shim common functions

[beanumber]

• Input functions from tidyverse (read_csv(), etc.) and read.csv()
• Output functions (write_csv(), write.csv(), etc.)
• library() and require() (@hadley)
• source()

Do a handful of these manually, and then consider a function factory or some kind of tidy evaluation.

Log the paths and calls to a CSV file (i.e., .fertile.csv)

In interactive mode, throw an error if the path is bad.

Have a sliding scale of strictness

[beanumber]

you could also mine https://github.com/hadley/strict for ideas/implementations

[beanumber]

setwd() should probably throw an error (in the strict case)

[beanumber]

ggplot2::save()?

[hadley]

ggsave()?

[beanumber]

Yes -- my bad.

[beanumber]

TODO: use a function factory to create a whole slew of functions at once.

[beanumber]

https://adv-r.hadley.nz/function-factories.html

[beanumber]

https://adv-r.hadley.nz/function-factories.html#another-approach

[beanumber]

https://adv-r.hadley.nz/function-factories.html#moving-a-list-to-the-global-environment

[beanumber]

Alright, @hadley @jennybc the first step is to admit you have a problem. Mine is that I still don't understand tidyeval.

I'm trying to do this:

#' Function factory

shim_log_input <- function(.f) {
  rlang::new_function(
    rlang::exprs(file = ),
    rlang::expr({
      log_push(file, !!.f)
      check_path(file)
      .f(file)
    }),
    rlang::caller_env()
  )
}

funs_to_shim <- c(
  utils::read.csv,
  readr::read_csv
)

shims <- purrr::map(funs_to_shim, shim_log_input)
shims

• in log_push(file, !!.f), I want a string with the name of the function. !!.f is returning the function itself (which I sort of understand), but how do I get the name of the function as a string?
• in .f(file), I want to actually execute the function. But if I put !!.f(file) here, it seems to be running when the functions are created by map, which of course errors. That is, when I am simply trying to create the shimmed function for read.csv(), it seems to actually be trying to execute read.csv(), which fails.
• and then we have to deal with the ....
• and then we have to name the shimmed functions, and attach() them (!)
• and then we have to write the documentation. Can this be automated?
• and then we have to write a second function factory for output functions like write.csv() that require an object and a path? And another one for function like library() that don't involve paths at all?
• and then what happens when the argument names vary (e.g., file vs. filename)?

Is this the right approach?

[jennybc]

I played around with some of your subtasks separately here, e.g. recovering the name of .f, looking at its formals, wrapping a function, wrapping a list of functions. It's not all wired up together, but maybe this provides some ideas you can build on until @hadley swoops 🙂. .f cannot ever know its own name -- you have to recover that by quoting the input.

library(rlang)

function_report <- function(.f) {
  .f_quoted <- enquo(.f)
  list(
    .f_quoted,
    .f_name = expr_text(quo_get_expr(.f_quoted)),
    formal_names = fn_fmls_names(.f)
  )
}

function_report(read.csv)
#> [[1]]
#> <quosure>
#>   expr: ^read.csv
#>   env:  global
#> 
#> $.f_name
#> [1] "read.csv"
#> 
#> $formal_names
#> [1] "file"         "header"       "sep"          "quote"       
#> [5] "dec"          "fill"         "comment.char" "..."

# forgive this horrible name, I'm not feeling creative
foo <- function(.f) {
  function(...) {
    l <- list(...)
    cat("file = ", l[["file"]], "\n")
    .f(...)
  }
}

read.csv(file = system.file("extdata/mtcars.csv", package = "readr"), nrows = 3)
#>    mpg cyl disp  hp drat    wt  qsec vs am gear carb
#> 1 21.0   6  160 110 3.90 2.620 16.46  0  1    4    4
#> 2 21.0   6  160 110 3.90 2.875 17.02  0  1    4    4
#> 3 22.8   4  108  93 3.85 2.320 18.61  1  1    4    1

read.csv_wrapped <- foo(read.csv)
read.csv_wrapped(
  file = system.file("extdata/mtcars.csv", package = "readr"),
  nrows = 3
)
#> file =  /Users/jenny/R/library/3.5/readr/extdata/mtcars.csv
#>    mpg cyl disp  hp drat    wt  qsec vs am gear carb
#> 1 21.0   6  160 110 3.90 2.620 16.46  0  1    4    4
#> 2 21.0   6  160 110 3.90 2.875 17.02  0  1    4    4
#> 3 22.8   4  108  93 3.85 2.320 18.61  1  1    4    1

funs_to_wrap <- c(read.csv = utils::read.csv, read_csv = readr::read_csv)
wraps <- lapply(funs_to_wrap, foo)
wraps
#> $read.csv
#> function (...) 
#> {
#>     l <- list(...)
#>     cat("file = ", l[["file"]], "\n")
#>     .f(...)
#> }
#> <bytecode: 0x7fe1d44fa2d8>
#> <environment: 0x7fe1d43a39e0>
#> 
#> $read_csv
#> function (...) 
#> {
#>     l <- list(...)
#>     cat("file = ", l[["file"]], "\n")
#>     .f(...)
#> }
#> <bytecode: 0x7fe1d44fa2d8>
#> <environment: 0x7fe1d45008f8>

wraps$read_csv(
  file = system.file("extdata/mtcars.csv", package = "readr"),
  n_max = 3
)
#> file =  /Users/jenny/R/library/3.5/readr/extdata/mtcars.csv
#> Parsed with column specification:
#> cols(
#>   mpg = col_double(),
#>   cyl = col_integer(),
#>   disp = col_integer(),
#>   hp = col_integer(),
#>   drat = col_double(),
#>   wt = col_double(),
#>   qsec = col_double(),
#>   vs = col_integer(),
#>   am = col_integer(),
#>   gear = col_integer(),
#>   carb = col_integer()
#> )
#> # A tibble: 3 x 11
#>     mpg   cyl  disp    hp  drat    wt  qsec    vs    am  gear  carb
#>   <dbl> <int> <int> <int> <dbl> <dbl> <dbl> <int> <int> <int> <int>
#> 1  21       6   160   110  3.9   2.62  16.5     0     1     4     4
#> 2  21       6   160   110  3.9   2.88  17.0     0     1     4     4
#> 3  22.8     4   108    93  3.85  2.32  18.6     1     1     4     1

^{Created on 2018-10-15 by the reprex package (v0.2.1)}

[hadley]

I don't have time to explain anything right now, but hopefully this gets you slightly unblocked:

library(rlang)
shim_log_input <- function(.f) {
  rlang::new_function(
    rlang::exprs(file = ),
    rlang::expr({
      log_push(file, !!(deparse(.f)))
      check_path(file)
      (!!.f)(file)
    }),
    rlang::caller_env()
  )
}

funs_to_shim <- exprs(
  utils::read.csv,
  readr::read_csv
)

shims <- purrr::map(funs_to_shim, shim_log_input)
shims
#> [[1]]
#> function (file) 
#> {
#>     log_push(file, "utils::read.csv")
#>     check_path(file)
#>     utils::read.csv(file)
#> }
#> <environment: 0x7ffd4b301378>
#> 
#> [[2]]
#> function (file) 
#> {
#>     log_push(file, "readr::read_csv")
#>     check_path(file)
#>     readr::read_csv(file)
#> }
#> <environment: 0x7ffd4b301378>

(note that funs_to_shim now uses exprs())

[beanumber]

Thank you both for the very helpful responses. I think I got it to work in ec7dea12.

I'm still not quite sure what to do about the varying argument names.

[jennybc]

Your shimming function will presumably have to take the input function and the name of the argument you want to give special treatment, possibly defaulting to file (?).

[beanumber]

On second thought, @hadley advises doing these individually.