From 055c0d0d5a5bb0dc75904fe53df7f2e4f5732a8f Mon Sep 17 00:00:00 2001
From: aarne <aarne@cs.chalmers.se>
Date: Wed, 21 May 2008 09:26:44 +0000
Subject: GF/src is now for 2.9, and the new sources are in src-3.0 - keep it
 this way until the release of GF 3

---
 src-3.0/GF/Canon/Unlex.hs | 49 +++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 49 insertions(+)
 create mode 100644 src-3.0/GF/Canon/Unlex.hs

(limited to 'src-3.0/GF/Canon/Unlex.hs')

diff --git a/src-3.0/GF/Canon/Unlex.hs b/src-3.0/GF/Canon/Unlex.hs
new file mode 100644
index 000000000..dd93390e2
--- /dev/null
+++ b/src-3.0/GF/Canon/Unlex.hs
@@ -0,0 +1,49 @@
+----------------------------------------------------------------------
+-- |
+-- Module      : Unlex
+-- Maintainer  : AR
+-- Stability   : (stable)
+-- Portability : (portable)
+--
+-- > CVS $Date: 2005/04/21 16:21:32 $ 
+-- > CVS $Author: bringert $
+-- > CVS $Revision: 1.8 $
+--
+-- elementary text postprocessing. AR 21/11/2001
+-----------------------------------------------------------------------------
+
+module GF.Canon.Unlex (formatAsText, unlex, performBinds) where
+
+import GF.Data.Operations
+import GF.Data.Str
+
+import Data.Char
+import Data.List (isPrefixOf)
+
+formatAsText :: String -> String
+formatAsText = unwords . format . cap . words where
+  format ws = case ws of
+    w : c : ww | major c -> (w ++ c)      : format (cap ww)
+    w : c : ww | minor c -> (w ++ c)      : format ww
+    c     : ww | para  c -> "\n\n"        : format ww
+    w     : ww           -> w             : format ww
+    [] -> []
+  cap (p:(c:cs):ww) | para p = p : (toUpper c : cs) : ww
+  cap ((c:cs):ww) = (toUpper c : cs) : ww
+  cap [] = []
+  major = flip elem (map (:[]) ".!?") 
+  minor = flip elem (map (:[]) ",:;")
+  para  = (=="&-") 
+
+unlex :: [Str] -> String
+unlex = formatAsText . performBinds . concat . map sstr . take 1 ----
+
+-- | modified from GF/src/Text by adding hyphen
+performBinds :: String -> String
+performBinds = unwords . format . words where
+  format ws = case ws of
+    w : "-"  : u : ws -> format ((w ++ "-" ++ u) : ws)
+    w : "&+" : u : ws -> format ((w ++ u) : ws)
+    w : ws            -> w : format ws
+    []                -> []
+
-- 
cgit v1.2.3