Mercurial > emacs

comparison lisp/international/code-pages.el @ 52434:14a91089126a

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
(cp-make-translation-table) (cp-valid-codes): Made defsubsts. (cp-fix-safe-chars): Deleted. (mule-diag): Don't require. (windows-1251, windows-1252): Removed to cyrillic.el/european.el. (top-level): Check for defined coding system when defining cp... aliases. Change w32-add-charset-info test to avoid warning. (non-iso-charset-alist): Defvar when compiling. (cp-make-coding-system): Doc fix.
author Dave Love <fx@gnu.org>
date Thu, 04 Sep 2003 16:59:00 +0000
parents 695cf19ef79e
children af41ec2032d0
comparison
equal deleted inserted replaced
52433:875975e6fae5 52434:14a91089126a
31 ;; codepage.el. The utility `cp-make-coding-system' derives them from 31 ;; codepage.el. The utility `cp-make-coding-system' derives them from
32 ;; simple tables. 32 ;; simple tables.
33 33
34 ;; Those covered are: cp437, cp737, cp720, cp775, cp850, cp851, cp852, 34 ;; Those covered are: cp437, cp737, cp720, cp775, cp850, cp851, cp852,
35 ;; cp855, cp857, cp860, cp861, cp862, cp863, cp864, cp865, cp866, 35 ;; cp855, cp857, cp860, cp861, cp862, cp863, cp864, cp865, cp866,
36 ;; cp869, cp874, cp1125, windows-1250, windows-1251, windows-1252, 36 ;; cp869, cp874, cp1125, windows-1250, windows-1253, windows-1254,
37 ;; windows-1253, windows-1254, windows-1255, windows-1256, 37 ;; windows-1255, windows-1256, windows-1257, windows-1258, next,
38 ;; windows-1257, windows-1258, next, koi8-u, iso-8859-6, 38 ;; iso-8859-6, iso-8859-10, iso-8859-11, iso-8859-16, koi8-t,
39 ;; iso-8859-10, iso-8859-11, iso-8859-16, koi8-t, georgian-ps. This 39 ;; georgian-ps. This is meant to include all the single-byte ones
40 ;; is meant to include all the single-byte ones relevant to GNU (used 40 ;; relevant to GNU (used in glibc-defined locales); we don't yet get
41 ;; in glibc-defined locales); we don't yet get all the multibyte ones 41 ;; all the multibyte ones in base Emacs.
42 ;; in base Emacs.
43 42
44 ;; Note that various of these can clash with definitions in 43 ;; Note that various of these can clash with definitions in
45 ;; codepage.el; we try to avoid damage from that. A few CPs from 44 ;; codepage.el; we try to avoid damage from that. A few CPs from
46 ;; codepage.el (770, 773, 774) aren't covered (in the absence of 45 ;; codepage.el (770, 773, 774) aren't covered (in the absence of
47 ;; translation tables to Unicode). 46 ;; translation tables to Unicode).
52 ;; splitting the file and autoloading the coding systems if/when my 51 ;; splitting the file and autoloading the coding systems if/when my
53 ;; (or similar) autoloading code is installed. 52 ;; (or similar) autoloading code is installed.
54 53
55 ;;; Code: 54 ;;; Code:
56 55
57 (defun cp-make-translation-table (v) 56 ;; The defsubsts here are just so that language files can use
57 ;; `cp-make-coding-system' and not require functions from this file
58 ;; at runtime.
59
60 (defsubst cp-make-translation-table (v)
58 "Return a translation table made from 128-long vector V. 61 "Return a translation table made from 128-long vector V.
59 V comprises characters encodable by mule-utf-8." 62 V comprises characters encodable by mule-utf-8."
60 (let ((encoding-vector (make-vector 256 0))) 63 (let ((encoding-vector (make-vector 256 0)))
61 (dotimes (i 128) 64 (dotimes (i 128)
62 (aset encoding-vector i i)) 65 (aset encoding-vector i i))
73 (if c1 ; we encode that unicode 76 (if c1 ; we encode that unicode
74 (aset encode-table c c1))))) 77 (aset encode-table c c1)))))
75 ucs-mule-to-mule-unicode) 78 ucs-mule-to-mule-unicode)
76 tab))) 79 tab)))
77 80
78 (defun cp-valid-codes (v) 81 (defsubst cp-valid-codes (v)
79 "Derive a valid-codes list for translation vector V. 82 "Derive a valid-codes list for translation vector V.
80 See `make-coding-system'." 83 See `make-coding-system'."
81 (let (pairs 84 (let (pairs
82 (i 128) ; index into v 85 (i 128) ; index into v
83 (start 0) ; start of a valid range 86 (start 0) ; start of a valid range
92 (setq start nil)) 95 (setq start nil))
93 (setq i (1+ i))) 96 (setq i (1+ i)))
94 (if start (push (cons start end) pairs)) 97 (if start (push (cons start end) pairs))
95 (nreverse pairs))) 98 (nreverse pairs)))
96 99
97 (defun cp-fix-safe-chars (cs)
98 "This is an obsolete function.
99 It exists just for backward compatibility, and it does nothing.")
100 (make-obsolete 'cp-fix-safe-chars
101 "Unnecessary function. Calling it has no effect."
102 "21.3")
103
104 ;; Fix things that have been, or might be, done by codepage.el. 100 ;; Fix things that have been, or might be, done by codepage.el.
105 (eval-after-load "codepage" 101 (eval-after-load "codepage"
106 '(progn 102 '(progn
107 103
108 ;; Semi-dummy version for the stuff in codepage.el which we don't 104 ;; Semi-dummy version for the stuff in codepage.el which we don't
143 ;; loading ccl. 139 ;; loading ccl.
144 ;;;###autoload 140 ;;;###autoload
145 (defmacro cp-make-coding-system (name v &optional doc-string mnemonic) 141 (defmacro cp-make-coding-system (name v &optional doc-string mnemonic)
146 "Make coding system NAME for and 8-bit, extended-ASCII character set. 142 "Make coding system NAME for and 8-bit, extended-ASCII character set.
147 V is a 128-long vector of characters to translate the upper half of 143 V is a 128-long vector of characters to translate the upper half of
148 the charactert set. DOC-STRING and MNEMONIC are used as the 144 the character set. DOC-STRING and MNEMONIC are used as the
149 corresponding args of `make-coding-system'. If MNEMONIC isn't given, 145 corresponding args of `make-coding-system'. If MNEMONIC isn't given,
150 ?* is used." 146 ?* is used.
147 Return an updated `non-iso-charset-alist'."
151 (let* ((encoder (intern (format "encode-%s" name))) 148 (let* ((encoder (intern (format "encode-%s" name)))
152 (decoder (intern (format "decode-%s" name))) 149 (decoder (intern (format "decode-%s" name)))
153 (ccl-decoder 150 (ccl-decoder
154 (ccl-compile 151 (ccl-compile
155 `(4 152 `(4
201 (push (cdr elt) l) 198 (push (cdr elt) l)
202 (push (car elt) l)) 199 (push (car elt) l))
203 (list l))) 200 (list l)))
204 non-iso-charset-alist)))) 201 non-iso-charset-alist))))
205 202
203 (eval-when-compile (defvar non-iso-charset-alist))
206 204
207 ;; These tables were mostly derived by running somthing like 205 ;; These tables were mostly derived by running somthing like
208 ;; `recode -f cpxxx/..utf-8' on a binary file filled by 206 ;; `recode -f cpxxx/..utf-8' on a binary file filled by
209 ;; `(dotimes (i 128) (insert ?? ?\\ (+ 128 i) ?\n))' and then 207 ;; `(dotimes (i 128) (insert ?? ?\\ (+ 128 i) ?\n))' and then
210 ;; exchanging the ?\� entries for nil. iconv was used instead in some 208 ;; exchanging the ?\� entries for nil. iconv was used instead in some
2570 ?\ű 2568 ?\ű
2571 ?\ü 2569 ?\ü
2572 ?\ý 2570 ?\ý
2573 ?\ţ 2571 ?\ţ
2574 ?\˙]) 2572 ?\˙])
2575
2576 ;; be_BY, bg_BG
2577 (cp-make-coding-system
2578 windows-1251
2579 [?\Ђ
2580 ?\Ѓ
2581 ?\‚
2582 ?\ѓ
2583 ?\„
2584 ?\…
2585 ?\†
2586 ?\‡
2587 ?\€
2588 ?\‰
2589 ?\Љ
2590 ?\‹
2591 ?\Њ
2592 ?\Ќ
2593 ?\Ћ
2594 ?\Џ
2595 ?\ђ
2596 ?\‘
2597 ?\’
2598 ?\“
2599 ?\”
2600 ?\•
2601 ?\–
2602 ?\—
2603 nil
2604 ?\™
2605 ?\љ
2606 ?\›
2607 ?\њ
2608 ?\ќ
2609 ?\ћ
2610 ?\џ
2611 ?\ 
2612 ?\Ў
2613 ?\ў
2614 ?\Ј
2615 ?\¤
2616 ?\Ґ
2617 ?\¦
2618 ?\§
2619 ?\Ё
2620 ?\©
2621 ?\Є
2622 ?\«
2623 ?\¬
2624 ?\­
2625 ?\®
2626 ?\Ї
2627 ?\°
2628 ?\±
2629 ?\І
2630 ?\і
2631 ?\ґ
2632 ?\µ
2633 ?\¶
2634 ?\·
2635 ?\ё
2636 ?\№
2637 ?\є
2638 ?\»
2639 ?\ј
2640 ?\Ѕ
2641 ?\ѕ
2642 ?\ї
2643 ?\А
2644 ?\Б
2645 ?\В
2646 ?\Г
2647 ?\Д
2648 ?\Е
2649 ?\Ж
2650 ?\З
2651 ?\И
2652 ?\Й
2653 ?\К
2654 ?\Л
2655 ?\М
2656 ?\Н
2657 ?\О
2658 ?\П
2659 ?\Р
2660 ?\С
2661 ?\Т
2662 ?\У
2663 ?\Ф
2664 ?\Х
2665 ?\Ц
2666 ?\Ч
2667 ?\Ш
2668 ?\Щ
2669 ?\Ъ
2670 ?\Ы
2671 ?\Ь
2672 ?\Э
2673 ?\Ю
2674 ?\Я
2675 ?\а
2676 ?\б
2677 ?\в
2678 ?\г
2679 ?\д
2680 ?\е
2681 ?\ж
2682 ?\з
2683 ?\и
2684 ?\й
2685 ?\к
2686 ?\л
2687 ?\м
2688 ?\н
2689 ?\о
2690 ?\п
2691 ?\р
2692 ?\с
2693 ?\т
2694 ?\у
2695 ?\ф
2696 ?\х
2697 ?\ц
2698 ?\ч
2699 ?\ш
2700 ?\щ
2701 ?\ъ
2702 ?\ы
2703 ?\ь
2704 ?\э
2705 ?\ю
2706 ?\я]
2707 nil ?b)
2708
2709 (cp-make-coding-system
2710 windows-1252
2711 [?\€
2712 nil
2713 ?\‚
2714 ?\ƒ
2715 ?\„
2716 ?\…
2717 ?\†
2718 ?\‡
2719 ?\ˆ
2720 ?\‰
2721 ?\Š
2722 ?\‹
2723 ?\Œ
2724 nil
2725 ?\Ž
2726 ?\ž
2727 nil
2728 ?\‘
2729 ?\’
2730 ?\“
2731 ?\”
2732 ?\•
2733 ?\–
2734 ?\—
2735 ?\˜
2736 ?\™
2737 ?\š
2738 ?\›
2739 ?\œ
2740 nil
2741 nil
2742 ?\Ÿ
2743 ?\ 
2744 ?\¡
2745 ?\¢
2746 ?\£
2747 ?\¤
2748 ?\¥
2749 ?\¦
2750 ?\§
2751 ?\¨
2752 ?\©
2753 ?\ª
2754 ?\«
2755 ?\¬
2756 ?\­
2757 ?\®
2758 ?\¯
2759 ?\°
2760 ?\±
2761 ?\²
2762 ?\³
2763 ?\´
2764 ?\µ
2765 ?\¶
2766 ?\·
2767 ?\¸
2768 ?\¹
2769 ?\º
2770 ?\»
2771 ?\¼
2772 ?\½
2773 ?\¾
2774 ?\¿
2775 ?\À
2776 ?\Á
2777 ?\Â
2778 ?\Ã
2779 ?\Ä
2780 ?\Å
2781 ?\Æ
2782 ?\Ç
2783 ?\È
2784 ?\É
2785 ?\Ê
2786 ?\Ë
2787 ?\Ì
2788 ?\Í
2789 ?\Î
2790 ?\Ï
2791 ?\Ð
2792 ?\Ñ
2793 ?\Ò
2794 ?\Ó
2795 ?\Ô
2796 ?\Õ
2797 ?\Ö
2798 ?\×
2799 ?\Ø
2800 ?\Ù
2801 ?\Ú
2802 ?\Û
2803 ?\Ü
2804 ?\Ý
2805 ?\Þ
2806 ?\ß
2807 ?\à
2808 ?\á
2809 ?\â
2810 ?\ã
2811 ?\ä
2812 ?\å
2813 ?\æ
2814 ?\ç
2815 ?\è
2816 ?\é
2817 ?\ê
2818 ?\ë
2819 ?\ì
2820 ?\í
2821 ?\î
2822 ?\ï
2823 ?\ð
2824 ?\ñ
2825 ?\ò
2826 ?\ó
2827 ?\ô
2828 ?\õ
2829 ?\ö
2830 ?\÷
2831 ?\ø
2832 ?\ù
2833 ?\ú
2834 ?\û
2835 ?\ü
2836 ?\ý
2837 ?\þ
2838 ?\ÿ])
2839 2573
2840 (cp-make-coding-system 2574 (cp-make-coding-system
2841 windows-1253 2575 windows-1253
2842 [?\€ 2576 [?\€
2843 nil 2577 nil
4747 "ISO-8859-11. This is `thai-tis620' with the addition of no-break-space.") 4481 "ISO-8859-11. This is `thai-tis620' with the addition of no-break-space.")
4748 4482
4749 (dotimes (i 8) 4483 (dotimes (i 8)
4750 (let ((w (intern (format "windows-125%d" i))) 4484 (let ((w (intern (format "windows-125%d" i)))
4751 (c (intern (format "cp125%d" i)))) 4485 (c (intern (format "cp125%d" i))))
4752 (define-coding-system-alias c w) 4486 (if (coding-system-p c) ; 1251 is in cyrillic.el
4487 (define-coding-system-alias c w))
4753 ;; Compatibility with codepage.el, though cp... are not the 4488 ;; Compatibility with codepage.el, though cp... are not the
4754 ;; canonical names. 4489 ;; canonical names.
4755 (push (assoc w non-iso-charset-alist) non-iso-charset-alist))) 4490 (push (assoc w non-iso-charset-alist) non-iso-charset-alist)))
4756 4491
4757 ;; Use Unicode font under Windows. Jason Rumney fecit. 4492 ;; Use Unicode font under Windows. Jason Rumney fecit.
4758 (if (and (fboundp 'w32-add-charset-info) 4493 (if (fboundp 'w32-add-charset-info)
4759 (not (boundp 'w32-unicode-charset-defined))) 4494 (unless (boundp 'w32-unicode-charset-defined)
4760 (w32-add-charset-info "iso10646-1" 'w32-charset-ansi t)) 4495 (w32-add-charset-info "iso10646-1" 'w32-charset-ansi t)))
4761 4496
4762 (provide 'code-pages) 4497 (provide 'code-pages)
4763 4498
4764 ;;; arch-tag: 8b6e3c73-b271-4198-866d-ea6d0ceff1b2 4499 ;;; arch-tag: 8b6e3c73-b271-4198-866d-ea6d0ceff1b2
4765 ;;; code-pages.el ends here 4500 ;;; code-pages.el ends here