Easy way to extract text between defined set of strings in R - regex

Question

I have some text with defined labels and need to split the text according to the labels.
For example given the text with labels set {A, B, C..}
text <- c("A: how are you B: hello sir C: bye bye")
text2 <- c("USER COMMENTS: TEST PROC: Refer manual. SOLUTION: fix BIAS32 user:param", "TEST PROC: install spare unit. USER COMMENTS: hello sir SOLUTION: tighten bolt 12","TEST PROC: bye bye.")
I need to extract text "how are you", "hello sir" , etc.. corresponding to labels A, B, etc.
There is no specific order of the labels, certain labels could be missing and labels can be phrases (not just characters)
This is what I have so far to extract text corresponding to label A:
gsub("(.*A.*:)(.*)(B.*|C.*)","\\2",text,perl=TRUE)
But this does not work in so many cases!
I am looking for a solution where I can define a vector of labels such as
labels <- c("USER COMMENTS", "TEST PROC", "SOLUTION") # this is a big list!
and extract the text corresponding to these labels as below
USER COMMENTS are "", "hello sir"
TEST PROC are "Refer manual.", "install spare unit.","bye bye."
SOLUTION are "fix BIAS32 user:param", "tighten bolt 12"
etc..

Answer 1

I think I might have a solution based on Sharath's comment.
First, there's strsplit(), which can split a vector based on regex. In your case you could use:
labels2<-paste(labels,collapse="|")
[1] "USER COMMENTS|TEST PROC|SOLUTION"
If you apply strsplit on that:
splittedtext<-strsplit(text2,labels2)
[[1]]
[1] "" ": "
[3] ": Refer manual. " ": fix BIAS32 user:param"
[[2]]
[1] "" ": install spare unit. " ": hello sir "
[4] ": tighten bolt 12"
[[3]]
[1] "" ": bye bye."
Pretty much what you want, right? You could do some refining by adding ": " to the end of every index, and the first element is gibberish. So taking care of the latter:
splittedtext<-lapply(splittedtext,"[",-1)
That generates the problem that you must figure out to which label a comment applies. For that you could use regexpr() function in R.
pos=sapply(labels,regexpr,text2)
USER COMMENTS TEST PROC SOLUTION
[1,] 1 16 41
[2,] 32 1 57
[3,] -1 1 -1
Each cell represents the position in which said label [column] appear on string [row]. -1 denote that it does not appear on this string.
Now switch, -1 for NA, and rank the remaining numbers. That will give to you which string snippet represents that label.
pos=ifelse(pos==-1,NA,pos) #switch -1 for NA
pos=t(apply(pos,1,rank,na.last="keep"))
USER COMMENTS TEST PROC SOLUTION
[1,] 1 2 3
[2,] 2 1 3
[3,] NA 1 NA
Now it's just matching.