unglue: unglue

In moodymudskipper/unglue: Extract Matched Substrings Using a Pattern

unglue

Description

The functions unglue_data(), unglue(), unglue_vec() and unglue_unnest() extract matched substrings using a syntax inspired from glue::glue(). Simple cases don't require regex knowledge at all.

Usage

unglue(x, patterns, open = "{", close = "}", convert = FALSE, multiple = NULL)

unglue_data(
  x,
  patterns,
  open = "{",
  close = "}",
  convert = FALSE,
  multiple = NULL,
  na = NA_character_
)

unglue_vec(
  x,
  patterns,
  var = 1,
  open = "{",
  close = "}",
  convert = FALSE,
  multiple = NULL,
  na = NA_character_
)

unglue_unnest(
  data,
  col,
  patterns,
  open = "{",
  close = "}",
  remove = TRUE,
  convert = FALSE,
  multiple = NULL,
  na = NA_character_
)

Arguments

x	a character vector to unglue.
patterns	a character vector or a list of character vectors, if a list, items will be pasted using an empty separator ("").
open	The opening delimiter.
close	The closing delimiter.
convert	If TRUE, will convert columns of output using utils::type.convert() with parameter as.is = TRUE, alternatively, can be a converting function, such as readr::type_convert. Formula notation is supported if the package rlang is installed, so things like convert = ~type_convert(., numerals = "warn.loss") are possible.
multiple	The aggregation function to use if several subpatterns are named the same, by default no function is used and subpatterns named the same will match the same value. If a function is provided it will be fed the conflicting values as separate arguments. Formula notation is supported if the package rlang is installed.
na	string to use when there is no match
var	for unglue_vec(), the numeric index or the name of the subpattern to extract from
data	a data frame.
col	column containing the character vector to extract values from.
remove	whether to remove the column col once extraction is performed

Details

Depending on the task you might want:

• unglue_data() to return a data frame from a character vector, just as glue::glue_data() does in reverse

• unglue() to return a list of data frames containing the matches

• unglue_vec() to extract one value by element of x, chosen by index or by name.

• unglue_unnest() to extract value from a column of a data frame to new columns

To build the relevant regex pattern special characters will be escaped in the input pattern and the subpatterns will be replaced with ⁠(.*?)⁠ if in standard "{foo}" form. An alternate regular expression can be provided after = so that "{foo=\\d}" will be translated into "(\\d)".

Sometimes we might want to use regex to match a part of the text that won't be extracted, in these cases we just need to omit the name as in "{=\\d}".

unglue_unnest()'s name is a tribute to tidyr::unnest() because unglue_unnest(data, col, patterns) returns a similar output as dplyr::mutate(data, unglued = unglue(col, patterns)) %>% tidyr::unnest() (without requiring any extra package). It is also very close to tidyr::extract() and efforts were made to make the syntax consistent with the latter.

Value

For unglue()a list of one row data frames, for unglue_data a data frame, for unglue_unnest the data frame input with additional columns built from extracted values, for unglue_vec an atomic vector.

Examples

# using an awample from ?glue::glue
if(require(magrittr) && require(glue)) {
  glued_data <- mtcars %>% glue_data("{rownames(.)} has {hp} hp")
  unglue_data(glued_data, "{rownames(.)} has {hp} hp")
}

facts <- c("Antarctica is the largest desert in the world!",
"The largest country in Europe is Russia!",
"The smallest country in Europe is Vatican!",
"Disneyland is the most visited place in Europe! Disneyland is in Paris!",
"The largest island in the world is Green Land!")
facts_df <- data.frame(id = 1:5, facts)

patterns <- c("The {adjective} {place_type} in {bigger_place} is {place}!",
            "{place} is the {adjective} {place_type=[^ ]+} in {bigger_place}!{=.*}")
unglue_data(facts, patterns)

sentences <- c("666 is [a number]", "foo is [a word]",
              "42 is [the answer]", "Area 51 is [unmatched]")
patterns <- c("{number=\\d+} is [{what}]", "{word=\\D+} is [{what}]")
unglue_data(sentences, patterns)

unglue_unnest(facts_df, facts, patterns)
unglue_unnest(facts_df, facts, patterns, remove = FALSE)

moodymudskipper/unglue documentation built on Dec. 8, 2024, 9:07 p.m.