Incorrect output sankey diagram when using shiny - r

When I create a sankey diagram in a regular R session the output looks ok. The tooltip shows an arrow between the connections:
require(rCharts)
require(rjson)
links <- matrix(unlist(
rjson::fromJSON(
file = "http://bost.ocks.org/mike/sankey/energy.json"
)$links
),ncol = 3, byrow = TRUE)
nodes <- unlist(
rjson::fromJSON(
file = "http://bost.ocks.org/mike/sankey/energy.json"
)$nodes
)
links <- data.frame(links)
colnames(links) <- c("source", "target", "value")
links$source <- sapply(links$source, FUN = function(x) {return(as.character(nodes[x+1]))}) #x+1 since js starts at 0
links$target <- sapply(links$target, FUN = function(x) {return(nodes[x+1])}) #x+1 since js starts at 0
sankeyPlot <- rCharts$new()
sankeyPlot$setLib('http://timelyportfolio.github.io/rCharts_d3_sankey')
sankeyPlot$set(
data = links,
nodeWidth = 15,
nodePadding = 10,
layout = 32,
width = 960,
height = 500,
units = "TWh",
title = "Sankey Diagram"
)
sankeyPlot
When I create it in shiny, the arrow in the tooltip is replaced by unusual characters. Also below the plot an unusual character is printed. I needed to download the d3_sankey library to make the shiny app version work, so if you want to reproduce it you have to change the path in the setLib statement. How can this be fixed?
require(shiny)
runApp(list(
ui = pageWithSidebar(
headerPanel('Test'),
sidebarPanel( 'Test' ),
mainPanel(
chartOutput("Plot", 'C:/R-3.0.1/library/rCharts/libraries/sankey')
)
),
server = function(input, output, session){
output$Plot <- renderChart2({
sankeyPlot2 <- rCharts$new()
sankeyPlot2$setLib('C:/R-3.0.1/library/rCharts/libraries/sankey')
sankeyPlot2$set(
data = links,
nodeWidth = 15,
nodePadding = 10,
layout = 32,
width = 960,
height = 500,
units = "TWh",
title = "Sankey Diagram"
)
return(sankeyPlot2)
})
}
))
> sessionInfo()
R version 3.0.2 (2013-09-25)
Platform: x86_64-w64-mingw32/x64 (64-bit)
locale:
[1] LC_COLLATE=Dutch_Belgium.1252 LC_CTYPE=Dutch_Belgium.1252
[3] LC_MONETARY=Dutch_Belgium.1252 LC_NUMERIC=C
[5] LC_TIME=Dutch_Belgium.1252
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] shiny_0.8.0.99 rjson_0.2.13 rCharts_0.4.2
loaded via a namespace (and not attached):
[1] bitops_1.0-6 caTools_1.16 digest_0.6.4 grid_3.0.2
[5] httpuv_1.2.1 lattice_0.20-23 plyr_1.8 Rcpp_0.10.6
[9] RCurl_1.95-4.1 RJSONIO_1.0-3 tools_3.0.2 whisker_0.3-2
[13] xtable_1.7-1 yaml_2.1.10

The problem is related to character encoding in multiple files. This is how I resolved the issues on my Windows 7 machine.
Mouseover tooltip issue
An arrow character is used to construct the "link" between source and target. It occurs in these files:
example_build_network_sankey.html
layouts\chart.html
layouts\chart_static_title.html
layouts\chart.html
Replace the arrow with the ASCii characters ->
so the code looks like:
.text(function (d) { return d.source.name + " -> " + d.target.name + "\n" + format(d.value); });
Characters below the chart
 is found in:
\libraries\highlighters\prettify\css\sunburst.css
\layouts\chart.html
\libraries\widgets\d3_sankey\layouts\chart.html
I used the Search and Replace in Files facility in UltraEdit to replace this special character with a blank space. This one is tricky because I could not see the character in the UE editor. If I highlighted the blank space it appears as a backtick. The character is also found in jquery-1.8.2.min.js.

Related

R Shiny and Heatmaps - no apparent error for debugging

I am trying to create a script to generate heatmaps in R shiny. I attach the script below
library(shiny)
options(shiny.maxRequestSize = 50*1024^2)
ui <- fluidPage(
sidebarLayout(
sidebarPanel(
fileInput('expression_table','Expression file (xlsx format)'),
selectInput('normalization','Normalize the values with', choices = c('None','CPM','LogCPM','Z-score','Log only')),
fileInput('metadata','Metadata file (xlsx format)'),
textInput('annotcol', 'Annotation column of metadata file', value = NULL),
textInput('gene_list','List of genes to plot (1 gene per line)', value = NULL),
textInput('heatmap_title','Heatmap title', value = 'Heatmap title'),
selectInput('color_scale','Choose your color scale', choices = c('Blue-White-Red',
'Red-White-Blue',
'Green-Black-Purple',
'Purple-Black-Green')
),
sliderInput('colsize','Column labels size', min=0, max=20, step=1, value=8 ),
sliderInput('colkm','Number of Column clusters', min=0, max=10, step=1, value=1 ),
selectInput('coldend','Show column dendrogram', choices = c(TRUE,FALSE)),
sliderInput('rowsize','Row labels size', min=0, max=20, step=1, value=8 ),
sliderInput('rowkm','Number of Row clusters', min=0, max=10, step=1, value=1 ),
selectInput('rowdend','Show row dendrogram', choices = c(TRUE,FALSE)),
actionButton('click','Generate Heatmap')
),
mainPanel(
plotOutput('heatmap')
)
)
)
server <- function(input, output, session) {
library(ComplexHeatmap)
library(circlize)
library(dplyr)
library(openxlsx)
library(edgeR)
library(stringr)
expression <- reactive({
read.xlsx(input$expression_table, rowNames = TRUE, colNames=TRUE)
})
expression_normalized <- reactive({
if(input$normalization == 'CPM'){
as.data.frame(cpm(expression))
} else if(input$normalization == 'LogCPM'){
as.data.frame(log(cpm(expression)+1))
} else if(input$normalization == 'Z-score'){
as.data.frame(t(scale(t(cpm(expression)))))
} else if(input$normalization == 'None'){
read.xlsx(input$expression_table, rowNames = TRUE, colNames=TRUE)
} else if(input$normalization == 'Log only'){
as.data.frame(log(expression + 1))
}
})
metadata <- reactive({
read.xlsx(input$metadata, rowNames = TRUE, colNames=TRUE)
})
expression_isolated <- reactive({
if(!is.null(input$gene_list)){
genes_list <- unlist(strsplit(input$gene_list, split = '\n'))
expression_normalized %>% filter(rownames(.) %in% genes_list)
} else {
expression_normalized
}
})
CS <- reactive({
if(input$color_scale == 'Blue-White-Red'){
colorRamp2(c(min(expression_isolated),0,max(expression_isolated)),
c("steelblue3","white","firebrick3"))
} else if(input$color_scale == 'Red-White-Blue'){
colorRamp2(c(min(expression_isolated),0,max(expression_isolated)),
c("firebrick3","white","steelblue3"))
} else if(input$color_scale == 'Green-Black-Purple'){
colorRamp2(c(min(expression_isolated),0,max(expression_isolated)),
c("olivedrab3","black","mediumorchid3"))
} else if(input$color_scale == 'Purple-Black-Green'){
colorRamp2(c(min(expression_isolated),0,max(expression_isolated)),
c("mediumorchid3","black","olivedrab3"))
}
})
#output
eventReactive(input$click, {
output$heatmap <- renderPlot({
if(!is.null(input$annotcol)) {
Heatmap(as.matrix(expression_isolated),
col = CS,
row_names_gp = gpar(fontsize = input$rowsize),
column_names_gp = gpar(fontsize = input$colsize),
column_km = input$colkm,
row_km= input$rowkm,
show_row_dend = input$rowdend,
show_column_dend = input$coldend,
column_title = input$heatmap_title,
top_annotation = HeatmapAnnotation(Condition = metadata[,input$annotcol], which = 'column')
)
} else {
Heatmap(as.matrix(expression_isolated),
col = CS,
row_names_gp = gpar(fontsize = input$rowsize),
column_names_gp = gpar(fontsize = input$colsize),
column_km = input$colkm,
row_km= input$rowkm,
show_row_dend = input$rowdend,
show_column_dend = input$coldend,
column_title = input$heatmap_title
)
}
})
})
}
shinyApp(ui, server)`
The application runs fine and generates the UI. However, when I upload the excel file with the data and press 'Generate Heatmap' I get no response and no error.
I also used observeEvent instead of eventReactive. When I used observeEvent for the click handling I get the following error:
"Error in as.vector: cannot coerce type 'closure' to vector of type 'any'"
The file input is a standard excel file with rownames in the first column, headers in the first row and numeric values for data, i.e.
SAMPLE1 SAMPLE2 SAMPLE3
IFNI 10 11 13
IFNII 11 16 15
TP53 45 22 56
Anyone know what's going on?
> sessionInfo()
R version 4.2.0 (2022-04-22)
Platform: x86_64-apple-darwin17.0 (64-bit)
Running under: macOS Monterey 12.4
Matrix products: default
LAPACK: /Library/Frameworks/R.framework/Versions/4.2/Resources/lib/libRlapack.dylib
locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
attached base packages:
[1] grid stats graphics grDevices utils datasets methods base
other attached packages:
[1] stringr_1.5.0 edgeR_3.38.4 limma_3.52.4 openxlsx_4.2.5.1
[5] dplyr_1.0.10 circlize_0.4.15 ComplexHeatmap_2.13.1 shiny_1.7.4
loaded via a namespace (and not attached):
[1] Rcpp_1.0.10 locfit_1.5-9.7 lattice_0.20-45 png_0.1-8
[5] assertthat_0.2.1 digest_0.6.31 foreach_1.5.2 utf8_1.2.2
[9] mime_0.12 R6_2.5.1 stats4_4.2.0 pillar_1.8.1
[13] GlobalOptions_0.1.2 rlang_1.0.6 rstudioapi_0.14 jquerylib_0.1.4
[17] S4Vectors_0.36.1 GetoptLong_1.0.5 textshaping_0.3.6 compiler_4.2.0
[21] httpuv_1.6.8 systemfonts_1.0.4 pkgconfig_2.0.3 BiocGenerics_0.44.0
[25] shape_1.4.6 htmltools_0.5.4 tidyselect_1.2.0 tibble_3.1.8
[29] IRanges_2.32.0 codetools_0.2-18 matrixStats_0.63.0 fansi_1.0.4
[33] crayon_1.5.2 later_1.3.0 jsonlite_1.8.4 xtable_1.8-4
[37] lifecycle_1.0.3 DBI_1.1.3 magrittr_2.0.3 zip_2.2.2
[41] cli_3.6.0 stringi_1.7.12 cachem_1.0.6 promises_1.2.0.1
[45] doParallel_1.0.17 bslib_0.4.2 ellipsis_0.3.2 ragg_1.2.5
[49] generics_0.1.3 vctrs_0.5.2 rjson_0.2.21 RColorBrewer_1.1-3
[53] iterators_1.0.14 tools_4.2.0 glue_1.6.2 parallel_4.2.0
[57] fastmap_1.1.0 clue_0.3-63 colorspace_2.1-0 cluster_2.1.4
[61] memoise_2.0.1 sass_0.4.5
BLUF: change input$metadata to input$metadata$datapath.
If you read ?shiny::fileInput and go down to the "Server value:" section, it reads:
Server value:
A 'data.frame' that contains one row for each selected file, and
following columns:
'name' The filename provided by the web browser. This is *not* the
path to read to get at the actual data that was uploaded (see
'datapath' column).
'size' The size of the uploaded data, in bytes.
'type' The MIME type reported by the browser (for example,
'text/plain'), or empty string if the browser didn't know.
'datapath' The path to a temp file that contains the data that was
uploaded. This file may be deleted if the user performs
another upload operation.
By using read.xlsx(input$metadata, ...), you are passing a data.frame to read.xlsx, which obviously doesn't work. There will likely be nothing in the shiny app, since you're not capturing warnings or errors. You should have seen something like:
read.xlsx(data.frame(name="myfile", datapath="myfile.xlsx"))
# Error in file(description = xlsxFile) : invalid 'description' argument
Multiple recommendations:
Change your code to use input$metadata$datapath;
Use req and possible validate/need, since you can avoid trying to read NULL (which can happen due to the order of reactivity) and, if not met, sometimes you can provide clear verbiage in the app interface itself.
Use tryCatch for anything that is even somewhat out of your control. By doing so, you give yourself more information and ability to recover when something goes wrong.
At a minimum, this is your fix.
metadata <- reactive({
read.xlsx(input$metadata$datapath, rowNames = TRUE, colNames=TRUE)
})
But perhaps this app presents the other recommendations clearly:
library(shiny)
shinyApp(
ui = fluidPage(
fileInput("metadata", "MetaData"),
plotOutput("plot")
),
server = function(input, output, session) {
mydata <- reactive({
req(input$metadata)
res <- tryCatch(
openxlsx::read.xlsx(input$metadata$datapath, rowNames = TRUE, colNames = TRUE),
error = function(e) e
)
validate(
need(
!inherits(res, "error"),
paste("There was a problem reading your file:",
paste(conditionMessage(res), collapse = "; "))
)
)
})
output$plot <- renderPlot({
req(mydata())
plot(y ~ x, data = mydata())
})
}
)
The error message is in the place the plot should normally show up. If there are multiple components that use mydata(), the error message in this example will be shown in all of the components (that support validate/need rendering). You can limit this by moving the validate(..) statement out of mydata <- and into one (or more) dependent components, though all components would then need to either check for req(!inherits(mydata(), "error")) or some other way communicate what happened.

Shiny selectInput/pickerInput with long names should overflow sidebar

I want to built a shiny app with bs4Dash in sidebar layout. The sidebar contains a dropdown selection menu with items that have long names. Once I open the dropdown, I would like the full item name to be visible, i.e. overflowing the dashboard's body. The names are cut at the sidebar border by default (shiny::selectizeInput) or the dropdown content is right-aligned to the sidebar border and the start of the item name is off-screen to the left (shinyWidgets::pickerInput).
This is how the app looks (updated 2022-12-16):
I tried to apply the solution for a flexdashboard described
here,
but could not get it working.
Thanks for your help!
Here is a reproducible example of my app:
# app.R
library(shiny)
library(bs4Dash)
library(shinyWidgets)
vec_long_items <- sapply(1:10, function(i) {
paste("START", paste(sample(letters, 100, replace = TRUE), collapse = ""))
})
shinyApp(
ui = dashboardPage(
header = bs4DashNavbar(
title = "Long items to select", disable = TRUE, controlbarIcon = NULL
),
sidebar = bs4DashSidebar(
skin = "white",
shinyWidgets::pickerInput(
inputId = "in1", label = "shinyWidgets::pickerInput", choices = vec_long_items
),
shiny::selectInput(
inputId = "in2", label = "shiny::selectInput", choices = vec_long_items
)
),
body = dashboardBody(tableOutput("out_text"))
),
server = function(input, output, session) {
output$out_text <- renderTable(data.frame(items = vec_long_items))
},
options = list(launch.browser = FALSE)
)
My sessionInfo():
R version 4.1.0 (2021-05-18)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 19044)
Matrix products: default
locale:
[1] LC_COLLATE=German_Switzerland.1252 LC_CTYPE=German_Switzerland.1252 LC_MONETARY=German_Switzerland.1252 LC_NUMERIC=C LC_TIME=German_Switzerland.1252
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] bs4Dash_2.1.0 shiny_1.7.2
loaded via a namespace (and not attached):
[1] Rcpp_1.0.7 shinyWidgets_0.7.5 digest_0.6.29 later_1.3.0 mime_0.12 R6_2.5.1 lifecycle_1.0.2 xtable_1.8-4 jsonlite_1.8.0 magrittr_2.0.3
[11] cachem_1.0.6 rlang_1.0.5 cli_3.4.0 fontawesome_0.3.0 promises_1.2.0.1 jquerylib_0.1.4 bslib_0.4.0 ellipsis_0.3.2 tools_4.1.0 httpuv_1.6.5
[21] fastmap_1.1.0 compiler_4.1.0 memoise_2.0.1 htmltools_0.5.2 sass_0.4.2
Try this css:
css <- ".main-sidebar, .main-sidebar .sidebar .os-viewport, .os-host-overflow, .os-host-overflow>.os-padding {overflow: visible !important;}"
i.e.
body = dashboardBody(
tags$head(
tags$style(HTML(css))
),
and
shinyWidgets::pickerInput(
inputId = "in1", label = "shinyWidgets::pickerInput", choices = vec_long_items,
options = pickerOptions(dropdownAlignRight = TRUE)
),
I found a CSS property for pickerInput that improves the output. Change the body as follows:
# ...
body = dashboardBody(
tags$head(
tags$style(HTML("
.dropdown-menu.show {
display: contents;
}
"))
),
tableOutput("out_text")
)
# ...
Now there is a scrollbar to look at the complete element. Not what I wanted but already better.
I am still looking for an "overflow-solution" :-)

crosstab output getting displayed in viewer pane only and not in Shiny app

I am generating a cross table output (contingency table). My requirement is to have column percentages along with row and column totals. I found this package "sjPlot" which gives very good output with colour features and was meeting my requirement. I used this package to generate the output in Shiny. But to my surprise the output is getting generated only in the viewer pane and not in the app. Below is a small part of my code.
library("shiny")
library("sjPlot")
ui <- shinyUI(fluidPage(
tabPanel("First Page"),
mainPanel(tabsetPanel(id='mytabs',
tabPanel("Table",tags$b(tags$br("Table Summary" )),tags$br(),dataTableOutput("crosstab2"))
)
)
))
server <- shinyServer(function(input, output,session){
updateTabsetPanel(session = session
,inputId = 'myTabs')
output$crosstab2 <- renderDataTable({
with(mtcars,sjt.xtab(mpg, cyl,
var.labels = c("mpg","cyl"),
show.col.prc = T,
show.summary = F,
show.na = F,
wrap.labels = 50,
tdcol.col = "#f90477",
emph.total = T,
emph.color = "#3aaee0",
#use.viewer = T ##when this is false it generates a html output. Can we capture the html output and display in Shiny?
CSS = list(css.table = "border: 1px solid;",
css.tdata = "border: 1px solid;")))
})
print("created crosstab1")
})
shinyApp(ui = ui, server = server)
This generates the needed output in viewer pane. How can I get this output displayed in the app. I can not save this output as a image as in my actual program the variables are getting selected dynamically and the table output accordingly changes.
sjt.xtabinvisibly returns a list that contains the html and css used to generate the table that you see in the viewer.
You can use htmlOutput on the ui side and renderUI on the server side to display it.
In your ui.R, you can use:
htmlOutput("crosstab2")
In your server.R:
output$crosstab2 <- renderUI({
custom_table <- sjt.xtab(mtcars$mpg, mtcars$cyl,variableLabels = c("mpg","cyl"),showColPerc = T,
showSummary = F,showNA=F,highlightTotal = T,tdcol.col = "#f90477", highlightColor ="#3aaee0",
CSS = list(css.table = "border: 1px solid;",
css.tdata = "border: 1px solid;"))
HTML(custom_table$knitr)
})
Here's (part of) my sessionInfo()
R version 3.1.2 (2014-10-31)
Platform: x86_64-apple-darwin13.4.0 (64-bit)
locale:
[1] en_GB.UTF-8/en_GB.UTF-8/en_GB.UTF-8/C/en_GB.UTF-8/en_GB.UTF-8
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] sjPlot_1.9.2 shiny_0.13.1
I suspect we don't have the same version of sjPlot as the arguments of the sjt.xtab you use in your example have changed names.

Can't get interactive zooming to work with ggvis

I'm trying to get interactive zooming to work in ggvis, more in particular zooming using a brush. Judging from https://github.com/rstudio/ggvis/issues/143 I thought this should work.
I have the following shiny and ggvis code (is fully reproducible):
## ui.R
library(ggvis)
shinyUI(fluidRow(
uiOutput('ui_plot1'),
ggvisOutput("graph_plot1")
))
## server.R
shinyServer(function(input, output, session) {
domains <- reactiveValues(x = c(NA, NA), y = c(NA, NA))
zoom_brush = function(items, session, page_loc, plot_loc, ...) {
domains$x = c(200, 400)
}
plot = reactive({
mtcars %>%
ggvis(~disp, ~mpg) %>%
layer_points() %>%
scale_numeric('x', domain = domains$x, clamp = TRUE) %>%
handle_brush(zoom_brush)
}) %>% bind_shiny('graph_plot1', 'ui_plot1')
})
So, as soon as a brush is drawn, the domains reactive is changed, which in turn changes the domain of the x scale_numeric. If still have the following challenges:
Inside zoom_brush I get the coordinates of the brush, but in the pixel coordinate system of the plot not the domain coordinate system. How can I translate the pixels to the domain scale? In d3 I can simply use the range to scale transform functions, but I don't see how these are available in ggvis (via vega).
The handle_brush function only supports setting a on_move event handler. In this case I only want to trigger the zoom if the brush finishes, so the onmouseup event in the context of the brush. I fear that this is simply not possible right now?
Only when I set clamp = TRUE do I get an effective zoom. In not, the points outiside the domain are still shown and only the axes are set to the new domain. Is there an easy fix for this? Or should I make the dataset a reactive, and subset it based on the domain that was set by the brush?
I run the following R version and package versions.
> sessionInfo()
R version 3.1.1 (2014-07-10)
Platform: x86_64-apple-darwin10.8.0 (64-bit)
locale:
[1] C
attached base packages:
[1] stats graphics grDevices utils datasets methods base
other attached packages:
[1] ggvis_0.4.1 shiny_0.12.0
loaded via a namespace (and not attached):
[1] DBI_0.3.1 R6_2.0.1 Rcpp_0.11.6 assertthat_0.1 digest_0.6.8 dplyr_0.4.1 htmltools_0.2.6 httpuv_1.3.2
[9] jsonlite_0.9.16 lazyeval_0.1.10 magrittr_1.5 mime_0.3 parallel_3.1.1 tools_3.1.1 xtable_1.7-4
I think you need to subset your data: ggvis doesn't yet appear clever enough to ignore out of scale points. The following server.R works for me:
## server.R
shinyServer(function(input, output, session) {
domains <- reactiveValues(x = c(NA, NA), y = c(NA, NA))
mtcars_reactive <- reactive({
if (anyNA(domains$x))
mtcars
else
mtcars[mtcars$disp >= domains[["x"]][1] & mtcars$disp <= domains[["x"]][2], ]
})
zoom_brush = function(items, page_loc, session, ...) { # plot_loc
print(items)
message("page_loc")
print(page_loc)
print(session)
domains$x = c(200, 400)
}
reactive({
mtcars_reactive() %>%
ggvis(~disp, ~mpg) %>%
layer_points() %>%
handle_brush(zoom_brush)
}) %>% bind_shiny('graph_plot1', 'ui_plot1')
})

How to save plots that are made in a shiny app

I'm trying to figure out how to use downloadButton to save a plot with shiny. The example in the package demonstrates downloadButton/downloadHandler to save a .csv. I'm going to make a reproducible example based on that.
For ui.R
shinyUI(pageWithSidebar(
headerPanel('Downloading Data'),
sidebarPanel(
selectInput("dataset", "Choose a dataset:",
choices = c("rock", "pressure", "cars")),
downloadButton('downloadData', 'Download Data'),
downloadButton('downloadPlot', 'Download Plot')
),
mainPanel(
plotOutput('plot')
)
))
For server.R
library(ggplot2)
shinyServer(function(input, output) {
datasetInput <- reactive({
switch(input$dataset,
"rock" = rock,
"pressure" = pressure,
"cars" = cars)
})
plotInput <- reactive({
df <- datasetInput()
p <-ggplot(df, aes_string(x=names(df)[1], y=names(df)[2])) +
geom_point()
})
output$plot <- renderPlot({
print(plotInput())
})
output$downloadData <- downloadHandler(
filename = function() { paste(input$dataset, '.csv', sep='') },
content = function(file) {
write.csv(datatasetInput(), file)
}
)
output$downloadPlot <- downloadHandler(
filename = function() { paste(input$dataset, '.png', sep='') },
content = function(file) {
ggsave(file,plotInput())
}
)
})
If you're answering this question, you are probably familiar with this, but to get this working, save the above into separate scripts (ui.R and server.R into a folder (foo) within the working directory. To run the shiny app, run runApp("foo").
Using ggsave, I get an error message indicating that ggsave can't use the filename function (I think). If I use the standard graphics device (like below), the Download Plot works without an error, but it doesn't write the graphic.
Any tips to get downloadHandler working for writing plots would be appreciated.
Not sure if this question is still active but it's the first one that came up when searching for "saving plots in shiny app" so I wanted to quickly add how to get ggsave to work with downloadHandler along the lines of the original question.
The alternative strategies suggested by juba using direct output instead of ggsave and alternative strategy suggested by alexwhan himself both work great, this is just for those who absolutely want to use ggsave in the downloadHandler).
The problem reported by alexwhan is caused by ggsave trying to match the file extension to the correct graphics device. The temporary file, however, doesn't have an extension so the matching fails. This can be remedied by specifically setting the device in the ggsave function call, like so in the original code example (for a png):
output$downloadPlot <- downloadHandler(
filename = function() { paste(input$dataset, '.png', sep='') },
content = function(file) {
device <- function(..., width, height) grDevices::png(..., width = width, height = height, res = 300, units = "in")
ggsave(file, plot = plotInput(), device = device)
}
)
This call basically takes the device function for a png that ggsave assigns internally (you can look at the ggsave function code to see the syntax for jpg, pdf, etc). Perhaps, ideally, one could specify the file extension (if different from the file name - as is the case here for the temporary file) as a ggsave parameter but this option is currently not available in ggsave.
A minimal self-contained working example:
library(shiny)
library(ggplot2)
runApp(list(
ui = fluidPage(downloadButton('foo')),
server = function(input, output) {
plotInput = function() {
qplot(speed, dist, data = cars)
}
output$foo = downloadHandler(
filename = 'test.png',
content = function(file) {
device <- function(..., width, height) {
grDevices::png(..., width = width, height = height,
res = 300, units = "in")
}
ggsave(file, plot = plotInput(), device = device)
})
}
))
sessionInfo()
# R version 3.1.1 (2014-07-10)
# Platform: x86_64-pc-linux-gnu (64-bit)
#
# locale:
# [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
# [3] LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
# [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
# [7] LC_PAPER=en_US.UTF-8 LC_NAME=C
# [9] LC_ADDRESS=C LC_TELEPHONE=C
# [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
#
# attached base packages:
# [1] stats graphics grDevices utils datasets methods base
#
# other attached packages:
# [1] ggplot2_1.0.0 shiny_0.10.1
#
# loaded via a namespace (and not attached):
# [1] bitops_1.0-6 caTools_1.17 colorspace_1.2-4 digest_0.6.4
# [5] formatR_1.0 grid_3.1.1 gtable_0.1.2 htmltools_0.2.6
# [9] httpuv_1.3.0 labeling_0.2 MASS_7.3-34 munsell_0.4.2
# [13] plyr_1.8.1 proto_0.3-10 Rcpp_0.11.2 reshape2_1.4
# [17] RJSONIO_1.3-0 scales_0.2.4 stringr_0.6.2 tools_3.1.1
# [21] xtable_1.7-3
Update
As of ggplot2 version 2.0.0, the ggsave function supports character input for the device parameter, that means the temporary file created by the downloadHandler can now be saved with a direct call to ggsave by specifying that the extension to be used should be e.g. "pdf" (rather than passing in a device function). This simplifies the above example to the following
output$downloadPlot <- downloadHandler(
filename = function() { paste(input$dataset, '.png', sep='') },
content = function(file) {
ggsave(file, plot = plotInput(), device = "png")
}
)
I didn't manage to make it work with ggsave, but with a standard call to png() it seems to be okay.
I only changed the output$downloadPlot part of your server.R file :
output$downloadPlot <- downloadHandler(
filename = function() { paste(input$dataset, '.png', sep='') },
content = function(file) {
png(file)
print(plotInput())
dev.off()
})
Note that I had some problems with the 0.3 version of shiny, but it works with the latest from Github :
library(devtools)
install_github("shiny","rstudio")
Here's a solution that allows using ggsave for saving shiny plots. It uses a logical checkbox and text input to call ggsave(). Add this to the ui.R file inside sidebarPanel:
textInput('filename', "Filename"),
checkboxInput('savePlot', "Check to save")
Then add this to the server.R file instead of the current output$plot reactivePlot function:
output$plot <- reactivePlot(function() {
name <- paste0(input$filename, ".png")
if(input$savePlot) {
ggsave(name, plotInput(), type="cairo-png")
}
else print(plotInput())
})
A user can then type the desired filename in the textbox (without extension) and tick the checkbox to save in the app directory. Unchecking the box prints the plot again. I'm sure there are neater ways of doing this, but at least I can now use ggsave and cairo in windows for much nicer png graphics.
Please add any suggestions you may have.
This is old, but still the top hit when someone googles "R shiny save ggplot", so I will contribute another workaround. Very simple... call ggsave in the same function that displays your graph, which will save the graph as a file on the server.
output$plot <- renderPlot({
ggsave("plot.pdf", plotInput())
plotInput()
})
Then, use downloadHandler and use file.copy() to write data from the existing file to the "file" parameter.
output$dndPlot <- downloadHandler(
filename = function() {
"plot.pdf"
},
content = function(file) {
file.copy("plot.pdf", file, overwrite=TRUE)
}
)
Works for me.

Resources