Factor loses factor class when nrow(DT) == 1 and updated

Start by running the issue's one-row and two-row R reproductions with data.table, then trace the := update path highlighted by the verbose output. Done means the one-row update retains the factor class and original levels, with regression coverage for the reported case.

data.table unexpectedly, at least to me, drops a variable's factor class when there is only one row in the data.table and the variable is updated, even if the new value is a defined level of the original factor.

> dt <- data.table(x = c(TRUE, TRUE), y = factor(c(1, 2), levels = 1:3))
> dt[(x), y := 2, verbose = TRUE]
Detected that j uses these columns: [y]
Assigning to 2 row subset of 2 rows
RHS_list_of_columns == false
> dt
        x      y
   <lgcl> <fctr>
1:   TRUE      2
2:   TRUE      2

> dt <- data.table(x = c(TRUE), y = factor(c(1), levels = 1:3))
> dt[(x), y := 2, verbose = TRUE]
Detected that j uses these columns: [y]
Assigning to all 1 rows
RHS_list_of_columns == false
RHS for item 1 has been duplicated because NAMED==4 MAYBE_SHARED==1, but then is being plonked. length(values)==1; length(cols)==1)
> dt
        x     y
   <lgcl> <num>
1:   TRUE     2

This is causing problems for me as my downstream code expects to find the original levels from the variable. For what it is worth, this also happens with character variables as well.

> sessionInfo()
R version 4.2.1 (2022-06-23)
Platform: aarch64-apple-darwin20 (64-bit)
Running under: macOS Monterey 12.6

Matrix products: default
LAPACK: /Library/Frameworks/R.framework/Versions/4.2-arm64/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] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] data.table_1.14.9 

loaded via a namespace (and not attached):
 [1] Rcpp_1.0.10       compiler_4.2.1    later_1.3.0       urlchecker_1.0.1  prettyunits_1.1.1 profvis_0.3.7    
 [7] remotes_2.4.2     tools_4.2.1       digest_0.6.30     pkgbuild_1.3.1    pkgload_1.3.0     memoise_2.0.1    
[13] lifecycle_1.0.3   rlang_1.0.6       shiny_1.7.2       cli_3.6.0         rstudioapi_0.14   fastmap_1.1.0    
[19] stringr_1.5.0     fs_1.5.2          htmlwidgets_1.5.4 vctrs_0.5.2       devtools_2.4.5    glue_1.6.2       
[25] R6_2.5.1          processx_3.7.0    sessioninfo_1.2.2 callr_3.7.2       purrr_1.0.1       magrittr_2.0.3   
[31] ps_1.7.1          promises_1.2.0.1  ellipsis_0.3.2    htmltools_0.5.3   usethis_2.1.6     mime_0.12        
[37] xtable_1.8-4      httpuv_1.6.6      stringi_1.7.12    miniUI_0.1.1.1    cachem_1.0.6      crayon_1.5.2 

I searched through all the issues and couldn't find any that seemed to be similar. I think #2403 and #3922 are at least somewhat related, but my issue is that the factor levels are being dropped completely. Sorry if I missed something.

Thanks!