Deleted Added
full compact
charset.c (238730) charset.c (293190)
1/*
1/*
2 * Copyright (C) 1984-2012 Mark Nudelman
2 * Copyright (C) 1984-2015 Mark Nudelman
3 *
4 * You may distribute under the terms of either the GNU General Public
5 * License or the Less License, as specified in the README file.
6 *
7 * For more information, see the README file.
8 */
9
10

--- 450 unchanged lines hidden (view full) ---

461 strcpy(buf, "ESC");
462 else if (ch < 128 && control_char(ch))
463 {
464 if (!control_char(ch ^ 0100))
465 SNPRINTF1(buf, sizeof(buf), "^%c", ((char) ch) ^ 0100);
466 else
467 SNPRINTF1(buf, sizeof(buf), binfmt, (char) ch);
468 } else if (is_ubin_char(ch))
3 *
4 * You may distribute under the terms of either the GNU General Public
5 * License or the Less License, as specified in the README file.
6 *
7 * For more information, see the README file.
8 */
9
10

--- 450 unchanged lines hidden (view full) ---

461 strcpy(buf, "ESC");
462 else if (ch < 128 && control_char(ch))
463 {
464 if (!control_char(ch ^ 0100))
465 SNPRINTF1(buf, sizeof(buf), "^%c", ((char) ch) ^ 0100);
466 else
467 SNPRINTF1(buf, sizeof(buf), binfmt, (char) ch);
468 } else if (is_ubin_char(ch))
469 {
469 SNPRINTF1(buf, sizeof(buf), utfbinfmt, ch);
470 SNPRINTF1(buf, sizeof(buf), utfbinfmt, ch);
470 else
471 } else
471 {
472 {
472 int len;
473 char *p = buf;
473 if (ch >= 0x80000000)
474 if (ch >= 0x80000000)
474 {
475 len = 3;
476 ch = 0xFFFD;
477 } else
478 {
479 len = (ch < 0x80) ? 1
480 : (ch < 0x800) ? 2
481 : (ch < 0x10000) ? 3
482 : (ch < 0x200000) ? 4
483 : (ch < 0x4000000) ? 5
484 : 6;
485 }
486 buf[len] = '\0';
487 if (len == 1)
488 *buf = (char) ch;
489 else
490 {
491 *buf = ((1 << len) - 1) << (8 - len);
492 while (--len > 0)
493 {
494 buf[len] = (char) (0x80 | (ch & 0x3F));
495 ch >>= 6;
496 }
497 *buf |= ch;
498 }
475 ch = 0xFFFD; /* REPLACEMENT CHARACTER */
476 put_wchar(&p, ch);
477 *p = '\0';
499 }
500 return (buf);
501}
502
503/*
504 * Get the length of a UTF-8 character in bytes.
505 */
506 public int

--- 12 unchanged lines hidden (view full) ---

519 return 5;
520 if ((ch & 0xFE) == 0xFC)
521 return 6;
522 /* Invalid UTF-8 encoding. */
523 return 1;
524}
525
526/*
478 }
479 return (buf);
480}
481
482/*
483 * Get the length of a UTF-8 character in bytes.
484 */
485 public int

--- 12 unchanged lines hidden (view full) ---

498 return 5;
499 if ((ch & 0xFE) == 0xFC)
500 return 6;
501 /* Invalid UTF-8 encoding. */
502 return 1;
503}
504
505/*
527 * Is a UTF-8 character well-formed?
506 * Does the parameter point to the lead byte of a well-formed UTF-8 character?
528 */
529 public int
507 */
508 public int
530is_utf8_well_formed(s)
509is_utf8_well_formed(s, slen)
531 unsigned char *s;
510 unsigned char *s;
511 int slen;
532{
533 int i;
534 int len;
535
536 if (IS_UTF8_INVALID(s[0]))
537 return (0);
538
539 len = utf_len((char) s[0]);
512{
513 int i;
514 int len;
515
516 if (IS_UTF8_INVALID(s[0]))
517 return (0);
518
519 len = utf_len((char) s[0]);
520 if (len > slen)
521 return (0);
540 if (len == 1)
541 return (1);
542 if (len == 2)
543 {
544 if (s[0] < 0xC2)
545 return (0);
546 } else
547 {

--- 5 unchanged lines hidden (view full) ---

553
554 for (i = 1; i < len; i++)
555 if (!IS_UTF8_TRAIL(s[i]))
556 return (0);
557 return (1);
558}
559
560/*
522 if (len == 1)
523 return (1);
524 if (len == 2)
525 {
526 if (s[0] < 0xC2)
527 return (0);
528 } else
529 {

--- 5 unchanged lines hidden (view full) ---

535
536 for (i = 1; i < len; i++)
537 if (!IS_UTF8_TRAIL(s[i]))
538 return (0);
539 return (1);
540}
541
542/*
543 * Return number of invalid UTF-8 sequences found in a buffer.
544 */
545 public int
546utf_bin_count(data, len)
547 unsigned char *data;
548 int len;
549{
550 int bin_count = 0;
551 while (len > 0)
552 {
553 if (is_utf8_well_formed(data, len))
554 {
555 int clen = utf_len(*data);
556 data += clen;
557 len -= clen;
558 } else
559 {
560 /* Skip to next lead byte. */
561 bin_count++;
562 do {
563 ++data;
564 --len;
565 } while (len > 0 && !IS_UTF8_LEAD(*data));
566 }
567 }
568 return (bin_count);
569}
570
571/*
561 * Get the value of a UTF-8 character.
562 */
563 public LWCHAR
564get_wchar(p)
565 char *p;
566{
567 switch (utf_len(p[0]))
568 {

--- 132 unchanged lines hidden (view full) ---

701 ch = 0;
702 }
703 *pp = p;
704 return ch;
705}
706
707/*
708 * Unicode characters data
572 * Get the value of a UTF-8 character.
573 */
574 public LWCHAR
575get_wchar(p)
576 char *p;
577{
578 switch (utf_len(p[0]))
579 {

--- 132 unchanged lines hidden (view full) ---

712 ch = 0;
713 }
714 *pp = p;
715 return ch;
716}
717
718/*
719 * Unicode characters data
720 * Actual data is in the generated *.uni files.
709 */
721 */
710struct wchar_range { LWCHAR first, last; };
711
722
712/*
713 * Characters with general category values
714 * Mn: Mark, Nonspacing
715 * Me: Mark, Enclosing
716 * Last synched with
717 * <http://www.unicode.org/Public/5.0.0/ucd/UnicodeData-5.0.0d7.txt>
718 * dated 2005-11-30T00:58:48Z
719 */
720static struct wchar_range comp_table[] = {
721 { 0x0300, 0x036F} /* Mn */, { 0x0483, 0x0486} /* Mn */,
722 { 0x0488, 0x0489} /* Me */,
723 { 0x0591, 0x05BD} /* Mn */, { 0x05BF, 0x05BF} /* Mn */,
724 { 0x05C1, 0x05C2} /* Mn */, { 0x05C4, 0x05C5} /* Mn */,
725 { 0x05C7, 0x05C7} /* Mn */, { 0x0610, 0x0615} /* Mn */,
726 { 0x064B, 0x065E} /* Mn */, { 0x0670, 0x0670} /* Mn */,
727 { 0x06D6, 0x06DC} /* Mn */,
728 { 0x06DE, 0x06DE} /* Me */,
729 { 0x06DF, 0x06E4} /* Mn */, { 0x06E7, 0x06E8} /* Mn */,
730 { 0x06EA, 0x06ED} /* Mn */, { 0x0711, 0x0711} /* Mn */,
731 { 0x0730, 0x074A} /* Mn */, { 0x07A6, 0x07B0} /* Mn */,
732 { 0x07EB, 0x07F3} /* Mn */, { 0x0901, 0x0902} /* Mn */,
733 { 0x093C, 0x093C} /* Mn */, { 0x0941, 0x0948} /* Mn */,
734 { 0x094D, 0x094D} /* Mn */, { 0x0951, 0x0954} /* Mn */,
735 { 0x0962, 0x0963} /* Mn */, { 0x0981, 0x0981} /* Mn */,
736 { 0x09BC, 0x09BC} /* Mn */, { 0x09C1, 0x09C4} /* Mn */,
737 { 0x09CD, 0x09CD} /* Mn */, { 0x09E2, 0x09E3} /* Mn */,
738 { 0x0A01, 0x0A02} /* Mn */, { 0x0A3C, 0x0A3C} /* Mn */,
739 { 0x0A41, 0x0A42} /* Mn */, { 0x0A47, 0x0A48} /* Mn */,
740 { 0x0A4B, 0x0A4D} /* Mn */, { 0x0A70, 0x0A71} /* Mn */,
741 { 0x0A81, 0x0A82} /* Mn */, { 0x0ABC, 0x0ABC} /* Mn */,
742 { 0x0AC1, 0x0AC5} /* Mn */, { 0x0AC7, 0x0AC8} /* Mn */,
743 { 0x0ACD, 0x0ACD} /* Mn */, { 0x0AE2, 0x0AE3} /* Mn */,
744 { 0x0B01, 0x0B01} /* Mn */, { 0x0B3C, 0x0B3C} /* Mn */,
745 { 0x0B3F, 0x0B3F} /* Mn */, { 0x0B41, 0x0B43} /* Mn */,
746 { 0x0B4D, 0x0B4D} /* Mn */, { 0x0B56, 0x0B56} /* Mn */,
747 { 0x0B82, 0x0B82} /* Mn */, { 0x0BC0, 0x0BC0} /* Mn */,
748 { 0x0BCD, 0x0BCD} /* Mn */, { 0x0C3E, 0x0C40} /* Mn */,
749 { 0x0C46, 0x0C48} /* Mn */, { 0x0C4A, 0x0C4D} /* Mn */,
750 { 0x0C55, 0x0C56} /* Mn */, { 0x0CBC, 0x0CBC} /* Mn */,
751 { 0x0CBF, 0x0CBF} /* Mn */, { 0x0CC6, 0x0CC6} /* Mn */,
752 { 0x0CCC, 0x0CCD} /* Mn */, { 0x0CE2, 0x0CE3} /* Mn */,
753 { 0x0D41, 0x0D43} /* Mn */, { 0x0D4D, 0x0D4D} /* Mn */,
754 { 0x0DCA, 0x0DCA} /* Mn */, { 0x0DD2, 0x0DD4} /* Mn */,
755 { 0x0DD6, 0x0DD6} /* Mn */, { 0x0E31, 0x0E31} /* Mn */,
756 { 0x0E34, 0x0E3A} /* Mn */, { 0x0E47, 0x0E4E} /* Mn */,
757 { 0x0EB1, 0x0EB1} /* Mn */, { 0x0EB4, 0x0EB9} /* Mn */,
758 { 0x0EBB, 0x0EBC} /* Mn */, { 0x0EC8, 0x0ECD} /* Mn */,
759 { 0x0F18, 0x0F19} /* Mn */, { 0x0F35, 0x0F35} /* Mn */,
760 { 0x0F37, 0x0F37} /* Mn */, { 0x0F39, 0x0F39} /* Mn */,
761 { 0x0F71, 0x0F7E} /* Mn */, { 0x0F80, 0x0F84} /* Mn */,
762 { 0x0F86, 0x0F87} /* Mn */, { 0x0F90, 0x0F97} /* Mn */,
763 { 0x0F99, 0x0FBC} /* Mn */, { 0x0FC6, 0x0FC6} /* Mn */,
764 { 0x102D, 0x1030} /* Mn */, { 0x1032, 0x1032} /* Mn */,
765 { 0x1036, 0x1037} /* Mn */, { 0x1039, 0x1039} /* Mn */,
766 { 0x1058, 0x1059} /* Mn */, { 0x135F, 0x135F} /* Mn */,
767 { 0x1712, 0x1714} /* Mn */, { 0x1732, 0x1734} /* Mn */,
768 { 0x1752, 0x1753} /* Mn */, { 0x1772, 0x1773} /* Mn */,
769 { 0x17B7, 0x17BD} /* Mn */, { 0x17C6, 0x17C6} /* Mn */,
770 { 0x17C9, 0x17D3} /* Mn */, { 0x17DD, 0x17DD} /* Mn */,
771 { 0x180B, 0x180D} /* Mn */, { 0x18A9, 0x18A9} /* Mn */,
772 { 0x1920, 0x1922} /* Mn */, { 0x1927, 0x1928} /* Mn */,
773 { 0x1932, 0x1932} /* Mn */, { 0x1939, 0x193B} /* Mn */,
774 { 0x1A17, 0x1A18} /* Mn */, { 0x1B00, 0x1B03} /* Mn */,
775 { 0x1B34, 0x1B34} /* Mn */, { 0x1B36, 0x1B3A} /* Mn */,
776 { 0x1B3C, 0x1B3C} /* Mn */, { 0x1B42, 0x1B42} /* Mn */,
777 { 0x1B6B, 0x1B73} /* Mn */, { 0x1DC0, 0x1DCA} /* Mn */,
778 { 0x1DFE, 0x1DFF} /* Mn */, { 0x20D0, 0x20DC} /* Mn */,
779 { 0x20DD, 0x20E0} /* Me */,
780 { 0x20E1, 0x20E1} /* Mn */,
781 { 0x20E2, 0x20E4} /* Me */,
782 { 0x20E5, 0x20EF} /* Mn */, { 0x302A, 0x302F} /* Mn */,
783 { 0x3099, 0x309A} /* Mn */, { 0xA806, 0xA806} /* Mn */,
784 { 0xA80B, 0xA80B} /* Mn */, { 0xA825, 0xA826} /* Mn */,
785 { 0xFB1E, 0xFB1E} /* Mn */, { 0xFE00, 0xFE0F} /* Mn */,
786 { 0xFE20, 0xFE23} /* Mn */, { 0x10A01, 0x10A03} /* Mn */,
787 { 0x10A05, 0x10A06} /* Mn */, { 0x10A0C, 0x10A0F} /* Mn */,
788 { 0x10A38, 0x10A3A} /* Mn */, { 0x10A3F, 0x10A3F} /* Mn */,
789 { 0x1D167, 0x1D169} /* Mn */, { 0x1D17B, 0x1D182} /* Mn */,
790 { 0x1D185, 0x1D18B} /* Mn */, { 0x1D1AA, 0x1D1AD} /* Mn */,
791 { 0x1D242, 0x1D244} /* Mn */, { 0xE0100, 0xE01EF} /* Mn */,
792};
723#define DECLARE_RANGE_TABLE_START(name) \
724 static struct wchar_range name##_array[] = {
725#define DECLARE_RANGE_TABLE_END(name) \
726 }; struct wchar_range_table name##_table = { name##_array, sizeof(name##_array)/sizeof(*name##_array) };
793
727
794/*
795 * Special pairs, not ranges.
796 */
728DECLARE_RANGE_TABLE_START(compose)
729#include "compose.uni"
730DECLARE_RANGE_TABLE_END(compose)
731
732DECLARE_RANGE_TABLE_START(ubin)
733#include "ubin.uni"
734DECLARE_RANGE_TABLE_END(ubin)
735
736DECLARE_RANGE_TABLE_START(wide)
737#include "wide.uni"
738DECLARE_RANGE_TABLE_END(wide)
739
740/* comb_table is special pairs, not ranges. */
797static struct wchar_range comb_table[] = {
798 {0x0644,0x0622}, {0x0644,0x0623}, {0x0644,0x0625}, {0x0644,0x0627},
799};
800
741static struct wchar_range comb_table[] = {
742 {0x0644,0x0622}, {0x0644,0x0623}, {0x0644,0x0625}, {0x0644,0x0627},
743};
744
801/*
802 * Characters with general category values
803 * Cc: Other, Control
804 * Cf: Other, Format
805 * Cs: Other, Surrogate
806 * Co: Other, Private Use
807 * Cn: Other, Not Assigned
808 * Zl: Separator, Line
809 * Zp: Separator, Paragraph
810 * Last synched with
811 * <http://www.unicode.org/Public/5.0.0/ucd/UnicodeData-5.0.0d7.txt>
812 * dated 2005-11-30T00:58:48Z
813 */
814static struct wchar_range ubin_table[] = {
815 { 0x0000, 0x0007} /* Cc */,
816 { 0x000B, 0x000C} /* Cc */,
817 { 0x000E, 0x001A} /* Cc */,
818 { 0x001C, 0x001F} /* Cc */,
819 { 0x007F, 0x009F} /* Cc */,
820#if 0
821 { 0x00AD, 0x00AD} /* Cf */,
822#endif
823 { 0x0370, 0x0373} /* Cn */, { 0x0376, 0x0379} /* Cn */,
824 { 0x037F, 0x0383} /* Cn */, { 0x038B, 0x038B} /* Cn */,
825 { 0x038D, 0x038D} /* Cn */, { 0x03A2, 0x03A2} /* Cn */,
826 { 0x03CF, 0x03CF} /* Cn */, { 0x0487, 0x0487} /* Cn */,
827 { 0x0514, 0x0530} /* Cn */, { 0x0557, 0x0558} /* Cn */,
828 { 0x0560, 0x0560} /* Cn */, { 0x0588, 0x0588} /* Cn */,
829 { 0x058B, 0x0590} /* Cn */, { 0x05C8, 0x05CF} /* Cn */,
830 { 0x05EB, 0x05EF} /* Cn */, { 0x05F5, 0x05FF} /* Cn */,
831#if 0
832 { 0x0600, 0x0603} /* Cf */,
833#endif
834 { 0x0604, 0x060A} /* Cn */, { 0x0616, 0x061A} /* Cn */,
835 { 0x061C, 0x061D} /* Cn */, { 0x0620, 0x0620} /* Cn */,
836 { 0x063B, 0x063F} /* Cn */, { 0x065F, 0x065F} /* Cn */,
837#if 0
838 { 0x06DD, 0x06DD} /* Cf */,
839#endif
840 { 0x070E, 0x070E} /* Cn */,
841#if 0
842 { 0x070F, 0x070F} /* Cf */,
843#endif
844 { 0x074B, 0x074C} /* Cn */, { 0x076E, 0x077F} /* Cn */,
845 { 0x07B2, 0x07BF} /* Cn */, { 0x07FB, 0x0900} /* Cn */,
846 { 0x093A, 0x093B} /* Cn */, { 0x094E, 0x094F} /* Cn */,
847 { 0x0955, 0x0957} /* Cn */, { 0x0971, 0x097A} /* Cn */,
848 { 0x0980, 0x0980} /* Cn */, { 0x0984, 0x0984} /* Cn */,
849 { 0x098D, 0x098E} /* Cn */, { 0x0991, 0x0992} /* Cn */,
850 { 0x09A9, 0x09A9} /* Cn */, { 0x09B1, 0x09B1} /* Cn */,
851 { 0x09B3, 0x09B5} /* Cn */, { 0x09BA, 0x09BB} /* Cn */,
852 { 0x09C5, 0x09C6} /* Cn */, { 0x09C9, 0x09CA} /* Cn */,
853 { 0x09CF, 0x09D6} /* Cn */, { 0x09D8, 0x09DB} /* Cn */,
854 { 0x09DE, 0x09DE} /* Cn */, { 0x09E4, 0x09E5} /* Cn */,
855 { 0x09FB, 0x0A00} /* Cn */, { 0x0A04, 0x0A04} /* Cn */,
856 { 0x0A0B, 0x0A0E} /* Cn */, { 0x0A11, 0x0A12} /* Cn */,
857 { 0x0A29, 0x0A29} /* Cn */, { 0x0A31, 0x0A31} /* Cn */,
858 { 0x0A34, 0x0A34} /* Cn */, { 0x0A37, 0x0A37} /* Cn */,
859 { 0x0A3A, 0x0A3B} /* Cn */, { 0x0A3D, 0x0A3D} /* Cn */,
860 { 0x0A43, 0x0A46} /* Cn */, { 0x0A49, 0x0A4A} /* Cn */,
861 { 0x0A4E, 0x0A58} /* Cn */, { 0x0A5D, 0x0A5D} /* Cn */,
862 { 0x0A5F, 0x0A65} /* Cn */, { 0x0A75, 0x0A80} /* Cn */,
863 { 0x0A84, 0x0A84} /* Cn */, { 0x0A8E, 0x0A8E} /* Cn */,
864 { 0x0A92, 0x0A92} /* Cn */, { 0x0AA9, 0x0AA9} /* Cn */,
865 { 0x0AB1, 0x0AB1} /* Cn */, { 0x0AB4, 0x0AB4} /* Cn */,
866 { 0x0ABA, 0x0ABB} /* Cn */, { 0x0AC6, 0x0AC6} /* Cn */,
867 { 0x0ACA, 0x0ACA} /* Cn */, { 0x0ACE, 0x0ACF} /* Cn */,
868 { 0x0AD1, 0x0ADF} /* Cn */, { 0x0AE4, 0x0AE5} /* Cn */,
869 { 0x0AF0, 0x0AF0} /* Cn */, { 0x0AF2, 0x0B00} /* Cn */,
870 { 0x0B04, 0x0B04} /* Cn */, { 0x0B0D, 0x0B0E} /* Cn */,
871 { 0x0B11, 0x0B12} /* Cn */, { 0x0B29, 0x0B29} /* Cn */,
872 { 0x0B31, 0x0B31} /* Cn */, { 0x0B34, 0x0B34} /* Cn */,
873 { 0x0B3A, 0x0B3B} /* Cn */, { 0x0B44, 0x0B46} /* Cn */,
874 { 0x0B49, 0x0B4A} /* Cn */, { 0x0B4E, 0x0B55} /* Cn */,
875 { 0x0B58, 0x0B5B} /* Cn */, { 0x0B5E, 0x0B5E} /* Cn */,
876 { 0x0B62, 0x0B65} /* Cn */, { 0x0B72, 0x0B81} /* Cn */,
877 { 0x0B84, 0x0B84} /* Cn */, { 0x0B8B, 0x0B8D} /* Cn */,
878 { 0x0B91, 0x0B91} /* Cn */, { 0x0B96, 0x0B98} /* Cn */,
879 { 0x0B9B, 0x0B9B} /* Cn */, { 0x0B9D, 0x0B9D} /* Cn */,
880 { 0x0BA0, 0x0BA2} /* Cn */, { 0x0BA5, 0x0BA7} /* Cn */,
881 { 0x0BAB, 0x0BAD} /* Cn */, { 0x0BBA, 0x0BBD} /* Cn */,
882 { 0x0BC3, 0x0BC5} /* Cn */, { 0x0BC9, 0x0BC9} /* Cn */,
883 { 0x0BCE, 0x0BD6} /* Cn */, { 0x0BD8, 0x0BE5} /* Cn */,
884 { 0x0BFB, 0x0C00} /* Cn */, { 0x0C04, 0x0C04} /* Cn */,
885 { 0x0C0D, 0x0C0D} /* Cn */, { 0x0C11, 0x0C11} /* Cn */,
886 { 0x0C29, 0x0C29} /* Cn */, { 0x0C34, 0x0C34} /* Cn */,
887 { 0x0C3A, 0x0C3D} /* Cn */, { 0x0C45, 0x0C45} /* Cn */,
888 { 0x0C49, 0x0C49} /* Cn */, { 0x0C4E, 0x0C54} /* Cn */,
889 { 0x0C57, 0x0C5F} /* Cn */, { 0x0C62, 0x0C65} /* Cn */,
890 { 0x0C70, 0x0C81} /* Cn */, { 0x0C84, 0x0C84} /* Cn */,
891 { 0x0C8D, 0x0C8D} /* Cn */, { 0x0C91, 0x0C91} /* Cn */,
892 { 0x0CA9, 0x0CA9} /* Cn */, { 0x0CB4, 0x0CB4} /* Cn */,
893 { 0x0CBA, 0x0CBB} /* Cn */, { 0x0CC5, 0x0CC5} /* Cn */,
894 { 0x0CC9, 0x0CC9} /* Cn */, { 0x0CCE, 0x0CD4} /* Cn */,
895 { 0x0CD7, 0x0CDD} /* Cn */, { 0x0CDF, 0x0CDF} /* Cn */,
896 { 0x0CE4, 0x0CE5} /* Cn */, { 0x0CF0, 0x0CF0} /* Cn */,
897 { 0x0CF3, 0x0D01} /* Cn */, { 0x0D04, 0x0D04} /* Cn */,
898 { 0x0D0D, 0x0D0D} /* Cn */, { 0x0D11, 0x0D11} /* Cn */,
899 { 0x0D29, 0x0D29} /* Cn */, { 0x0D3A, 0x0D3D} /* Cn */,
900 { 0x0D44, 0x0D45} /* Cn */, { 0x0D49, 0x0D49} /* Cn */,
901 { 0x0D4E, 0x0D56} /* Cn */, { 0x0D58, 0x0D5F} /* Cn */,
902 { 0x0D62, 0x0D65} /* Cn */, { 0x0D70, 0x0D81} /* Cn */,
903 { 0x0D84, 0x0D84} /* Cn */, { 0x0D97, 0x0D99} /* Cn */,
904 { 0x0DB2, 0x0DB2} /* Cn */, { 0x0DBC, 0x0DBC} /* Cn */,
905 { 0x0DBE, 0x0DBF} /* Cn */, { 0x0DC7, 0x0DC9} /* Cn */,
906 { 0x0DCB, 0x0DCE} /* Cn */, { 0x0DD5, 0x0DD5} /* Cn */,
907 { 0x0DD7, 0x0DD7} /* Cn */, { 0x0DE0, 0x0DF1} /* Cn */,
908 { 0x0DF5, 0x0E00} /* Cn */, { 0x0E3B, 0x0E3E} /* Cn */,
909 { 0x0E5C, 0x0E80} /* Cn */, { 0x0E83, 0x0E83} /* Cn */,
910 { 0x0E85, 0x0E86} /* Cn */, { 0x0E89, 0x0E89} /* Cn */,
911 { 0x0E8B, 0x0E8C} /* Cn */, { 0x0E8E, 0x0E93} /* Cn */,
912 { 0x0E98, 0x0E98} /* Cn */, { 0x0EA0, 0x0EA0} /* Cn */,
913 { 0x0EA4, 0x0EA4} /* Cn */, { 0x0EA6, 0x0EA6} /* Cn */,
914 { 0x0EA8, 0x0EA9} /* Cn */, { 0x0EAC, 0x0EAC} /* Cn */,
915 { 0x0EBA, 0x0EBA} /* Cn */, { 0x0EBE, 0x0EBF} /* Cn */,
916 { 0x0EC5, 0x0EC5} /* Cn */, { 0x0EC7, 0x0EC7} /* Cn */,
917 { 0x0ECE, 0x0ECF} /* Cn */, { 0x0EDA, 0x0EDB} /* Cn */,
918 { 0x0EDE, 0x0EFF} /* Cn */, { 0x0F48, 0x0F48} /* Cn */,
919 { 0x0F6B, 0x0F70} /* Cn */, { 0x0F8C, 0x0F8F} /* Cn */,
920 { 0x0F98, 0x0F98} /* Cn */, { 0x0FBD, 0x0FBD} /* Cn */,
921 { 0x0FCD, 0x0FCE} /* Cn */, { 0x0FD2, 0x0FFF} /* Cn */,
922 { 0x1022, 0x1022} /* Cn */, { 0x1028, 0x1028} /* Cn */,
923 { 0x102B, 0x102B} /* Cn */, { 0x1033, 0x1035} /* Cn */,
924 { 0x103A, 0x103F} /* Cn */, { 0x105A, 0x109F} /* Cn */,
925 { 0x10C6, 0x10CF} /* Cn */, { 0x10FD, 0x10FF} /* Cn */,
926 { 0x115A, 0x115E} /* Cn */, { 0x11A3, 0x11A7} /* Cn */,
927 { 0x11FA, 0x11FF} /* Cn */, { 0x1249, 0x1249} /* Cn */,
928 { 0x124E, 0x124F} /* Cn */, { 0x1257, 0x1257} /* Cn */,
929 { 0x1259, 0x1259} /* Cn */, { 0x125E, 0x125F} /* Cn */,
930 { 0x1289, 0x1289} /* Cn */, { 0x128E, 0x128F} /* Cn */,
931 { 0x12B1, 0x12B1} /* Cn */, { 0x12B6, 0x12B7} /* Cn */,
932 { 0x12BF, 0x12BF} /* Cn */, { 0x12C1, 0x12C1} /* Cn */,
933 { 0x12C6, 0x12C7} /* Cn */, { 0x12D7, 0x12D7} /* Cn */,
934 { 0x1311, 0x1311} /* Cn */, { 0x1316, 0x1317} /* Cn */,
935 { 0x135B, 0x135E} /* Cn */, { 0x137D, 0x137F} /* Cn */,
936 { 0x139A, 0x139F} /* Cn */, { 0x13F5, 0x1400} /* Cn */,
937 { 0x1677, 0x167F} /* Cn */, { 0x169D, 0x169F} /* Cn */,
938 { 0x16F1, 0x16FF} /* Cn */, { 0x170D, 0x170D} /* Cn */,
939 { 0x1715, 0x171F} /* Cn */, { 0x1737, 0x173F} /* Cn */,
940 { 0x1754, 0x175F} /* Cn */, { 0x176D, 0x176D} /* Cn */,
941 { 0x1771, 0x1771} /* Cn */, { 0x1774, 0x177F} /* Cn */,
942#if 0
943 { 0x17B4, 0x17B5} /* Cf */,
944#endif
945 { 0x17DE, 0x17DF} /* Cn */, { 0x17EA, 0x17EF} /* Cn */,
946 { 0x17FA, 0x17FF} /* Cn */, { 0x180F, 0x180F} /* Cn */,
947 { 0x181A, 0x181F} /* Cn */, { 0x1878, 0x187F} /* Cn */,
948 { 0x18AA, 0x18FF} /* Cn */, { 0x191D, 0x191F} /* Cn */,
949 { 0x192C, 0x192F} /* Cn */, { 0x193C, 0x193F} /* Cn */,
950 { 0x1941, 0x1943} /* Cn */, { 0x196E, 0x196F} /* Cn */,
951 { 0x1975, 0x197F} /* Cn */, { 0x19AA, 0x19AF} /* Cn */,
952 { 0x19CA, 0x19CF} /* Cn */, { 0x19DA, 0x19DD} /* Cn */,
953 { 0x1A1C, 0x1A1D} /* Cn */, { 0x1A20, 0x1AFF} /* Cn */,
954 { 0x1B4C, 0x1B4F} /* Cn */, { 0x1B7D, 0x1CFF} /* Cn */,
955 { 0x1DCB, 0x1DFD} /* Cn */, { 0x1E9C, 0x1E9F} /* Cn */,
956 { 0x1EFA, 0x1EFF} /* Cn */, { 0x1F16, 0x1F17} /* Cn */,
957 { 0x1F1E, 0x1F1F} /* Cn */, { 0x1F46, 0x1F47} /* Cn */,
958 { 0x1F4E, 0x1F4F} /* Cn */, { 0x1F58, 0x1F58} /* Cn */,
959 { 0x1F5A, 0x1F5A} /* Cn */, { 0x1F5C, 0x1F5C} /* Cn */,
960 { 0x1F5E, 0x1F5E} /* Cn */, { 0x1F7E, 0x1F7F} /* Cn */,
961 { 0x1FB5, 0x1FB5} /* Cn */, { 0x1FC5, 0x1FC5} /* Cn */,
962 { 0x1FD4, 0x1FD5} /* Cn */, { 0x1FDC, 0x1FDC} /* Cn */,
963 { 0x1FF0, 0x1FF1} /* Cn */, { 0x1FF5, 0x1FF5} /* Cn */,
964 { 0x1FFF, 0x1FFF} /* Cn */,
965 { 0x200B, 0x200F} /* Cf */,
966 { 0x2028, 0x2028} /* Zl */,
967 { 0x2029, 0x2029} /* Zp */,
968 { 0x202A, 0x202E} /* Cf */,
969 { 0x2060, 0x2063} /* Cf */,
970 { 0x2064, 0x2069} /* Cn */,
971 { 0x206A, 0x206F} /* Cf */,
972 { 0x2072, 0x2073} /* Cn */, { 0x208F, 0x208F} /* Cn */,
973 { 0x2095, 0x209F} /* Cn */, { 0x20B6, 0x20CF} /* Cn */,
974 { 0x20F0, 0x20FF} /* Cn */, { 0x214F, 0x2152} /* Cn */,
975 { 0x2185, 0x218F} /* Cn */, { 0x23E8, 0x23FF} /* Cn */,
976 { 0x2427, 0x243F} /* Cn */, { 0x244B, 0x245F} /* Cn */,
977 { 0x269D, 0x269F} /* Cn */, { 0x26B3, 0x2700} /* Cn */,
978 { 0x2705, 0x2705} /* Cn */, { 0x270A, 0x270B} /* Cn */,
979 { 0x2728, 0x2728} /* Cn */, { 0x274C, 0x274C} /* Cn */,
980 { 0x274E, 0x274E} /* Cn */, { 0x2753, 0x2755} /* Cn */,
981 { 0x2757, 0x2757} /* Cn */, { 0x275F, 0x2760} /* Cn */,
982 { 0x2795, 0x2797} /* Cn */, { 0x27B0, 0x27B0} /* Cn */,
983 { 0x27BF, 0x27BF} /* Cn */, { 0x27CB, 0x27CF} /* Cn */,
984 { 0x27EC, 0x27EF} /* Cn */, { 0x2B1B, 0x2B1F} /* Cn */,
985 { 0x2B24, 0x2BFF} /* Cn */, { 0x2C2F, 0x2C2F} /* Cn */,
986 { 0x2C5F, 0x2C5F} /* Cn */, { 0x2C6D, 0x2C73} /* Cn */,
987 { 0x2C78, 0x2C7F} /* Cn */, { 0x2CEB, 0x2CF8} /* Cn */,
988 { 0x2D26, 0x2D2F} /* Cn */, { 0x2D66, 0x2D6E} /* Cn */,
989 { 0x2D70, 0x2D7F} /* Cn */, { 0x2D97, 0x2D9F} /* Cn */,
990 { 0x2DA7, 0x2DA7} /* Cn */, { 0x2DAF, 0x2DAF} /* Cn */,
991 { 0x2DB7, 0x2DB7} /* Cn */, { 0x2DBF, 0x2DBF} /* Cn */,
992 { 0x2DC7, 0x2DC7} /* Cn */, { 0x2DCF, 0x2DCF} /* Cn */,
993 { 0x2DD7, 0x2DD7} /* Cn */, { 0x2DDF, 0x2DFF} /* Cn */,
994 { 0x2E18, 0x2E1B} /* Cn */, { 0x2E1E, 0x2E7F} /* Cn */,
995 { 0x2E9A, 0x2E9A} /* Cn */, { 0x2EF4, 0x2EFF} /* Cn */,
996 { 0x2FD6, 0x2FEF} /* Cn */, { 0x2FFC, 0x2FFF} /* Cn */,
997 { 0x3040, 0x3040} /* Cn */, { 0x3097, 0x3098} /* Cn */,
998 { 0x3100, 0x3104} /* Cn */, { 0x312D, 0x3130} /* Cn */,
999 { 0x318F, 0x318F} /* Cn */, { 0x31B8, 0x31BF} /* Cn */,
1000 { 0x31D0, 0x31EF} /* Cn */, { 0x321F, 0x321F} /* Cn */,
1001 { 0x3244, 0x324F} /* Cn */, { 0x32FF, 0x32FF} /* Cn */,
1002 { 0x4DB6, 0x4DBF} /* Cn */, { 0x9FBC, 0x9FFF} /* Cn */,
1003 { 0xA48D, 0xA48F} /* Cn */, { 0xA4C7, 0xA6FF} /* Cn */,
1004 { 0xA71B, 0xA71F} /* Cn */, { 0xA722, 0xA7FF} /* Cn */,
1005 { 0xA82C, 0xA83F} /* Cn */, { 0xA878, 0xABFF} /* Cn */,
1006 { 0xD7A4, 0xD7FF} /* Cn */,
1007 { 0xD800, 0xDFFF} /* Cs */,
1008 { 0xE000, 0xF8FF} /* Co */,
1009 { 0xFA2E, 0xFA2F} /* Cn */, { 0xFA6B, 0xFA6F} /* Cn */,
1010 { 0xFADA, 0xFAFF} /* Cn */, { 0xFB07, 0xFB12} /* Cn */,
1011 { 0xFB18, 0xFB1C} /* Cn */, { 0xFB37, 0xFB37} /* Cn */,
1012 { 0xFB3D, 0xFB3D} /* Cn */, { 0xFB3F, 0xFB3F} /* Cn */,
1013 { 0xFB42, 0xFB42} /* Cn */, { 0xFB45, 0xFB45} /* Cn */,
1014 { 0xFBB2, 0xFBD2} /* Cn */, { 0xFD40, 0xFD4F} /* Cn */,
1015 { 0xFD90, 0xFD91} /* Cn */, { 0xFDC8, 0xFDEF} /* Cn */,
1016 { 0xFDFE, 0xFDFF} /* Cn */, { 0xFE1A, 0xFE1F} /* Cn */,
1017 { 0xFE24, 0xFE2F} /* Cn */, { 0xFE53, 0xFE53} /* Cn */,
1018 { 0xFE67, 0xFE67} /* Cn */, { 0xFE6C, 0xFE6F} /* Cn */,
1019 { 0xFE75, 0xFE75} /* Cn */, { 0xFEFD, 0xFEFE} /* Cn */,
1020 { 0xFEFF, 0xFEFF} /* Cf */,
1021 { 0xFF00, 0xFF00} /* Cn */, { 0xFFBF, 0xFFC1} /* Cn */,
1022 { 0xFFC8, 0xFFC9} /* Cn */, { 0xFFD0, 0xFFD1} /* Cn */,
1023 { 0xFFD8, 0xFFD9} /* Cn */, { 0xFFDD, 0xFFDF} /* Cn */,
1024 { 0xFFE7, 0xFFE7} /* Cn */, { 0xFFEF, 0xFFF8} /* Cn */,
1025 { 0xFFF9, 0xFFFB} /* Cf */,
1026 { 0xFFFE, 0xFFFF} /* Cn */, { 0x1000C, 0x1000C} /* Cn */,
1027 { 0x10027, 0x10027} /* Cn */, { 0x1003B, 0x1003B} /* Cn */,
1028 { 0x1003E, 0x1003E} /* Cn */, { 0x1004E, 0x1004F} /* Cn */,
1029 { 0x1005E, 0x1007F} /* Cn */, { 0x100FB, 0x100FF} /* Cn */,
1030 { 0x10103, 0x10106} /* Cn */, { 0x10134, 0x10136} /* Cn */,
1031 { 0x1018B, 0x102FF} /* Cn */, { 0x1031F, 0x1031F} /* Cn */,
1032 { 0x10324, 0x1032F} /* Cn */, { 0x1034B, 0x1037F} /* Cn */,
1033 { 0x1039E, 0x1039E} /* Cn */, { 0x103C4, 0x103C7} /* Cn */,
1034 { 0x103D6, 0x103FF} /* Cn */,
1035 { 0x1049E, 0x1049F} /* Cn */, { 0x104AA, 0x107FF} /* Cn */,
1036 { 0x10806, 0x10807} /* Cn */, { 0x10809, 0x10809} /* Cn */,
1037 { 0x10836, 0x10836} /* Cn */, { 0x10839, 0x1083B} /* Cn */,
1038 { 0x1083D, 0x1083E} /* Cn */, { 0x10840, 0x108FF} /* Cn */,
1039 { 0x1091A, 0x1091E} /* Cn */, { 0x10920, 0x109FF} /* Cn */,
1040 { 0x10A04, 0x10A04} /* Cn */, { 0x10A07, 0x10A0B} /* Cn */,
1041 { 0x10A14, 0x10A14} /* Cn */, { 0x10A18, 0x10A18} /* Cn */,
1042 { 0x10A34, 0x10A37} /* Cn */, { 0x10A3B, 0x10A3E} /* Cn */,
1043 { 0x10A48, 0x10A4F} /* Cn */, { 0x10A59, 0x11FFF} /* Cn */,
1044 { 0x1236F, 0x123FF} /* Cn */, { 0x12463, 0x1246F} /* Cn */,
1045 { 0x12474, 0x1CFFF} /* Cn */, { 0x1D0F6, 0x1D0FF} /* Cn */,
1046 { 0x1D127, 0x1D129} /* Cn */,
1047 { 0x1D173, 0x1D17A} /* Cf */,
1048 { 0x1D1DE, 0x1D1FF} /* Cn */, { 0x1D246, 0x1D2FF} /* Cn */,
1049 { 0x1D357, 0x1D35F} /* Cn */, { 0x1D372, 0x1D3FF} /* Cn */,
1050 { 0x1D455, 0x1D455} /* Cn */, { 0x1D49D, 0x1D49D} /* Cn */,
1051 { 0x1D4A0, 0x1D4A1} /* Cn */, { 0x1D4A3, 0x1D4A4} /* Cn */,
1052 { 0x1D4A7, 0x1D4A8} /* Cn */, { 0x1D4AD, 0x1D4AD} /* Cn */,
1053 { 0x1D4BA, 0x1D4BA} /* Cn */, { 0x1D4BC, 0x1D4BC} /* Cn */,
1054 { 0x1D4C4, 0x1D4C4} /* Cn */, { 0x1D506, 0x1D506} /* Cn */,
1055 { 0x1D50B, 0x1D50C} /* Cn */, { 0x1D515, 0x1D515} /* Cn */,
1056 { 0x1D51D, 0x1D51D} /* Cn */, { 0x1D53A, 0x1D53A} /* Cn */,
1057 { 0x1D53F, 0x1D53F} /* Cn */, { 0x1D545, 0x1D545} /* Cn */,
1058 { 0x1D547, 0x1D549} /* Cn */, { 0x1D551, 0x1D551} /* Cn */,
1059 { 0x1D6A6, 0x1D6A7} /* Cn */, { 0x1D7CC, 0x1D7CD} /* Cn */,
1060 { 0x1D800, 0x1FFFF} /* Cn */, { 0x2A6D7, 0x2F7FF} /* Cn */,
1061 { 0x2FA1E, 0xE0000} /* Cn */,
1062 { 0xE0001, 0xE0001} /* Cf */,
1063 { 0xE0002, 0xE001F} /* Cn */,
1064 { 0xE0020, 0xE007F} /* Cf */,
1065 { 0xE0080, 0xE00FF} /* Cn */, { 0xE01F0, 0xEFFFF} /* Cn */,
1066 { 0xF0000, 0xFFFFD} /* Co */,
1067 { 0xFFFFE, 0xFFFFF} /* Cn */,
1068 {0x100000,0x10FFFD} /* Co */,
1069 {0x10FFFE,0x10FFFF} /* Cn */,
1070 {0x110000,0x7FFFFFFF} /* ISO 10646?? */
1071};
1072
745
1073/*
1074 * Double width characters
1075 * W: East Asian Wide
1076 * F: East Asian Full-width
1077 * Unassigned code points may be included when they allow ranges to be merged.
1078 * Last synched with
1079 * <http://www.unicode.org/Public/5.0.0/ucd/EastAsianWidth-5.0.0d2.txt>
1080 * dated 2005-11-08T01:32:56Z
1081 */
1082static struct wchar_range wide_table[] = {
1083 { 0x1100, 0x115F} /* W */, { 0x2329, 0x232A} /* W */,
1084 { 0x2E80, 0x2FFB} /* W */,
1085 { 0x3000, 0x3000} /* F */,
1086 { 0x3001, 0x303E} /* W */, { 0x3041, 0x4DB5} /* W */,
1087 { 0x4E00, 0x9FBB} /* W */, { 0xA000, 0xA4C6} /* W */,
1088 { 0xAC00, 0xD7A3} /* W */, { 0xF900, 0xFAD9} /* W */,
1089 { 0xFE10, 0xFE19} /* W */, { 0xFE30, 0xFE6B} /* W */,
1090 { 0xFF01, 0xFF60} /* F */, { 0xFFE0, 0xFFE6} /* F */,
1091 { 0x20000, 0x2FFFD} /* W */, { 0x30000, 0x3FFFD} /* W */,
1092};
1093
1094 static int
746 static int
1095is_in_table(ch, table, tsize)
747is_in_table(ch, table)
1096 LWCHAR ch;
748 LWCHAR ch;
1097 struct wchar_range table[];
1098 int tsize;
749 struct wchar_range_table *table;
1099{
1100 int hi;
1101 int lo;
1102
1103 /* Binary search in the table. */
750{
751 int hi;
752 int lo;
753
754 /* Binary search in the table. */
1104 if (ch < table[0].first)
755 if (ch < table->table[0].first)
1105 return 0;
1106 lo = 0;
756 return 0;
757 lo = 0;
1107 hi = tsize - 1;
758 hi = table->count - 1;
1108 while (lo <= hi)
1109 {
1110 int mid = (lo + hi) / 2;
759 while (lo <= hi)
760 {
761 int mid = (lo + hi) / 2;
1111 if (ch > table[mid].last)
762 if (ch > table->table[mid].last)
1112 lo = mid + 1;
763 lo = mid + 1;
1113 else if (ch < table[mid].first)
764 else if (ch < table->table[mid].first)
1114 hi = mid - 1;
1115 else
1116 return 1;
1117 }
1118 return 0;
1119}
1120
1121/*
1122 * Is a character a UTF-8 composing character?
1123 * If a composing character follows any char, the two combine into one glyph.
1124 */
1125 public int
1126is_composing_char(ch)
1127 LWCHAR ch;
1128{
765 hi = mid - 1;
766 else
767 return 1;
768 }
769 return 0;
770}
771
772/*
773 * Is a character a UTF-8 composing character?
774 * If a composing character follows any char, the two combine into one glyph.
775 */
776 public int
777is_composing_char(ch)
778 LWCHAR ch;
779{
1129 return is_in_table(ch, comp_table, (sizeof(comp_table) / sizeof(*comp_table)));
780 return is_in_table(ch, &compose_table);
1130}
1131
1132/*
1133 * Should this UTF-8 character be treated as binary?
1134 */
1135 public int
1136is_ubin_char(ch)
1137 LWCHAR ch;
1138{
781}
782
783/*
784 * Should this UTF-8 character be treated as binary?
785 */
786 public int
787is_ubin_char(ch)
788 LWCHAR ch;
789{
1139 return is_in_table(ch, ubin_table, (sizeof(ubin_table) / sizeof(*ubin_table)));
790 return is_in_table(ch, &ubin_table);
1140}
1141
1142/*
1143 * Is this a double width UTF-8 character?
1144 */
1145 public int
1146is_wide_char(ch)
1147 LWCHAR ch;
1148{
791}
792
793/*
794 * Is this a double width UTF-8 character?
795 */
796 public int
797is_wide_char(ch)
798 LWCHAR ch;
799{
1149 return is_in_table(ch, wide_table, (sizeof(wide_table) / sizeof(*wide_table)));
800 return is_in_table(ch, &wide_table);
1150}
1151
1152/*
1153 * Is a character a UTF-8 combining character?
1154 * A combining char acts like an ordinary char, but if it follows
1155 * a specific char (not any char), the two combine into one glyph.
1156 */
1157 public int

--- 15 unchanged lines hidden ---
801}
802
803/*
804 * Is a character a UTF-8 combining character?
805 * A combining char acts like an ordinary char, but if it follows
806 * a specific char (not any char), the two combine into one glyph.
807 */
808 public int

--- 15 unchanged lines hidden ---