msyds/gf-rgl
1
0
Fork 0
forked from GitHub/gf-rgl
Code Pull Requests Activity

extraction of adjectives in Icelandic wiktionary

Browse Source
This commit is contained in:
Aarne Ranta
2019-08-08 12:11:38 +02:00
parent 4263f57d7e
commit 892d59c89f
1 changed files with 28 additions and 15 deletions
Download Patch File Download Diff File Expand all files Collapse all files

43
src/icelandic/wiktionary/CheckWithWiktionary.hs
View File

@@ -138,29 +138,32 @@ actInf v@(i:_) = last (words i)
jumpToIcelandic ls = dropWhile (\l -> not (isPrefixOf "<h2>" l && isPrefixOf "Icelandic" (untag l))) ls
-------------------------------
-- just retrieving ------------
----------------------------------------------------------------------
-- just retrieving, instead of checking existing GF files ------------
-- to be run in verbs/
getAllWiktVerbs = do
vs <- readFile "wikt-verbs.txt" >>= return . lines
writeFile "v.tmp" ""
mapM_ (\v -> getWiktVerb v >>= appendFile "v.tmp" . unlines . emitGF) vs
-- to be run in wiktionary/, with subdirs nouns/ adjectives/ verbs/
-- to be run in adjectives/
getAllWiktAdjectives = do
vs <- readFile "wikt-verbs.txt" >>= return . lines
mapM_ (\v -> getWiktAdjective v >>= putStrLn . unlines . emitGF) vs
vs <- readFile "adjectives/wikt-adjectives.txt" >>= return . lines
writeFile "a.tmp" ""
mapM_ (\v -> getWiktAdjective "adjectives/" v >>= appendFile "a.tmp" . unlines . emitGF) vs
getAllWiktVerbs = do
vs <- readFile "verbs/wikt-verbs.txt" >>= return . lines
writeFile "v.tmp" ""
mapM_ (\v -> getWiktVerb "verbs/" v >>= appendFile "v.tmp" . unlines . emitGF) vs
-- return ([relevant Wikt lines], (fun,cat,lin), message)
getWiktWord :: Int -> (String -> [String] -> ([String],((String,String,String),Message))) -> FilePath -> IO ([String],((String,String,String),Message))
getWiktWord number check file = do
s <- readFile file >>= return . map untag . take number . getTD . jumpToIcelandic . lines
getWiktWord :: Int -> (String -> [String] -> ([String],((String,String,String),Message))) -> FilePath -> FilePath -> IO ([String],((String,String,String),Message))
getWiktWord number check dir file = do
s <- readFile (dir++file) >>= return . map untag . take number . getTD . jumpToIcelandic . lines
return $ check file s
getWiktNoun = getWiktWord 17 checkNoun
getWiktAdjective = getWiktWord 120 noCheck
getWiktAdjective = getWiktWord 120 checkAdjective
getWiktVerb = getWiktWord 75 checkVerb
noCheck :: String -> [String] -> ([String],((String,String,String),Message))
@@ -169,12 +172,22 @@ noGF = ("--","--","--")
checkNoun noun forms = noCheck noun forms ----
checkAdjective adj forms = case length forms of
n | n < 24 -> (forms, (noGF, MBad (adj ++ " A: only " ++ show (length forms) ++ " lines")))
n | n < 120 -> (forms, checkZZ adj (adj ++ "_A", "A", "mkA", [forms!!0, forms!!1]))
_ -> (forms, checkZZ adj (adj ++ "_A", "A", "mkA", [forms!!0, forms!!1,forms!!48]))
checkZZ w (fun,cat,lin,args) =
if elem "ZZ" (map (wform . words) args)
then (noGF, MBad (w ++ " " ++ cat ++ ": missing forms in data"))
else ((fun,cat, app lin args),MGood w)
checkVerb verb forms =
if length forms < 75
then (forms, (noGF, MBad (verb ++ ": only " ++ show (length forms) ++ " lines")))
then (forms, (noGF, MBad (verb ++ " V: only " ++ show (length forms) ++ " lines")))
else case unexpectedWikLines forms of
[] -> (forms, ((verb ++ "_V", "V", app "mkV" [verb, forms!!5, forms!!18, forms!!74, forms!!1]), MGood verb)) ----
us -> (forms, (noGF, MBad (verb ++ ": unexpected lines " ++ show (length us))))
us -> (forms, (noGF, MBad (verb ++ " V: unexpected lines " ++ show (length us))))
data Message =
MGood String