summaryrefslogtreecommitdiff
path: root/src/compiler/GF/Speech/GSL.hs
diff options
context:
space:
mode:
authorkrasimir <krasimir@chalmers.se>2009-12-13 18:50:29 +0000
committerkrasimir <krasimir@chalmers.se>2009-12-13 18:50:29 +0000
commitf85232947e74ee7ef8c7b0ad2338212e7e68f1be (patch)
tree667b886a5e3a4b026a63d4e3597f32497d824761 /src/compiler/GF/Speech/GSL.hs
parentd88a865faff59c98fc91556ff8700b10ee5f2df8 (diff)
reorganize the directories under src, and rescue the JavaScript interpreter from deprecated
Diffstat (limited to 'src/compiler/GF/Speech/GSL.hs')
-rw-r--r--src/compiler/GF/Speech/GSL.hs95
1 files changed, 95 insertions, 0 deletions
diff --git a/src/compiler/GF/Speech/GSL.hs b/src/compiler/GF/Speech/GSL.hs
new file mode 100644
index 000000000..8f26ea64c
--- /dev/null
+++ b/src/compiler/GF/Speech/GSL.hs
@@ -0,0 +1,95 @@
+----------------------------------------------------------------------
+-- |
+-- 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.Option
+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 :: Options -> PGF -> CId -> String
+gslPrinter opts pgf cnc = renderStyle st $ prGSL $ makeNonLeftRecursiveSRG opts 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