From b1402e8bd6a68a891b00a214d6cf184d66defe19 Mon Sep 17 00:00:00 2001
From: aarne <unknown>
Date: Mon, 22 Sep 2003 13:16:55 +0000
Subject: Founding the newly structured GF2.0 cvs archive.

---
 src/GF/Canon/Unlex.hs | 37 +++++++++++++++++++++++++++++++++++++
 1 file changed, 37 insertions(+)
 create mode 100644 src/GF/Canon/Unlex.hs

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

diff --git a/src/GF/Canon/Unlex.hs b/src/GF/Canon/Unlex.hs
new file mode 100644
index 000000000..f665f4c85
--- /dev/null
+++ b/src/GF/Canon/Unlex.hs
@@ -0,0 +1,37 @@
+module Unlex where
+
+import Operations
+import Str
+
+import Char
+import List (isPrefixOf)
+
+-- elementary text postprocessing. AR 21/11/2001
+
+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  = (=="<p>") 
+
+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