bug with the font parameter for annotation

[stla]

I want to add an annotation to a plotly heatmap. There is a bug with the font parameter. I've followed the example given here.

library(plotly)

### heatmap data ### 
x <- 1:3; y <- 3:5
dd <- transform(expand.grid(x=x, y=y), z=x+y)

### plotly graphic ###
pp <- plot_ly(data=dd, x=~x, y=~y, z=~z, type="heatmap",
              hoverinfo = 'text',
              text = ~paste('<em>x</em>: ', dd$x, 
                            '</br> <em>y</em>: ', dd$y,
                            '</br> <em>z</em>: ', dd$z)) %>%
  layout(xaxis=list(title="<em>xaxis title</em>"))

### add annotation ###
pp <- pp %>% 
  add_annotations(  x = 1,
                    y = 4,
                    text = "<em>MY ANNOTATION</em>",
                    xref = "x",
                    yref = "y",
                    showarrow = TRUE,
                    arrowhead = 7,
                    ax = 10,
                    ay = 30,
                    font = list(color = 'red',
                                family = 'sans serif',
                                size = 24)
  )

That does not yield the expected result:

In fact, the problem lies in the fact that plotly_build generates three annotation layouts, one for each parameter of the font list:

> bpp <- plotly_build(pp)
> annotation <- bpp$x$layout$annotations
> length(annotation)
[1] 3
> sapply(annotation, function(x) x$font)
[1] "red"        "sans serif" "24"  

Thus, here is a way to get the expected result:

myannotation <- annotation[[1]]
myannotation$font <-  list(color = 'red',
                           family = 'sans serif',
                           size = 24)
bpp$x$layout$annotations <- myannotation
bpp

> packageVersion("plotly")
[1] ‘4.5.2’

[stla]

This bug does not occur anymore with plotly 4.5.6.

But there's still something strange, there is some replicated code:

> bpp <- plotly_build(pp)
> annotation <- bpp$x$layout$annotations
> length(annotation)
[1] 3
> identical(annotation[[1]], annotation[[2]])
[1] TRUE
> identical(annotation[[2]], annotation[[3]])
[1] TRUE

We get exactly the same plot if we do:

bpp$x$layout$annotations <- annotation[[1]]
bpp