Using `stringr::str_split()` with dplyr to extract and combine strings

Question

Regular expression buffs, I need your help with this.

I have a string of the form: s <- "Variable Name (Shortcut) (Variable Unit)".

I need to extract two things: a) Everything BEFORE (Variable Unit); b) Variable Unit from inside the last set of parantheses. How? #rstats

https://twitter.com/IsabellaGhement/status/1597653720841736194?s=20&t=ICQ5FEcBplmVPZIVT0hxOQ

Code

library(dplyr)
library(tidyr)
library(stringr)

elements <- c("Total Zinc (T) (mu/g)", "PH (%)", "Total Iron (mu/g)")

# We're going to put this in a tibble so we can take advantage of dplyr.
elements_tbl <- tibble(element = elements)

# str_split() approach --------------------------------------------------------

# This approach minimizes the amount of regex you have to write, and lets dplyr
# do the heavy lifting. It's verbose but quicker to write and easier to
# understand.

# These next two lines are just so we can get nicer column names when we
# unnest_wider.
element_section_names <- c("name", "shortcut", "unit")
elements_column_names <- c(
  setdiff(names(elements_tbl), "strings"), element_section_names
)

# The element strings don't have any indicator of what is a unit versus
# shorthand, so we need to specify this ourselves to be able to distinguish
# them.
element_units <- c("(%)", "(mu/g)")

elements_tbl %>%
  # This creates a list column which we can unnest to get one column for each
  # part of the string.
  mutate(strings = str_split(element, " (?=\\()")) %>%
  unnest_wider(strings, names_repair = ~ elements_column_names) %>%
  # Then we can wrangle the new columns containing the three parts of our
  # string to get the desired result.
  mutate(
    unit = case_when(
      shortcut %in% element_units ~ shortcut,
      TRUE ~ unit
    ),
    shortcut = case_when(
      shortcut %in% element_units ~ NA_character_,
      TRUE ~ shortcut
    ),
    name_shortcut = case_when(
      !is.na(shortcut) ~ paste(name, shortcut),
      is.na(shortcut) ~ name
    )
  )

# str_extract() approach ------------------------------------------------------

# This approach minimizes the amount of data wrangling you have to do, and uses
# regex for the heavy lifting. It's compact but slower to write and harder to
# understand. Using stringr::str_view() helps while figuring out the patterns
# you need.

elements_tbl %>%
  mutate(
    name = str_extract(element, ".+?(?= \\()"),
    shortcut = str_extract(element, "\\(.+\\)(?= \\()"),
    unit = str_extract(element, "(\\([^\\(\\)]+\\))(?! )"),
    name_shortcut = str_extract(element, ".+(?= \\()")
  )

Result

library(dplyr)
library(tidyr)
library(stringr)

elements <- c("Total Zinc (T) (mu/g)", "PH (%)", "Total Iron (mu/g)")

# We're going to put this in a tibble so we can take advantage of dplyr.
elements_tbl <- tibble(element = elements)

# str_split() approach --------------------------------------------------------

# This approach minimizes the amount of regex you have to write, and lets dplyr
# do the heavy lifting. It's verbose but quicker to write and easier to
# understand.

# These next two lines are just so we can get nicer column names when we
# unnest_wider.
element_section_names <- c("name", "shortcut", "unit")
elements_column_names <- c(
  setdiff(names(elements_tbl), "strings"), element_section_names
)

# The element strings don't have any indicator of what is a unit versus
# shorthand, so we need to specify this ourselves to be able to distinguish
# them.
element_units <- c("(%)", "(mu/g)")

elements_tbl %>%
  # This creates a list column which we can unnest to get one column for each
  # part of the string.
  mutate(strings = str_split(element, " (?=\\()")) %>%
  unnest_wider(strings, names_repair = ~ elements_column_names) %>%
  # Then we can wrangle the new columns containing the three parts of our
  # string to get the desired result.
  mutate(
    unit = case_when(
      shortcut %in% element_units ~ shortcut,
      TRUE ~ unit
    ),
    shortcut = case_when(
      shortcut %in% element_units ~ NA_character_,
      TRUE ~ shortcut
    ),
    name_shortcut = case_when(
      !is.na(shortcut) ~ paste(name, shortcut),
      is.na(shortcut) ~ name
    )
  )
#> New names:
#> • `` -> `...1`
#> • `` -> `...2`
#> • `` -> `...3`
#> New names:
#> New names:
#> • `` -> `...1`
#> • `` -> `...2`
#> # A tibble: 3 × 5
#>   element               name       shortcut unit   name_shortcut 
#>   <chr>                 <chr>      <chr>    <chr>  <chr>         
#> 1 Total Zinc (T) (mu/g) Total Zinc (T)      (mu/g) Total Zinc (T)
#> 2 PH (%)                PH         <NA>     (%)    PH            
#> 3 Total Iron (mu/g)     Total Iron <NA>     (mu/g) Total Iron

# str_extract() approach ------------------------------------------------------

# This approach minimizes the amount of data wrangling you have to do, and uses
# regex for the heavy lifting. It's compact but slower to write and harder to
# understand. Using stringr::str_view() helps while figuring out the patterns
# you need.

elements_tbl %>%
  mutate(
    name = str_extract(element, ".+?(?= \\()"),
    shortcut = str_extract(element, "\\(.+\\)(?= \\()"),
    unit = str_extract(element, "(\\([^\\(\\)]+\\))(?! )"),
    name_shortcut = str_extract(element, ".+(?= \\()")
  )
#> # A tibble: 3 × 5
#>   element               name       shortcut unit   name_shortcut 
#>   <chr>                 <chr>      <chr>    <chr>  <chr>         
#> 1 Total Zinc (T) (mu/g) Total Zinc (T)      (mu/g) Total Zinc (T)
#> 2 PH (%)                PH         <NA>     (%)    PH            
#> 3 Total Iron (mu/g)     Total Iron <NA>     (mu/g) Total Iron

^{Created on 2022-11-29 by the reprex package (v2.0.1)}

mccarthy-m-g/2022-11-29_str-split.md

Question

Code

Result