TinyDB raised permission denied when db.json is read-only

[wisaTong]

Description

My setup mounts PYTHAINLP_DATA_DIR from a central server in read-only mode.

Expected results

If db.json existed and it is read-only, there should be no problem reading it.

Current results

TinyDB raised permission denied when db.json is read-only

Traceback (most recent call last):
  File "[REDACTED]/train.py", line 142, in <module>
    main(args)
  File "[REDACTED]/train.py", line 83, in main
    from featurizer import Featurizer
  File "[REDACTED]/featurizer.py", line 1, in <module>
    from pythainlp.tokenize import newmm
  File "[REDACTED]/venv/lib/python3.9/site-packages/pythainlp/__init__.py", line 39, in <module>
    from pythainlp.soundex import soundex
  File "[REDACTED]/venv/lib/python3.9/site-packages/pythainlp/soundex/__init__.py", line 15, in <module>
    from pythainlp.soundex.lk82 import lk82
  File "[REDACTED]/venv/lib/python3.9/site-packages/pythainlp/soundex/lk82.py", line 10, in <module>
    from pythainlp.util import remove_tonemark
  File "[REDACTED]/venv/lib/python3.9/site-packages/pythainlp/util/__init__.py", line 68, in <module>
    from pythainlp.util.keywords import find_keyword, rank
  File "[REDACTED]/venv/lib/python3.9/site-packages/pythainlp/util/keywords.py", line 5, in <module>
    from pythainlp.corpus import thai_stopwords
  File "[REDACTED]/venv/lib/python3.9/site-packages/pythainlp/corpus/__init__.py", line 55, in <module>
    TinyDB(_CORPUS_DB_PATH).close()
  File "[REDACTED]/venv/lib/python3.9/site-packages/tinydb/database.py", line 94, in __init__
    self._storage: Storage = storage(*args, **kwargs)
  File "[REDACTED]/venv/lib/python3.9/site-packages/tinydb/storages.py", line 102, in __init__
    touch(path, create_dirs=create_dirs)
  File "[REDACTED]/venv/lib/python3.9/site-packages/tinydb/storages.py", line 31, in touch
    with open(path, 'a'):
PermissionError: [Errno 13] Permission denied: './datadir/pythainlp-data/db.json'

Steps to reproduce

1. change data dir permission to read-only chmod 0444 /your/dir/here/*
2. import ulmfit preprocess.
3. observe the error.

from pythainlp.ulmfit import preprocess

Context

Your environment

• PyThaiNLP version: 3.0.0
• Python version: 3.9.10
• Operating system and version (distro, 32/64-bit): Linux gentoo-grill 5.16.4-gentoo
• More info (Docker, VM, etc.):

Possible solution

A little digging and I found that TinyDB my has option for access modes msiemens/tinydb#136 (comment)

Files

datadir's content:

pythainlp-data/
├── db.json
├── itos_lstm.pkl
└── thwiki_model_lstm.pth

db.json:

{"_default": {"1": {"name": "wiki_lm_lstm", "version": "0.32", "filename": "thwiki_model_lstm.pth"}, "2": {"name": "wiki_itos_lstm", "version": "0.32", "filename": "itos_lstm.pkl"}}}

[wannaphong]

Now, PyThaiNLP v3.0.1 was release. You can setting PYTHAINLP_READ_MODE environment variable with value is 1.