CODEWITHSUNDEEP
regexrstringgrepl
Naveen Mathew
Naveen Mathew

Reputation: 372

Find matching patterns from list of patterns using grepl

I used grepl to check whether a string contains any of the patterns from a set of patterns (I used '|' to separate the patterns). Reverse search didn't help. How to identify the set of patterns that match?

Additional information: This can be solved by writing a loop, but it is very time consuming as my set has > 100,000 strings. Can it be optimized?

Eg: Let the string be a <- "Hello"

pattern <- c("ll", "lo", "hl")

pattern1 <- paste(pattern, collapse="|") # "ll|lo|hl"

grepl(a, pattern=pattern1) # returns TRUE

grepl(pattern, pattern=a) # returns FALSE 'n' times - n is 3 here

Upvotes: 4

Views: 3197

Answers (2)

Rorschach
Rorschach

Reputation: 32456

You can also use base R with a lookahead expression, (?=), since the patterns overlap. With gregexpr you can extract the match location for each grouped pattern as a matrix.

## changed your string so the second pattern matches twice
a <- "Hellolo"
pattern <- c("ll", "lo", "hl")
pattern1 <- sprintf("(?=(%s))", paste(pattern, collapse=")|(")) #  "(?=(ll)|(lo)|(hl))"

attr(gregexpr(pattern1, a, perl=T)[[1]], "capture.start")
# [1,] 3 0 0
# [2,] 0 4 0
# [3,] 0 6 0

Each column of the matrix corresponds to the patterns, so pattern 2 matched positions 4 and 6 in the test string, pattern 1 matched at position 3, and so on.

Upvotes: 1

Colonel Beauvel
Colonel Beauvel

Reputation: 31181

You are looking for str_detect from package stringr:

library(stringr)

str_detect(a, pattern)
#[1]  TRUE  TRUE FALSE

In case you have multiple strings like a = c('hello','hola','plouf') you can do:

lapply(a, function(u) pattern[str_detect(u, pattern)])

Upvotes: 8

Related Questions