Hypothetical re-write of RVerbalExpressions

rverbalexpressions_notes.R

library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union

# Utilities
rx <- function() character(0L)

`~` <- function(...) {
  args <- as.list(sys.call())[-1]
  args <- paste0("rx() %>% ", args)
  args <- sapply(args, function(x) eval(parse(text = x)), USE.NAMES = FALSE)
  paste0(args, collapse = "")
}

vals <- `~`

sanitize <- function(x) {
  if(missing(x))
    stop("The 'value' argument is missing. Did you forget to start the rx chain with rx()?")
  esc <- c(".", "|", "*", "?", "+", "(", ")", "{", "}", "^", "$", "\\", ":", "=", "[", "]")
  gsub(paste0("([\\", paste0(collapse = "\\", esc), "])"), "\\\\\\1", x, perl = TRUE)
}

# Literals
rx_literal <- function(.data, string) {
  paste0(.data, sanitize(string))
}

# Character classes/sets
rx_one_of <- function(.data, ...) {
  paste0(.data, "[", ..., "]")
}

rx_none_of <- function(.data, ...) {
  paste0(.data, "[^", ..., "]")
}

# The dot
rx_anything <- function(.data) {
  paste0(.data, ".")
}

# Anchors
rx_start_of_line <- function(.data) {
  paste0(.data, "^")
}

rx_end_of_line <- function(.data) {
  paste0(.data, "$")
}

rx_group <- function(.data, ...) {
  paste0(.data, "(", ..., ")")
}

rx_either_of <- function(.data, ...) {
  paste0(.data, paste(..., sep = "|"))
}

rx() %>%
  rx_literal("a") %>%
  rx_one_of("b") %>%
  rx_none_of("c") %>%
  rx_anything() %>%
  rx_start_of_line() %>%
  rx_end_of_line() %>%
  rx_group(
    ~ rx_literal("a"),
    ~ rx_none_of("b")
  ) %>%
  rx_either_of("a", "b")
#> [1] "a[b][^c].^$(a[^b])a|b"

# with tilde
rx() %>%
  rx_group(
    ~ rx_either_of("cat", "dog")
  ) %>%
  rx_literal(" food") %>%
  stringr::str_extract(c("cat food", "dog food", "fish food"), .)
#> [1] "cat food" "dog food" NA

# with vals
rx() %>%
  rx_group(
    vals(rx_either_of("cat", "dog"))
  ) %>%
  rx_literal(" food") %>%
  stringr::str_extract(c("cat food", "dog food", "fish food"), .)
#> [1] "cat food" "dog food" NA