summaryrefslogtreecommitdiff
path: root/treebanks/susanne/SusanneFormat.hs
diff options
context:
space:
mode:
authorkr.angelov <kr.angelov@gmail.com>2013-12-05 10:05:33 +0000
committerkr.angelov <kr.angelov@gmail.com>2013-12-05 10:05:33 +0000
commit05854280181f5ad30939a4fc533d42560103d23f (patch)
tree124ef170431dbc6455a25824a8752e7ce81fbf20 /treebanks/susanne/SusanneFormat.hs
parent106b41a2cb562b26a477af5372ddf09fe8f79ca7 (diff)
more on the Susanne treebank
Diffstat (limited to 'treebanks/susanne/SusanneFormat.hs')
-rw-r--r--treebanks/susanne/SusanneFormat.hs85
1 files changed, 62 insertions, 23 deletions
diff --git a/treebanks/susanne/SusanneFormat.hs b/treebanks/susanne/SusanneFormat.hs
index 052f95978..3eb3187e2 100644
--- a/treebanks/susanne/SusanneFormat.hs
+++ b/treebanks/susanne/SusanneFormat.hs
@@ -1,43 +1,82 @@
-module SusanneFormat(Tag,Id,Word,Lemma,ParseTree(..),readTreebank) where
+module SusanneFormat(Tag,Id,Word,Lemma,ParseTree(..),readTreebank,readTag) where
+import PGF(CId)
import Data.Char
type Tag = String
+type Mods = String
+type Fn = String
+type Index = Int
type Id = String
type Word = String
type Lemma = String
data ParseTree
- = Phrase Tag [ParseTree]
+ = Phrase Tag Mods Fn Index [ParseTree]
| Word Id Tag Word Lemma
+ | App CId [ParseTree]
data ParseTreePos
= Root
- | At ParseTreePos Tag [ParseTree]
+ | At ParseTreePos ([ParseTree] -> ParseTree) [ParseTree]
instance Show ParseTree where
- show (Phrase tag ts) = "["++tag++" "++unwords (map show ts)++"]"
- show (Word _ tag w _) = "["++tag++" "++w++"]"
+ show (Phrase tag mods fn idx ts)
+ | tag == "" = "["++fn++show idx++" "++unwords (map show ts)++"]"
+ | fn == "" && idx == 0 = "["++tag++mods++" "++unwords (map show ts)++"]"
+ | otherwise = "["++tag++mods++":"++fn++show idx++" "++unwords (map show ts)++"]"
+ show (Word _ tag w _) = "["++tag++" "++w++"]"
+ show (App f ts)
+ | null ts = show f
+ | otherwise = "("++show f++" "++unwords (map show ts)++")"
readTreebank ls = readLines Root (map words ls)
readLines p [] = []
readLines p ([id,_,tag,w,l,parse]:ls) =
- readParse p (Word id tag w l) parse ls
-
-readParse p w [] ls = readLines p ls
-readParse p w ('[':cs) ls =
- case break (not . isTagChar) cs of
- (tag,cs) -> readParse (At p tag []) w cs ls
-readParse (At p tag ts) w ('.':cs) ls =
- readParse (At p tag (w:ts)) w cs ls
-readParse (At p tag ts) w cs ls =
- case break (not . isTagChar) cs of
- (tag,']':cs) -> let t = Phrase tag (reverse ts)
- in case p of
- Root -> t : readLines p ls
- At p tag ts -> readParse (At p tag (t:ts)) w cs ls
- _ -> error cs
-
-isTagChar c =
- isLetter c || isDigit c || elem c ":&+-%@=?\"*!"
+ readParse (Word id tag w l) p parse ls
+
+readParse w p [] ls = readLines p ls
+readParse w p ('[':cs) ls =
+ case readTag w cs of
+ (fn,cs) -> readParse w (At p fn []) cs ls
+readParse w (At p fn ts) ('.':cs) ls =
+ readParse w (At p fn (w:ts)) cs ls
+readParse w (At p fn ts) cs ls =
+ case readTag w cs of
+ (_,']':cs) -> let t = fn (reverse ts)
+ in case p of
+ Root -> t : readLines p ls
+ At p fn ts -> readParse w (At p fn (t:ts)) cs ls
+ _ -> readError w
+
+readTag w cs@(c1:c2:_) -- word tag on phrase level
+ | isUpper c1 && isUpper c2 =
+ case break (\c -> not (isLetter c || isDigit c)) cs of
+ (tag,cs) -> case break (\c -> not (elem c "?*%!\"=+-&@")) cs of
+ (mods,cs) -> case cs of
+ (':':c:cs) | isLetter c -> case break (not . isDigit) cs of
+ (ds,cs) -> (Phrase tag mods [c] (if null ds then 0 else read ds),cs)
+ | isDigit c -> case break (not . isDigit) (c:cs) of
+ (ds,cs) -> (Phrase tag mods "" (if null ds then 0 else read ds),cs)
+ _ -> (Phrase tag mods "" 0,cs)
+readTag w (c:cs) -- phrase tag
+ | isUpper c = let tag = [c]
+ in case break (\c -> not (isLetter c || isDigit c || elem c "?*%!\"=+-&@")) cs of
+ (mods,cs) -> case cs of
+ (':':c:cs) | isLetter c -> case break (not . isDigit) cs of
+ (ds,cs) -> (Phrase tag mods [c] (if null ds then 0 else read ds),cs)
+ | isDigit c -> case break (not . isDigit) (c:cs) of
+ (ds,cs) -> (Phrase tag mods "" (if null ds then 0 else read ds),cs)
+ _ -> (Phrase tag mods "" 0,cs)
+ | isLower c = let tag = []
+ mods = []
+ in case break (not . isDigit) cs of
+ (ds,cs) -> (Phrase tag mods [c] (if null ds then 0 else read ds),cs)
+ | isDigit c = let tag = []
+ mods = []
+ in case break (not . isDigit) cs of
+ (ds,cs) -> (Phrase tag mods [] (read ds),cs)
+readTag w cs = readError w
+
+readError (Word id _ _ _) = error id