(defglobal **character-database** ,character-database)
(defglobal **character-decompositions** ,decompositions)
(defglobal **character-long-decompositions** ,long-decompositions)
+ (defglobal **character-primary-compositions**
+ (let ((table (make-hash-table))
+ (info ,(read-ub8-vector (file "comp" "dat"))))
+ (flet ((code (j)
+ (dpb (aref info (* 4 j))
+ (byte 8 24)
+ (dpb (aref info (+ (* 4 j) 1))
+ (byte 8 16)
+ (dpb (aref info (+ (* 4 j) 2))
+ (byte 8 8)
+ (aref info (+ (* 4 j) 3)))))))
+ #!+sb-unicode
+ (dotimes (i (/ (length info) 12))
+ (setf (gethash (dpb (code (* 3 i)) (byte 21 21)
+ (code (1+ (* 3 i))))
+ table)
+ (code-char (code (+ (* 3 i) 2)))))
+ table)))
(defun !character-database-cold-init ()
(setf **character-database** ,character-database))
,(with-open-file (stream (file "ucd-names" "lisp-expr")
#!+sb-doc
"The argument must be a character object; UPPER-CASE-P returns T if the
argument is an upper-case character, NIL otherwise."
- (< (ucd-value-0 char) 4))
+ (< (ucd-value-0 char) 5))
(defun lower-case-p (char)
#!+sb-doc
"The argument must be a character object; LOWER-CASE-P returns T if the
argument is a lower-case character, NIL otherwise."
- (< 3 (ucd-value-0 char) 8))
+ (< 4 (ucd-value-0 char) 9))
(defun both-case-p (char)
#!+sb-doc
"The argument must be a character object. BOTH-CASE-P returns T if the
argument is an alphabetic character and if the character exists in both upper
and lower case. For ASCII, this is the same as ALPHA-CHAR-P."
- (< (ucd-value-0 char) 8))
+ (< (ucd-value-0 char) 9))
(defun digit-char-p (char &optional (radix 10.))
#!+sb-doc
(defmacro equal-char-code (character)
(let ((ch (gensym)))
`(let ((,ch ,character))
- (if (= (ucd-value-0 ,ch) 0)
+ (if (< (ucd-value-0 ,ch) 5)
(ucd-value-1 ,ch)
(char-code ,ch)))))
#!+sb-doc
"Return CHAR converted to upper-case if that is possible. Don't convert
lowercase eszet (U+DF)."
- (if (< 3 (ucd-value-0 char) 8)
+ (if (< 4 (ucd-value-0 char) 9)
(code-char (ucd-value-1 char))
char))
(defun char-downcase (char)
#!+sb-doc
"Return CHAR converted to lower-case if that is possible."
- (if (< (ucd-value-0 char) 4)
+ (if (< (ucd-value-0 char) 5)
(code-char (ucd-value-1 char))
char))
(go again)))
(apply 'concatenate 'string (nreverse result))))
-#+nil
(defun primary-composition (char1 char2)
- (when (and (char= char1 #\e)
- (char= char2 #\combining_acute_accent))
- #\latin_small_letter_e_with_acute))
+ (let ((c1 (char-code char1))
+ (c2 (char-code char2)))
+ (cond
+ ((gethash (dpb (char-code char1) (byte 21 21) (char-code char2))
+ **character-primary-compositions**))
+ ((and (<= #x1100 c1) (<= c1 #x1112)
+ (<= #x1161 c2) (<= c2 #x1175))
+ (let ((lindex (- c1 #x1100))
+ (vindex (- c2 #x1161)))
+ (code-char (+ #xac00 (* lindex 588) (* vindex 28)))))
+ ((and (<= #xac00 c1) (<= c1 #.(+ #xac00 11171))
+ (<= #x11a8 c2) (<= c2 #x11c2)
+ (= 0 (rem (- c1 #xac00) 28)))
+ (code-char (+ c1 (- c2 #x11a7)))))))
;;; This implements a sequence data structure, specialized for
;;; efficient deletion of characters at an index, along with tolerable
(labels ()
(let* ((result (list (list 0 (length string) string)))
(previous-starter-index (position 0 string :key #'ucd-ccc))
- (i (1+ previous-starter-index)))
- (when (= i (length string))
+ (i (and previous-starter-index (1+ previous-starter-index))))
+ (when (or (not i) (= i (length string)))
(return-from canonically-compose string))
(tagbody
again
- (when (and (> (- i previous-starter-index) 2)
+ (when (and (>= (- i previous-starter-index) 2)
;; test for Blocked (Unicode 3.11 para. D115)
+ ;;
+ ;; (assumes here that string has sorted combiners,
+ ;; so can look back just one step)
(>= (ucd-ccc (lref result (1- i)))
(ucd-ccc (lref result i))))
(when (= (ucd-ccc (lref result i)) 0)
(defun normalize-string (string &optional (form :nfd))
(declare (type (member :nfd :nfkd :nfc :nfkc) form))
+ #!-sb-unicode
+ (etypecase string
+ ((array nil (*)) string)
+ (string
+ (ecase form
+ ((:nfc :nfkc) string)
+ ((:nfd :nfkd) (error "Cannot normalize to ~A form in #-SB-UNICODE builds" form)))))
+ #!+sb-unicode
(etypecase string
- #!+sb-unicode
(base-string string)
- ((or (array character (*)) #!-sb-unicode base-string)
+ ((array character (*))
(ecase form
+ ((:nfc)
+ (canonically-compose (sort-combiners (decompose-string string))))
((:nfd)
(sort-combiners (decompose-string string)))
+ ((:nfkc)
+ (canonically-compose (sort-combiners (decompose-string string :compatibility))))
((:nfkd)
(sort-combiners (decompose-string string :compatibility)))))
((array nil (*)) string)))