1
0
forked from GitHub/gf-core

"Committed_by_peb"

This commit is contained in:
peb
2005-04-14 17:38:36 +00:00
parent b63b29a247
commit 95c6e8a58f
11 changed files with 302 additions and 97 deletions

View File

@@ -5,9 +5,9 @@
-- Stability : (stable)
-- Portability : (portable)
--
-- > CVS $Date: 2005/04/11 13:53:38 $
-- > CVS $Date: 2005/04/14 18:38:36 $
-- > CVS $Author: peb $
-- > CVS $Revision: 1.15 $
-- > CVS $Revision: 1.16 $
--
-- This module prints a CFG as a Nuance GSL 2.0 grammar.
--
@@ -19,16 +19,20 @@ module PrGSL (gslPrinter) where
import SRG
import Ident
import GF.OldParsing.CFGrammar
import GF.OldParsing.Utilities (Symbol(..))
import GF.OldParsing.GrammarTypes
import GF.Printing.PrintParser
-- import GF.OldParsing.CFGrammar
-- import GF.OldParsing.Utilities (Symbol(..))
-- import GF.OldParsing.GrammarTypes
-- import GF.Printing.PrintParser
import GF.Formalism.CFG
import GF.Formalism.Utilities (Symbol(..))
import GF.Conversion.Types
import GF.Infra.Print
import Option
import Data.Char (toUpper,toLower)
gslPrinter :: Ident -- ^ Grammar name
-> Options -> CFGrammar -> String
-> Options -> CGrammar -> String
gslPrinter name opts cfg = prGSL srg ""
where srg = makeSRG name opts cfg
@@ -55,13 +59,13 @@ firstToUpper :: String -> String
firstToUpper [] = []
firstToUpper (x:xs) = toUpper x : xs
rmPunct :: [Symbol String Tokn] -> [Symbol String Tokn]
rmPunct :: [Symbol String Token] -> [Symbol String Token]
rmPunct [] = []
rmPunct (Tok t:ss) | all isPunct (prt t) = rmPunct ss
rmPunct (s:ss) = s : rmPunct ss
-- Nuance does not like upper case characters in tokens
showToken :: Tokn -> String
showToken :: Token -> String
showToken t = map toLower (prt t)
isPunct :: Char -> Bool

View File

@@ -5,9 +5,9 @@
-- Stability : (stable)
-- Portability : (portable)
--
-- > CVS $Date: 2005/04/11 13:53:39 $
-- > CVS $Date: 2005/04/14 18:38:36 $
-- > CVS $Author: peb $
-- > CVS $Revision: 1.9 $
-- > CVS $Revision: 1.10 $
--
-- This module prints a CFG as a JSGF grammar.
--
@@ -21,14 +21,18 @@ module PrJSGF (jsgfPrinter) where
import SRG
import Ident
import GF.OldParsing.CFGrammar
import GF.OldParsing.Utilities (Symbol(..))
import GF.OldParsing.GrammarTypes
import GF.Printing.PrintParser
-- import GF.OldParsing.CFGrammar
-- import GF.OldParsing.Utilities (Symbol(..))
-- import GF.OldParsing.GrammarTypes
-- import GF.Printing.PrintParser
import GF.Formalism.CFG
import GF.Formalism.Utilities (Symbol(..))
import GF.Conversion.Types
import GF.Infra.Print
import Option
jsgfPrinter :: Ident -- ^ Grammar name
-> Options -> CFGrammar -> String
-> Options -> CGrammar -> String
jsgfPrinter name opts cfg = prJSGF srg ""
where srg = makeSRG name opts cfg
@@ -53,7 +57,7 @@ prJSGF (SRG{grammarName=name,startCat=start,origStartCat=origStart,rules=rs})
prSymbol (Tok t) = wrap "\"" (prtS t) "\""
prCat c = showChar '<' . showString c . showChar '>'
rmPunct :: [Symbol String Tokn] -> [Symbol String Tokn]
rmPunct :: [Symbol String Token] -> [Symbol String Token]
rmPunct [] = []
rmPunct (Tok t:ss) | all isPunct (prt t) = rmPunct ss
rmPunct (s:ss) = s : rmPunct ss

View File

@@ -5,9 +5,9 @@
-- Stability : (stable)
-- Portability : (portable)
--
-- > CVS $Date: 2005/04/11 13:53:39 $
-- > CVS $Date: 2005/04/14 18:38:36 $
-- > CVS $Author: peb $
-- > CVS $Revision: 1.11 $
-- > CVS $Revision: 1.12 $
--
-- Representation of, conversion to, and utilities for
-- printing of a general Speech Recognition Grammar.
@@ -21,10 +21,14 @@
module SRG where
import Ident
import GF.OldParsing.CFGrammar
import GF.OldParsing.Utilities (Symbol(..))
import GF.OldParsing.GrammarTypes
import GF.Printing.PrintParser
-- import GF.OldParsing.CFGrammar
-- import GF.OldParsing.Utilities (Symbol(..))
-- import GF.OldParsing.GrammarTypes
-- import GF.Printing.PrintParser
import GF.Formalism.CFG
import GF.Formalism.Utilities (Symbol(..))
import GF.Conversion.Types
import GF.Infra.Print
import TransformCFG
import Option
@@ -40,7 +44,7 @@ data SRG = SRG { grammarName :: String -- ^ grammar name
}
data SRGRule = SRGRule String String [SRGAlt] -- ^ SRG category name, original category name
-- and productions
type SRGAlt = [Symbol String Tokn]
type SRGAlt = [Symbol String Token]
-- | SRG category name and original name
type CatName = (String,String)
@@ -49,7 +53,7 @@ type CatNames = FiniteMap String String
makeSRG :: Ident -- ^ Grammar name
-> Options -- ^ Grammar options
-> CFGrammar -- ^ A context-free grammar
-> CGrammar -- ^ A context-free grammar
-> SRG
makeSRG i opts gr = SRG { grammarName = name,
startCat = start,
@@ -71,11 +75,11 @@ cfgRulesToSRGRule names rs@(r:_) = SRGRule cat origCat rhs
renameCat (Cat c) = Cat (lookupFM_ names c)
renameCat t = t
ruleCat :: Rule n c t -> c
ruleCat (Rule c _ _) = c
ruleCat :: CFRule c n t -> c
ruleCat (CFRule c _ _) = c
ruleRhs :: Rule n c t -> [Symbol c t]
ruleRhs (Rule _ r _) = r
ruleRhs :: CFRule c n t -> [Symbol c t]
ruleRhs (CFRule _ r _) = r
mkCatNames :: String -- ^ Category name prefix
-> [String] -- ^ Original category names

View File

@@ -5,22 +5,28 @@
-- Stability : (stable)
-- Portability : (portable)
--
-- > CVS $Date: 2005/04/11 13:53:39 $
-- > CVS $Date: 2005/04/14 18:38:36 $
-- > CVS $Author: peb $
-- > CVS $Revision: 1.9 $
-- > CVS $Revision: 1.10 $
--
-- This module does some useful transformations on CFGs.
--
-- FIXME: remove cycles
--
-- peb thinks: most of this module should be moved to GF.Conversion...
-----------------------------------------------------------------------------
module TransformCFG (makeNice, CFRule_) where
import Ident
import GF.OldParsing.CFGrammar
import GF.OldParsing.Utilities (Symbol(..))
import GF.OldParsing.GrammarTypes
import GF.Printing.PrintParser
-- import GF.OldParsing.CFGrammar
-- import GF.OldParsing.Utilities (Symbol(..))
-- import GF.OldParsing.GrammarTypes
-- import GF.Printing.PrintParser
import GF.Formalism.CFG
import GF.Formalism.Utilities (Symbol(..), mapSymbol)
import GF.Conversion.Types
import GF.Infra.Print
import Data.FiniteMap
import Data.List
@@ -30,63 +36,65 @@ import Debug.Trace
-- | not very nice to get replace the structured CFCat type with a simple string
type CFRule_ = Rule CFName String Tokn
type CFRule_ = CFRule Cat_ Name Token
type Cat_ = String
type CFRules = FiniteMap String [CFRule_]
type CFRules = FiniteMap Cat_ [CFRule_]
makeNice :: CFGrammar -> [CFRule_]
makeNice :: CGrammar -> [CFRule_]
makeNice = concat . eltsFM . makeNice' . groupProds . cfgToCFRules
where makeNice' = removeLeftRecursion . removeEmptyCats
cfgToCFRules :: CFGrammar -> [CFRule_]
cfgToCFRules cfg = [Rule (catToString c) (map symb r) n | Rule c r n <- cfg]
where symb (Cat c) = Cat (catToString c)
symb (Tok t) = Tok t
cfgToCFRules :: CGrammar -> [CFRule_]
cfgToCFRules cfg = [CFRule (catToString c) (map symb r) n | CFRule c r n <- cfg]
where symb = mapSymbol catToString id
-- symb (Cat c) = Cat (catToString c)
-- symb (Tok t) = Tok t
catToString = prt
-- | Group productions by their lhs categories
groupProds :: [CFRule_] -> CFRules
groupProds = addListToFM_C (++) emptyFM . map (\rs -> (ruleCat rs,[rs]))
where ruleCat (Rule c _ _) = c
where ruleCat (CFRule c _ _) = c
-- | Remove productions which use categories which have no productions
removeEmptyCats :: CFRules -> CFRules
removeEmptyCats rss = listToFM $ fix removeEmptyCats' $ fmToList rss
where
removeEmptyCats' :: [(String,[CFRule_])] -> [(String,[CFRule_])]
removeEmptyCats' :: [(Cat_,[CFRule_])] -> [(Cat_,[CFRule_])]
removeEmptyCats' rs = k'
where
keep = filter (not . null . snd) rs
allCats = nub [c | (_,r) <- rs, Rule _ rhs _ <- r, Cat c <- rhs]
allCats = nub [c | (_,r) <- rs, CFRule _ rhs _ <- r, Cat c <- rhs]
emptyCats = filter (nothingOrNull . flip lookup rs) allCats
k' = map (\ (c,xs) -> (c, filter (not . anyUsedBy emptyCats) xs)) keep
anyUsedBy :: [String] -> CFRule_ -> Bool
anyUsedBy ss (Rule _ r _) = or [c `elem` ss | Cat c <- r]
anyUsedBy :: [Cat_] -> CFRule_ -> Bool
anyUsedBy ss (CFRule _ r _) = or [c `elem` ss | Cat c <- r]
removeLeftRecursion :: CFRules -> CFRules
removeLeftRecursion rs = listToFM $ concatMap removeDirectLeftRecursion $ map handleProds $ fmToList rs
where
handleProds (c, r) = (c, concatMap handleProd r)
handleProd (Rule ai (Cat aj:alpha) n) | aj < ai =
handleProd (CFRule ai (Cat aj:alpha) n) | aj < ai =
-- FIXME: this will give multiple rules with the same name
[Rule ai (beta ++ alpha) n | Rule _ beta _ <- fromJust (lookupFM rs aj)]
[CFRule ai (beta ++ alpha) n | CFRule _ beta _ <- fromJust (lookupFM rs aj)]
handleProd r = [r]
removeDirectLeftRecursion :: (String,[CFRule_]) -- ^ All productions for a category
-> [(String,[CFRule_])]
removeDirectLeftRecursion :: (Cat_,[CFRule_]) -- ^ All productions for a category
-> [(Cat_,[CFRule_])]
removeDirectLeftRecursion (a,rs) | null dr = [(a,rs)]
| otherwise = [(a, as), (a', a's)]
where
a' = a ++ "'" -- FIXME: this might not be unique
(dr,nr) = partition isDirectLeftRecursive rs
as = maybeEndWithA' nr
is = [Rule a' (tail r) n | Rule _ r n <- dr]
is = [CFRule a' (tail r) n | CFRule _ r n <- dr]
a's = maybeEndWithA' is
maybeEndWithA' xs = xs ++ [Rule c (r++[Cat a']) n | Rule c r n <- xs]
maybeEndWithA' xs = xs ++ [CFRule c (r++[Cat a']) n | CFRule c r n <- xs]
isDirectLeftRecursive :: CFRule_ -> Bool
isDirectLeftRecursive (Rule c (Cat c':_) _) = c == c'
isDirectLeftRecursive (CFRule c (Cat c':_) _) = c == c'
isDirectLeftRecursive _ = False