Quarto Is Multi-lingual! – Kazuharu Yanagimoto

A common misunderstanding about Quarto is that we cannot use multiple languages within a document. Indeed, Jupyter cannot use multiple languages within a document, and we usually use the jupyter engine for Python and Julia (and it is officially supported.)

However, knitr has already been able to execute multiple languages tons of years ago, so why can’t we do it in Quarto?

Let’s see the following example with palmerpenguins::penguins of Horst, Hill, and Gorman (2020)

R

library(palmerpenguins)
library(ggplot2)

ggplot(data = penguins, aes(x = flipper_length_mm)) +
  geom_histogram(aes(fill = species), 
                 alpha = 0.5, 
                 position = "identity") +
  scale_fill_manual(values = c("darkorange","purple","cyan4")) +
  labs(x = "Flipper length (mm)",
       y = "Frequency",
       title = "Penguin Flipper Lengths",
       fill = NULL) +
  theme_minimal() +
  theme(panel.grid.major.x = element_blank(),
        panel.grid.minor = element_blank(),
        legend.position = c(0.9, 0.9))

Python

To run a python code on knitr, install.packages("reticulate")

from palmerpenguins import load_penguins
import seaborn as sns
import matplotlib.pyplot as plt

penguins = load_penguins()

plt.clf()
sns.histplot(data=penguins, x='flipper_length_mm',
                hue='species', palette=['#FF8C00', '#159090', '#A034F0'])

plt.xlabel("Flipper length (mm)") 
plt.ylabel("Frequncy")
plt.legend(title = "")
plt.show()

Julia

To run a julia code on knitr, run install.packages("JuliaCall"). PalmerPenguins.load() asks you to download the data, so you have to set ENV["DATADEPS_ALWAYS_ACCEPT"] = true to automatically accept it.

using PalmerPenguins
using StatsPlots
using DataFrames
ENV["DATADEPS_ALWAYS_ACCEPT"] = true;

penguins = DataFrame(PalmerPenguins.load());

histogram(
    [penguins[penguins.species .== species, :].flipper_length_mm for species in ["Adelie", "Chinstrap", "Gentoo"]],
    label = ["Adelie" "Chinstrap" "Gentoo"],
    fillcolor = [:darkorange :purple :cyan4],
    fillalpha = 0.5,
    xlabel = "Flipper length (mm)",
    ylabel = "Frequency",
    title = "Penguin Flipper Lengths",
    legend = :topright,
    grid = false,
    legendfontsize = 7,
    legendtitle = nothing,
    bar_width = 5,
    size = (480, 320),
    dpi = 300,
    tickfontsize = 6,
    guidefontsize = 10,
    titlefontsize = 12,
    margin=Plots.Measures.Length(:mm, 2.0)
)

Stata

Douglas Hemken’s Statamarkdown allows us to use Stata on Rmarkdown (so as on Quarto!)

# devtools::install_github("Hemken/Statamarkdown")
library(Statamarkdown)
haven::write_dta(penguins, here::here("blog/2023/04/30/quarto_multi_lang/palmerpenguins.dta"))

Thanks to Statamarkdown, we can easily set the working directory to the directory where the .qmd file exists by just calling cd in Stata.

Stata

cd

C:\Users\kazuh\GitHub\kazuyanagimoto.github.io\blog\2023\04\30\quarto_multi_lan
> g

Then, you can run Stata code in a code chunk.

use palmerpenguins.dta, clear

twoway (hist body_mass_g if species == 1, color(orange%40)) ///
  (hist body_mass_g if species == 2, color(emerald%40)) ///
  (hist body_mass_g if species == 3, color(purple%40)), ///
   xtitle("Body Mass (g)") ytitle("Frequency") ///
   legend(order(1 "Adelie" 2 "Chinstrap" 3 "Gentoo") pos(1) ring(0) col(1)) ///
   plotregion(fcolor(white)) graphregion(fcolor(white))
graph export "figure/stata/hist.svg", replace

But you have to manually export an image file for a plot and add it into a document…

Tabset

I think, one of the most useful case is when you want to support multiple source codes in a textbook or lecture notes. Quarto supports Tabset layout and it’s so smart!

penguins |>
  ggplot(aes(x = flipper_length_mm, y = body_mass_g,
             color = species, shape = species)) +
  geom_point(size = 3) +
  scale_color_manual(values = c("darkorange","purple","cyan4")) +
  labs(x = "Flipper Length (mm)", y = "Body Mass (g)", color = NULL, shape = NULL) +
  theme_minimal() +
  theme(legend.position = c(0.9, 0.1),
        panel.grid.major = element_blank())

plt.clf()

sns.scatterplot(data=penguins, x='flipper_length_mm', y='body_mass_g',
                hue='species', style="species",
                palette=['#FF8C00', '#159090', '#A034F0'])

plt.xlabel('Flipper Length (mm)')
plt.ylabel('Body Mass (g)')
plt.legend(title = "")
plt.show()

@df penguins scatter(
  :flipper_length_mm,
  :body_mass_g,
  group = :species,
  markersize = 5,
  markershape = [:circle :utriangle :rect ],
  markerstrokecolor = :white,
  palette = ["#FF8C00", "#159090", "#A034F0"],
  xaxis = "Flipper Length (mm)",
  yaxis = "Body Mass (g)",
  size = (480, 320),
  dpi = 300,
  legendfontsize = 8,
  tickfontsize = 6,
  guidefontsize = 10,
  margin=Plots.Measures.Length(:mm, 2.0)
)

use data/palmerpenguins, clear

separate body_mass_g, by(species)
twoway scatter body_mass_g? flipper_length_mm, ///
   mcolor(orange emerald purple) mlab("") ///
   msymbol(O T S) ///
   xtitle("Flipper Length (mm)") ytitle("Body Mass (g)") ///
   legend(order(1 "Adelie" 2 "Chinstrap" 3 "Gentoo") pos(11) ring(0) col(1)) ///
   plotregion(fcolor(white)) graphregion(fcolor(white))
qui graph export "figure/stata/scatter.svg", replace

Have a happy Quarto life 🥂!

References

Horst, Allison Marie, Alison Presmanes Hill, and Kristen B Gorman. 2020. Palmerpenguins: Palmer Archipelago (Antarctica) Penguin Data. Manual. https://doi.org/10.5281/zenodo.3960218.

--- title: "Quarto Is Multi-lingual!" date: 2023-4-30 author: Kazuharu Yanagimoto categories: [Julia, Python, R, Quarto, Stata] image: img/matthias-heyde-8HLMLrkyLvE-unsplash.jpg twitter-card: image: img/matthias-heyde-8HLMLrkyLvE-unsplash.jpg knitr: true execute: warning: false message: false --- A common misunderstanding about Quarto is that we cannot use multiple languages within a document. Indeed, _Jupyter_ cannot use multiple languages within a document, and we usually use the jupyter engine for Python and Julia (and it is officially supported.) However, _knitr_ has already been able to execute multiple languages tons of years ago, so why can't we do it in Quarto? Let's see the following example with `palmerpenguins::penguins` of @horst2020 ### R ```{r} #| fig-width: 6 #| fig-height: 4 #| fig-align: center library(palmerpenguins) library(ggplot2) ggplot(data = penguins, aes(x = flipper_length_mm)) + geom_histogram(aes(fill = species), alpha = 0.5, position = "identity") + scale_fill_manual(values = c("darkorange","purple","cyan4")) + labs(x = "Flipper length (mm)", y = "Frequency", title = "Penguin Flipper Lengths", fill = NULL) + theme_minimal() + theme(panel.grid.major.x = element_blank(), panel.grid.minor = element_blank(), legend.position = c(0.9, 0.9)) ``` ### Python To run a python code on _knitr_, `install.packages("reticulate")` ```{python} #| fig-width: 6 #| fig-height: 4 #| fig-align: center from palmerpenguins import load_penguins import seaborn as sns import matplotlib.pyplot as plt penguins = load_penguins() plt.clf() sns.histplot(data=penguins, x='flipper_length_mm', hue='species', palette=['#FF8C00', '#159090', '#A034F0']) plt.xlabel("Flipper length (mm)") plt.ylabel("Frequncy") plt.legend(title = "") plt.show() ``` ### Julia To run a julia code on _knitr_, run `install.packages("JuliaCall")`. `PalmerPenguins.load()` asks you to download the data, so you have to set `ENV["DATADEPS_ALWAYS_ACCEPT"] = true` to automatically accept it. ```{julia} #| fig-width: 6 #| fig-height: 4 #| fig-align: center using PalmerPenguins using StatsPlots using DataFrames ENV["DATADEPS_ALWAYS_ACCEPT"] = true; penguins = DataFrame(PalmerPenguins.load()); histogram( [penguins[penguins.species .== species, :].flipper_length_mm for species in ["Adelie", "Chinstrap", "Gentoo"]], label = ["Adelie" "Chinstrap" "Gentoo"], fillcolor = [:darkorange :purple :cyan4], fillalpha = 0.5, xlabel = "Flipper length (mm)", ylabel = "Frequency", title = "Penguin Flipper Lengths", legend = :topright, grid = false, legendfontsize = 7, legendtitle = nothing, bar_width = 5, size = (480, 320), dpi = 300, tickfontsize = 6, guidefontsize = 10, titlefontsize = 12, margin=Plots.Measures.Length(:mm, 2.0) ) ``` ### Stata Douglas Hemken's [Statamarkdown](https://github.com/hemken/Statamarkdown) allows us to use Stata on Rmarkdown (so as on Quarto!) ```{r, filename = "R"} # devtools::install_github("Hemken/Statamarkdown") library(Statamarkdown) haven::write_dta(penguins, here::here("blog/2023/04/30/quarto_multi_lang/palmerpenguins.dta")) ``` Thanks to `Statamarkdown`, we can easily set the working directory to the directory where the `.qmd` file exists by just calling `cd` in Stata. ```{stata, filename = "Stata"} cd ``` Then, you can run Stata code in a code chunk. ```{stata, results="hide"} use palmerpenguins.dta, clear twoway (hist body_mass_g if species == 1, color(orange%40)) /// (hist body_mass_g if species == 2, color(emerald%40)) /// (hist body_mass_g if species == 3, color(purple%40)), /// xtitle("Body Mass (g)") ytitle("Frequency") /// legend(order(1 "Adelie" 2 "Chinstrap" 3 "Gentoo") pos(1) ring(0) col(1)) /// plotregion(fcolor(white)) graphregion(fcolor(white)) graph export "figure/stata/hist.svg", replace ``` ![](img/stata/hist.svg) But you have to manually export an image file for a plot and add it into a document... ## Tabset I think, one of the most useful case is when you want to support multiple source codes in a textbook or lecture notes. Quarto supports [Tabset layout](https://quarto.org/docs/interactive/layout.html#tabset-panel) and it's so smart! ::: {.panel-tabset} ## R ```{r} #| fig-width: 6 #| fig-height: 4 #| fig-align: center penguins |> ggplot(aes(x = flipper_length_mm, y = body_mass_g, color = species, shape = species)) + geom_point(size = 3) + scale_color_manual(values = c("darkorange","purple","cyan4")) + labs(x = "Flipper Length (mm)", y = "Body Mass (g)", color = NULL, shape = NULL) + theme_minimal() + theme(legend.position = c(0.9, 0.1), panel.grid.major = element_blank()) ``` ## Python ```{python} #| fig-width: 6 #| fig-height: 4 #| fig-align: center plt.clf() sns.scatterplot(data=penguins, x='flipper_length_mm', y='body_mass_g', hue='species', style="species", palette=['#FF8C00', '#159090', '#A034F0']) plt.xlabel('Flipper Length (mm)') plt.ylabel('Body Mass (g)') plt.legend(title = "") plt.show() ``` ## Julia ```{julia} @df penguins scatter( :flipper_length_mm, :body_mass_g, group = :species, markersize = 5, markershape = [:circle :utriangle :rect ], markerstrokecolor = :white, palette = ["#FF8C00", "#159090", "#A034F0"], xaxis = "Flipper Length (mm)", yaxis = "Body Mass (g)", size = (480, 320), dpi = 300, legendfontsize = 8, tickfontsize = 6, guidefontsize = 10, margin=Plots.Measures.Length(:mm, 2.0) ) ``` ## Stata ```{stata, results="hide"} use data/palmerpenguins, clear separate body_mass_g, by(species) twoway scatter body_mass_g? flipper_length_mm, /// mcolor(orange emerald purple) mlab("") /// msymbol(O T S) /// xtitle("Flipper Length (mm)") ytitle("Body Mass (g)") /// legend(order(1 "Adelie" 2 "Chinstrap" 3 "Gentoo") pos(11) ring(0) col(1)) /// plotregion(fcolor(white)) graphregion(fcolor(white)) qui graph export "figure/stata/scatter.svg", replace ``` ` ![](img/stata/scatter.svg){fig-align=center} ::: Have a happy Quarto life 🥂!