hangul syllables

2026-02-12 09:45:52 -07:00
parent 0ba6518630
commit fd36692419
8 changed files with 197 additions and 81 deletions
--- a/scripts/syllables-to-jamo
+++ b/scripts/syllables-to-jamo
@@ -0,0 +1,44 @@
+#!/usr/bin/env bb
+
+(require '[babashka.fs :as fs]
+         '[clojure.string :as str])
+
+(defn syllables->jamo [s]
+  "Convert Hangul syllables in string S to their jamo components."
+  (->> s
+       (map int)
+       (mapcat
+        (fn [c]
+          (if (<= 0xAC00 c 0xD7A3)
+            ;; Hangul syllable - decompose
+            (let [code (- c 0xAC00)
+                  lead (quot code (* 21 28))
+                  medial (quot (mod code (* 21 28)) 28)
+                  final (mod code 28)
+                  lead-jamo (+ 0x1100 lead)
+                  medial-jamo (+ 0x1161 medial)
+                  final-jamo (if (> final 0) (+ 0x11A7 final) nil)]
+              (remove nil? [lead-jamo medial-jamo final-jamo]))
+            ;; Not a Hangul syllable
+            [c])))
+       (map char)
+       (apply str)))
+
+
+(defn -main [& args]
+  (if (seq args)
+    (doseq [f args]
+      (let [x (-> (slurp f)
+                  (str/replace #"\p{block=HangulSyllables}+"
+                               syllables->jamo))]
+        (spit f x)))
+    (loop [line (read-line)]
+      (when line
+        (-> line
+            (str/replace #"\p{block=HangulSyllables}+"
+                         syllables->jamo)
+            println)
+        (recur (read-line))))))
+
+(when (= *file* (System/getProperty "babashka.file"))
+  (apply -main *command-line-args*))