diff --git a/Makefile b/Makefile
index 3b161fd1..1210c94f 100644
--- a/Makefile
+++ b/Makefile
@@ -184,12 +184,19 @@ data/%.txt:
 build-lib/chibi/char-set/derived.scm: data/UnicodeData.txt data/DerivedCoreProperties.txt chibi-scheme$(EXE)
 	$(CHIBI) tools/extract-unicode-props.scm --default > $@
 
+build-lib/chibi/char-set/width.scm: data/UnicodeData.txt data/EastAsianWidth.txt chibi-scheme$(EXE)
+	$(CHIBI) tools/extract-unicode-props.scm Zero-Width=Mn > $@
+	$(CHIBI) tools/extract-unicode-props.scm -d data/EastAsianWidth.txt Full-Width=F@1,W@1 Ambiguous-Width=A@1 >> $@
+
 lib/chibi/char-set/ascii.scm: build-lib/chibi/char-set/derived.scm chibi-scheme$(EXE)
 	$(CHIBI) -Abuild-lib tools/optimize-char-sets.scm --ascii chibi.char-set.compute > $@
 
 lib/chibi/char-set/full.scm: build-lib/chibi/char-set/derived.scm chibi-scheme$(EXE)
 	$(CHIBI) -Abuild-lib tools/optimize-char-sets.scm chibi.char-set.compute > $@
 
+lib/chibi/show/width.scm: build-lib/chibi/char-set/width.scm chibi-scheme$(EXE)
+	$(CHIBI) -Abuild-lib tools/optimize-char-sets.scm --predicate chibi.char-set.width > $@
+
 lib/scheme/char/case-offsets.scm: data/UnicodeData.txt chibi-scheme$(EXE) all-libs
 	$(CHIBI) tools/extract-case-offsets.scm $< > $@
 
diff --git a/lib/chibi/iset/optimize.scm b/lib/chibi/iset/optimize.scm
index 8657cdb5..887696c6 100644
--- a/lib/chibi/iset/optimize.scm
+++ b/lib/chibi/iset/optimize.scm
@@ -168,3 +168,23 @@
                     ,(iset-bits iset)
                     ,(iset->code (iset-left iset))
                     ,(iset->code (iset-right iset)))))
+
+;; uses only if, <, <=, >, and SRFI 151 bit-set?
+(define (iset->code/lambda iset)
+  (define (code iset)
+    (and iset
+         (if (and (not (iset-left iset))
+                  (not (iset-right iset))
+                  (not (iset-bits iset)))
+             `(<= ,(iset-start iset) n ,(iset-end iset))
+             `(if (< n ,(iset-start iset))
+               ,(code (iset-left iset))
+               ,(if (and (not (iset-right iset)) (not (iset-bits iset)))
+                    `(<= n ,(iset-end iset))
+                    `(if (> n ,(iset-end iset))
+                         ,(code (iset-right iset))
+                         ,(if (iset-bits iset)
+                              `(bit-set? (- n ,(iset-start iset))
+                                         ,(iset-bits iset))
+                              #t)))))))
+  `(lambda (n) ,(code iset)))
diff --git a/lib/chibi/iset/optimize.sld b/lib/chibi/iset/optimize.sld
index 4a2f52bc..7ad11c61 100644
--- a/lib/chibi/iset/optimize.sld
+++ b/lib/chibi/iset/optimize.sld
@@ -17,4 +17,5 @@
         (bitwise-and (%mask size) (arithmetic-shift n (- position)))))))
   (include "optimize.scm")
   (export
-   iset-balance iset-balance! iset-optimize iset-optimize! iset->code))
+   iset-balance iset-balance! iset-optimize iset-optimize!
+   iset->code iset->code/lambda))
diff --git a/lib/chibi/show-test.sld b/lib/chibi/show-test.sld
index 8880ac14..f337bfc8 100644
--- a/lib/chibi/show-test.sld
+++ b/lib/chibi/show-test.sld
@@ -742,7 +742,13 @@ def | 6
       (test "日本語"
             (show #f (as-unicode (with ((pad-char #\〜)) (padded/both 5 "日本語")))))
       (test "日本語 col: 6"
-            (show #f (as-unicode "日本語" (fn ((col)) (each " col: " col)))))
+            (show #f (as-unicode "日本語" (fn (col) (each " col: " col)))))
+      (test "日本語ΠΜΕ col: 9"
+            (show #f (as-unicode "日本語ΠΜΕ" (fn (col) (each " col: " col)))))
+      (test "日本語ΠΜΕ col: 12"
+          (show #f (with ((ambiguous-is-wide? #t))
+                     (as-unicode "日本語ΠΜΕ"
+                                 (fn (col) (each " col: " col))))))
 
       ;; from-file
       ;; for reference, filesystem-test relies on creating files under /tmp
diff --git a/lib/chibi/show/unicode.scm b/lib/chibi/show/unicode.scm
index 3cee213d..d4c665f9 100644
--- a/lib/chibi/show/unicode.scm
+++ b/lib/chibi/show/unicode.scm
@@ -1,134 +1,70 @@
 ;; unicode.scm -- Unicode character width and ANSI escape support
-;; Copyright (c) 2006-2017 Alex Shinn.  All rights reserved.
+;; Copyright (c) 2006-2020 Alex Shinn.  All rights reserved.
 ;; BSD-style license: http://synthcode.com/license.txt
 
-;; a condensed non-spacing mark range from UnicodeData.txt (chars with
-;; the Mn property) - generated partially by hand, should automate
-;; this better
-
-(define low-non-spacing-chars
-  (bytevector
-#xff #xff #xff #xff #xff #xff #xff #xff #xff #xff #xff #xff #xff #xff    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-#x78    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0 #xfe #xff #xff #xff #xff #xff #x1f    0    0    0    0    0    0    0
-   0    0 #x3f    0    0    0    0    0    0 #xf8 #xff #x01    0    0 #x01    0
-   0    0    0    0    0    0    0    0    0    0 #xc0 #xff #xff #x3f    0    0
-   0    0 #x02    0    0    0 #xff #xff #xff #x07    0    0    0    0    0    0
-   0    0    0    0 #xc0 #xff #x01    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-#x06    0    0    0    0    0    0 #x10 #xfe #x21 #x1e    0 #x0c    0    0    0
-#x02    0    0    0    0    0    0 #x10 #x1e #x20    0    0 #x0c    0    0    0
-#x06    0    0    0    0    0    0 #x10 #xfe #x3f    0    0    0    0 #x03    0
-#x06    0    0    0    0    0    0 #x30 #xfe #x21    0    0 #x0c    0    0    0
-#x02    0    0    0    0    0    0 #x90 #x0e #x20 #x40    0    0    0    0    0
-#x04    0    0    0    0    0    0    0    0 #x20    0    0    0    0    0    0
-   0    0    0    0    0    0    0 #xc0 #xc1 #xff #x7f    0    0    0    0    0
-   0    0    0    0    0    0    0 #x10 #x40 #x30    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0 #x0e #x20    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0 #x04 #x7c    0    0    0    0    0
-   0    0    0    0    0    0 #xf2 #x07 #x80 #x7f    0    0    0    0    0    0
-   0    0    0    0    0    0 #xf2 #x1f    0 #x3f    0    0    0    0    0    0
-   0    0    0 #x03    0    0 #xa0 #x02    0    0    0    0    0    0 #xfe #x7f
-#xdf    0 #xff #xff #xff #xff #xff #x1f #x40    0    0    0    0    0    0    0
-   0    0    0    0    0 #xe0 #xfd #x02    0    0    0 #x03    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0 #x1c    0    0    0 #x1c    0    0    0 #x0c    0    0    0 #x0c    0
-   0    0    0    0    0    0 #x80 #x3f #x40 #xfe #x0f #x20    0    0    0    0
-   0 #x38    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0 #x02    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0 #x87 #x01 #x04 #x0e    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0    0    0    0    0    0    0
-   0    0    0    0    0    0    0    0    0    0 #xff #x1f #xe2 #x07
-       ))
-
-(define (unicode-char-width c)
-  (let ((ci (char->integer c)))
+(define (unicode-char-width ch ambiguous-is-wide?)
+  (let ((ci (char->integer ch)))
     (cond
-      ;; hand-checked ranges from EastAsianWidth.txt
-      ((<= #x1100 ci #x115F) 2) ; Hangul
-      ((<= #x2E80 ci #x4DB5) 2) ; CJK
-      ((<= #x4E00 ci #xA4C6) 2)
-      ((<= #xAC00 ci #xD7A3) 2) ; Hangul
-      ((<= #xF900 ci #xFAD9) 2) ; CJK compat
-      ((<= #xFE10 ci #xFE6B) 2)
-      ((<= #xFF01 ci #xFF60) 2)
-      ((<= #xFFE0 ci #xFFE6) 2)
-      ((<= #x20000 ci #x30000) 2)
-      ;; non-spacing mark (Mn) ranges from UnicodeData.txt
-      ((<= #x0300 ci #x3029)
-       ;; inlined bit-vector-ref for portability
-       (let* ((i (- ci #x0300))
-              (byte (quotient i 8))
-              (off (remainder i 8)))
-         (if (zero? (bitwise-and (bytevector-u8-ref low-non-spacing-chars byte)
-                                 (arithmetic-shift 1 off)))
-             1
-             0)))
-      ((<= #x302A ci #x302F) 0)
-      ((<= #x3099 ci #x309A) 0)
-      ((= #xFB1E ci) 0)
-      ((<= #xFE00 ci #xFE23) 0)
-      ((<= #x1D167 ci #x1D169) 0)
-      ((<= #x1D17B ci #x1D182) 0)
-      ((<= #x1D185 ci #x1D18B) 0)
-      ((<= #x1D1AA ci #x1D1AD) 0)
-      ((<= #xE0100 ci #xE01EF) 0)
-      (else 1))))
+     ((char-set:zero-width? ci)
+      0)
+     ((char-set:full-width? ci)
+      2)
+     ((and ambiguous-is-wide? (char-set:ambiguous-width? ci))
+      2)
+     (else
+      1))))
+
+(define (unicode-terminal-width/aux str start end ambiguous-is-wide?)
+  (let lp1 ((sc start) (width 0))
+    (if (string-cursor>=? sc end)
+        width
+        (let ((c (string-ref/cursor str sc)))
+          (cond
+           ;; ANSI escapes
+           ;; TODO: maintain a state machine so the escape can be
+           ;; spread across multiple strings
+           ((and (= 27 (char->integer c)) ; esc
+                 (string-cursor<? (string-cursor-next str sc) end)
+                 (eqv? #\[ (string-ref/cursor str (string-cursor-next str sc))))
+            (let lp2 ((sc (string-cursor-forward str sc 2)))
+              (cond ((string-cursor>=? sc end) width)
+                    ((memv (string-ref/cursor str sc) '(#\m #\newline))
+                     (lp1 (string-cursor-next str sc) width))
+                    (else (lp2 (string-cursor-next str sc))))))
+           ;; fast-path ASCII
+           ((char<=? c #\~)
+            (lp1 (string-cursor-next str sc) (+ width 1)))
+           ;; unicode
+           (else
+            (lp1 (string-cursor-next str sc)
+                 (+ width (unicode-char-width c ambiguous-is-wide?)))))))))
+
+(define (cursor-arg str x)
+  (if (string-cursor? x) x (string-index->cursor str x)))
+
+;; convert args to cursors internally for efficiency
 
 (define (unicode-terminal-width str . o)
-  (let ((start (if (pair? o) (car o) 0))
-        (end (if (and (pair? o) (pair? (cdr o)))
-                 (cadr o)
-                 (string-length str))))
-    (let lp1 ((i start) (width 0))
-      (if (>= i end)
-          width
-          (let ((c (string-ref str i)))
-            (cond
-              ;; ANSI escapes
-              ((and (= 27 (char->integer c)) ; esc
-                    (< (+ i 1) end)
-                    (eqv? #\[ (string-ref str (+ i 1))))
-               (let lp2 ((i (+ i 2)))
-                 (cond ((>= i end) width)
-                       ((memv (string-ref str i) '(#\m #\newline))
-                        (lp1 (+ i 1) width))
-                       (else (lp2 (+ i 1))))))
-              ;; unicode characters
-              ((>= (char->integer c) #x80)
-               (lp1 (+ i 1) (+ width (unicode-char-width c))))
-              ;; normal ASCII
-              (else (lp1 (+ i 1) (+ width 1)))))))))
+  (let ((start (cursor-arg str (if (pair? o)
+                                   (car o)
+                                   (string-cursor-start str))))
+        (end (cursor-arg str (if (and (pair? o) (pair? (cdr o)))
+                                 (cadr o)
+                                 (string-cursor-end str)))))
+    (unicode-terminal-width/aux str start end #f)))
+
+(define (unicode-terminal-width/wide str . o)
+  (let ((start (cursor-arg str (if (pair? o)
+                                   (car o)
+                                   (string-cursor-start str))))
+        (end (cursor-arg str (if (and (pair? o) (pair? (cdr o)))
+                                 (cadr o)
+                                 (string-cursor-end str)))))
+    (unicode-terminal-width/aux str start end #t)))
 
 (define (as-unicode . args)
-  (with ((string-width unicode-terminal-width))
-    (each-in-list args)))
+  (fn (ambiguous-is-wide?)
+    (with ((string-width (if ambiguous-is-wide?
+                             unicode-terminal-width/wide
+                             unicode-terminal-width)))
+      (each-in-list args))))
diff --git a/lib/chibi/show/unicode.sld b/lib/chibi/show/unicode.sld
index 88e84ab3..435e9ba9 100644
--- a/lib/chibi/show/unicode.sld
+++ b/lib/chibi/show/unicode.sld
@@ -1,5 +1,5 @@
 
 (define-library (chibi show unicode)
-  (import (scheme base) (chibi show base) (srfi 151))
-  (export as-unicode unicode-terminal-width)
-  (include "unicode.scm"))
+  (import (scheme base) (chibi show base) (srfi 130) (srfi 151))
+  (export as-unicode unicode-terminal-width unicode-terminal-width/wide)
+  (include "width.scm" "unicode.scm"))
diff --git a/lib/srfi/166.sld b/lib/srfi/166.sld
index 2eb2df4f..a9c9e1a6 100644
--- a/lib/srfi/166.sld
+++ b/lib/srfi/166.sld
@@ -19,14 +19,14 @@
    ;; state variables
    port row col width output writer string-width pad-char ellipsis
    radix precision decimal-sep decimal-align sign-rule
-   comma-sep comma-rule word-separator?
+   comma-sep comma-rule word-separator? ambiguous-is-wide?
    ;; pretty
    pretty pretty-shared pretty-simply pretty-color
    ;; columnar
    columnar tabular wrapped wrapped/list wrapped/char
    justified from-file line-numbers show-columns
    ;; unicode
-   as-unicode unicode-terminal-width
+   as-unicode unicode-terminal-width unicode-terminal-width/wide
    upcased downcased
    ;; color
    as-red as-blue as-green as-cyan as-yellow
diff --git a/lib/srfi/166/base.scm b/lib/srfi/166/base.scm
index e3a1f020..2ae9857b 100644
--- a/lib/srfi/166/base.scm
+++ b/lib/srfi/166/base.scm
@@ -38,6 +38,7 @@
   (output output-default)
   (string-width substring-length)
   (word-separator? char-whitespace?)
+  (ambiguous-is-wide? #f)
   (ellipsis "")
   (decimal-align #f)
   (decimal-sep #f)
diff --git a/lib/srfi/166/base.sld b/lib/srfi/166/base.sld
index dc7da4f5..177ea01b 100644
--- a/lib/srfi/166/base.sld
+++ b/lib/srfi/166/base.sld
@@ -48,7 +48,7 @@
    ;; state variables
    port row col width output writer string-width pad-char ellipsis
    radix precision decimal-sep decimal-align sign-rule
-   comma-sep comma-rule word-separator?
+   comma-sep comma-rule word-separator? ambiguous-is-wide?
    )
   (include "base.scm")
   (include "../../chibi/show/write.scm")
diff --git a/lib/srfi/166/test.sld b/lib/srfi/166/test.sld
index 583b4018..98146e79 100644
--- a/lib/srfi/166/test.sld
+++ b/lib/srfi/166/test.sld
@@ -741,6 +741,12 @@ def | 6
             (show #f (as-unicode (with ((pad-char #\〜)) (padded/both 5 "日本語")))))
       (test "日本語 col: 6"
             (show #f (as-unicode "日本語" (fn (col) (each " col: " col)))))
+      (test "日本語ΠΜΕ col: 9"
+            (show #f (as-unicode "日本語ΠΜΕ" (fn (col) (each " col: " col)))))
+      (test "日本語ΠΜΕ col: 12"
+          (show #f (with ((ambiguous-is-wide? #t))
+                     (as-unicode "日本語ΠΜΕ"
+                                 (fn (col) (each " col: " col))))))
 
       ;; from-file
       ;; for reference, filesystem-test relies on creating files under /tmp
diff --git a/lib/srfi/166/unicode.sld b/lib/srfi/166/unicode.sld
index 5284244e..56729106 100644
--- a/lib/srfi/166/unicode.sld
+++ b/lib/srfi/166/unicode.sld
@@ -1,5 +1,6 @@
 
 (define-library (srfi 166 unicode)
-  (import (scheme base) (srfi 151) (srfi 166 base))
-  (export-all)
-  (include "../../chibi/show/unicode.scm"))
+  (import (scheme base) (srfi 130) (srfi 151) (srfi 166 base))
+  (export as-unicode unicode-terminal-width unicode-terminal-width/wide)
+  (include "../../chibi/show/width.scm"
+           "../../chibi/show/unicode.scm"))
diff --git a/tools/extract-unicode-props.scm b/tools/extract-unicode-props.scm
index 0ab57a8e..9f203bde 100755
--- a/tools/extract-unicode-props.scm
+++ b/tools/extract-unicode-props.scm
@@ -15,6 +15,7 @@
 ;; A value can be any of:
 ;;
 ;;   Property_Name: all unicode characters with the given derived property
+;;   Prop@N: a property matched at a given field (instead of default category)
 ;;   Xx: all unicode characters with the given general category
 ;;   X: all unicode characters with any general category X*
 ;;   NNNN: a single unicode value in hex format
@@ -25,7 +26,7 @@
 ;; Unicode property name.
 ;;
 ;; Assumes the files UnicodeData.txt and DerivedCoreProperties.txt are
-;; in the data/ current directory, unless overridden with the --data or
+;; in the ./data/ directory, unless overridden with the --data or
 ;; --derived options.
 
 (import (chibi) (chibi io) (chibi string))
@@ -35,8 +36,9 @@
     (for-each (lambda (x) (display x err)) args)
     (newline err)))
 
-;; Parse UnicodeData.txt for characters matching a given class.
-(define (extract-char-set-category cat data)
+;; Parse UnicodeData.txt or other semi-colon-delimited TSV file for
+;; characters matching a given class in a given field.
+(define (extract-char-set-category cat field data)
   (define (join-to-range n ls)
     (cond
      ((null? ls)
@@ -62,22 +64,43 @@
            ((or (equal? line "") (eqv? #\# (string-ref line 0)))
             (lp ranges))
            (else
-            (let ((ls (string-split line #\; 4)))
+            (let* ((line (substring-cursor line
+                                           (string-cursor-start line)
+                                           (string-find line #\#)))
+                   (ls (map string-trim (string-split line #\;))))
               (cond
-               ((< (length ls) 3)
+               ((<= (length ls) field)
                 (warn "invalid UnicodeData line: " line)
                 (lp ranges))
                (else
-                (let ((ch (string->number (car ls) 16))
+                (let ((ch (if (string-contains (car ls) "..")
+                              (let* ((sc (string-contains (car ls) "..")))
+                                (cons (string->number
+                                       (substring-cursor
+                                        (car ls)
+                                        (string-cursor-start (car ls))
+                                        sc)
+                                       16)
+                                      (string->number
+                                       (substring-cursor
+                                        (car ls)
+                                        (string-cursor-forward (car ls) sc 2))
+                                       16)))
+                              (string->number (car ls) 16)))
                       (name (cadr ls))
-                      (ch-cat (car (cddr ls))))
+                      (ch-cat (list-ref ls field)))
                   (cond
-                   ((or (not ch) (not (= 2 (string-length ch-cat))))
+                   ((not (or (integer? ch)
+                             (and (pair? ch)
+                                  (integer? (car ch))
+                                  (integer? (cdr ch)))))
                     (warn "invalid UnicodeData line: " line))
                    ((if (char? cat)
                         (eqv? cat (string-ref ch-cat 0))
                         (equal? cat ch-cat))
-                    (lp (join-to-range ch ranges)))
+                    (lp (if (pair? ch)
+                            (cons ch ranges)
+                            (join-to-range ch ranges))))
                    (else
                     (lp ranges))))))))))))))
 
@@ -131,13 +154,19 @@
             (error "invalid character range, expected NNNN-MMMM, got: " def))))
      ((string->number def 16)
       => (lambda (start) `(char-set ,(integer->char start))))
+     ((string-find? def #\@)
+      (let* ((sc (string-find def #\@))
+             (cat (substring-cursor def (string-cursor-start def) sc))
+             (field (string->number
+                     (substring-cursor def (string-cursor-next def sc)))))
+        (extract-char-set-category cat field data)))
      ((and (= 1 (string-length def))
            (char-upper-case? (string-ref def 0)))
-      (extract-char-set-category (string-ref def 0) data))
+      (extract-char-set-category (string-ref def 0) 2 data))
      ((and (= 2 (string-length def))
            (char-upper-case? (string-ref def 0))
            (char-lower-case? (string-ref def 1)))
-      (extract-char-set-category def data))
+      (extract-char-set-category def 2 data))
      ;; derived properties
      ((and (> (string-length def) 1)
            (eqv? #\: (string-ref def 0)))
diff --git a/tools/optimize-char-sets.scm b/tools/optimize-char-sets.scm
index d020bf1b..8eebcfcc 100644
--- a/tools/optimize-char-sets.scm
+++ b/tools/optimize-char-sets.scm
@@ -21,13 +21,15 @@
     (remove (lambda (i) (hash-table-exists? ls2-tab i)) ls1)))
 
 (let ((args (command-line)))
-  (let lp ((ls (cdr args)) (ascii? #f))
+  (let lp ((ls (cdr args)) (ascii? #f) (predicate? #f))
     (cond
      ((and (pair? ls) (not (equal? "" (car ls)))
            (eqv? #\- (string-ref (car ls) 0)))
       (cond
        ((member (car ls) '("-a" "--ascii"))
-        (lp (cdr ls) #t))
+        (lp (cdr ls) #t predicate?))
+       ((member (car ls) '("-p" "--predicate"))
+        (lp (cdr ls) ascii? #t))
        (else (error "unknown option" (car ls)))))
      ((or (null? ls) (pair? (cdr ls)))
       (error "usage: optimize-char-sets.scm [--ascii] module.name"))
@@ -78,8 +80,13 @@
                          (newline (current-error-port))
                          (error "optimized iset is different"))))
                  (display "  writing\n" (current-error-port))
-                 (write `(define ,exp
-                           (immutable-char-set ,(iset->code iset2))))
+                 (write
+                  (if predicate?
+                      `(define ,(string->symbol
+                                 (string-append (symbol->string exp) "?"))
+                         ,(iset->code/lambda iset2))
+                      `(define ,exp
+                         (immutable-char-set ,(iset->code iset2)))))
                  (newline)
                  (newline)
                  (display "  done\n" (current-error-port)))))))