From @jimhester in #36:

FWIW it would be nice if reprex() also accepted a pre-formatted string directly, like you get from clipr::read_clip().

Because I think this package will soon be the top google hit for reprex

draft one and put it in vignettes/releases/reprex-1.0.0

I'm about push something that changes how expression input is converted to a character vector. The leading whitespace can be suboptimal. To be clear, I'm not sure this is really new. The previous expression handling probably also mangled whitespace, just in a different way.

Bottom line: given source in a character vector or file, is there a way to re-indent it nicely?

@jimhester do you know?

It would be nice to be able to paste a reprex into a github issue as a folded block.

The HTML is a little tricky to get right, but this could be addressed by incorporating it on the reprex side. I was thinking something like:

reprex(venue = 'gh', folded = TRUE)

Would you consider a PR?

Hi Jenny.

First, thank you for the package, it is promising! I just used this package to post a question on stack overflow using the following code:

reprex({
  library(lme4)
  
  set.seed(101)
  
  sleepstudy$Subject <- as.character(sleepstudy$Subject)
  ss <- sleepstudy[sleepstudy$Subject == "308",]
  
  m1 <- lmer(Reaction ~ Days + (1 | Subject), ss)
}, venue = "so")

Since SO uses 4 spaces to make bloc code, I was wondering if the output in the preview pane could already include these 4 spaces so we can copy/paste directly on SO?

Happy new year,
Phil

See my so post that I attempted to make with reprex(venue = "so") but the path hardcoding of the outputted .pngs does not currently work and so I have to upload each image manually. I realize this is early -- but very neat so far!

Here's a sketch from me and @wch

x <- quote({
  # This is a comment
  a +      b
})

src <- attr(x, "srcref")
src_1 <- src[[1]]
src_n <- src[[length(src)]]

lines <-  attr(src_1, "srcfile")$lines
lines <- lines[seq(src_1[1], src_n[3])]
lines[1] <- substring(lines[1], src_1[5] + 1, nchar(lines[1]))

if (lines[1] == "" || grepl("^\\s+$", lines[1])) {
  lines <- lines[-1]
}

I wrote an answer for SO in the form of a full Rmd document for convenience, but knitr produces md not fullly compatible (code blocks, for one thing). It would be great if reprex included knitr hooks / rmarkdown custom flavoured md output format for SO. (there may well be one out there already, my memory and google haven't been very helpful). @yihui ?

I just added this to my snippets file:

snippet reprex
    reprex::reprex({
      $1
    })

That way I can type reprex<tab> and get the full code block

If my R console contains this:

> x <- rnorm(10)
> mean(x)
[1] 0.4189699

And I copy the first two lines so that my clipboard contains the following character string:

"> x <- rnorm(10)
> mean(x)"

And I call reprex(), then I get this as my output:

> x <- rnorm(10)
> mean(x)
#> Error: <text>:1:1: unexpected '>'
#> 1: >
#>     ^

A nice feature here would be to automatically strip out values of

options("prompt")
#> $prompt
#> [1] "> "

when they appear at the start of a newline.

Cool package!

fall back to sessionInfo() is no devtools

Images are one of the most difficult aspects of a reproducible example; they tend to stop an answer flow cold, as it involves saving an image, uploading it to something like Imgur, then pasting in a link to that.

Is there anything we can do to make this process easier? The ideal solution would be for knitr to upload generated figures to Imgur (which could be done with a custom knitr hook), but you need an API token to upload programmatically, which is an obstacle.

Perhaps at least the figures generated could be stuck in some easy-to-reach place so they can be uploaded.

(I know I've generated a pile of issues today; most of them are just for me to come back to later)

Given that, for CRAN prep, it's useful to know that your package will pass even when the clipboard is not available, I've simulated this on Travis using a build matrix

https://github.com/mdlincoln/clipr/blob/93329a6d96ae6fd00cb0d920f5b861b482b56ae8/.travis.yml

This should run tests both with and without DISPLAY being configured properly.

Or remove everything related to that. Right now we are neither here nor there.

http://rstudio.github.io/rstudioaddins/

I think the nicest implementation would replace the currently selected code with the reprexed version. To make it work if called multiple times, you could first strip any lines that start with #' (maybe reprex already does this?)

https://github.com/MangoTheCat/callr

It appears my procrastination on moving this package along may have paid off big time.

This causes foo.R to be overwritten by decorated version. Rename it?

reprex(infile = "foo.R", outfile = "foo")

Hi,
I found your package very usefull, however, when you run code that produces plots they are not printed when you copy and past to stack overflow or github.

I might have some ideas on how to make this process easier for the user, generating url links to each plots automatically, so the final output will include the plots when pasted on online.

Can I fork this repo and suggest some changes?

I am also having some problems with the current dev version:

Error in reprex("teste.R") :
não foi possível encontrar a função "construct_filename"

Cheers,
Gustavo

If you copy reprex output from github or SO, there should be a function to strip out the #> lines and get just the code back.

This example:

# note that you can include newlines in those brackets
reprex({
  x <- 1:4
  y <- 2:5
  x + y
 })

doesn't actually work.

I'm not sure if this is a double up of #28, but I was thinking it would be really nifty to make a keyboard shortcut, so you could highlight some code, and then do something like Cmd + Shift + R and it would copy the text, then run reprex?

Thank you for writing a package I use almost every day!

I used to be able to use reprex, but now for some reason it seems unable to use the clipboard, even though xclip is installed on my Ubuntu 16.04 machine. When I launch reprex() I get:

No input provided via `x` or `input` and clipboard is not available.

Any idea what might be going on? I'd be happy to provide any information you might need for debugging purposes. I'm using the last github version of reprex.

Currently, if I run

reprex::reprex({x <- rnorm(100) ; mean(x)}, venue = 'gh', show = FALSE)

I get the code for

x <- rnorm(100)
mean(x)
#> [1] 0.06771625

which is wonderful! However, changing the comment character, e.g. to ## is more complicated, requiring (as suggested on Twitter) copying

#+ include=FALSE
knitr::opts_chunk$set(comment = '##')

#+ realstuff
x <- rnorm(100)
mean(x)

and then calling reprex::reprex() to get the code for

x <- rnorm(100)
mean(x)
## [1] 0.04698201

To make this process simpler, could we get a comment character parameter in reprex akin to formatR::tidy_eval's prefix?

Why? Because I couldn't use reprex to do this:

getwd()
#> [1] "/Users/jenny/rrr/googlesheets"
rstudioapi::getActiveProject()
#> [1] "/Users/jenny/rrr/googlesheets"
devtools::package_file()
#> [1] "/Users/jenny/rrr/googlesheets"
testthat::test_path()
#> [1] "tests/testthat"
testthat::test_path("for_reference/gap_googlesheet.rds")
#> [1] "tests/testthat/for_reference/gap_googlesheet.rds"

setwd(testthat::test_path())
getwd()
#> [1] "/Users/jenny/rrr/googlesheets/tests/testthat"
testthat::test_path()
#> Error in if (!file.exists(path)) { : argument is of length zero
testthat::test_path("for_reference/gap_googlesheet.rds")
#> [1] "for_reference/gap_googlesheet.rds"

https://twitter.com/noamross/status/742710818534522880

cc @noamross

right now the produced .R is original source

Per this twitter discussion, it would be useful to be able to generate an output .R file that is runnable but has outputs stored in comments.

One slight difference in the output will be that image links would have to be commented.

I bet reprex could package this up nicely ... a gist tweeted by @AmeliaMN on code highlighting for Keynote presentations:

https://gist.github.com/jimbojsb/1630790

I'm going to paste the body here in case the gist goes away. But the comments look pretty useful, e.g. using alternative highlighters like Pygments..

Step 0:

Get Homebrew installed on your mac if you don't already have it

Step 1:

Install highlight. "brew install highlight". (This brings down Lua and Boost as well)

Step 2:

Highlight your source. "highlight -O rtf myfile.php | pbcopy". This will highlight the source code in RTF format and copy the result of that operation to your Mac's clipboard.

Step 3:

Paste into your slide in Keynote. Keynote will create a new text box with your highlighted code. You can adjust the width as needed, and adjust the font size with cmd+ and cmd-

@muschellij2 correctly pointed out on twitter that current R version is probably unnecessarily high.

Roll this back and see what happens? Possible version: 3.0.2, which he proposed because it's what CRAN devtools requires.

in your readme

In theory, this is the domain of gistr, but people also tend to post .R or .Rmd and NOT resulting .md ... is there a gap in tools or is this just an unfortunate behavior pattern? I don't always want to render these files to see what someone's trying to show me.

but a .R or .Rmd file can go up un-rendered via gist_create() if you do knit=FALSE (which is the default)

One common thing users forget to do in a reproducible example is to share the variables they've already defined (for example, datasets they're working with). You can turn a variable into a reproducible chunk of code with dput.

If we already have a few variables defined, it would be good to be able to add them to the start of the reprex code automatically. Here's a hello-world version using lazy_eval:

define_vars <- function(...) {
  vs <- lazyeval::lazy_dots(...)

  varnames <- unname(sapply(vs, function(v) as.character(v$expr)))

  dputs <- sapply(vs, function(v) {
    v$expr <- substitute(capture.output(dput(x)), list(x = v$expr))
    lazyeval::lazy_eval(v)
  })

  paste(varnames, dputs, sep = " <- ")
}

This takes one or more variables as its arguments and turns them into code that defines those variables. For example:

a <- 1:5
d <- data.frame(a = 1:5, b = 2)

varcode <- define_vars(a, d)

# turns it into code
cat(varcode, sep = "\n")
#> a <- 1:5
#> d <- c("structure(list(a = 1:5, b = c(2, 2, 2, 2, 2)), .Names = c(\"a\", ", "\"b\"), row.names = c(NA, -5L), class = \"data.frame\")")

In the end version of reprex, I imagine having a vars function, which creates an argument suitable for passing in:

y <- 1:100
reprex({y + 2}, vars = vars(y))
# even though it wasn't included in the reprex, resulting code would start with
# y <- 1:100

Or just wrap with a list (though that takes a little expression-parsing internally to get the names out):

reprex({y + 2}, vars = list(y))

If you like this feature idea I'll add it to the next PR.

(Fun fact; after I finished this example I used reprex to share it. Works like a charm!)

Now this would require a strategy more like formatR::tidy_eval().

Might be nice to (optionally) include sessionInfo() or devtools::session_info() at the end

Put output format into YAML, instead of in the call to render().

if it's going to be there at all

I have this R code:

library(plm)
data(Grunfeld, package="plm")
class(Grunfeld)

copying to clipboard and executing reprex() yields:

library(plm)
data(Grunfeld, package="plm")
class(Grunfeld)
#> Error in eval(expr, envir, enclos): object 'Grunfeld' not found

Something very simple like mean(c(1:5)) works, also iris dataset works:

data(iris)
class(iris)
#> [1] "data.frame"

Maybe the wrong environment is referred to for dataset from packages?

Session info --------------------------------------------------------------------------------------------------------------------------------------------------------------
 setting  value                       
 version  R version 3.2.2 (2015-08-14)
 system   x86_64, mingw32             
 ui       RStudio (0.99.738)          
 language (EN)                        
 collate  German_Germany.1252         
 tz       Europe/Berlin               
 date     2015-10-28                  

Packages ------------------------------------------------------------------------------------------------------------------------------------------------------------------
 package   * version    date       source                         
 bdsmatrix   1.3-2      2014-08-22 CRAN (R 3.2.2)                 
 clipr       0.2.0      2015-10-06 CRAN (R 3.2.2)                 
 curl        0.9.3      2015-08-25 CRAN (R 3.2.2)                 
 devtools  * 1.9.1      2015-09-11 CRAN (R 3.2.2)                 
 digest      0.6.8      2014-12-31 CRAN (R 3.2.2)                 
 evaluate    0.8        2015-09-18 CRAN (R 3.2.2)                 
 formatR     1.2.1      2015-09-18 CRAN (R 3.2.2)                 
 Formula   * 1.2-1      2015-04-07 CRAN (R 3.1.3)                 
 htmltools   0.2.6      2014-09-08 CRAN (R 3.2.2)                 
 httr        1.0.0      2015-06-25 CRAN (R 3.2.2)                 
 knitr       1.11       2015-08-14 CRAN (R 3.2.2)                 
 lattice     0.20-33    2015-07-14 CRAN (R 3.2.2)                 
 magrittr    1.5        2014-11-22 CRAN (R 3.2.2)                 
 MASS        7.3-43     2015-07-16 CRAN (R 3.2.2)                 
 memoise     0.2.1      2014-04-22 CRAN (R 3.2.2)                 
 nlme        3.1-122    2015-08-19 CRAN (R 3.2.2)                 
 plm       * 1.4-0      2013-12-28 CRAN (R 3.2.2)                 
 R6          2.1.1      2015-08-19 CRAN (R 3.2.2)                 
 reprex    * 0.0.0.9001 2015-10-28 Github (jennybc/reprex@826ddf4)
 rmarkdown   0.8.1      2015-10-10 CRAN (R 3.2.2)                 
 sandwich    2.3-4      2015-09-24 CRAN (R 3.2.2)                 
 stringi     1.0-1      2015-10-22 CRAN (R 3.2.2)                 
 stringr     1.0.0      2015-04-30 CRAN (R 3.2.2)                 
 zoo         1.7-12     2015-03-16 CRAN (R 3.1.3)   

If put on the clipboard, this reprex code can exploit the usual knitr::spin tricks for text insertion and setting chunk options. Wonder if there's any way this could be retained for expression input?

#' ## Hey I'm a heading
(x <- 1:4)
#' blah blah
#+ eval = FALSE
y <- 2:5
x + y

yields this:

Hey I'm a heading

(x <- 1:4)
#> [1] 1 2 3 4

blah blah

y <- 2:5
x + y

Hadley's suggestion on twitter

Currently the function works in the current directory using a base filename along the lines of 08-25-2015_REPREX_gh. It keeps them there if outfiles = TRUE.

I suspect that:

• Most users won't want to keep the input and output files, especially since the goal is to get it into GitHub or SO
• Those that do would be at least as happy providing their own filename (so they can refer to it later); while they can specify

Thus, I'd propose working in a temporary file unless the user specifically provides an output file base (in which case the .R and .md will be appended). This saves some of the utility work in the reprex function, and lets the user choose the filename without being appended with a date.

reprex(reprex(1 + 1))

gives the error

 Error in env$metadata <- list() : 
    cannot change value of locked binding for 'metadata' 

(I wanted to reprex this example but I couldn't!!)

You can find the overflow package here. Have not done much development on it in a while, but @richierocks had made some interesting contributions for the soanswer function.

We've had clipboard issues too, but here's our implementation of readClip and writeClip.

Re: the addin. I'm not implementing infile at this point. Notes.

shiny::fileInput()

• http://shiny.rstudio.com/gallery/file-upload.html
• Not ideal because it "uploads" the file to a temp file. Also doesn't seem to start pointed at current working directory for the R process, which I would want.

shinyFiles package

• Brings a dependency on RJSONIO

Hello world

## a comment
(x <- 1:4)
#> [1] 1 2 3 4

hi again

y <- 2:5
x + y
#> [1] 3 5 7 9

devtools::session_info()
#> Session info -------------------------------------------------------------
#>  setting  value                       
#>  version  R version 3.3.1 (2016-06-21)
#>  system   x86_64, darwin13.4.0        
#>  ui       X11                         
#>  language (EN)                        
#>  collate  en_CA.UTF-8                 
#>  tz       America/Vancouver           
#>  date     2017-01-02
#> Packages -----------------------------------------------------------------
#>  package   * version     date       source                            
#>  backports   1.0.4       2016-10-24 cran (@1.0.4)                     
#>  devtools    1.12.0.9000 2017-01-02 Github (hadley/devtools@1ce84b0)  
#>  digest      0.6.10      2016-08-02 cran (@0.6.10)                    
#>  evaluate    0.10        2016-10-11 cran (@0.10)                      
#>  htmltools   0.3.5       2016-03-21 CRAN (R 3.2.4)                    
#>  knitr       1.15.1      2016-11-22 CRAN (R 3.3.2)                    
#>  magrittr    1.5         2014-11-22 CRAN (R 3.2.0)                    
#>  memoise     1.0.0.9001  2016-12-18 Github (hadley/memoise@884d565)   
#>  pkgbuild    0.0.0.9000  2016-11-22 Github (r-pkgs/pkgbuild@65eace0)  
#>  pkgload     0.0.0.9000  2016-11-22 Github (r-pkgs/pkgload@def2b10)   
#>  Rcpp        0.12.8.3    2017-01-02 Github (RcppCore/Rcpp@f0442a0)    
#>  rmarkdown   1.3         2017-01-02 Github (rstudio/rmarkdown@3276760)
#>  rprojroot   1.1         2016-11-15 Github (krlmlr/rprojroot@1745481) 
#>  stringi     1.1.2       2016-10-01 CRAN (R 3.3.0)                    
#>  stringr     1.1.0       2016-08-19 CRAN (R 3.3.0)                    
#>  withr       1.0.2       2016-06-20 cran (@1.0.2)                     
#>  yaml        2.1.14      2016-11-12 cran (@2.1.14)

For instance:

y <- 10
reprex(y + 1)

produces

y + 1
#> [1] 11

To fix, we could add the argument envir = new.env() to the render function (I can add to my PR #6 )

Right now, if I put this code on the clipboard:

plot(1:100)

and bring my internet connection down, I get this result:

> library(reprex)
> reprex()
 Show Traceback

 Rerun with Debug
 Error in reprex_(r_file, venue, show, upload.fun) : 
Cannot render this code. Maybe the clipboard contents are not what you think?
Error in function (type, msg, asError = TRUE)  : 
  Could not resolve host: api.imgur.com 

If input is given as foo.R, there should be a way to say that the base name for outfiles should be foo. Maybe by setting outfile = NA (vs the default of NULL or giving an actual value).