CLI commands produce no output when run in Docker container

[tassa-yoniso-manasi-karoto]

Description

When running PyThaiNLP CLI commands inside a Docker container, no output is produced even though the commands execute without errors.

The Python API works correctly, suggesting the issue is specific to the CLI implementation.

Expected results

$ docker run --rm pythainlp thainlp tokenize word "สวัสดีครับ"
สวัสดี|ครับ|

Current results

$ docker run --rm pythainlp thainlp tokenize word "สวัสดีครับ"
$ 

(No output is produced)

Steps to reproduce

1. Build PyThaiNLP Docker image:

   git clone https://github.com/PyThaiNLP/pythainlp.git
   cd pythainlp
   # Apply WORKDIR fix to Dockerfile first
   docker build -t pythainlp .

2. Run any CLI command:

   docker run --rm pythainlp thainlp tokenize word "สวัสดีครับ"
   docker run --rm pythainlp thainlp data catalog
   docker run --rm pythainlp thainlp soundex "วรรณ"

3. Observe that no output is produced

More info

• Python API works correctly: docker run --rm pythainlp python -c "from pythainlp.tokenize import word_tokenize; print(word_tokenize('สวัสดีครับ'))" outputs ['สวัสดี', 'ครับ']
• Environment has proper UTF-8 encoding (verified with locale and sys.stdout.encoding)
• Issue occurs with both -t (TTY) and without TTY allocation
• Tested with PYTHONUNBUFFERED=1 environment variable - no effect

Environment:

• Docker version: 28.3.1
• PyThaiNLP version: 5.1.0
• Python version: 3.12 (from python:3.12 base image)
• Host OS: Linux 6.12

[tassa-yoniso-manasi-karoto]

Also, I have a library that aims at unifying many NLP providers for various languages into a single library interface for Go.
To make it possible I wrote intermediary libraries for Ichiran (morphological analyzer for Japanese) and aksharamukha (transliteration engine for many Indic languages) that use Docker Compose' SDK.

I wanted to write a library to expose go library binding for pythainlp using its CI but since it doesn't seem to work right now I guess I will use the python library directly.

So my question is: Is the python API stable at this point?

[wannaphong]

Also, I have a library that aims at unifying many NLP providers for various languages into a single library interface for Go. To make it possible I wrote intermediary libraries for Ichiran (morphological analyzer for Japanese) and aksharamukha (transliteration engine for many Indic languages) that use Docker Compose' SDK.

I wanted to write a library to expose go library binding for pythainlp using its CI but since it doesn't seem to work right now I guess I will use the python library directly.

So my question is: Is the python API stable at this point?

Yes, it is stable if you use the API in PyThaiNLP 5.1.X, which is released on PyPI. For GitHub, it is the development branch and is still not stable. However, we don't change much of the API except for changing the behavior of the function. We are trying to use semantic versioning and keep it.

[wannaphong]

Fixed #1154