Activity

1. Jamongss

   changed the title [-]Improve encoding/decoding error handling in compat.py[/-] [+]compat.py - 한글 인코딩/디코딩 에러 예외처리[/+] on Aug 25, 2025
2. Jamongss commented on Aug 25, 2025

   

   Jamongss

   on Aug 25, 2025

   Last edited by Jamongss

   Author

   More actions

   [ENG]

   Summary

   • Environment: Python 3.10 / Supervisor 4.2.5
   • When running supervisorctl tail -f, Korean characters are displayed as garbled text due to missing error handling in encoding/decoding.
   • Specifically, with the command supervisorctl> tail -f service_name, previously written log messages are corrupted when being read back, while new logs printed after the command runs are displayed correctly.

   Root Cause

   • When tail -f is executed, the initial portion of the log (head length) is read as bytes.
   • If a multibyte Korean character is cut in the middle, decoding fails and the corrupted string is printed as-is.
   • This corruption spreads: not only the truncated first character but also the rest of the string is displayed incorrectly.
   • Newly appended log lines are streamed and displayed correctly.
   • The decode logic does not handle errors, leading to UnicodeDecodeError or broken text output.

   Modification

   • File: compat.py

   AS-IS

     - 54 line: return s.encode(encoding)
     - 60 line: return s.decode(encoding)
   

   TO-BE

     - 54 line: return s.encode(encoding, errors='replace')
     - 60 line: return s.decode(encoding, errors='replace')
   

   After the Fix

   • When invalid bytes are encountered, they are replaced with the Unicode replacement character �, while the rest of the string is decoded correctly.
   • Ensures continuity of log output by gracefully handling corrupted sequences.
   • Prevents UnicodeDecodeError / UnicodeEncodeError caused by malformed encoding or truncated multibyte characters.
3. mnaberez

   changed the title [-]compat.py - 한글 인코딩/디코딩 에러 예외처리[/-] [+]tail -f garbled output when log has Korean characters[/+] on Aug 25, 2025
4. mnaberez commented on Aug 26, 2025

   

   mnaberez

   on Aug 26, 2025

   Member

   More actions

   Patch suggested above and in #1694:

   diff --git a/supervisor/compat.py b/supervisor/compat.py
   index e741f47fb..596727173 100644
   --- a/supervisor/compat.py
   +++ b/supervisor/compat.py
   @@ -51,13 +51,14 @@ def as_bytes(s, encoding='utf8'):
            if isinstance(s, bytes):
                return s
            else:
   -            return s.encode(encoding)
   +            return s.encode(encoding, errors='replace')
    
        def as_string(s, encoding='utf8'):
            if isinstance(s, str):
                return s
            else:
   -            return s.decode(encoding)
   +            # 인입된 bytes에서 깨진 부분은 '�'로 표시하고 나머지 부분은 계속 decode 진행
   +            return s.decode(encoding, errors='replace')
    
        def is_text_stream(stream):
            import _io
5. mnaberez

   added

   logging

   on Aug 26, 2025