I thought that getting the ptype should be possible without accessing the vector. This currently fails for date, time, and logical columns.
Important for duckplyr because we want to do things differently depending on the column type but without materializing. I can work around if I must, ideally, I'd just rely on vctrs.
CC @hannes.
options(conflicts.policy = list(warn = FALSE))
library(duckplyr)
#> ✔ Overwriting dplyr methods with duckplyr methods.
#> ℹ Turn off with `duckplyr::methods_restore()`.
Sys.setenv(DUCKPLYR_FORCE = TRUE)
data.frame(a = Sys.Date()) |>
pull() |>
vctrs::vec_ptype()
#> materializing:
#> ---------------------
#> --- Relation Tree ---
#> ---------------------
#> Projection [a as a]
#> r_dataframe_scan(0x150fb1fb0)
#>
#> ---------------------
#> -- Result Columns --
#> ---------------------
#> - a (DATE)
#> Date of length 0
data.frame(a = clock::date_time_set_zone(Sys.time(), "UTC")) |>
pull() |>
vctrs::vec_ptype()
#> materializing:
#> ---------------------
#> --- Relation Tree ---
#> ---------------------
#> r_dataframe_scan(0x130efe8b0)
#>
#> ---------------------
#> -- Result Columns --
#> ---------------------
#> - a (TIMESTAMP)
#>
#> materializing:
#> ---------------------
#> --- Relation Tree ---
#> ---------------------
#> Projection [a as a]
#> r_dataframe_scan(0x130efe8b0)
#>
#> ---------------------
#> -- Result Columns --
#> ---------------------
#> - a (TIMESTAMP)
#> POSIXct of length 0
data.frame(a = FALSE) |>
pull() |>
vctrs::vec_ptype()
#> materializing:
#> ---------------------
#> --- Relation Tree ---
#> ---------------------
#> Projection [a as a]
#> r_dataframe_scan(0x127b87498)
#>
#> ---------------------
#> -- Result Columns --
#> ---------------------
#> - a (BOOLEAN)
#> logical(0)
data.frame(a = 1) |>
pull() |>
vctrs::vec_ptype()
#> numeric(0)
data.frame(a = 1L) |>
pull() |>
vctrs::vec_ptype()
#> integer(0)
data.frame(a = "x") |>
pull() |>
vctrs::vec_ptype()
#> character(0)
I thought that getting the ptype should be possible without accessing the vector. This currently fails for date, time, and logical columns.
Important for duckplyr because we want to do things differently depending on the column type but without materializing. I can work around if I must, ideally, I'd just rely on vctrs.
CC @hannes.
Created on 2024-10-23 with reprex v2.1.1