Remove text inside brackets, parens, and/or braces

Question

I am in need of a function that extracts any type of bracket ie (), [], {} and the information in between. I created it and get it to do what I want but I get an annoying w

Answer 1

Suppose that brackets are not nested and that we have this test data:

x <- c("a (bb) [ccc]{d}e", "x[a]y")

Then using strapply in gsubfn we have this two-line solution which first translates all parentheses and square brackets to brace brackets and then processes that:

library(gsubfn)    

xx <- chartr("[]()", "{}{}", x)
s <- strapply(xx, "{([^}]*)}", c)

The result of the above is the following list:

> s
[[1]]
[1] "bb"  "ccc" "d"  

[[2]]
[1] "a"

Answer 2

Maybe this function is a little more straight-forward? Or at least more compact.

bracketXtract <-
    function(txt, br = c("(", "[", "{", "all"), with=FALSE)
{
    br <- match.arg(br)
    left <-        # what pattern are we looking for on the left?
        if ("all" == br) "\\(|\\{|\\["
        else sprintf("\\%s", br)
    map <-         # what's the corresponding pattern on the right?
        c(`\\(`="\\)", `\\[`="\\]", `\\{`="\\}",
          `\\(|\\{|\\[`="\\)|\\}|\\]")
    fmt <-         # create the appropriate regular expression
        if (with) "(%s).*?(%s)"
        else "(?<=%s).*?(?=%s)"
    re <- sprintf(fmt, left, map[left])
    regmatches(txt, gregexpr(re, txt, perl=TRUE))    # do it!
}

No need to lapply; the regular expression functions are vectorized in that way. This fails with nested parentheses; likely regular expressions won't be a good solution if that's important. Here we are in action:

> txt <- c("I love chicken [unintelligible]!",
+          "Me too! (laughter) It's so good.[interupting]",
+          "Yep it's awesome {reading}.",
+          "Agreed.")
> bracketXtract(txt, "all")
[[1]]
[1] "unintelligible"

[[2]]
[1] "laughter"    "interupting"

[[3]]
[1] "reading"

[[4]]
character(0)

This fits without trouble into a data.frame.

> examp2 <- data.frame(var1=1:4)
> examp2$text <- c("I love chicken [unintelligible]!",
+                  "Me too! (laughter) It's so good.[interupting]",
+                  "Yep it's awesome {reading}.", "Agreed.")
> examp2$text2<-bracketXtract(examp2$text, 'all')
> examp2
  var1                                          text                 text2
1    1              I love chicken [unintelligible]!        unintelligible
2    2 Me too! (laughter) It's so good.[interupting] laughter, interupting
3    3                   Yep it's awesome {reading}.               reading
4    4                                       Agreed.                      

The warning you were seeing has to do with trying to stick a matrix into a data frame. I think the answer is "don't do that".

> df = data.frame(x=1:2)
> df$y = matrix(list(), 2, 2)
> df
  x    y
1 1 NULL
2 2 NULL
Warning message:
In format.data.frame(x, digits = digits, na.encode = FALSE) :
  corrupt data frame: columns will be truncated or padded with NAs

Answer 3

My thought had been to make 6 (implicitly vectorized) helper functions, but I will be studying Martin's code instead, since he is much better at this than I:

rm.curlybkt.no <-function(x) gsub("(\\{).*(\\})", "\\1\\2", x, perl=TRUE)
rm.rndbkt.no <-  function(x) gsub("(\\().*(\\))", "\\1\\2", x, perl=TRUE)
rm.sqrbkt.no <-  function(x) gsub("(\\[).*(\\])", "\\1\\2", x, perl=TRUE)

rm.rndbkt.in <- function(x) gsub("\\(.*\\)", "", x)
rm.curlybkt.in <- function(x) gsub("\\{.*\\}", "", x)
rm.sqrbkt.in   <- function(x) gsub("\\[.*\\]", "", x)

Answer 4

Give this a shot. I prefer the stringr package! :)

bracketXtract <- function(string, bracket = "all", include.bracket = TRUE){
  # Load stringr package
  require(stringr)

  # Regular expressions for your brackets
  rgx = list(square = "\\[\\w*\\]", curly  = "\\{\\w*\\}", round  = "\\(\\w*\\)")
  rgx['all'] = sprintf('(%s)|(%s)|(%s)', rgx$square, rgx$curly, rgx$round)

  # Ensure you have the correct bracket name
  stopifnot(bracket %in% names(rgx))

  # Find your matches
  matches = str_extract_all(string, pattern = rgx[[bracket]])[[1]]

  # Remove brackets from results if needed
  if(!include.bracket) 
    matches = sapply(matches, function(m) substr(m, 2, nchar(m)-1))

  unname(matches)
}



j <- "What kind of cheese isn't your cheese? {wonder} Nacho cheese! [groan] (Laugh)"  
bracketXtract(j)
# [1] "{wonder}" "[groan]"  "(Laugh)" 
bracketXtract(j, bracket = "square")
# [1] "[groan]"
bracketXtract(j, include.bracket = F)
# [1] "wonder" "groan"  "Laugh"