Add pythainlp.tools.safe_print

Handle Unicode print on console

Author: bact

pythainlp/tools/__init__.py

@@ -6,10 +6,11 @@
     "get_full_data_path",
     "get_pythainlp_data_path",
     "get_pythainlp_path",
+    "safe_print",
     "warn_deprecation",
 ]
 
-from pythainlp.tools.core import warn_deprecation
+from pythainlp.tools.core import safe_print, warn_deprecation
 from pythainlp.tools.path import (
     PYTHAINLP_DEFAULT_DATA_DIR,
     get_full_data_path,

pythainlp/tools/core.py

@@ -5,6 +5,7 @@
 Generic support functions for PyThaiNLP.
 """
 
+import sys
 import warnings
 
 
@@ -13,8 +14,7 @@ def warn_deprecation(
     replacing_func: str = "",
     version: str = "",
 ):
-    """
-    Warn about the deprecation of a function.
+    """Warn about the deprecation of a function.
 
     :param str deprecated_func: Name of the deprecated function.
     :param str replacing_func: Name of the function to use instead (optional).
@@ -28,3 +28,19 @@ def warn_deprecation(
     if replacing_func:
         message += f" Please use '{replacing_func}' instead."
     warnings.warn(message, DeprecationWarning, stacklevel=2)
+
+
+def safe_print(text: str):
+    """Print text to console, handling UnicodeEncodeError.
+
+    :param text: Text to print.
+    :type text: str
+    """
+    try:
+        print(text)
+    except UnicodeEncodeError:
+        print(
+            text.encode(sys.stdout.encoding, errors="replace").decode(
+                sys.stdout.encoding
+            )
+        )