Translation of dplyr::if_else() breaks if .data$var is used

[mrcaseb]

I am using dplyr inside a package and therefore do .data$var to avoid tidy evaluation R CMD check NOTEs

dtplyr translates this falsely inside a dplyr::if_else()

library(data.table)
library(dtplyr)
library(dplyr, warn.conflicts = FALSE)

tib_lazy <- tibble::tibble(num = 1:5) %>% 
  dtplyr::lazy_dt() %>% 
  dplyr::mutate(
    # don't use .data$ -> works well
    lvl = dplyr::if_else(num <= 3, "low", "high"),
    # use .data$ in if_else -> the string ".data$" remains in the query
    lvl_.data = dplyr::if_else(.data$num <= 3, "low", "high"),
    # use .data$ in base ifelse -> works but doesn't call fifelse
    lvl_alt = ifelse(.data$num <= 3, "low", "high"),
    # use .data$ outside if_else works well
    num_alt = .data$num * 10
  )

# lvl_.data is NA
# note the string ".data$" in the fifelse query of the lvl_.data variable

tib_lazy
#> Source: local data table [5 x 5]
#> Call:   copy(`_DT1`)[, `:=`(lvl = fifelse(num <= 3, "low", "high"), lvl_.data = fifelse(.data$num <= 
#>     3, "low", "high"), lvl_alt = ifelse(num <= 3, "low", "high"), 
#>     num_alt = num * 10)]
#> 
#>     num lvl   lvl_.data lvl_alt num_alt
#>   <int> <chr> <chr>     <chr>     <dbl>
#> 1     1 low   <NA>      low          10
#> 2     2 low   <NA>      low          20
#> 3     3 low   <NA>      low          30
#> 4     4 high  <NA>      high         40
#> 5     5 high  <NA>      high         50
#> 
#> # Use as.data.table()/as.data.frame()/as_tibble() to access results

^{Created on 2021-03-08 by the reprex package (v1.0.0)}

[hadley]

Somewhat more minimal reprex:

library(dtplyr)
library(dplyr, warn.conflicts = FALSE)

dt <- lazy_dt(tibble(x = 1:5))

dt %>% 
  mutate(y = dplyr::if_else(.data$x <= 3, 1, 2)) %>% 
  show_query()
#> copy(`_DT1`)[, `:=`(y = fifelse(.data$x <= 3, 1, 2))]

^{Created on 2021-03-08 by the reprex package (v1.0.0)}

I also see:

library(dtplyr)
library(dplyr, warn.conflicts = FALSE)

dt <- lazy_dt(tibble(x = 1:5))

dt %>% 
  mutate(y = dplyr::if_else(.data$x <= 3, 1, 2))
#> Error in fifelse(.data$x <= 3, 1, 2): could not find function "fifelse"

^{Created on 2021-03-08 by the reprex package (v1.0.0)}

which suggests we forgot to add fifelse to dt_env().

And seems like it's worth translating ifelse() too.