php preg_match_all equivalent - regex

Question

I am looking for an R equivalent to PHP's preg_match_all function.
Objective:
Search a single string (not a vector of several strings) for a regexp pattern
Return a matrix of matches
Example:
Assume the following flat string without delimitation.
"This is a sample string written like a paragraph. In this string two sets of information exist. Each set contains two variables. We want to extract the sets and variables within those sets. Each information set is formatted the same way. The first set is Title: Sir; Last Name: John; and the second set is Title: Mr.; Last Name: Smith."
Using a regular expression pattern similar to
"Title: ([^;]*?); Last Name: ([^;.]*?)"
I would like to produce the following matrix from the above string:
[ ][,1] [,2]
[1,] Sir John
[2,] Mr. Smith
I have successfully accomplished this in PHP on a remote server using the preg_match_all function; however, the text files I am accessing are relatively large (not huge but slow to upload anyways). Building this in R will save a significant amount of time.
I have read up on use of grep, etc. in R but every example I have found searches for patterns in a vector and I have been unable to generate the matrix as described above.
I have also played with the stringr package but again I have not been successful generating a matrix.
This seems like a common task to me so I am sure someone smarter than me has found a solution before.

Answer 1

Consider the following option using regmatches :
x <- 'This is a sample string written like a paragraph. In this string two sets of information exist. Each set contains two variables. We want to extract the sets and variables within those sets. Each information set is formatted the same way. The first set is Title: Sir; Last Name: John; and the second set is Title: Mr.; Last Name: Smith.'
m <- regmatches(x, gregexpr('(?i)Title: \\K[^;]+|Last Name: \\K[^;.]+', x, perl=T))
matrix(unlist(m), ncol=2, byrow=T)
Output:
[,1] [,2]
[1,] "Sir" "John"
[2,] "Mr." "Smith"

Answer 2

For some reason there doesn't seem to be an easy way to extract captured matches in base (I wish regmatches also worked with captured groups but it does not). I ended up writing my own you can find it at regcapturedmatches.R. it will work with
a <- "The first set is Title: Sir and Last Name: John; and the second set is Title: Mr. and Last Name: Smith."
m<-gregexpr("Title: ([^;]*) and Last Name: ([^;.]*)", a, perl=T, ignore.case=T)
regcapturedmatches(a,m)[[1]]
This will return
[,1] [,2]
[1,] "Sir" "John"
[2,] "Mr." "Smith"
(I added the [[1]] because you said you would only operate on one string at a time. The function can operate on a vector and will return results in a list. Really, in R, every thing is a vector so there is no such thing as a "single" string, you just have a vector of strings with length 1.)
Of course this method is only as good as your regular expression. I had to modify your sample data a bit so your expression would match more than one Title/Name.

Answer 3

Here is a stringr version:
library(stringr)
str_match_all(x, pattern)
Produces:
[[1]]
[,1] [,2] [,3]
[1,] "Title: Sir and Last Name: John" "Sir" "John"
[2,] "Title: Mr. and Last Name: Smith" "Mr." "Smith"
Note that I had to edit your text so that the second one is also of form "and Last Name:". To get your matrix you can just do:
result[[1]][[-1]] # assumes the above is in `result`
One limitation of this is it uses regexec, which doesn't support perl regular expressions.