summaryrefslogtreecommitdiff
path: root/old-examples/uusisuomi/kotus/Kotus.hs
diff options
context:
space:
mode:
authoraarne <aarne@cs.chalmers.se>2008-06-25 16:54:35 +0000
committeraarne <aarne@cs.chalmers.se>2008-06-25 16:54:35 +0000
commite9e80fc389365e24d4300d7d5390c7d833a96c50 (patch)
treef0b58473adaa670bd8fc52ada419d8cad470ee03 /old-examples/uusisuomi/kotus/Kotus.hs
parentb96b36f43de3e2f8b58d5f539daa6f6d47f25870 (diff)
changed names of resource-1.3; added a note on homepage on release
Diffstat (limited to 'old-examples/uusisuomi/kotus/Kotus.hs')
-rw-r--r--old-examples/uusisuomi/kotus/Kotus.hs55
1 files changed, 55 insertions, 0 deletions
diff --git a/old-examples/uusisuomi/kotus/Kotus.hs b/old-examples/uusisuomi/kotus/Kotus.hs
new file mode 100644
index 000000000..ad63a1e14
--- /dev/null
+++ b/old-examples/uusisuomi/kotus/Kotus.hs
@@ -0,0 +1,55 @@
+import List
+import Char
+
+kotus = "sanat.xxmmll"
+
+main = do
+ ss <- readFile kotus >>= return . lines
+ let ws = [w | Just w <- map analyse ss]
+ writeFile "kotus.gf" $ unlines $ treat ws
+-- mapM putStrLn $ treat ws
+
+treat = map mkLin . entries
+
+entries = zip [10001..] . filter isNoun
+
+isNoun x = ((<5) . read . take 1 . fst) x && (all isAlpha . snd) x
+
+mkLin (n,(pa,ex)) =
+ "fun n" ++ show n ++ "_" ++ ex ++ " : N ;\n" ++
+ "lin n" ++ show n ++ "_" ++ ex ++ " = ud d" ++ pa ++ " \"" ++ ex ++ "\" ;"
+
+-- treat = map mkRule . paradigms
+
+mkRule ((pa,ex),nu) =
+ " " ++ pos ++ pa ++ " : Str -> " ++ poss ++
+ "Forms -- " ++ show nu ++ " " ++ ex ++ "\n = \\s -> ;"
+ where
+ (pos,poss) = if read (take 2 pa) < 52 then ("d","N") else ("c","V")
+
+-- paradigms = map info . groupByFst . sort
+
+info x = (last x, length x)
+--info = last
+
+-- groupByFst = groupBy (\ x y -> fst x == fst y)
+
+-- <st><s>aaloe</s><t><tn>3</tn></t></st>
+-- <st><s>vuoksi</s><hn>1</hn><t><tn>7</tn></t></st>
+-- <st><s>visiitti</s><t><tn>5</tn><av>C</av></t></st>
+
+analyse s =
+ let
+ rest = drop 7 s
+ (word,end) = span (/='<') rest
+ lst = drop 6 $ dropWhile (/='t') end
+ (num,gr) = span isDigit lst
+ para = (replicate (2 - length num) '0' ++ num) ++ ['A' | isPrefixOf "av" (drop 6 gr)]
+ in case num of
+ "" -> Nothing
+ "0" -> Nothing
+ _ | length num > 2 -> Nothing
+ _ -> if last word == 't' then Nothing else Just (para,word)
+
+sub cs s = isPrefixOf cs s || isPrefixOf cs (drop 1 s)
+