Issues previewing/rendering a combined {python} and {r} Quarto document

[gethinwade]

Hi,

I'm working in a .qmd document within Positron that has one Python chunk and one R chunk. I have two separate / mutually exclusive issues when trying to preview the output:

1. If I don't include the code "engine: jupyter" in the YAML header, I receive a ModuleNotFoundError. I'm using a conda virtual environment in Positron with all the necessary packages installed. It's worth noting that this error is only raised for pandas and matplotlib, not for numpy. If I exclude those packages, the code runs normally.
2. If I instead specify "engine: jupyter" in the YAML header, the Python code runs fine, but my R code doesn't produce any output in the preview.

Any help in diagnosing this issue would be greatly appreciated!

My code

---
title: "Testing"
format: pdf
---

```{python}
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

print("hello world")

```{r}
y <- c(199,205,196,200,218,220,215,223,237,234,235,230,250,248,253,246)
x <- c(10,13,14,16,22,24,27,29,31,33,36,37,39,40,41,43)

linear_model <- lm(y~x) 
print(summary(linear_model))

plot(x, y, xlab='X', ylab='Y', 
     main='Scatter Plot and Regression Line')
abline(linear_model,col="green")

Error message in the first case (ModuleNotFoundError)

processing file: Testing.qmd
1/4                  
2/4 [unnamed-chunk-1]
Error in `py_call_impl()`:
! ModuleNotFoundError: No module named 'pandas'
Run `reticulate::py_last_error()` for details.
Backtrace:
     x
  1. \-global .main()
  4.   \-execute(...)
  5.     \-rmarkdown::render(...)
  6.       \-knitr::knit(knit_input, knit_output, envir = envir, quiet = quiet)
  7.         \-knitr:::process_file(text, output)
  8.           +-xfun:::handle_error(...)
  9.           +-base::withCallingHandlers(...)
  10.           \-knitr:::process_group(group)
  11.             \-knitr:::call_block(x)
 12.               \-knitr:::block_exec(params)
 13.                 +-knitr:::in_input_dir(engine(options))
 14.                 | \-knitr:::in_dir(input_dir(), expr)
 15.                 \-knitr (local) engine(options)
 16.                   \-reticulate::eng_python(options)
 17.                     \-reticulate:::py_compile_eval(snippet, compile_mode)
 18.                       +-reticulate::py_capture_output(...)
 19.                       | +-base::tryCatch(...)
 20.                       | | \-base (local) tryCatchList(expr, classes, parentenv, handlers)
 21.                       | \-base::force(expr)
 22.                       \-builtins$eval(compiled, globals, locals)
 23.                         \-reticulate:::py_call_impl(callable, call_args$unnamed, call_args$named)

Quitting from Testing.qmd:6-44 [unnamed-chunk-1]
Execution halted

Issue in the second case (no output from the R chunk in the preview)

[juliasilge]

The only way currently to combine R and Python in one Quarto document is to use knitr and reticulate; you can't use engine: jupyter because there isn't support for multiple language Jupyter kernels or multiple Jupyter kernels per document.

It sounds like when you use knitr for your Quarto document, reticulate can't find the right Python environment. I usually do this by having a uv virtual environment right there in the project/workspace; I find that reticulate and Quarto do a good job of finding it in this situation. If you need to go a step further to get reticulate to find the right Python environment, check out these reticulate docs:
https://rstudio.github.io/reticulate/articles/versions.html

I also have found this blog post really helpful for combining R and Python in one Quarto document:
https://nrennie.rbind.io/blog/combining-r-and-python-with-reticulate-and-quarto/

[gethinwade]

Thank you, I will try setting up a virtual environment in the project for reticulate to to find.