stden/document-tools repository overview

⁠stden/document-tools

Create PDF files with Typst and LibreOffice, and inspect them with Poppler tools.

⁠Overview

Use this image for document builds, report conversion, and PDF checks. It contains:

• Typst for document layout and PDF generation.
• LibreOffice for headless document conversion.
• Poppler tools, for example pdfinfo, pdftotext, and pdftoppm.
• DejaVu and Liberation fonts.
• Python and uv for document scripts.

The working directory is /workspace. The default command starts Bash. The LibreOffice package runs only on linux/amd64.

⁠Tags

View all tags on Docker Hub⁠. The latest tag changes when a new build passes its tests.

⁠Pull the image

docker pull stden/document-tools:latest

⁠Create a PDF with Typst

Run these commands in the directory where you want to save the document:

printf '= Hello from Docker\nThis PDF was built with Typst.\n' > report.typ

docker run --rm -v "$PWD":/workspace stden/document-tools:latest \
  typst compile report.typ report.pdf

The mount saves report.pdf on your host. If your document uses Typst packages, network access can be necessary at their first use.

⁠Convert an office document to PDF

Put report.docx in the current directory. Then run this command:

docker run --rm -v "$PWD":/workspace stden/document-tools:latest \
  libreoffice -env:UserInstallation=file:///tmp/lo-profile --headless \
  --convert-to pdf --outdir /workspace report.docx

The same command can convert supported office files, such as .odt, .xlsx, and .pptx. Change the input file name. LibreOffice can also convert a text file to DOCX:

docker run --rm -v "$PWD":/workspace stden/document-tools:latest \
  libreoffice -env:UserInstallation=file:///tmp/lo-profile --headless \
  --convert-to docx --outdir /workspace notes.txt

Each command uses a temporary LibreOffice profile inside its container. Supply the fonts that your source document uses. Missing fonts can change the layout.

⁠Inspect a PDF

Show the page count, the page size, and the document metadata:

docker run --rm -v "$PWD":/workspace stden/document-tools:latest \
  pdfinfo report.pdf

Extract the text and keep the page layout:

docker run --rm -v "$PWD":/workspace stden/document-tools:latest \
  pdftotext -layout report.pdf report.txt

Create a PNG image of the first page for a visual check:

docker run --rm -v "$PWD":/workspace stden/document-tools:latest \
  pdftoppm -f 1 -singlefile -png -r 150 report.pdf report-page

The last command saves report-page.png. Text extraction works on PDF text. It does not do OCR on scanned pages.

⁠Use your own scripts

Mount the project directory and run your Python script:

docker run --rm -v "$PWD":/workspace stden/document-tools:latest \
  python check_report.py

Install extra Python libraries in a child image that starts with FROM stden/document-tools:latest. All generated files in /workspace stay on the host. The default container user is root, so on Linux the output files can belong to root.

⁠Source and checks

The smoke test creates a Typst PDF, converts text to DOCX and PDF with LibreOffice, and checks the extracted text.

• Dockerfile⁠
• Smoke test⁠
• Overview source⁠
• Repository and Docker tutorial⁠