Fix base dependencies

[what-in-the-nim]

What does this changes

Remove pandas, numpy and pyyaml from base dependencies

What was wrong

As described in issue #1184

How this fixes it

After investigating, these three libraries are used in two modules, benchmark and misspell. In misspell, we can totally remove numpy out since we can achieve the same functionality by use built-in python module math and random (might be quicker too, since we remove numpy overhead). In benchmark, we can use lazy imports so that running help in parser does not raise import errors. The modules are only imported at execution time, which improves flexibility and avoids failures during help or argument parsing. I also added the warning that the user should install pythainlp[benchmarks] if they want to use the thainlp benchmark

Fixes

• Remove base dependencies
• Make misspell module, use built-in python library instead of numpy
• Make benchmark module lazy import and warning to install extra dependency.

Your checklist for this pull request

• Passed code styles and structures
• Passed code linting checks and unit test

[Copilot]

Pull request overview

This PR removes pandas, numpy, and PyYAML from the base dependencies of PyThaiNLP to make the package more lightweight. These libraries are now only required when using specific features like benchmarking.

Changes:

• Removed pandas, numpy, and PyYAML from base requirements in setup.py
• Refactored misspell module to use built-in Python math and random instead of numpy
• Added lazy imports with error handling to the benchmark CLI module

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
setup.py	Removed PYYAML, PANDAS, and NUMPY from base requirements list
pythainlp/tools/misspell.py	Replaced numpy functions with Python built-ins (math.floor, random.sample, random.choice)
pythainlp/cli/benchmark.py	Added lazy imports for yaml and word_tokenization with helpful error message

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[sonarqubecloud]

Quality Gate failed

Failed conditions
2 Security Hotspots

See analysis details on SonarQube Cloud

[what-in-the-nim]

SonarQube and Codacy errors on pseudo random generator (random.choice etc.)

Since this is not a security-sensitive context, we should be fine here.

[coveralls]

coverage: 52.644% (-0.3%) from 52.947%
when pulling 33ba93b on what-in-the-nim:fix-base-dependencies
into 8819cfe on PyThaiNLP:dev.

[bact]

LGTM. Thank you.

Yes, numpy just for the floor and random are too overkill.

[wannaphong]

Thank you!