5

I have an Rmarkdown with a simple scatter plot (a map for instance), and I would like users to be able to provide some arbitrary x and y coordinates via an input and have those plotted on the graph (in red in the example below). The problem is, I don't have a shiny server so I cannot rely on that option. Is there a implement this, for instance, via javascript or something?

This is what I have:

---
title: "Untitled"
output: html_document
---

```{r setup, include=FALSE}
library(ggplot2)
library(plotly)
```

```{r fig.height=4, fig.width=4}
X <- data.frame(x = 1:10, y = 1:10)
gg <- ggplot(X, aes(x, y)) + geom_point()
ggplotly(gg)
```

This is what I am looking for:

enter image description here


Edit
The example above is a simplification. In reality, the grid is 360x240 and the coordinates can only be integers.

Edit 2 @JohanRosa already provided a nice answer by rebuilding the plot entirely on plotly.js. However, my ggplot is in fact quite complexe and I have many of them. It would therefore be quite complicated for me to rebuild each of them into plotly.js. This is the reason I am looking for an solution that can work directly on the ggplot(ly) that I have.

mat
  • 2,412
  • 5
  • 31
  • 69
  • 1
    You're probably aware you can host a Shiny app for free at [shinyapps.io](https://www.shinyapps.io) without needing to have your own server? Would only be a good option if your app doesn't have to be private and you don't need >25 hours total computation time/month. – zephryl Mar 08 '22 at 17:19
  • @zephryl I am aware yes, and I need more than >25h and neither can afford to pay for the service, that is why I am looking for an alternative. – mat Mar 09 '22 at 07:45

3 Answers3

7

We can use htmlwidgets::onRender to inject custom JS code into your ggplotly object.

I reused @JohanRosa's inputs (thanks! +1) and provided an id to the container div to listen on the inputs. Furthermore I'm using Plotly.restyle to avoid redrawing the plot.

Please check the following:

---
title: "ggplotly user inputs"
output: html_document
---
    
:::{#inputcontainerid .input-container}

:::{.xs}
### X coordinate
<input type='number' value=5 id='x1' class='x'>
<input type='number' value=2.5 id='x2' class='x'>
<input type='number' value=7.5 id='x3' class='x'>
:::

:::{.ys}
### Y coordinate
<input type='number' value=10 id='y1'>
<input type='number' value=5 id='y2'>
<input type='number' value=2.5 id='y3'>
:::

:::

<!-- css configuration to arrange the inputs -->
```{css, echo = FALSE}
input {
  display: block;
}

.xs, .ys {
  display: inline-block;
}
```

```{r setup, include=FALSE}
library(ggplot2)
library(plotly)
library(htmlwidgets)
```

```{r out.width='100%', echo=FALSE}
X <- data.frame(x = 1:10, y = 1:10)

JS <- "
function(el, x){
  var id = el.getAttribute('id');
  var gd = document.getElementById(id);
  
  let defaultInputs = {
    x: [$('#x1').val(), $('#x2').val(), $('#x3').val()],
    y: [$('#y1').val(), $('#y2').val(), $('#y3').val()],
    mode: 'markers',
    type: 'scatter',
    name: 'user'
  };
  
  Plotly.addTraces(gd, defaultInputs);
  
  document.getElementById('inputcontainerid').addEventListener('input', function(event){
    let userInputs = {
      x: [[$('#x1').val(), $('#x2').val(), $('#x3').val()]],
      y: [[$('#y1').val(), $('#y2').val(), $('#y3').val()]]
    };
    Plotly.restyle(gd, userInputs, 1);
  });
}
"

gg <- ggplot(X, aes(x, y)) + geom_point()
ggplotly(gg) %>% 
  layout(xaxis = list(autorange = TRUE), yaxis = list(autorange = TRUE)) %>%
  onRender(jsCode = JS)
```

result

For additional infos please see chapter 5 "Event handling in JavaScript" from Carson Sievert's book Interactive web-based data visualization with R, plotly, and shiny.

ismirsehregal
  • 30,045
  • 5
  • 31
  • 78
1

I did it using plotlty.js directly in JavaScript. I think with this you can advance.

---
output: html_document
---

<!-- This is a container for your inputs -->

:::{.input-container}

:::{.xs}
### X coordinate
<input type='number' value=5 id='x1' class='x'>
<input type='number' value=2.5 id='x2' class='x'>
<input type='number' value=7.5 id='x3' class='x'>
:::

:::{.ys}
### Y coordinate
<input type='number' value=10 id='y1'>
<input type='number' value=5 id='y2'>
<input type='number' value=2.5 id='y3'>
:::

:::

<!-- 
I did it using a submit button, I have to read more to make it totally reactive
-->
<input type='button' id='plot' value='Update points' class='btn btn-primary'>


<!-- The next div is a placeholder for the plot -->
<div id="tester" style="width:600px;height:250px;"></div>

<!-- You have to include the plolty.js script -->
<script src="https://cdn.plot.ly/plotly-2.9.0.min.js"></script>


<!-- css configuration to arrange the inputs -->
```{css, echo = FALSE}
input {
  display: block;
}

.xs, .ys {
  display: inline-block;
}
```

<!-- This is the magic, the Js code -->
<!-- language: lang-js -->
```{js, echo=FALSE}

// Get the html element that should contain the plot
plot = document.getElementById('tester');

// Create an object with the default data
let var1 = {
  x: [1, 2, 3, 4, 5, 6, 7, 8, 9, 10],
  x: [1, 2, 3, 4, 5, 6, 7, 8, 9, 10],
  mode: 'markers',
  type: 'scatter',
  name: 'deafault'
}


let layout = {
  xaxis: {
    range: [ 0, 10.5 ]
  },
  yaxis: {
    range: [0, 10.5]
  },
  title:'Testing'
};

let data = [var1]

// Default plot
Plotly.newPlot(plot, data, layout);


// Using jQuery add an event listener on click to de button
// that way when the user click on it the plot will be updated
$('#plot').click(function(){
  let userInputs = {
    x: [$('#x1').val(), $('#x2').val(), $('#x3').val()],
    y: [$('#y1').val(), $('#y2').val(), $('#y3').val()],
    mode: 'markers',
    type: 'scatter',
    name: 'user'
  }
  
  data = [var1, userInputs] 
  
  Plotly.newPlot(plot, data, layout);
})
``` 

enter image description here

Johan Rosa
  • 2,797
  • 10
  • 18
  • Thanks Johan, this indeed works. I wish however to use it on my ggplot, as the one that I have is quite complexe and I am not sure I will be able to reproduce it with plotly.js. If there no answer in a few days from now, I'll accept yours. – mat Mar 07 '22 at 13:49
  • 1
    maybe there is a way to combine this approach with the package crosstalk? – Mike Mar 09 '22 at 15:16
  • 2
    I will take a look. I did not know about that package. – Johan Rosa Mar 09 '22 at 15:29
0

This may not be what you want but you can do this by adding a runtime of shiny in your yaml

---
title: "Untitled"
output: html_document
runtime: shiny
---

```{r setup, include=FALSE}
library(ggplot2)
library(plotly)
library(shiny)
```

```{r shinyInputs}

shiny::numericInput('someInput', "Some Number", value = 5)
shiny::numericInput('someInput2', "Some Number2", value = 2)

plotlyOutput('gg')

```

```{r fig.height=4, fig.width=4}
X <- data.frame(x = 1:10, y = 1:10)

output$gg <- renderPlotly({
  temp <- tibble::tibble(x = input$someInput, y = input$someInput2)
  ggplotly(ggplot(X, aes(x, y)) + geom_point() + geom_point(data = temp, aes(x =         
x, y = y), color = 'red'))
  })
```
MattO
  • 171
  • 1
  • 6
  • Thank you but unfortunately this is indeed not what I want as I am looking for a solution that does not rely shiny (as mentioned in my post). – mat Mar 04 '22 at 21:23