R：正则表达式疯狂（stringi）

Question

I have a vector of strings that look like this:

G30(H).G3(M).G0(L).Replicate(1)

Iterating over c("H", "M", "L") , I would like to extract G30 (for " H "), G3 (for " M ") and G0 (for " L ").

My various attempts have me confused - the regex101.com debugger, eg indicates that (\\w*)\\(M\\) works just fine, but transferring that to R fails ...

Answer 1

Using the stringi package and the outer() function:

library(stringi)

strings <- c(
  "G30(H).G3(M).G0(L).Replicate(1)",
  "G5(M).G11(L).G6(H).Replicate(9)",
  "G10(M).G6(H).G8(M).Replicate(200)"  # No "L", repeated "M"
)
targets  <- c("H", "M", "L")
patterns <- paste0("\\w+(?=\\(", targets, "\\))")
matches  <- outer(strings, patterns, FUN = stri_extract_first_regex)
colnames(matches) <- targets
matches
#      H     M    L    
# [1,] "G30" "G3" "G0" 
# [2,] "G6"  "G5" "G11"
# [3,] "G6"  "G10" NA

This ignores any instances of a target letter past the first, gives you an NA when the target's not found, and returns everything in a simple matrix. The regular expressions stored in patterns match substrings like XX(Y) , where Y is the target letter and XX is any number of word characters.

Answer 2

I am pretty sure there are better solutions, but this works...

jnk <- 'G30(H).G3(M).G0(L).Replicate(1)'
patter <- '([^\\(]+)\\(H\\)\\.([^\\(]+)\\(M\\)\\.([^\\(]+)\\(L\\)\\.Replicate\\(\\d+\\)'
H <- sub(patter, '\\1', jnk)
M <- sub(patter, '\\2', jnk)
L <- sub(patter, '\\3', jnk)

EDIT:

Actually, I found once a very nice function parse.one which makes it possible to search more in a python like regular expression way...

Have a look at this:

parse.one <- function(res, result) {
  m <- do.call(rbind, lapply(seq_along(res), function(i) {
    if(result[i] == -1) return("")
    st <- attr(result, "capture.start")[i, ]
    substring(res[i], st, st + attr(result, "capture.length")[i, ] - 1)
  }))
  colnames(m) <- attr(result, "capture.names")
  m
}
jnk <- 'G30(H).G3(M).G0(L).Replicate(1)'
pattern <- '(?<H>[^\\(]+)\\(H\\)\\.(?<M>[^\\(]+)\\(M\\)\\.(?<L>[^\\(]+)\\(L\\)\\.Replicate\\(\\d+\\)'
parse.one(jnk, regexpr(pattern, jnk, perl=TRUE))

Result looks like this:

> parse.one(jnk, regexpr(pattern, jnk, perl=TRUE))
     H     M    L   
[1,] "G30" "G3" "G0"

Answer 3

If the order is always the same, an alternative might be to split the strings. For instance:

string <- "G30(H).G3(M).G0(L).Replicate(1)"
tmp <- str_split(string, "\\.")[[1]]
lapply(tmp[1:3], function(x) str_split(x, "\\(")[[1]][1])
[[1]]
[1] "G30"

[[2]]
[1] "G3"

[[3]]
[1] "G0"

Answer 4

If codes (eg, 'G30') preceding the tags(eg, '(H).') or the order of the tags in the string are allowed to change (different letters or length), you may want to try a more flexible solution based on regexpr() .

aa <-paste("G30(H).G3(M).G0(L).Replicate(",1:10,")", sep="")
my.tags <- c("H","M", "L")

extr.data <- lapply(my.tags, (function(tag){
  pat <-  paste("\\(", tag, "\\)\\.", sep="")
  pos <- regexpr(paste("(^|\\.)([[:alnum:]])*", pat ,sep=""), aa)
  out <- substr(aa, pos, (pos+attributes(pos)$match.length - 4 - length(tag)))  
  gsub("(^\\.)", "", out) 
}))
names(extr.data) <- my.tags
extr.data

Answer 5

I'm going to assume that the functions (G...) are variable and the inputs are variable. This does assume that your functions start with a G and your input is always a letter.

parse = function(arb){
  tmp = stringi::stri_extract_all_regex(arb,"G.*?\\([A-Z]\\)")[[1]]
  unlist(lapply(lapply(tmp,strsplit,"\\)|\\("),function(x){
    output = x[[1]][1]
    names(output) = x[[1]][2]
    return(output)
  }))
}

This first parses out all the G functions with their inputs. Then, each of those is split into their function part and their input part. This is the put into a character vector of functions named for their input.

parse("G30(H).G3(M).G0(L).Replicate(1)")
>     H     M     L 
  "G30"  "G3"  "G0"

Or

parse("G35(L).G31(P).G02(K).Replicate(1)")
>     L     P     K 
  "G35" "G31" "G02"