nlparrot

commit 9b85e20d877c0b1c35b2b39ef5d86bb2a22fd702
parent 70cb172f11e6ac4c3c9700573a3e8be7f23cf45b
Author: Stefan Koch <programming@stefan-koch.name>
Date:   Sun, 27 Aug 2023 13:57:24 +0200

accept multiple possible languages to choose from

Diffstat:
M
pyproject.toml
 | 
2
+-
M
src/nlparrot/server.py
 | 
13
+++++--------

2 files changed, 6 insertions(+), 9 deletions(-)
diff --git a/pyproject.toml b/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "nlparrot"
-version = "0.0.1"
+version = "0.0.2"
 dependencies = [
     'classla',
     'pyphen',
diff --git a/src/nlparrot/server.py b/src/nlparrot/server.py
@@ -23,9 +23,6 @@ if __name__ == "__main__":
     current_language = None
     tokenizers = None
 
-    # TODO: Receive supported languages from client program
-    supported_languages = ["hr", "ja"]
-
     with Listener(**get_listen_address()) as listener:
         print("Listening for connection ...")
 
@@ -33,12 +30,12 @@ if __name__ == "__main__":
             with listener.accept() as conn:
                 task: dict = conn.recv()
 
-                if task["language"] is None:
-                    language = detect_language(task["text"], supported_languages=supported_languages)
-                    print(f'Auto-detected language "{language}"')
-                else:
-                    language = task["language"]
+                if len(task["languages"]) == 1:
+                    language = task["languages"][0]
                     print(f'Received language "{language}"')
+                else:
+                    language = detect_language(task["text"], supported_languages=task["languages"])
+                    print(f'Auto-detected language "{language}"')
 
                 if tokenizers is None or current_language != language:
                     tokenizers = get_tokenizers(language)