(interactive)
(encode-hz-region (point-min) (point-max)))
-;; The following sets up a translation table (big5-to-cns) from Big 5
-;; to CNS encoding, using some auxiliary functions to make the code
-;; more readable.
-
-;; Many kudos to Himi! The used code has been adapted from his
-;; mule-ucs package.
-
-(defun big5-to-flat-code (num)
- "Convert NUM in Big 5 encoding to a `flat code'.
-0xA140 will be mapped to position 0, 0xA141 to position 1, etc.
-There are no gaps in the flat code."
-
- (let ((hi (/ num 256))
- (lo (% num 256)))
- (+ (* 157 (- hi #xa1))
- (- lo (if (>= lo #xa1) 98 64)))))
-
-(defun flat-code-to-big5 (num)
- "Convert NUM from a `flat code' to Big 5 encoding.
-This is the inverse function of `big5-to-flat-code'."
-
- (let ((hi (/ num 157))
- (lo (% num 157)))
- (+ (* 256 (+ hi #xa1))
- (+ lo (if (< lo 63) 64 98)))))
-
-(defun euc-to-flat-code (num)
- "Convert NUM in EUC encoding (in GL representation) to a `flat code'.
-0x2121 will be mapped to position 0, 0x2122 to position 1, etc.
-There are no gaps in the flat code."
-
- (let ((hi (/ num 256))
- (lo (% num 256)))
- (+ (* 94 (- hi #x21))
- (- lo #x21))))
-
-(defun flat-code-to-euc (num)
- "Convert NUM from a `flat code' to EUC encoding (in GL representation).
-The inverse function of `euc-to-flat-code'. The high and low bytes are
-returned in a list."
-
- (let ((hi (/ num 94))
- (lo (% num 94)))
- (list (+ hi #x21) (+ lo #x21))))
-
-(defun expand-euc-big5-alist (alist)
- "Create a translation table and fills it with data given in ALIST.
-Elements of ALIST can be either given as
-
- ((euc-charset . startchar) . (big5-range-begin . big5-range-end))
-
-or as
-
- (euc-character . big5-charcode)
-
-The former maps a range of glyphs in an EUC charset (where STARTCHAR
-is in GL representation) to a certain range of Big 5 encoded
-characters, the latter maps a single glyph. Glyphs which can't be
-mapped will be represented with the byte 0xFF.
-
-The return value is the filled translation table."
-
- (let (chartable
- elem
- result
- char
- big5
- i
- end
- codepoint
- charset)
- (setq chartable (make-char-table 'translation-table #xFF))
- (while alist
- (setq elem (car alist)
- char (car elem)
- big5 (cdr elem)
- alist (cdr alist))
- (cond ((and (consp char)
- (consp big5))
- (setq i (big5-to-flat-code (car big5))
- end (big5-to-flat-code (cdr big5))
- codepoint (euc-to-flat-code (cdr char))
- charset (car char))
- (while (>= end i)
- (aset chartable
- (decode-big5-char (flat-code-to-big5 i))
- (apply (function make-char)
- charset
- (flat-code-to-euc codepoint)))
- (setq i (1+ i)
- codepoint (1+ codepoint)))
- )
- ((and (char-valid-p char)
- (numberp big5))
- (setq i (decode-big5-char big5))
- (aset chartable i char)
- )
- (t
- (error "Unknown slot type: %S" elem)
- )
- )
- )
- ;; the return value
- chartable
- )
-)
-
-;; All non-CNS encodings are commented out.
-
-(define-translation-table 'big5-to-cns
- (expand-euc-big5-alist
- '(
- ;; Symbols
- ((chinese-cns11643-1 . #x2121) . (#xA140 . #xA1F5))
- (?\e$(G"X\e(B . #xA1F6)
- (?\e$(G"W\e(B . #xA1F7)
- ((chinese-cns11643-1 . #x2259) . (#xA1F8 . #xA2AE))
- ((chinese-cns11643-1 . #x2421) . (#xA2AF . #xA3BF))
- ;; Control codes (vendor dependent)
- ((chinese-cns11643-1 . #x4221) . (#xA3C0 . #xA3E0))
- ;; Level 1 Ideographs
- ((chinese-cns11643-1 . #x4421) . (#xA440 . #xACFD))
- (?\e$(GWS\e(B . #xACFE)
- ((chinese-cns11643-1 . #x5323) . (#xAD40 . #xAFCF))
- ((chinese-cns11643-1 . #x5754) . (#xAFD0 . #xBBC7))
- ((chinese-cns11643-1 . #x6B51) . (#xBBC8 . #xBE51))
- (?\e$(GkP\e(B . #xBE52)
- ((chinese-cns11643-1 . #x6F5C) . (#xBE53 . #xC1AA))
- ((chinese-cns11643-1 . #x7536) . (#xC1AB . #xC2CA))
- (?\e$(Gu5\e(B . #xC2CB)
- ((chinese-cns11643-1 . #x7737) . (#xC2CC . #xC360))
- ((chinese-cns11643-1 . #x782E) . (#xC361 . #xC3B8))
- (?\e$(Gxe\e(B . #xC3B9)
- (?\e$(Gxd\e(B . #xC3BA)
- ((chinese-cns11643-1 . #x7866) . (#xC3BB . #xC455))
- (?\e$(Gx-\e(B . #xC456)
- ((chinese-cns11643-1 . #x7962) . (#xC457 . #xC67E))
- ;; Symbols
- ((chinese-cns11643-1 . #x2621) . (#xC6A1 . #xC6BE))
- ;; Radicals
- (?\e$(G'#\e(B . #xC6BF)
- (?\e$(G'$\e(B . #xC6C0)
- (?\e$(G'&\e(B . #xC6C1)
- (?\e$(G'(\e(B . #xC6C2)
- (?\e$(G'-\e(B . #xC6C3)
- (?\e$(G'.\e(B . #xC6C4)
- (?\e$(G'/\e(B . #xC6C5)
- (?\e$(G'4\e(B . #xC6C6)
- (?\e$(G'7\e(B . #xC6C7)
- (?\e$(G':\e(B . #xC6C8)
- (?\e$(G'<\e(B . #xC6C9)
- (?\e$(G'B\e(B . #xC6CA)
- (?\e$(G'G\e(B . #xC6CB)
- (?\e$(G'N\e(B . #xC6CC)
- (?\e$(G'S\e(B . #xC6CD)
- (?\e$(G'T\e(B . #xC6CE)
- (?\e$(G'U\e(B . #xC6CF)
- (?\e$(G'Y\e(B . #xC6D0)
- (?\e$(G'Z\e(B . #xC6D1)
- (?\e$(G'a\e(B . #xC6D2)
- (?\e$(G'f\e(B . #xC6D3)
- (?\e$(G()\e(B . #xC6D4)
- (?\e$(G(*\e(B . #xC6D5)
- (?\e$(G(c\e(B . #xC6D6)
- (?\e$(G(l\e(B . #xC6D7)
- ;; Diacritical Marks
- ; ((japanese-jisx0208 . #x212F) . (#xC6D8 . #xC6D9))
- ;; Japanese Kana Supplement
- ; ((japanese-jisx0208 . #x2133) . (#xC6DA . #xC6E3))
- ;; Japanese Hiragana
- ; ((japanese-jisx0208 . #x2421) . (#xC6E7 . #xC77A))
- ;; Japanese Katakana
- ; ((japanese-jisx0208 . #x2521) . (#xC77B . #xC7F2))
- ;; Cyrillic Characters
- ; ((japanese-jisx0208 . #x2721) . (#xC7F3 . #xC854))
- ; ((japanese-jisx0208 . #x2751) . (#xC855 . #xC875))
- ;; Special Chinese Characters
- (?\e$(J!#\e(B . #xC879)
- (?\e$(J!$\e(B . #xC87B)
- (?\e$(J!*\e(B . #xC87D)
- (?\e$(J!R\e(B . #xC8A2)
-
- ;; JIS X 0208 NOT SIGN (cf. U+00AC)
- ; (?\e$B"L\e(B . #xC8CD)
- ;; JIS X 0212 BROKEN BAR (cf. U+00A6)
- ; (?\e$(D"C\e(B . #xC8CE)
-
- ;; GB 2312 characters
- ; (?\e$A!d\e(B . #xC8CF)
- ; (?\e$A!e\e(B . #xC8D0)
- ;;;;; C8D1 - Japanese `(\e$B3t\e(B)'
- ; (?\e$A!m\e(B . #xC8D2)
- ;;;;; C8D2 - Tel.
-
- ;; Level 2 Ideographs
- ((chinese-cns11643-2 . #x2121) . (#xC940 . #xC949))
- (?\e$(GDB\e(B . #xC94A);; a duplicate of #xA461
- ((chinese-cns11643-2 . #x212B) . (#xC94B . #xC96B))
- ((chinese-cns11643-2 . #x214D) . (#xC96C . #xC9BD))
- (?\e$(H!L\e(B . #xC9BE)
- ((chinese-cns11643-2 . #x217D) . (#xC9BF . #xC9EC))
- ((chinese-cns11643-2 . #x224E) . (#xC9ED . #xCAF6))
- (?\e$(H"M\e(B . #xCAF7)
- ((chinese-cns11643-2 . #x2439) . (#xCAF8 . #xD6CB))
- (?\e$(H>c\e(B . #xD6CC)
- ((chinese-cns11643-2 . #x3770) . (#xD6CD . #xD779))
- (?\e$(H?j\e(B . #xD77A)
- ((chinese-cns11643-2 . #x387E) . (#xD77B . #xDADE))
- (?\e$(H7o\e(B . #xDADF)
- ((chinese-cns11643-2 . #x3E64) . (#xDAE0 . #xDBA6))
- ((chinese-cns11643-2 . #x3F6B) . (#xDBA7 . #xDDFB))
- (?\e$(HAv\e(B . #xDDFC);; a duplicate of #xDCD1
- ((chinese-cns11643-2 . #x4424) . (#xDDFD . #xE8A2))
- ((chinese-cns11643-2 . #x554C) . (#xE8A3 . #xE975))
- ((chinese-cns11643-2 . #x5723) . (#xE976 . #xEB5A))
- ((chinese-cns11643-2 . #x5A29) . (#xEB5B . #xEBF0))
- (?\e$(HUK\e(B . #xEBF1)
- ((chinese-cns11643-2 . #x5B3F) . (#xEBF2 . #xECDD))
- (?\e$(HW"\e(B . #xECDE)
- ((chinese-cns11643-2 . #x5C6A) . (#xECDF . #xEDA9))
- ((chinese-cns11643-2 . #x5D75) . (#xEDAA . #xEEEA))
- (?\e$(Hd/\e(B . #xEEEB)
- ((chinese-cns11643-2 . #x6039) . (#xEEEC . #xF055))
- (?\e$(H]t\e(B . #xF056)
- ((chinese-cns11643-2 . #x6243) . (#xF057 . #xF0CA))
- (?\e$(HZ(\e(B . #xF0CB)
- ((chinese-cns11643-2 . #x6337) . (#xF0CC . #xF162))
- ((chinese-cns11643-2 . #x6430) . (#xF163 . #xF16A))
- (?\e$(Hga\e(B . #xF16B)
- ((chinese-cns11643-2 . #x6438) . (#xF16C . #xF267))
- (?\e$(Hi4\e(B . #xF268)
- ((chinese-cns11643-2 . #x6573) . (#xF269 . #xF2C2))
- ((chinese-cns11643-2 . #x664E) . (#xF2C3 . #xF374))
- ((chinese-cns11643-2 . #x6762) . (#xF375 . #xF465))
- ((chinese-cns11643-2 . #x6935) . (#xF466 . #xF4B4))
- (?\e$(HfM\e(B . #xF4B5)
- ((chinese-cns11643-2 . #x6962) . (#xF4B6 . #xF4FC))
- ((chinese-cns11643-2 . #x6A4C) . (#xF4FD . #xF662))
- (?\e$(HjK\e(B . #xF663)
- ((chinese-cns11643-2 . #x6C52) . (#xF664 . #xF976))
- ((chinese-cns11643-2 . #x7167) . (#xF977 . #xF9C3))
- (?\e$(Hqf\e(B . #xF9C4)
- (?\e$(Hr4\e(B . #xF9C5)
- (?\e$(Hr@\e(B . #xF9C6)
- ((chinese-cns11643-2 . #x7235) . (#xF9C7 . #xF9D1))
- ((chinese-cns11643-2 . #x7241) . (#xF9D2 . #xF9D5))
-
- ;; Additional Ideographs
- (?\e$(IC7\e(B . #xF9D6)
- (?\e$(IOP\e(B . #xF9D7)
- (?\e$(IDN\e(B . #xF9D8)
- (?\e$(IPJ\e(B . #xF9D9)
- (?\e$(I,]\e(B . #xF9DA)
- (?\e$(I=~\e(B . #xF9DB)
- (?\e$(IK\\e(B . #xF9DC)
- )
- )
-)
+;;;###autoload
+(defun post-read-decode-hz (len)
+ (let ((pos (point))
+ (buffer-modified-p (buffer-modified-p))
+ last-coding-system-used)
+ (prog1
+ (decode-hz-region pos (+ pos len))
+ (set-buffer-modified-p buffer-modified-p))))
+;;;###autoload
+(defun pre-write-encode-hz (from to)
+ (let ((buf (current-buffer)))
+ (set-buffer (generate-new-buffer " *temp*"))
+ (if (stringp from)
+ (insert from)
+ (insert-buffer-substring buf from to))
+ (let (last-coding-system-used)
+ (encode-hz-region 1 (point-max)))
+ nil))
;;
(provide 'china-util)
projects / emacs.git / commitdiff