R function to extract a vector of substrings from a list or vector of strings, based on a regexpr

regexSubStrings.r

# A function that takes a vector or a list of strings and a regex pattern, 
# and returns a vector of strings that matches the regex patterns. Don't
# have parameter checking, error handling, etc. 

# The key lesson learned is that regmatches() is badly designed
# as it silently drops any non-matched elements. As a result, the length
# of the returned vector may nor may not be the same as the input. 
# had to use the good-o substring() trick.

# The second lesson is that when the regex pattern contains several (), such 
# as the example below, the order doesn't affect regexpr(pattern, x), but when
# doing gsub(pattern, "\\1", x), the \1 is NOT the first () that matched, but
# the first () in the pattern. As a result, I had to contatinate the \1\2,etc.
# which is really a hack. I don't know a better solution. 

regexSubStrings <-function(x, pattern) {
  #x<-unlist(x)
  y<-regexpr(pattern, x)
  str<-substring(unlist(x), y, y + attr(y, "match.length")-1) 
  # @@ can't use regmatches, because it does not return anything for non-matches
  #str<-ifelse(y>0, regmatches(unlist(x), y), "")
  #str<-regmatches(unlist(x), y)
  str<-gsub(pattern, "\\1\\2\\3\\4\\5\\6", str)
  return(str)
}

# in my data frame "trunk", trunk$extInfo is a vector of lists, with a single element in the list.   
# unlist(lapply(test, regexSubStrings))
test<-list("this is control:URL_0", "id is bad", "@id=`link_123`+link_234")
pattern <-"control:([a-zA-Z0-9_]+)|@id=`([a-zA-Z0-9_]+)|event:(taskStart)"
recoveredEvents<-regexSubStrings(test, pattern)