DOCX Document Loader

[github-actions]

Description

Add a DocxLoader that reads .docx files using python-docx and returns a Document per file (concatenating all paragraphs).

Motivation

Word documents are ubiquitous in business workflows. Supporting DOCX opens the framework to HR, legal, and operational use-cases.

Acceptance criteria

• Class DocxLoader in ragframework/document/loaders.py
• Subclasses DocumentLoader from ragframework/base.py
• Uses python-docx — add it to the [docx] extra in pyproject.toml
• Raises LoaderError on failure; import guarded with helpful message
• Unit tests in tests/test_document/test_loaders.py
• Exported from ragframework/document/__init__.py
• CHANGELOG.md updated under [Unreleased]

Files to touch

• ragframework/document/loaders.py — add DocxLoader
• ragframework/document/__init__.py — export it
• pyproject.toml — add python-docx to [docx] extra
• tests/test_document/test_loaders.py — add tests

Resources

• python-docx docs
• Existing loaders (TextFileLoader, MarkdownLoader) in ragframework/document/loaders.py as reference