Join GitHub today
GitHub is home to over 40 million developers working together to host and review code, manage projects, and build software together.
Sign upwrite_dta can't save labelled class without labels #442
Comments
This comment was marked as resolved.
This comment was marked as resolved.
|
It is the same for |
This comment was marked as resolved.
This comment was marked as resolved.
|
Wrote this function thanks to the workaround suggested by @Gootjes. Hope someone would find this useful. library(tibble)
library(dplyr)
library(magrittr)
library(haven)
fix_haven_labelled_write_problem <- function(df){
# Provides a workaround to the problem of writing haven-labelled variables (with null `labels`
# and non-null `label`) to Stata / SPSS files. Returns a dataframe compatible with functions
# such as `write_sav`.
# Reference: https://github.com/tidyverse/haven/issues/442
# df: dataframe
# vector of problematic column names
cnames <- df %>%
select_if(~ is.labelled(.) & is.null(attributes(.)$labels)) %>%
colnames
# convert the problematic columns back to their generic data types
df_fixed <- df %>%
mutate_if(~ is.labelled(.) & is.null(attributes(.)$labels) & typeof(.) == 'character',
as.character) %>%
mutate_if(~ is.labelled(.) & is.null(attributes(.)$labels) & typeof(.) == 'integer',
as.integer) %>%
mutate_if(~ is.labelled(.) & is.null(attributes(.)$labels) & typeof(.) == 'double',
as.double)
# add an attribute called `label` to the respective columns
for(cname in cnames){
attributes(df_fixed[[cname]])$label <- df[[cname]] %>% attributes %$% label
}
return(df_fixed)
}
df <- tibble(x=labelled(1:20, labels = NULL, label = 'This is clearly a integer'))
df %>% write_dta(tempfile(fileext = '.dta'), version = 14)
#> Error in write_dta_(data, normalizePath(path, mustWork = FALSE), version = stata_file_format(version)): Not compatible with requested type: [type=NULL; target=integer].
df %>% fix_haven_labelled_write_problem %>% write_dta(tempfile(fileext = '.dta'), version = 14)
# worksCreated on 2019-08-26 by the reprex package (v0.3.0) |
This comment has been minimized.
This comment has been minimized.
|
I think the issue arises in this defineVariable function:
https://github.com/tidyverse/haven/blob/7c24314b7a39f045d95719b21de553bbb2b7c865/src/DfWriter.cpp#L223
The function assumes the existence of a labelset if a variable is of class
haven_labelled. Later on in the function, the function is perfectly happy
to set labelset to a NULL value. Should be an easy fix: check and allow for
a NULL labelset.
…On Mon, Aug 26, 2019, 05:23 nbchan ***@***.***> wrote:
Wrote this function thanks to the workaround suggested by @Gootjes
<https://github.com/Gootjes>. Hope someone would find this useful.
fix_haven_labelled_write_problem <- function(df){
# Provides a workaround to the problem of writing haven-labelled variables (with null `labels`
# and non-null `label`) to Stata / SPSS files. Returns a dataframe compatible with functions
# such as `write_sav`.
# Reference: #442
# df: dataframe
# vector of problematic column names
cnames <- df %>%
select_if(~ is.labelled(.) & is.null(attributes(.)$labels)) %>%
colnames
# convert the problematic columns back to their generic data types
df_fixed <- df %>%
mutate_if(~ is.labelled(.) & is.null(attributes(.)$labels) & typeof(.) == 'character',
as.character) %>%
mutate_if(~ is.labelled(.) & is.null(attributes(.)$labels) & typeof(.) == 'integer',
as.integer) %>%
mutate_if(~ is.labelled(.) & is.null(attributes(.)$labels) & typeof(.) == 'double',
as.double)
# add an attribute called `label` to the respective columns
for(cname in cnames){
attributes(df_fixed[[cname]])$label <- df[[cname]] %>% attributes %$% label
}
return(df_fixed)
}
Created on 2019-08-26 by the reprex package <https://reprex.tidyverse.org>
(v0.3.0)
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#442?email_source=notifications&email_token=AJAX3UZVDDI54FMMHPGXTMDQGNEEVA5CNFSM4G65GXU2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOD5DE2XI#issuecomment-524701021>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AJAX3U22F7X4LASKGKY2ZEDQGNEEVANCNFSM4G65GXUQ>
.
|
This comment has been minimized.
This comment has been minimized.
|
Minimal reprex: library(haven)
x <- labelled(1:3, labels = NULL, label = 'x')
df <- data.frame(x = x)
write_dta(df, tempfile(fileext = '.dta'), version = 14)
#> Error in get_result(output = out, options) :
#> callr subprocess failed: could not start R, exited with non-zero status, has crashed or was killed |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
According to the documentation of
labelled, the argumentlabelsaccepts aNULL, which is reasonable, given that you might want to save a column with nolabelsbut an actuallabel.write_dtathrows a weird error when saving with thelabelsargument toNULL.