From a7080a47b30616aa85f7556d6bb57c37d810e81d Mon Sep 17 00:00:00 2001
From: Tan Kian-ting <chenjt30@gmail.com>
Date: Thu, 23 Oct 2025 05:52:34 +0800
Subject: [PATCH] add cjk-splitter pass

---
 expander.rkt | 21 +++++++++++++++++++++
 test1.ug     |  5 +++--
 2 files changed, 24 insertions(+), 2 deletions(-)

diff --git a/expander.rkt b/expander.rkt
index 2224d2f..211e3c7 100644
--- a/expander.rkt
+++ b/expander.rkt
@@ -69,12 +69,33 @@
                                  ["body" . ,remove-newline-between-cjk-aux])))
 
 
+(define [splitter-cjk-aux cont]
+  (display (list-ref cont 1))
+  (define [splitter-cjk-unit x]
+  (define tmp (regexp-match* #rx"[\u4E00-\u9FFF\u3400-\u4DBF\uF900-\uFAFF\U00020000-\U0002EBEF\u3000-\u303F·，。！？：」』》】］〕』〗〉｝「『《〔［【『〖〈｛…—]" x #:gap-select? #t))
+  (filter (lambda (x) [not (equal? x "")]) tmp))
+  (define [splitter-cjk-proc res rem]
+    (cond
+      [(eq? rem '()) (reverse res)]
+      [(string? [car rem]) [splitter-cjk-proc [append (reverse (splitter-cjk-unit [car rem])) res] (cdr rem)]]
+      [else [splitter-cjk-proc [cons (car rem) res] (cdr rem)]]
+      ))
+  (list-set cont 1 (splitter-cjk-proc '() (list-ref cont 1))))
+
+(define splitter-cjk (make-hash `(["lang" . "zh"]
+                                 ["body" . ,splitter-cjk-aux])))
+
+
+
+  
+
 (define-syntax append-plugin!
   (syntax-rules ()
     [(_ ls plugin) (set! ls (reverse (cons plugin (reverse ls))))]))
   
 [append-plugin! plugin-list two-newlines-into-par]
 [append-plugin! plugin-list remove-newline-between-cjk]
+[append-plugin! plugin-list splitter-cjk]
 
   (define (main-text txt)
     (begin
diff --git a/test1.ug b/test1.ug
index 9363ced..813293b 100644
--- a/test1.ug
+++ b/test1.ug
@@ -1,8 +1,9 @@
-#lang uahgi-ng
+#lang uahgi2
 {@display|`10.3}
 {@dict?|@main-config}
+{@set-main-config|lang|{@list|zh}}
 {@set-main-config|paper-width|`2480}
 {@display|{@dict-ref|@main-config|paper-width}}
 {@main-text|
-天地人123我為貓}
+{@hd2|我是貓 I'm a cat}天地人123我為貓}
 {@display|@main-frame}
\ No newline at end of file