-
DescriptionI've tried to mix different content: markdown headers, static graphs and html widgets (tables and graphs). The reason behind this is to build parametrised reports where the titles will be the comparisons and there might be different types of output generated. I've tried tons of options because I wasn't sure about how to approach this, and so far, only one option (out of 26) seems to be working. It's important to mention that what I see as output in the IDE does not match what I see as output after doing Quarto rendering (likely because it treats each part separately). I come here to see if there's a "correct"/official way to do this, and if not, I'll use method 23, which is renderTags + cat(x$html). Working solution: |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 7 replies
-
|
@joanbadia Please provide a small reproducible example. There is no way someone can help you with a one thousand lines of code. You likely need |
Beta Was this translation helpful? Give feedback.
-
|
The reproducible example is just the first part of the code... and knitr::knit_child was already attempted in method n. 20 and did not work, thought I might have not used it correctly. The reason the code is so long it's because I actually had to spend a good 2-3 hours testing many ways to do mixed content rendering... it's not something that can be easily done or be easily found on documentation. Updated the post with shorter code. |
Beta Was this translation helpful? Give feedback.
-
|
There are many examples of using knit_child on the internet. You can search in the discussions to find those examples.
The documentation for knitr/rmarkdown tricks: You need to understand that if you don't provide a small and simple reproducible example, the likelihood of you getting help is quite low.
I get that but asking others to spend the same amount of time to read and test your one thousand lines of code won't work. |
Beta Was this translation helpful? Give feedback.
-
|
From my understanding of your example this is how it would be done
This child doc would be used in main doc through ```{r}
#| results: asis
#| echo: false
#| message: false
#| warning: false
for(content_type in content_types) {
res <- knitr::knit_child(
"_child.qmd",
quiet = TRUE
)
cat(res, sep = "\n")
}
```Full example ---
title: "Markdown header + HTML Widget + Static Graph"
format:
html:
embed-resources: true
html-table-processing: none
---
```{r}
#| label: setup
#| echo: false
#| message: false
library(dplyr)
library(reactable)
library(htmltools)
library(purrr)
library(knitr)
library(ggplot2)
library(plotly)
# Create test reactable function
create_test_table <- function(data) {
reactable(
data,
defaultPageSize = 5,
striped = TRUE,
compact = TRUE,
style = list(fontSize = "12px"),
theme = reactableTheme(
color = "#333",
backgroundColor = "#f9f9f9",
borderColor = "#ddd"
)
)
}
# Create small ggplot
create_ggplot <- function(data) {
ggplot(data, aes(x = Category, y = Value)) +
geom_col(fill = "steelblue", alpha = 0.7) +
theme_minimal() +
theme(
text = element_text(size = 8),
axis.text = element_text(size = 6),
plot.title = element_text(size = 9)
) +
labs(title = "Values by Category")
}
# Create small plotly
create_plotly <- function(data) {
plot_ly(data, x = ~Score, y = ~Value, type = 'scatter', mode = 'markers',
marker = list(size = 6, color = 'red', opacity = 0.7)) %>%
layout(
title = list(text = "Score vs Value", font = list(size = 10)),
xaxis = list(title = list(text = "Score", font = list(size = 8))),
yaxis = list(title = list(text = "Value", font = list(size = 8))),
width = 300,
height = 200,
margin = list(l = 40, r = 20, t = 40, b = 30)
) %>%
config(displayModeBar = FALSE)
}
# Generate single test dataset
set.seed(42)
test_data <- data.frame(
ID = paste0("Item_", 1:8),
Name = paste("Test", letters[1:8]),
Value = round(runif(8, 10, 100), 1),
Category = sample(c("A", "B", "C"), 8, replace = TRUE),
Score = round(rnorm(8, 50, 15), 1)
)
# Content types for realistic loop testing
content_types <- c("HTML table", "static graph", "html graph")
```
## Creating Section with `knit_child()`
```{r}
#| results: asis
#| echo: false
for(content_type in content_types) {
res <- knitr::knit_child(
"_child.qmd",
quiet = TRUE
)
cat(res, sep = "\n")
}
```
This could be done different in the child document with several chunk and conditional evaluation based on
The important part of all this is that I don't know what are you others 20+ methods, but this is how I would do it. Hope it helps |
Beta Was this translation helpful? Give feedback.
-
|
Thank you, I am trying to focus on writing parametrised reports and I need an extensive study of how Quarto/Rmarkdown/knitr handles mixed type content, because there are some things I could resolve with 2 for loops (one for comparisons one for methods) but it seems it struggles with mixed content. It also seems some type of outputs are a bit more resistant to mixed types (kable > reactable), so that's why I tried to test so many methods at once, to know more about how mixed type is handled and then focusing on one way. This is for example, a real application of this problem (GSEA output display), where in a single tabset I can put all the methods with titles, table and a graph. Sometimes this graph will be plotly, sometimes it won't, that's why I was trying to learn more about it. I felt at the time that a knit_child for just printing output was a bit overkill, but I'll think about it, and maybe move all the plotting code there too. I also have the option to use "text = " too, thanks for the advice. |
Beta Was this translation helpful? Give feedback.
-
|
You should not mix Instead your That's how "mixed content" is supposed to be handled in Here you are relying on some |
Beta Was this translation helpful? Give feedback.
-
|
@joanbadia The key is to understand that knitr outputs cells using a custom printing method which is described in one of the vignette. You can see This is the function and associated methods that handle complex object printing. They do more than just The knitting context is important for formatting the cell output, and this is why Also, there is some limitation in evaluation, where only first level printing of objects works, but nesting printing inside for-loop is not something that will be catch by the evaluation process in knitr to be formatted for output. I hope this clarifies a little why |
Beta Was this translation helpful? Give feedback.
-
|
Yes, thanks to both of you I understood why you were recommending me knit_child, the main purpose being properly adding the entire code chunks that will trigger knit_print() instead of trying to directly print. To rely on the knitting process entirely. Thanks to your advices I got a working solution for my real problem application. One using purrr::map and the other using for loops, but they're essentially the same thing. There were many things I was doing wrong (besides mixing prints and cats), but the end result now as it's dependent on the knitting process it does accept kable, reactable or datatable. It'll also allow me to do multiple things in a flexible manner (like saving plots in PDF format) amongst others. Final results:Solution with for loopsSolution with map |
Beta Was this translation helpful? Give feedback.
-
|
Glad it works for you now ! |
Beta Was this translation helpful? Give feedback.
From my understanding of your example this is how it would be done
_child.qmdThis child doc would be used in main doc through
knitr::knit_child()Full example