1
0
forked from GitHub/gf-core

new unlexers ; work on Romance

This commit is contained in:
aarne
2005-03-31 14:47:42 +00:00
parent 70deda4132
commit 0151c5c418
2 changed files with 22 additions and 12 deletions

View File

@@ -5,9 +5,9 @@
-- Stability : (stable) -- Stability : (stable)
-- Portability : (portable) -- Portability : (portable)
-- --
-- > CVS $Date: 2005/02/18 19:21:16 $ -- > CVS $Date: 2005/03/31 15:47:43 $
-- > CVS $Author: peb $ -- > CVS $Author: aarne $
-- > CVS $Revision: 1.6 $ -- > CVS $Revision: 1.7 $
-- --
-- elementary text postprocessing. AR 21\/11\/2001. -- elementary text postprocessing. AR 21\/11\/2001.
-- --
@@ -22,6 +22,8 @@ module Text (untokWithXML,
formatAsTextLit, formatAsTextLit,
formatAsCodeLit, formatAsCodeLit,
formatAsText, formatAsText,
formatAsHTML,
formatAsLatex,
formatAsCode, formatAsCode,
performBinds, performBinds,
unStringLit, unStringLit,
@@ -56,20 +58,26 @@ formatAsTextLit = formatAsText . unwords . map unStringLit . words
formatAsCodeLit :: String -> String formatAsCodeLit :: String -> String
formatAsCodeLit = formatAsCode . unwords . map unStringLit . words formatAsCodeLit = formatAsCode . unwords . map unStringLit . words
formatAsText :: String -> String formatAsText,formatAsHTML,formatAsLatex :: String -> String
formatAsText = unwords . format . cap . words where formatAsText = formatAsTextGen (=="&-") (=="&-")
formatAsHTML = formatAsTextGen ((=="<") . take 1) (const False)
formatAsLatex = formatAsTextGen ((=="\\") . take 1) (const False)
formatAsTextGen :: (String -> Bool) -> (String -> Bool) -> String -> String
formatAsTextGen tag para = unwords . format . cap . words where
format ws = case ws of format ws = case ws of
w : c : ww | major c -> (w ++ c) : format (cap ww) w : c : ww | major c -> format $ (w ++ c) :(cap ww)
w : c : ww | minor c -> (w ++ c) : format ww w : c : ww | minor c -> format $ (w ++ c) : ww
p : c : ww | openp p -> format $ (p ++ c) :ww
c : ww | para c -> "\n\n" : format ww c : ww | para c -> "\n\n" : format ww
w : ww -> w : format ww w : ww -> w : format ww
[] -> [] [] -> []
cap (p:(c:cs):ww) | para p = p : (toUpper c : cs) : ww cap (p:ww) | tag p = p : cap ww
cap ((c:cs):ww) = (toUpper c : cs) : ww cap ((c:cs):ww) = (toUpper c : cs) : ww
cap [] = [] cap [] = []
major = flip elem (map singleton ".!?") major = flip elem (map singleton ".!?")
minor = flip elem (map singleton ",:;") minor = flip elem (map singleton ",:;)")
para = (=="&-") openp = all (flip elem "(")
formatAsCode :: String -> String formatAsCode :: String -> String
formatAsCode = rend 0 . words where formatAsCode = rend 0 . words where

View File

@@ -5,9 +5,9 @@
-- Stability : (stable) -- Stability : (stable)
-- Portability : (portable) -- Portability : (portable)
-- --
-- > CVS $Date: 2005/03/29 13:26:37 $ -- > CVS $Date: 2005/03/31 15:47:43 $
-- > CVS $Author: aarne $ -- > CVS $Author: aarne $
-- > CVS $Revision: 1.49 $ -- > CVS $Revision: 1.50 $
-- --
-- A database for customizable GF shell commands. -- A database for customizable GF shell commands.
-- --
@@ -376,6 +376,8 @@ customUntokenizer =
[ [
(strCI "unwords", const $ id) -- DEFAULT (strCI "unwords", const $ id) -- DEFAULT
,(strCI "text", const $ formatAsText) ,(strCI "text", const $ formatAsText)
,(strCI "html", const $ formatAsHTML)
,(strCI "latex", const $ formatAsLatex)
,(strCI "code", const $ formatAsCode) ,(strCI "code", const $ formatAsCode)
,(strCI "concat", const $ filter (not . isSpace)) ,(strCI "concat", const $ filter (not . isSpace))
,(strCI "textlit", const $ formatAsTextLit) ,(strCI "textlit", const $ formatAsTextLit)