337+ ES.Escape 0x72 {-r-}

337

338

339

339{-# NOINLINE placeholder #-}

340{-# NOINLINE placeholder #-}

340−placeholder :: Utf8.Chunk

341+placeholder :: ES.Chunk

341placeholder =

342placeholder =

342− Utf8.CodePoint 0xFFFD {- replacement character -}

343+ ES.CodePoint 0xFFFD {-replacement character-}