mcpField
T
translate-ja

translate-ja

github.com/k5-motDocuments093 installsUpdated 6w ago
Repository

About translate-ja

translate-ja is a document translation skill that normalizes PDFs, Word files, PowerPoint, HTML, and images into Docling schema JSON, then cleans, chunks, and translates the content into Japanese. It can also produce combined Markdown and Word docx output.

Key features

  1. 01Normalize documents to Docling schema JSON
  2. 02Correct structure and clean content
  3. 03Chunk documents and translate into Japanese
  4. 04Combine output as Markdown
  5. 05Convert output to Word docx
  6. 06Use a CSV terminology dictionary

Use cases

  1. 01Translate source documents such as PDFs
  2. 02Apply a terminology dictionary to specialist material
  3. 03Organize translated content as Markdown
  4. 04Create translated Word documents

Install

Copy the command and run it in a terminal.

$npx skills add https://github.com/k5-mot/skills --skill translate-ja

After installing, type/translate-jain Claude Code to confirm it loaded.

Bundled files

References and scripts installed along with SKILL.md.

README.md
run.ps1
run.sh
SPEC_v0.md
SPEC_v1.md
SPEC_v2.md
template.docx
template.dotx
template.v1.dotx
scripts/
chunk_docling_json.py
clean_doc.py
concat_chunks.py
config.py
convert_md_to_docx_with_docling.py
convert_to_md_remote.py
io_utils.py
preprocess_doc_with_docling.py
realign_doc_struct_with_llm.py
translate_chunks.py
translate_ja.py
tests/
test_chunk_docling_json.py
test_clean_doc.py
test_concat_chunks.py
test_convert_to_md_remote.py
test_realign_doc_struct_with_llm.py
test_translate_ja.py

Details

From the project README

translate-ja

PDF、Word、PowerPoint、HTML、画像などの入力文書を Docling schema JSON に正規化し、構造補正、清掃、チャンク化、日本語翻訳、Markdown 連結、Word docx 変換までを実行する。

詳細仕様、Docling artifacts、runner挙動、Chunk JSONL採用理由は SPEC_v0.md を参照する。

使い方

./skills/translate-ja/run.sh --input ./docs/source/source.pdf

用語辞書を使う場合:

./skills/translate-ja/run.sh --input ./docs/source/source.pdf --dictionary-csv ./docs/source/dictionary.csv

辞書 CSV は UTF-8 で、列は english, japanese, genre, description とする。

注意

  • Docling Serve と OpenAI 互換 API が必要。外部サービスなしの PDF フォールバックは行わない。
  • .env または環境変数に DOCLING_SERVER_URL, DOCLING_API_KEY, OPENAI_BASE_URL, OPENAI_API_KEY, OPENAI_MODEL を設定する。
  • ログには原文や翻訳文が含まれうるため、機密文書では保存先のアクセス制御を確認する。

Comments

Markdown supported