Inverting a regex pattern in R to match all left and between a giving set of strings-CodePudding

Hello everyone I hope you guys are having a good one,

I have multiple and long strings of text in a dataset, I am trying to capture all text between , after and before a set of words, I will refere to this words as keywords

keywords= UHJ, uhj, AXY, axy, YUI, yui, OPL, opl, UJI, uji

if I have the following string:

UHJ This is only a test to AXY check regex in a YUI educational context so OPL please be kind UJI

The following regex will easily match my keywords:

UHJ|uhj|AXY|axy|YUI|yui|OPL|opl|UJI|uji

but since I am interested in capturing eveyrthing in between after and before those words, I am in some way wanting to capture the invert of my regex so that I can have something like this:

I have tried the following:

[^UHJ|uhj|AXY|axy|YUI|yui|OPL|opl|UJI|uji]

with no luck and in the future the keywords may change so please if you know a regex that would work in R that can achive my desired output

CodePudding user response：

The simplest solution is probably just to split by your pattern. (Note this includes an empty string if the text starts with a keyword.)

x <- "UHJ This is only a test to AXY check regex in a YUI educational context so OPL please be kind UJI"

strsplit(x, "UHJ|uhj|AXY|axy|YUI|yui|OPL|opl|UJI|uji")
# [[1]]
# [1] ""                         " This is only a test to "
# [3] " check regex in a "       " educational context so "
# [5] " please be kind "

Other options would be to use regmatches() with invert = TRUE. (This includes empty strings if the text starts or ends with a keyword.)

regmatches(
  x,
  gregexpr("UHJ|uhj|AXY|axy|YUI|yui|OPL|opl|UJI|uji", x, perl = TRUE),
  invert = TRUE
)
# [[1]]
# [1] ""                         " This is only a test to "
# [3] " check regex in a "       " educational context so "
# [5] " please be kind "

Or stringr::str_extract_all() with your pattern in both a lookbehind and a lookahead. (This doesn't include empty strings.)

library(stringr)

str_extract_all(
  x, 
  "(?<=UHJ|uhj|AXY|axy|YUI|yui|OPL|opl|UJI|uji). ?(?=UHJ|uhj|AXY|axy|YUI|yui|OPL|opl|UJI|uji)"
)
# [[1]]
# [1] " This is only a test to " " check regex in a "      
# [3] " educational context so " " please be kind "