mirror of
https://github.com/GrammaticalFramework/gf-core.git
synced 2026-04-15 15:59:32 -06:00
95 lines
2.4 KiB
Haskell
95 lines
2.4 KiB
Haskell
----------------------------------------------------------------------
|
|
-- |
|
|
-- Module : GF.Speech.GSL
|
|
--
|
|
-- This module prints a CFG as a Nuance GSL 2.0 grammar.
|
|
--
|
|
-----------------------------------------------------------------------------
|
|
|
|
module GF.Speech.GSL (gslPrinter) where
|
|
|
|
import GF.Data.Utilities
|
|
import GF.Speech.CFG
|
|
import GF.Speech.SRG
|
|
import GF.Speech.RegExp
|
|
import GF.Infra.Ident
|
|
import PGF.CId
|
|
import PGF.Data
|
|
|
|
import Data.Char (toUpper,toLower)
|
|
import Data.List (partition)
|
|
import Text.PrettyPrint.HughesPJ
|
|
|
|
width :: Int
|
|
width = 75
|
|
|
|
gslPrinter :: PGF -> CId -> String
|
|
gslPrinter pgf cnc = renderStyle st $ prGSL $ makeSimpleSRG pgf cnc
|
|
where st = style { lineLength = width }
|
|
|
|
prGSL :: SRG -> Doc
|
|
prGSL srg = header $++$ mainCat $++$ foldr ($++$) empty (map prRule (srgRules srg))
|
|
where
|
|
header = text ";GSL2.0" $$
|
|
comment ("Nuance speech recognition grammar for " ++ srgName srg) $$
|
|
comment ("Generated by GF")
|
|
mainCat = text ".MAIN" <+> prCat (srgStartCat srg)
|
|
prRule (SRGRule cat rhs) = prCat cat <+> union (map prAlt rhs)
|
|
-- FIXME: use the probability
|
|
prAlt (SRGAlt mp _ rhs) = prItem rhs
|
|
|
|
|
|
prItem :: SRGItem -> Doc
|
|
prItem = f
|
|
where
|
|
f (REUnion xs) = (if null es then empty else text "?") <> union (map f nes)
|
|
where (es,nes) = partition isEpsilon xs
|
|
f (REConcat [x]) = f x
|
|
f (REConcat xs) = text "(" <> fsep (map f xs) <> text ")"
|
|
f (RERepeat x) = text "*" <> f x
|
|
f (RESymbol s) = prSymbol s
|
|
|
|
union :: [Doc] -> Doc
|
|
union [x] = x
|
|
union xs = text "[" <> fsep xs <> text "]"
|
|
|
|
prSymbol :: Symbol SRGNT Token -> Doc
|
|
prSymbol = symbol (prCat . fst) (doubleQuotes . showToken)
|
|
|
|
-- GSL requires an upper case letter in category names
|
|
prCat :: Cat -> Doc
|
|
prCat = text . firstToUpper
|
|
|
|
|
|
firstToUpper :: String -> String
|
|
firstToUpper [] = []
|
|
firstToUpper (x:xs) = toUpper x : xs
|
|
|
|
{-
|
|
rmPunctCFG :: CGrammar -> CGrammar
|
|
rmPunctCFG g = [CFRule c (filter keepSymbol ss) n | CFRule c ss n <- g]
|
|
|
|
keepSymbol :: Symbol c Token -> Bool
|
|
keepSymbol (Tok t) = not (all isPunct (prt t))
|
|
keepSymbol _ = True
|
|
-}
|
|
|
|
-- Nuance does not like upper case characters in tokens
|
|
showToken :: Token -> Doc
|
|
showToken = text . map toLower
|
|
|
|
isPunct :: Char -> Bool
|
|
isPunct c = c `elem` "-_.:;.,?!()[]{}"
|
|
|
|
comment :: String -> Doc
|
|
comment s = text ";" <+> text s
|
|
|
|
|
|
-- Pretty-printing utilities
|
|
|
|
emptyLine :: Doc
|
|
emptyLine = text ""
|
|
|
|
($++$) :: Doc -> Doc -> Doc
|
|
x $++$ y = x $$ emptyLine $$ y
|