unicode-truncate.clj

(defn truncate-safely-to [s bytelimit]
  "Truncate a UTF-8 string to the longest valid UTF-8 string less than bytelimit bytes"
  (if (< bytelimit 1) "" ; empty string
    (let [utf8 (.getBytes s "UTF-8")
          eos (- bytelimit 1)
          idx (if (<= (alength utf8) bytelimit)
                bytelimit
                (if (= 0 (bit-and (aget utf8 eos) 0x80))
                  eos
                  (first                               ;; get the character
                    (drop-while                        ;; before any at the end
                      #(let [c (aget utf8 %)]          ;; that have 0x80 set and 0x40 clear
                         (and (> (bit-and 0x80 c) 0)   ;; which indicates a non-initial byte
                              (= (bit-and 0x40 c) 0))) ;; of a multibyte utf-8 character
                      (range eos 0 -1))) ;; iterate back from the end
                  ))]
      (java.util.Arrays/copyOfRange utf8 0 idx))))
  
(let [ustr "This is ⟑ ┳∈Ⓢⓣ"]
  (map #(prn (truncate-safely-to ustr %)) (range 1 27)))

Some places I want to put a string have a byte-length limit. It's good to avoid cutting off UTF-8 strings mid-codepoint.