error after uploading .RData file

Question

error after uploading .RData file

shirley-jin opened this issue 5 years ago · 21 comments

Hi,

Thanks for the great package! I just tried to use it for the first time, but met a problem after uploading my .RData file into browser. The error info shown in console is as below, and the webpage will be covered in translucent gray. May I know how to solve this? Thanks a lot!!

Warning: Error in map$get: key must be not be "" or NA
  141: map$get
  140: .values$get
  139: .subset2(x, "impl")$get
  138: [[.reactivevalues
  136: <reactive:tlabels> [/Library/Frameworks/R.framework/Versions/3.6/Resources/library/stminsights/app/app.R#837]
  120: tlabels
  114: <reactive:topicDocs> [/Library/Frameworks/R.framework/Versions/3.6/Resources/library/stminsights/app/app.R#1291]
   98: topicDocs
   97: renderDataTable
   96: func
   83: origRenderFunc
   82: output$tlabel
    2: runApp
    1: run_stminsights
Warning: Error in map$get: key must be not be "" or NA
  123: <Anonymous>
Warning: Error in map$get: key must be not be "" or NA
  49: <Anonymous>

Answer 1 · 2019-11-03T14:02:40.000Z

Hi,

I have not seen this error before. Have checked that your .RData file is structured as described in the package vignette and github readme? It needs to include

one or several stm objects.
one or several estimateEffect objects.
an object out which was used to fit your stm models.

Naming is important, meaning that if stminsights cannot find an out object, it will not work properly.

Answer 2 · 2019-11-03T14:12:49.000Z

Hi,

I have not seen this error before. Have checked that your .RData file is structured as described in the package vignette and github readme? It needs to include

one or several stm objects.

one or several estimateEffect objects.

an object out which was used to fit your stm models.

Naming is important, meaning that if stminsights cannot find an out object, it will not work properly.

Hi,

Thanks for your reply. I checked my .RData file, it does include several estimateEffect objects, stm objects, and an out object. The display of the webpage is like below. I tried to reinstall it several times, but it seems that this error keeps happening.

Answer 3 · 2019-11-03T14:30:55.000Z

Okay, then I need to dig a lil deeper. Could you share with me any .RData file which leads to this error? I would try to reproduce it locally to get a better understanding for what is going on.

Answer 4 · 2019-11-03T15:19:02.000Z

Hi,

Github does not allow to upload file larger than 10 MB, so I sent it to you by email. Please let me know if you need any other info on this. Thanks.

Answer 5 · 2019-11-03T15:43:30.000Z

Thanks, got it! Your .RData file came with some encoding issues and you stored your main textual variable as a factor variable. I used the following script to fix it:

load("bcorp.RData")

# https://stackoverflow.com/questions/34024654/reading-rdata-file-with-different-encoding
fix.encoding <- function(df, originalEncoding = "ISO-8859-1 ") {
  numCols <- ncol(df)
  df <- data.frame(df)
  for (col in 1:numCols)
  {
    if(class(df[, col]) == "character"){
      Encoding(df[, col]) <- originalEncoding
    }
    
    if(class(df[, col]) == "factor"){
      Encoding(levels(df[, col])) <- originalEncoding
    }
  }
  return(as_tibble(df))
}

out$meta <- fix.encoding(out$meta)
out$meta$document <-  as.character(out$meta$document)

rm(processed, storage, fix.encoding)
save.image("bcorp_clean.RData")

Please try it out and let me know whether this solves your problem (it works for me).

Answer 6 · 2019-11-04T05:59:53.000Z

Hi,

Thanks for the solution. I just tried the script to fix the encoding issue, but the same error still occurred. The error information shown in RStudio console after I uploaded the .RData file is the same as that I mentioned in the first post. I also tried it on both Windows and Mac, but still got the same error.

Warning: Error in map$get: key must be not be "" or NA
  141: map$get
  140: .values$get
  139: .subset2(x, "impl")$get
  138: [[.reactivevalues
  136: <reactive:tlabels> [/Library/Frameworks/R.framework/Versions/3.6/Resources/library/stminsights/app/app.R#837]
  120: tlabels
  114: <reactive:topicDocs> [/Library/Frameworks/R.framework/Versions/3.6/Resources/library/stminsights/app/app.R#1291]
   98: topicDocs
   97: renderDataTable
   96: func
   83: origRenderFunc
   82: output$tlabel
    2: runApp
    1: run_stminsights
Warning: Error in map$get: key must be not be "" or NA
  123: <Anonymous>
Warning: Error in map$get: key must be not be "" or NA
  49: <Anonymous>

Answer 7 · 2019-11-04T15:09:17.000Z

Okay, then I would please need a little bit more info. Could you send me the output of sessionInfo() after loading all the packages you use for one of your machines where the error occurs? Also, can you try using the .RData file I just sent you via e-mail and check whether the error still occurs?

Answer 8 · 2019-11-04T15:28:42.000Z

Thanks!! I tried the .RData file you sent me, but still got the same error. I copied the output of sessionInfo() below.

R version 3.6.1 (2019-07-05)
Platform: x86_64-apple-darwin15.6.0 (64-bit)
Running under: macOS Catalina 10.15.1

Matrix products: default
BLAS:   /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/3.6/Resources/lib/libRlapack.dylib

Random number generation:
 RNG:     Mersenne-Twister 
 Normal:  Inversion 
 Sample:  Rounding 
 
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] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] readr_1.3.1          dplyr_0.8.3          tibble_2.1.3         ggrepel_0.8.1       
 [5] stringr_1.4.0        ggraph_2.0.0         tidygraph_1.1.2      shinydashboard_0.7.1
 [9] shinyBS_0.61         shiny_1.4.0          ggplot2_3.2.1        stm_1.3.3           
[13] stminsights_0.3.0   

loaded via a namespace (and not attached):
 [1] tidyselect_0.2.5   shinyjs_1.0        purrr_0.3.3        graphlayouts_0.5.0 lattice_0.20-38   
 [6] colorspace_1.4-1   vctrs_0.2.0        htmltools_0.4.0    viridisLite_0.3.0  rlang_0.4.0       
[11] pillar_1.4.2       later_1.0.0        glue_1.3.1         withr_2.1.2        tweenr_1.0.1      
[16] lifecycle_0.1.0    munsell_0.5.0      gtable_0.3.0       fastmap_1.0.1      httpuv_1.5.2      
[21] Rcpp_1.0.2         xtable_1.8-4       backports_1.1.5    scales_1.0.0       promises_1.1.0    
[26] jsonlite_1.6       mime_0.7           farver_1.1.0       gridExtra_2.3      ggforce_0.3.1     
[31] hms_0.5.2          digest_0.6.21      stringi_1.4.3      polyclip_1.10-0    grid_3.6.1        
[36] tools_3.6.1        magrittr_1.5       lazyeval_0.2.2     crayon_1.3.4       tidyr_1.0.0       
[41] pkgconfig_2.0.3    zeallot_0.1.0      MASS_7.3-51.4      Matrix_1.2-17      data.table_1.12.6 
[46] assertthat_0.2.1   viridis_0.5.1      R6_2.4.0           igraph_1.2.4.1     compiler_3.6.1

Answer 9 · 2019-11-05T15:40:04.000Z

Okay it seems that there are two main problems.

Unrelated to stminsights code, the latest version of htmltools is not compatible with shiny and the way it handles data tables:
rstudio/htmltools#143
This leads to datatables holding e.g. probability and frex terms to only include single characters.
Installing the development version of shiny should fix this:
devtools::install_github("rstudio/shiny")
Something that changed in Shiny 1.3- 1.4. affected the timings of reactive functions waiting for an input. The error you included above results from the reactive function tlabels() looking for UI elements for topic labels. At the time it is looking for these labels, they do not yet exist, which leads to the Error in map$get: key must be not be "" or NA problem. I already fixed this for launching the app (locally, not in github yet), but when one switches to a topic model with a different number of topics, the problem still persists.

I will keep looking for a solution, but if other folks such as @jonneguyt, @cpsievert or @daattali have an idea what exactly is going on here, I would be super happy about any hints.

Answer 10 · 2019-11-05T16:23:08.000Z

Ran into the same issue a week ago, which I seemed to have on OSX (10.14), but not on windows (10). Currently don't have time to look into this, but it may be OS related. I was unable to find a quick fix that did not have knock on effects (see below).
I thought it was related the the format of variables, but upon changing them I still had the issue. I found a cheap hack that messed up the tables, so my fix was to simply use Windows in the meantime. If you need a file that will reproduce it, happy to share.
With my cheap hack there was an issue with keywords table (the frex, score, etc.), that seemed to be related to the first aspect that you mention. Upon changing the follow

for (i in tnames) { topiclabels[[i]] <- i }
the formatting of the table was off (one character per entry, number of entries equalling i). Given that I haven't helped to develop this part, I am out of depth as to what's happening. Happy to provide more input, but currently unable to chase this down.

Answer 11 · 2019-11-05T16:26:05.000Z

Thanks for your fast response! I am afraid it is not only an OS issue though because I can reproduce the error on a Windows machine. The table formatting issues you mention should be fixed after installing the dev version of shiny (see above).

Answer 12 · 2019-11-05T17:11:45.000Z

Can confirm that pulling the latest version of shiny makes the table formatting issues go away... So indeed not OS related (I think my windows machine was lagging updates for several packages). No clue what else is happening, but the cheap hack above makes things work. Other than that, out of depth...

Answer 13 · 2019-11-19T14:04:23.000Z

Okay it seems that there are two main problems.

Unrelated to stminsights code, the latest version of htmltools is not compatible with shiny and the way it handles data tables:
rstudio/htmltools#143
This leads to datatables holding e.g. probability and frex terms to only include single characters.
Installing the development version of shiny should fix this:
devtools::install_github("rstudio/shiny")

Something that changed in Shiny 0.3- 0.4. affected the timings of reactive functions waiting for an input. The error you included above results from the reactive function tlabels() looking for UI elements for topic labels. At the time it is looking for these labels, they do not yet exist, which leads to the Error in map$get: key must be not be "" or NA problem. I already fixed this for launching the app (locally, not in github yet), but when one switches to a topic model with a different number of topics, the problem still persists.

I will keep looking for a solution, but if other folks such as @jonneguyt, @cpsievert or @daattali have an idea what exactly is going on here, I would be super happy about any hints.

Hello. Thank you very much for the 'stminsights' package.
There is same issue for my computer, so I update 'shiny' package on github.
Unfortunately, I encounter the same problem.
How can I solve the problem?

Answer 14 · 2019-11-19T16:17:50.000Z

I am working on it and let you know as soon as I found a solution. In the meantime, what you could also try is downgrading shiny, e.g. to 1.3:

require(devtools)
install_version("shiny", version = "1.3.0", repos = "http://cran.us.r-project.org")

Answer 15 · 2019-11-19T23:54:25.000Z

I am working on it and let you know as soon as I found a solution. In the meantime, what you could also try is downgrading shiny, e.g. to 1.3:
require(devtools)
install_version("shiny", version = "1.3.0", repos = "http://cran.us.r-project.org")

Thank you for reply.
This solution saves me.

Answer 16 · 2019-12-23T22:05:29.000Z

Hello,
I also became to have the problem with stminsights (this month).
I have read your comments and downgraded shiny, but still have the same:

Could you recommend me how to solve this problem.

Thanks in advance.

Answer 17 · 2020-01-29T16:30:05.000Z

I just pushed an update to stminsights which hopefully fixed all the bugs introduced by the Shiny changes. Please try reinstalling with devtools::install_github("cschwem2er/stminsights") and let me know if this works for you.
Also, in case the app does not crash, please try to compare the output of the app with that returned by the stm package and let me know when you find anything suspicious.

Answer 18 · 2020-01-30T14:30:09.000Z

Thanks a lot for reply.
But still have a problems even with installing process:

When I run devtools::install_github("cschwem2er/stminsights"), I receive the propositions:

What option better to choose?
I tried 1, 2 and 3 and always have an error:

When I choose 1 or 2:

When I choose 3:

Do you have any ideas about it?

Answer 19 · 2020-01-30T14:34:15.000Z

@NinaRizun This current build of stminsights relies on development versions of packages that are only available on Github. So you need 1) Rtools properly installed and 2) to choose the first option when installing stminsights from Github.
I suggest to restart your R environment and then trying it again. The error message you receive suggests that the installation failed because you had some of the packages currently loaded ("are in use and will not be installed").

Answer 20 · 2020-02-03T15:34:03.000Z

Works for me. One minor note, it seems that it needs the latest development version of shiny. For me, this resulted in an error without the option to automatically update all packages. I had to manually install the dev version of shiny, which can be done by
devtools::install_github("rstudio/shiny")

Answer 21 · 2020-02-03T15:51:18.000Z

Thanks for the heads-up @jonneguyt. Yes, unfortunately the development version of Shiny is still necessary as their CRAN version does not yet have the bug fixed which messes up the output of data tables.