Skip to content

has2k1/qparameterize

Repository files navigation

qparameterize

Parameterize Quarto documents for batch rendering with multiple parameter sets for the jupyter engine.

The default uses papermill.

However, when rendering to HTML with Quarto's Jupyter engine, each execution overwrites output files (figures, cell outputs) because filenames like cell-output-1 are only unique within a single file, not across parameterized runs.

This tool solves that by generating separate .qmd files with unique labels, so Quarto can render them all without conflicts.

Why an alternative to quarto's default parameterizing?

The default is excellent for parameterizing Jupyter notebooks when generating single-file outputs e.g. notebooks, PDFs or contained HTML pages.

See Quarto documentation on parameters for more context.

The Jupyter engine generates output filenames with prefixes like cell-output-1, cell-output-2 for figures and other artifacts. These prefixes are the automatic labels of the cells. Even when cells are labelled, those labels are only unique within the file. When you render multiple parameterized versions, the last run overwrites all previous outputs.

e.g. If you render a file map.qmd with parameters to html

quarto render map.qmd -P country:afghanistan -o afghanistan.html
...
quarto render map.qmd -P country:zimbabwe -o zimbabwe.html

The artifacts for all renderings will be stored in _site/map/. If the first cell of map.qmd has an image output it will be saved as _site/map/figure-html/cell-output-1.png for all the renderings. All the HTML files will show display images from the last rendering!

Solution

This package addresses the issue by:

  1. Substituting parameter values in the parameters block
  2. Adding unique labels to all code blocks based on parameter values
  3. Outputs a parameterized qmd file

With this approach, the parameterizing is separated from the rendering. When rendering, Quarto will know of all the qmd source files.

Workflow

  1. Create a .qmd with a parameters block
  2. Run qparameterize for each parameter set → generates input--param-values.qmd
  3. Run quarto render on all generated files

Installation

Requires Python 3.11+.

# With uv
uv pip install qparameterize

# With pip
pip install qparameterize

Usage

Define a parameters block

In your .qmd file, mark a code block with the parameters tag:

```{python}
#| tags: [parameters]
animal = "cat"
count = 5
```

Run the tool

# Using key:value pairs
uv run qparameterize input.qmd -P animal:dog -P count:10

# Using a YAML file
uv run qparameterize input.qmd --execute-params params.yaml

# With custom output filename
uv run qparameterize input.qmd -P animal:dog -P count:10 -o ten-dogs.qmd

Example YAML params file

animal: dog
count: 10

Output

A new file with parameter values in the filename:

input.qmd → input--dog-10.qmd

The output file has:

  • Substituted parameter values
  • Unique labels on all code blocks (e.g., #| label: dog-10--1)

Cleaning up

To delete all previously generated files:

uv run qparameterize --clean

The generated files are tracked in .quarto/qparameterize.txt.

In A Quarto Project

Parameterize in the pre-render step and clean up in the post-render.

project:
  type: website
  pre-render:
    - uv run qparameterize input.qmd -P animal:dog -P count:10
    - uv run qparameterize input.qmd -P animal:elephant -P count:7
  post-render:
    - uv run qparameterize --clean

Development

# Lint
uv run --group dev ruff check src/

# Test
uv run --group dev pytest

About

Parameterise a Quarto document for the Jupyter Engine

Resources

License

Stars

Watchers

Forks

Packages

 
 
 

Contributors