To start: This question might need more detail/clarification. On my first reading of the question, I believed the OP was asking for a general search function (ex. If find or regex was fed a sequence of 5 words, it would display all the matches.)
If I do not interpret the desired as a general search, then I believe you are asking for the function, given N as input, to iteratively:
- If there are N words at point and ahead, take them as input. If not, return output and exit.
- Capture the search results of this input within the document as output.
- Move forward one word.
(In other words, something similar to what writing assistant programs do to improve writing quality by reducing repeated phrases.)
I am unsure of what the desired output is. For simplicity, the output is a set of marks pushed to the mark ring. This way, you can use the mark ring to iterate through the results for editing/viewing as is best suited to whichever use case you have.
We can implement a function like this using regex to search for each phrase with word length N:
(defun search-repeating-N-words ()
(interactive)
(let (N x regex-search)
(setq N (read-number "How many words to search?: " 5))
(setq x 1)
(save-excursion
;; The search at the end will execute, even if there aren't N words left.
;; Given that it only searches forward, this is inefficient but not erroneous.
(while (< x (length (buffer-string))
(save-excursion
(let (p1 p2 (case-fold-search t))
(setq p1 x)
;; After search N words forward,
;; set end point as index of those last char of those words
(dotimes (y N (setq p2 (point)))
(skip-chars-forward "_a-z0-9"))
(setq regex-search
(buffer-substring-no-properties p1 p2)))
;; Only forward search is necessary. If it was repeated behind,
;; we would have caught it in the previous iterations.
;; For the record, this implementation does introduce
;; the problem where the same repeated phrase is
;; captured by multiple earlier searches.
(save-excursion
(while (search-forward regex-search nil t)
(let (p2)
(setq p2 (point))
(goto-char (- p2 (length regex-search)))
(push-mark p2))))
(setq x (+ x (skip-chars-forward "_a-zA-Z0-9") 1))))))))
