Fix XML output from stanza which had multiple roots + add a demonstrator...

Fix XML output from stanza which had multiple roots + add a demonstrator haskell script that was used to fix the problem in place

Fix XML output from stanza which had multiple roots + add a demonstrator...
Fix XML output from stanza which had multiple roots + add a demonstrator haskell script that was used to fix the problem in place
37a652cc · Alice Brenon · cc30e47d · 37a652cc · 37a652cc
Commit 37a652cc authored 2 years ago by Alice Brenon
--- a/scripts/reroot.hs
+++ b/scripts/reroot.hs
+#!/usr/bin/env -S runhaskell --ghc-arg="-i lib"
+{-# LANGUAGE OverloadedStrings #-}
+
+import Data.Text (Text)
+import System.Environment (getArgs)
+import System.Script (syntax)
+import Text.Editor (editAll)
+
+wrap :: [Text] -> [Text]
+wrap text = "<body>": (("\t"<>) <$> text) ++ ["</body>"]
+
+main :: IO ()
+main = getArgs >>= run
+  where
+    run [target] = getContents >>= editAll wrap target
+    run _ = syntax "TARGET_DIRECTORY"
--- a/scripts/stanza-annotator.py
+++ b/scripts/stanza-annotator.py
@@ -33,20 +33,22 @@ class Annotator:

    def encode(self, file, annotation):
        with open(f'{self.target}/{file}', 'w') as target:
+            print('<text>', file=target)
            for sentence in annotation.sentences:
                self.encode_sentence(sentence, target)
+            print('</text>', file=target)

    def encode_sentence(self, sentence, target):
-        print('<s>', file=target)
+        print('\t<s>', file=target)
        for token in sentence.tokens:
            self.encode_token(token, target)
-        print('</s>', file=target)
+        print('\t</s>', file=target)

    def encode_token(self, token, target):
        form = token.text
        lemma = '+'.join(map(lambda w: w.lemma, token.words))
        upos = '+'.join(map(lambda w: w.upos, token.words))
-        print(f'\t<w lemma="{lemma}" pos="{upos}">{form}</w>', file=target)
+        print(f'\t\t<w lemma="{lemma}" pos="{upos}">{form}</w>', file=target)

 if __name__ == '__main__':
    Annotator(sys.argv[1], sys.argv[2]).run()