Merge pull request #1218 from PyThaiNLP/copilot/fix-coverage-parser-error

Fix Python 3.13 SyntaxError: global declaration before variable use in tokenizer modules

Author: bact

pythainlp/tokenize/budoux.py

@@ -49,9 +49,9 @@ def segment(text: str) -> list[str]:
         return []
 
     # Thread-safe lazy initialization
+    global _parser
     with _parser_lock:
         if _parser is None:
-            global _parser
             _parser = _init_parser()
         parser = _parser
 

pythainlp/tokenize/oskut.py

@@ -38,10 +38,10 @@ def segment(text: str, engine: str = "ws") -> list[str]:
         return []
 
     # Thread-safe model loading
+    global _DEFAULT_ENGINE
     with _engine_lock:
         if engine != _DEFAULT_ENGINE:
             # Need to update global state and reload model
-            global _DEFAULT_ENGINE
             _DEFAULT_ENGINE = engine
             oskut.load_model(engine=_DEFAULT_ENGINE)
 

pythainlp/tokenize/sefr_cut.py

@@ -37,10 +37,10 @@ def segment(text: str, engine: str = "ws1000") -> list[str]:
         return []
 
     # Thread-safe model loading
+    global _DEFAULT_ENGINE
     with _engine_lock:
         if engine != _DEFAULT_ENGINE:
             # Need to update global state and reload model
-            global _DEFAULT_ENGINE
             _DEFAULT_ENGINE = engine
             sefr_cut.load_model(engine=_DEFAULT_ENGINE)
 

pythainlp/tokenize/wtsplit.py

@@ -33,9 +33,9 @@ def _tokenize(
     thread-safety guarantees.
     """
     # Thread-safe model loading
+    global _MODEL, _MODEL_NAME
     with _model_lock:
         if _MODEL_NAME != model:
-            global _MODEL, _MODEL_NAME
             _MODEL = WtP(model_name_or_model=model)
             _MODEL_NAME = model
         model_instance = _MODEL