Detect a length&alphanumeric pattern in a vector of strings - r

Question

I have a character vector that stores different sentences. Some sentences include an alphanumeric string of a fixed length (15 characters), but some don't. Such string(s) of interest might vary in their combination of alphanumeric characters, but regardless will always be comprised of:
upper case letters
lower case letters
digits
no special characters
and will always be length of 15.
In some vector elements, there will be leading and trailing blank space before/after the string of interest. However, in other elements the string might show immediately at the beginning, or otherwise at the end. The most complex situation is when the string of interest shows without any spaces before/after, which means that it's embedded within another string.
I want to take such a vector and manipulate to return a new vector of the same length, but:
in elements that contained a string of interest, return only the string of interest.
in elements that did not contain a string of interest, return NA.
Example
set.seed(2020)
library(stringi)
library(stringr)
vector_of_strings <- stri_rand_strings(n = 100, length = 15, pattern = "[A-Za-z0-9]")
my_sentences <-
c(
str_interp("my sentece contains ${sample(vector_of_strings, size = 1)}"),
str_interp("${sample(vector_of_strings, size = 1)} is in my sentence"),
str_interp("sometimes - ${sample(vector_of_strings, size = 1)} - it shows like this"),
str_interp("other times it could be${sample(vector_of_strings, size = 1)}without any space before or after"),
"occasionally there's no string of interest, so such element should become 'NA'"
)
my_sentences
## [1] "my sentece contains 8OarR1YUGPBoRfi"
## [2] "WoV8ym3WB2zg2TD is in my sentence"
## [3] "sometimes - pmMk73q0L73qKUa - it shows like this"
## [4] "other times it could be1qvzWei5FxPtRGXwithout any space before or after"
## [5] "occasionally there's no string of interest, so such element should become 'NA'"
How can I take my_sentences and have it return the following?
[1] "8OarR1YUGPBoRfi"
[2] "WoV8ym3WB2zg2TD"
[3] "pmMk73q0L73qKUa"
[4] "1qvzWei5FxPtRGX"
[5] NA
EDIT
Based on ekoam's comment, I wonder whether the following mechanism could be utilized.
(1) First step, test whether any of the strings in vector_of_strings exists in my_sentences's elements. If yes, return the string of interest.
(2) Else, if no match, test whether any combination of alphanumeric and 15-character length exists. If there's a definitive single match, return the matched string.
(3).a. Else, if there's more than one possible match, return all possible strings.
(3).b. Else, if there's no match whatsoever, return NA.
For the sake of this example, let's assume that step (1) above is based on matching against sample(vector_of_strings, size = 50). This is for leaving some room for no match (to be able to move forward to step (2).
And just to make it clearer as of the desired output, I'm trying to get it all in a single vector that is of the same length as the original my_sentences, and the output(s) of the "mechanism" described above are at the respective element positions of the original vector.

Answer 1

This is not straightforwardly addressed. All the conditions you mention are easily operationalizable but one: that the string of interest can be embedded within a larger string. What you could do is extract all words that have the right combination of character types but allow for their length to go beyond 15:
library(stringr)
x <- str_extract(my_sentences, "\\b[A-Za-z0-9]{15,}\\b")
x
[1] "heLvIQabKdDTrBC" "KpxeqHQ0Z94X6vG" "UNMcDuUDzPsRU7s" "beZccQAS3rCFZ5UO7without"
[5] NA
This way you would at least be sure not to have overlooked the embedded target strings. If the number of such strings is not too large you could then in a second step isolate the larger-than-15-char strings and remove the unwanted bits:
x[which(nchar(x) > 15)]
[1] "beZccQAS3rCFZ5UO7without"