mentipy

mentipy is a lightweight, self-hosted classroom-polling library for LaTeX presentations. It generates LaTeX snippets and QR codes for audience questions, then runs a local HTTP server that collects responses in a local JSON store.

It is meant for lectures where you want Mentimeter-style interaction without a cloud service: put polls in the lecture source, compile the slides or handout, serve the poll locally, and let students answer by scanning the QR code.

Basic Workflow

1. Render questions from PythonTeX or from the mentipy render CLI.
2. Compile the LaTeX document so the generated QR codes are embedded.
3. Start the polling server with mentipy serve --store ./mentipy.json.
4. Let respondents scan the QR codes and answer the questions.
5. Watch or export answers with mentipy watch, mentipy export, or the live results page.

The examples below keep state in ./mentipy.json and write QR artifacts in mentipy-obj/.

Authoring a Poll

Both authoring paths produce the same kind of LaTeX fragment. Use the Python helpers when your document build already runs PythonTeX; use the CLI when you want to inspect or redirect the generated LaTeX yourself.

PythonTeX-style authoring:

from mentipy.latex import mc

print(
    mc(
        "How confident do you feel about the course goals?",
        ["Very confident", "Somewhat confident", "Not yet"],
        base_url="http://localhost:8080",
        qr_dir="mentipy-obj",
        layout="article+qr",
    )
)

The matching CLI command:

mentipy render mc "How confident do you feel about the course goals?" \
  "Very confident" "Somewhat confident" "Not yet" \
  --base-url http://localhost:8080 \
  --qr-dir mentipy-obj \
  --layout article+qr

The CLI prints LaTeX to standard output, so you can redirect it into a file:

mentipy render mc "Ready for the next section?" \
  "Yes" "Need a recap" "Not yet" > poll.tex

Question Kinds

Use the question kind that matches the response shape you want to collect.

Open-text questions can also collect uploaded files. Pass fence="python" or --fence python when uploaded code should be rendered as a language-tagged Markdown code block on the results page.

Word-cloud questions default to render="svg". Installing mentipy[wordcloud] lets that mode use the optional wordcloud layout engine; without it, mentipy falls back to its built-in SVG renderer.

Layout and LaTeX Integration

Rendering options control the LaTeX wrapper without changing the poll identity.

The named LaTeX environment must already be defined by your document class or preamble. For example, a lecture note template can use environment="exercise" to reuse its existing exercise styling and counters.

Serving Polls

After the questions have been registered during rendering, start the local server:

mentipy serve --store ./mentipy.json

Use --results when respondents should land on the live results page after submitting an answer:

mentipy serve --store ./mentipy.json --results

Terminal helpers use question hash prefixes. List registered questions first, then watch or export one question by prefix:

mentipy list --store ./mentipy.json
mentipy watch f379 --store ./mentipy.json
mentipy export f379 --format csv --store ./mentipy.json

Respondent URLs and QR Codes

You can pass base_url / --base-url while rendering, or store a default:

mentipy config set base_url http://localhost:8080

If no base URL is configured, mentipy resolves the current LAN address from the serving port. Because QR codes are generated when the document compiles, mentipy serve may regenerate remembered QR images and ask you to recompile the slides so the PDF embeds the updated codes.

For public access, mentipy serve can expose the local server in two modes:

mentipy serve --store ./mentipy.json --public upnp
mentipy config set ssh_tunnel user@example.org:8080
mentipy serve --store ./mentipy.json --public ssh

The UPnP mode requires the optional mentipy[public] dependency and a router that accepts temporary port mappings. The SSH mode uses an SSH reverse tunnel; the remote SSH server must allow the forwarded port to be reached publicly.

The polling server is intentionally simple. Anyone with the URL can submit answers, and with --results they can view the live tally. Treat a public poll URL like a shared room link, not like a private admin interface.

Discovering Commands

Useful entry points while learning the tool:

mentipy render --help
python3 -m pydoc mentipy.latex
mentipy tutorial list
mentipy tutorial run using-tutorials

The tutorial sequence mirrors the usage introduction: authoring paths, question kinds, layout and environment choices, respondent URLs, and serving/publication.