data frame splicing in j

[eutwt]

closes #454

This appears to work but I'm not sure how I feel about adding it. I'm not sure if there's a better option for j when there's more than one ... argument (second example below). I'm also assuming R doesn't make a copy of the elements in the inner list but haven't tested that.

Right now I only made the change for summarise()

devtools::load_all('~/Documents/dtplyr')
#> ℹ Loading dtplyr

mtcars |> 
  dtplyr::lazy_dt() |> 
  dplyr::summarise(
    stats::quantile(mpg, probs = c(0.5, 0.9)) |> 
      tibble::as_tibble_row(),
    .by = cyl
  )
#> Source: local data table [3 x 3]
#> Call:   `_DT1`[, tibble::as_tibble_row(stats::quantile(mpg, probs = c(0.5, 
#>     0.9))), keyby = .(cyl)]
#> 
#>     cyl `50%` `90%`
#>   <dbl> <dbl> <dbl>
#> 1     4  26    32.4
#> 2     6  19.7  21.2
#> 3     8  15.2  18.3
#> 
#> # Use as.data.table()/as.data.frame()/as_tibble() to access results

mtcars |> 
  dtplyr::lazy_dt() |> 
  dplyr::summarise(
    stats::quantile(mpg, probs = c(0.5, 0.9)) |> 
      tibble::as_tibble_row(),
    z = 1,
    .by = cyl
  )
#> Source: local data table [3 x 4]
#> Call:   `_DT2`[, unlist(.(tibble::as_tibble_row(stats::quantile(mpg, 
#>     probs = c(0.5, 0.9))), z = list(1)), recursive = FALSE), 
#>     keyby = .(cyl)]
#> 
#>     cyl `50%` `90%`     z
#>   <dbl> <dbl> <dbl> <dbl>
#> 1     4  26    32.4     1
#> 2     6  19.7  21.2     1
#> 3     8  15.2  18.3     1
#> 
#> # Use as.data.table()/as.data.frame()/as_tibble() to access results

^{Created on 2023-12-01 with reprex v2.0.2}

[eutwt]

I didn't realize this when I made the commits above but when there is no by we don't need to list() + unlist() because data.table already splices the output. Just need to handle the names.

library(data.table)

df <- data.frame(a = 1, b = 2)
lst <- as.list(df)
dt <- data.table(x = 'x')
# spliced
dt[, .(df)]
#>     df.a  df.b
#>    <num> <num>
#> 1:     1     2
dt[, .(df, y = 'y')]
#>     df.a  df.b      y
#>    <num> <num> <char>
#> 1:     1     2      y

^{Created on 2023-12-02 with reprex v2.0.2}

But this doesn't work with by

df <- data.frame(a = 1, b = 2)
lst <- as.list(df)
dt <- data.table(x = 'x')

dt[, .(df), by = x]
#> Error in `[.data.table`(dt, , .(df), by = x): All items in j=list(...) should be atomic vectors or lists. If you are trying something like j=list(.SD,newcol=mean(colA)) then use := by group instead (much quicker), or cbind or merge afterwards.

[markfairbanks]

Somehow this is possible now in dbplyr including with more complicated cases than calls to data frame creation functions. I've never had the time to investigate, so maybe it's a mechanism that wouldn't work in dtplyr.

pacman::p_load(dbplyr, dplyr)

new_cols <- tibble(c = 3, d = 4)

query <- memdb_frame(a = 1, b = 2)

query %>%
  mutate(new_cols) %>%
  show_query()
#> <SQL>
#> SELECT `dbplyr_001`.*, 3.0 AS `c`, 4.0 AS `d`
#> FROM `dbplyr_001`