lycharsets.c

基于rtos开发的浏览器！

#include "HTUtils.h"
#include "tcp.h"
#include "HTCJK.h"
#include "HTMLDTD.h"

#include "LYGlobalDefs.h"
#include "UCMap.h"
#include "UCDefs.h"
#include "LYCharSets.h"
#include "LYCharUtils.h"
#include "HTFont.h"
#include "GridText.h"
#include "LYCurses.h"
#include "LYStrings.h"

#include "LYexit.h"
#include "LYLeaks.h"

extern BOOL HTPassEightBitRaw;
extern BOOL HTPassEightBitNum;
extern BOOL HTPassHighCtrlRaw;
extern BOOL HTPassHighCtrlNum;
extern HTCJKlang HTCJK;
PUBLIC HTkcode kanji_code = NOKANJI;
PUBLIC BOOLEAN LYHaveCJKCharacterSet = FALSE;
extern void UCInit NOARGS;
extern int UCInitialized;
PUBLIC int LYNumCharsets = 0; /* Will be initialized later by UC_Register. */

/*
 *  INSTRUCTIONS for adding new character sets which do not have
 *		 Unicode tables.
 *
 *  Currently we only declare some charset's properties here
 *  (such as MIME names, etc.), it does not include real mapping.
 *
 *  [We hope you need not correct/add old-style mapping
 *  as in ISO_LATIN1[] or SevenBitApproximations[] any more -
 *  it works now via new chartrans mechanism, but kept for compatibility only:
 *  we should cleanup the stuff, but this is not so easy...]
 *
 *  There is a place marked "Add your new character sets HERE" in this file.
 *  Make up a character set and add it in the same
 *  style as the ISO_LATIN1 set below, giving it a unique name.
 *
 *  Add the name of the set to LYCharSets.
 *  Similarly add the appropriate information to the tables below:
 *  LYchar_set_names, LYCharSet_UC, LYlowest_eightbit.
 *  These 4 tables all MUST have the same order.
 *  (And this is the order you will see in Lynx Options Menu,
 *  which is why few unicode-based charsets are listed here).
 *
 */

/*	Entity values -- for ISO Latin 1 local representation
**
**	This MUST match exactly the table referred to in the DTD!
*/
PRIVATE char * ISO_Latin1[] = {
	"\306", /* capital AE diphthong (ligature) (Æ) - AElig */
	"\301", /* capital A, acute accent (Á) - Aacute */
	"\302", /* capital A, circumflex accent (Â) - Acirc */
	"\300", /* capital A, grave accent (À) - Agrave */
	"\305", /* capital A, ring - Aring (Å) */
	"\303", /* capital A, tilde - Atilde (Ã) */
	"\304", /* capital A, dieresis or umlaut mark (Ä) - Auml */
	"\307", /* capital C, cedilla - Ccedil (Ç) */
	"\320", /* capital Eth or D with stroke (Ð) - Dstrok */
	"\320", /* capital Eth, Icelandic (Ð) - ETH */
	"\311", /* capital E, acute accent (É) - Eacute */
	"\312", /* capital E, circumflex accent (Ê) - Ecirc */
	"\310", /* capital E, grave accent (È) - Egrave */
	"\313", /* capital E, dieresis or umlaut mark (Ë) - Euml */
	"\315", /* capital I, acute accent (Í) - Iacute */
	"\316", /* capital I, circumflex accent (Î) - Icirc */
	"\314", /* capital I, grave accent (È) - Igrave */
	"\317", /* capital I, dieresis or umlaut mark (Ï) - Iuml */
	"\321", /* capital N, tilde (Ñ) - Ntilde */
	"\323", /* capital O, acute accent (Ó) - Oacute */
	"\324", /* capital O, circumflex accent (Ô) - Ocirc */
	"\322", /* capital O, grave accent (Ò) - Ograve */
	"\330", /* capital O, slash (Ø) - Oslash */
	"\325", /* capital O, tilde (Õ) - Otilde */
	"\326", /* capital O, dieresis or umlaut mark (Ö) - Ouml */
	"\336", /* capital THORN, Icelandic (Þ) - THORN */
	"\332", /* capital U, acute accent (Ú) - Uacute */
	"\333", /* capital U, circumflex accent (Û) - Ucirc */
	"\331", /* capital U, grave accent (Ù) - Ugrave */
	"\334", /* capital U, dieresis or umlaut mark (Ü) - Uuml */
	"\335", /* capital Y, acute accent (Ý) - Yacute */
	"\341", /* small a, acute accent (á) - aacute */
	"\342", /* small a, circumflex accent (â) - acirc */
	"\264", /* spacing acute (´) - acute */
	"\346", /* small ae diphthong (ligature) (æ) - aelig */
	"\340", /* small a, grave accent (à) - agrave */
	"\046", /* ampersand (&) - amp */
	"\345", /* small a, ring (å) - aring */
	"\343", /* small a, tilde (ã) - atilde */
	"\344", /* small a, dieresis or umlaut mark (ä) - auml */
	"\246", /* broken vertical bar (¦) - brkbar */
	"\246", /* broken vertical bar (¦) - brvbar */
	"\347", /* small c, cedilla (ç) - ccedil */
	"\270", /* spacing cedilla (¸) - cedil */
	"\242", /* cent sign (¢) - cent */
	"\251", /* copyright sign (©) - copy */
	"\244", /* currency sign (¤) - curren */
	"\260", /* degree sign (°) - deg */
	"\250", /* spacing dieresis (¨) - die */
	"\367", /* division sign (÷) - divide */
	"\351", /* small e, acute accent (é) - eacute */
	"\352", /* small e, circumflex accent (ê) - ecirc */
	"\350", /* small e, grave accent (è) - egrave */
	"-",	/* dash the width of emsp - emdash */
	"\002", /* emsp, em space - not collapsed NEVER CHANGE THIS - emsp */
	"-",	/* dash the width of ensp - endash */
	"\002", /* ensp, en space - not collapsed NEVER CHANGE THIS - ensp */
	"\360", /* small eth, Icelandic (ð) - eth */
	"\353", /* small e, dieresis or umlaut mark (ë) - euml */
	"\275", /* fraction 1/2 (½) - frac12 */
	"\274", /* fraction 1/4 (¼) - frac14 */
	"\276", /* fraction 3/4 (¾) - frac34 */
	"\076", /* greater than (>) - gt */
	"\257", /* spacing macron (¯) - hibar */
	"\355", /* small i, acute accent (í) - iacute */
	"\356", /* small i, circumflex accent (î) - icirc */
	"\241", /* inverted exclamation mark (¡) - iexcl */
	"\354", /* small i, grave accent (ì) - igrave */
	"\277", /* inverted question mark (¿) - iquest */
	"\357", /* small i, dieresis or umlaut mark (ï) - iuml */
	"\253", /* angle quotation mark, left («) - laquo */
	"\074", /* less than (<) - lt */
	"\257", /* spacing macron (¯) - macr */
	"-",	/* dash the width of emsp - mdash */
	"\265", /* micro sign (µ) - micro */
	"\267", /* middle dot (·) - middot */
	"\001", /* nbsp non-breaking space NEVER CHANGE THIS - nbsp */
	"-",	/* dash the width of ensp - ndash */
	"\254", /* negation sign (¬) - not */
	"\361", /* small n, tilde (ñ) - ntilde */
	"\363", /* small o, acute accent (ó) - oacute */
	"\364", /* small o, circumflex accent (ô) - ocirc */
	"\362", /* small o, grave accent (ò) - ograve */
	"\252", /* feminine ordinal indicator (ª) - ordf */
	"\272", /* masculine ordinal indicator (º) - ordm */
	"\370", /* small o, slash (ø) - oslash */
	"\365", /* small o, tilde (õ) - otilde */
	"\366", /* small o, dieresis or umlaut mark (ö) - ouml */
	"\266", /* paragraph sign (¶) - para */
	"\261", /* plus-or-minus sign (±) - plusmn */
	"\243", /* pound sign (£) - pound */
	"\042", /* quote '"' (") - quot */
	"\273", /* angle quotation mark, right (») - raquo */
	"\256", /* circled R registered sign (®) - reg */
	"\247", /* section sign (§) - sect */
	"\007", /* soft hyphen (­) NEVER CHANGE THIS - shy */
	"\271", /* superscript 1 (¹) - sup1 */
	"\262", /* superscript 2 (²) - sup2 */
	"\263", /* superscript 3 (³) - sup3 */
	"\337", /* small sharp s, German (sz ligature) (ß) - szlig */
	"\002", /* thin space - not collapsed NEVER CHANGE THIS - thinsp */
	"\376", /* small thorn, Icelandic (þ) - thorn */
	"\327", /* multiplication sign (×) - times */
	"(TM)", /* circled TM trade mark sign (™) - trade */
	"\372", /* small u, acute accent (ú) - uacute */
	"\373", /* small u, circumflex accent (û) - ucirc */
	"\371", /* small u, grave accent (ù) - ugrave */
	"\250", /* spacing dieresis (¨) - uml */
	"\374", /* small u, dieresis or umlaut mark (ü) - uuml */
	"\375", /* small y, acute accent (ý) - yacute */
	"\245", /* yen sign (¥) - yen */
	"\377", /* small y, dieresis or umlaut mark (ÿ) - yuml */
};

/*	Entity values -- 7 bit character approximations
**
**	This MUST match exactly the table referred to in the DTD!
*/
PUBLIC char * SevenBitApproximations[] = {
	"AE",	/* capital AE diphthong (ligature) (Æ) - AElig */
	"A",	/* capital A, acute accent (Á) - Aacute */
	"A",	/* capital A, circumflex accent (Â) - Acirc */
	"A",	/* capital A, grave accent (À) - Agrave */
	"A",	/* capital A, ring - Aring (Å) */
	"A",	/* capital A, tilde - Atilde (Ã) */
#ifdef LY_UMLAUT
	"Ae",	/* capital A, dieresis or umlaut mark (Ä) - Auml*/
#else
	"A",	/* capital A, dieresis or umlaut mark (Ä) - Auml*/
#endif /* LY_UMLAUT */
	"C",	/* capital C, cedilla (Ç) - Ccedil */
	"Dj",	/* capital D with stroke (Ð) - Dstrok */
	"DH",	/* capital Eth, Icelandic (Ð) - ETH */
	"E",	/* capital E, acute accent (É) - Eacute */
	"E",	/* capital E, circumflex accent (Ê) - Ecirc */
	"E",	/* capital E, grave accent (È) - Egrave */
	"E",	/* capital E, dieresis or umlaut mark (Ë) - Euml */
	"I",	/* capital I, acute accent (Í) - Iacute */
	"I",	/* capital I, circumflex accent (Î) - Icirc */
	"I",	/* capital I, grave accent (Ì) - Igrave */
	"I",	/* capital I, dieresis or umlaut mark (Ï) - Iuml */
	"N",	/* capital N, tilde - Ntilde (Ñ) */
	"O",	/* capital O, acute accent (Ó) - Oacute */
	"O",	/* capital O, circumflex accent (Ô) - Ocirc */
	"O",	/* capital O, grave accent (Ò) - Ograve */
	"O",	/* capital O, slash (Ø) - Oslash */
	"O",	/* capital O, tilde (Õ) - Otilde */
#ifdef LY_UMLAUT
	"Oe",	/* capital O, dieresis or umlaut mark (Ö) - Ouml */
#else
	"O",	/* capital O, dieresis or umlaut mark (Ö) - Ouml */
#endif /* LY_UMLAUT */
	"P",	/* capital THORN, Icelandic (Þ) - THORN */
	"U",	/* capital U, acute accent (Ú) - Uacute */
	"U",	/* capital U, circumflex accent (Û) - Ucirc */
	"U",	/* capital U, grave accent (Ù) - Ugrave */
#ifdef LY_UMLAUT
	"Ue",	/* capital U, dieresis or umlaut mark (Ü) - Uuml */
#else
	"U",	/* capital U, dieresis or umlaut mark (Ü) - Uuml */
#endif /* LY_UMLAUT */
	"Y",	/* capital Y, acute accent (Ý) - Yacute */
	"a",	/* small a, acute accent (á) - aacute */
	"a",	/* small a, circumflex accent (â) - acirc */
	"'",	/* spacing acute (´) - acute */
	"ae",	/* small ae diphthong (ligature) (æ) - aelig */
	"`a",	/* small a, grave accent (è) - agrave */
	"&",	/* ampersand (&) - amp */
	"a",	/* small a, ring (å) - aring */
	"a",	/* small a, tilde (ã) - atilde */
#ifdef LY_UMLAUT
	"ae",	/* small a, dieresis or umlaut mark (ä) - auml */
#else
	"a",	/* small a, dieresis or umlaut mark (ä) - auml */
#endif /* LY_UMLAUT */
	"|",	/* broken vertical bar (¦) - brkbar */
	"|",	/* broken vertical bar (¦) - brvbar */
	"c",	/* small c, cedilla (ç) - ccedil */
	",",	/* spacing cedilla (¸) - cedil */
	"-c-",	/* cent sign (¢) - cent */
	"(c)",	/* copyright sign (©) - copy */
	"CUR",	/* currency sign (¤) - curren */
	"DEG",	/* degree sign (°) - deg */
	"\042", /* spacing dieresis (¨) - die */
	"/",	/* division sign (÷) - divide */
	"e",	/* small e, acute accent (é) - eacute */
	"e",	/* small e, circumflex accent (ê) - ecirc */
	"e",	/* small e, grave accent (è) - egrave */
	"-",	/* dash the width of emsp - emdash */
	"\002", /* emsp NEVER CHANGE THIS - emsp */
	"-",	/* dash the width of ensp - endash */
	"\002", /* ensp NEVER CHANGE THIS - ensp */
	"dh",	/* small eth, Icelandic eth (ð) */
	"e",	/* small e, dieresis or umlaut mark (ë) - euml */
	" 1/2", /* fraction 1/2 (½) - frac12 */
	" 1/4", /* fraction 1/4 (¼) - frac14 */
	" 3/4", /* fraction 3/4 (¾) - frac34 */
	">",	/* greater than (>) - gt */
	"-",	/* spacing macron (¯) - hibar */
	"i",	/* small i, acute accent (í) - iacute */
	"i",	/* small i, circumflex accent (î) - icirc*/
	"!",	/* inverted exclamation mark (¡) - iexcl */
	"`i",	/* small i, grave accent (ì) - igrave */
	"?",	/* inverted question mark (¿) - iquest */
	"i",	/* small i, dieresis or umlaut mark (ï) - iuml */
	"<<",	/* angle quotation mark, left (&#171;) - laquo */
	"<",	/* less than - lt (&#60;) */
	"-",	/* spacing macron (&#175;) - macr */
	"-",	/* dash the width of emsp - mdash */
	"u",	/* micro sign (&#181;) - micro */
	".",	/* middle dot (&#183;) - middot */
	"\001", /* nbsp non-breaking space NEVER CHANGE THIS - nbsp */
	"-",	/* dash the width of ensp - ndash */
	"NOT",	/* negation sign (&#172;) - not */
	"n",	/* small n, tilde (&#241;) - ntilde */
	"o",	/* small o, acute accent (&#243;) - oacute */
	"o",	/* small o, circumflex accent (&#244;) - ocirc */
	"o",	/* small o, grave accent (&#242;) - ograve */
	"-a",	/* feminine ordinal indicator (&#170;) - ordf */
	"-o",	/* masculine ordinal indicator (&#186;) - ordm */
	"o",	/* small o, slash (&#248;) - oslash */
	"o",	/* small o, tilde (&#245;) - otilde */
#ifdef LY_UMLAUT
	"oe",	/* small o, dieresis or umlaut mark (&#246;) - ouml */
#else
	"o",	/* small o, dieresis or umlaut mark (&#246;) - ouml */
#endif /* LY_UMLAUT */
	"P:",	/* paragraph sign (&#182;) - para */
	"+-",	/* plus-or-minus sign (&#177;) - plusmn */
	"-L-",	/* pound sign (&#163;) - pound */
	"\"",	/* quote '"' (&#34;) - quot */
	">>",	/* angle quotation mark, right (&#187;) - raquo */
	"(R)",	/* circled R registered sign (&#174;) - reg */
	"S:",	/* section sign (&#167;) - sect */
	"\007", /* soft hyphen (&#173;) NEVER CHANGE THIS - shy */
	"^1",	/* superscript 1 (&#185;) - sup1 */
	"^2",	/* superscript 2 (&#178;) - sup2 */
	"^3",	/* superscript 3 (&#179;) - sup3 */
	"ss",	/* small sharp s, German (sz ligature) (&#223;) - szlig */
	"\002", /* thin space - not collapsed NEVER CHANGE THIS - thinsp */
	"p",	/* small thorn, Icelandic (&#254;) - thorn */
	"*",	/* multiplication sign (&#215;) - times */
	"(TM)", /* circled TM trade mark sign (&#8482;) - trade */
	"u",	/* small u, acute accent (&#250;) - uacute */
	"u",	/* small u, circumflex accent (&#251;) - ucirc */
	"u",	/* small u, grave accent (&#249;) - ugrave */
	"\042", /* spacing dieresis (&#168;) - uml */
#ifdef LY_UMLAUT
	"ue",	/* small u, dieresis or umlaut mark (&#252;) - uuml */
#else
	"u",	/* small u, dieresis or umlaut mark (&#252;) - uuml */
#endif /* LY_UMLAUT */
	"y",	/* small y, acute accent (&#253;) - yacute */
	"YEN",	/* yen sign (&#165;) - yen */