xmlunicode.c

libxml,在UNIX/LINUX下非常重要的一个库,为XML相关应用提供方便.目前上载的是最新版本,若要取得最新版本,请参考里面的readme.

    {0xfe68, 0xfe68}, {0xfe6a, 0xfe6b}, {0xff01, 0xff03}, {0xff05, 0xff07}, 
    {0xff0a, 0xff0a}, {0xff0c, 0xff0c}, {0xff0e, 0xff0f}, {0xff1a, 0xff1b}, 
    {0xff1f, 0xff20}, {0xff3c, 0xff3c}, {0xff61, 0xff61}, {0xff64, 0xff64} };
static xmlChLRange xmlPoL[] = {{0x10100, 0x10101}, {0x1039f, 0x1039f} };
static xmlChRangeGroup xmlPoG = {72,2,xmlPoS,xmlPoL};

static xmlChSRange xmlPsS[] = {{0x28, 0x28}, {0x5b, 0x5b}, {0x7b, 0x7b}, 
    {0xf3a, 0xf3a}, {0xf3c, 0xf3c}, {0x169b, 0x169b}, {0x201a, 0x201a}, 
    {0x201e, 0x201e}, {0x2045, 0x2045}, {0x207d, 0x207d}, {0x208d, 0x208d}, 
    {0x2329, 0x2329}, {0x23b4, 0x23b4}, {0x2768, 0x2768}, {0x276a, 0x276a}, 
    {0x276c, 0x276c}, {0x276e, 0x276e}, {0x2770, 0x2770}, {0x2772, 0x2772}, 
    {0x2774, 0x2774}, {0x27e6, 0x27e6}, {0x27e8, 0x27e8}, {0x27ea, 0x27ea}, 
    {0x2983, 0x2983}, {0x2985, 0x2985}, {0x2987, 0x2987}, {0x2989, 0x2989}, 
    {0x298b, 0x298b}, {0x298d, 0x298d}, {0x298f, 0x298f}, {0x2991, 0x2991}, 
    {0x2993, 0x2993}, {0x2995, 0x2995}, {0x2997, 0x2997}, {0x29d8, 0x29d8}, 
    {0x29da, 0x29da}, {0x29fc, 0x29fc}, {0x3008, 0x3008}, {0x300a, 0x300a}, 
    {0x300c, 0x300c}, {0x300e, 0x300e}, {0x3010, 0x3010}, {0x3014, 0x3014}, 
    {0x3016, 0x3016}, {0x3018, 0x3018}, {0x301a, 0x301a}, {0x301d, 0x301d}, 
    {0xfd3e, 0xfd3e}, {0xfe35, 0xfe35}, {0xfe37, 0xfe37}, {0xfe39, 0xfe39}, 
    {0xfe3b, 0xfe3b}, {0xfe3d, 0xfe3d}, {0xfe3f, 0xfe3f}, {0xfe41, 0xfe41}, 
    {0xfe43, 0xfe43}, {0xfe47, 0xfe47}, {0xfe59, 0xfe59}, {0xfe5b, 0xfe5b}, 
    {0xfe5d, 0xfe5d}, {0xff08, 0xff08}, {0xff3b, 0xff3b}, {0xff5b, 0xff5b}, 
    {0xff5f, 0xff5f}, {0xff62, 0xff62} };
static xmlChRangeGroup xmlPsG = {65,0,xmlPsS,NULL};

static xmlChSRange xmlSS[] = {{0x24, 0x24}, {0x2b, 0x2b}, {0x3c, 0x3e}, 
    {0x5e, 0x5e}, {0x60, 0x60}, {0x7c, 0x7c}, {0x7e, 0x7e}, {0xa2, 0xa9}, 
    {0xac, 0xac}, {0xae, 0xb1}, {0xb4, 0xb4}, {0xb6, 0xb6}, {0xb8, 0xb8}, 
    {0xd7, 0xd7}, {0xf7, 0xf7}, {0x2c2, 0x2c5}, {0x2d2, 0x2df}, 
    {0x2e5, 0x2ed}, {0x2ef, 0x2ff}, {0x374, 0x375}, {0x384, 0x385}, 
    {0x3f6, 0x3f6}, {0x482, 0x482}, {0x60e, 0x60f}, {0x6e9, 0x6e9}, 
    {0x6fd, 0x6fe}, {0x9f2, 0x9f3}, {0x9fa, 0x9fa}, {0xaf1, 0xaf1}, 
    {0xb70, 0xb70}, {0xbf3, 0xbfa}, {0xe3f, 0xe3f}, {0xf01, 0xf03}, 
    {0xf13, 0xf17}, {0xf1a, 0xf1f}, {0xf34, 0xf34}, {0xf36, 0xf36}, 
    {0xf38, 0xf38}, {0xfbe, 0xfc5}, {0xfc7, 0xfcc}, {0xfcf, 0xfcf}, 
    {0x17db, 0x17db}, {0x1940, 0x1940}, {0x19e0, 0x19ff}, {0x1fbd, 0x1fbd}, 
    {0x1fbf, 0x1fc1}, {0x1fcd, 0x1fcf}, {0x1fdd, 0x1fdf}, {0x1fed, 0x1fef}, 
    {0x1ffd, 0x1ffe}, {0x2044, 0x2044}, {0x2052, 0x2052}, {0x207a, 0x207c}, 
    {0x208a, 0x208c}, {0x20a0, 0x20b1}, {0x2100, 0x2101}, {0x2103, 0x2106}, 
    {0x2108, 0x2109}, {0x2114, 0x2114}, {0x2116, 0x2118}, {0x211e, 0x2123}, 
    {0x2125, 0x2125}, {0x2127, 0x2127}, {0x2129, 0x2129}, {0x212e, 0x212e}, 
    {0x2132, 0x2132}, {0x213a, 0x213b}, {0x2140, 0x2144}, {0x214a, 0x214b}, 
    {0x2190, 0x2328}, {0x232b, 0x23b3}, {0x23b7, 0x23d0}, {0x2400, 0x2426}, 
    {0x2440, 0x244a}, {0x249c, 0x24e9}, {0x2500, 0x2617}, {0x2619, 0x267d}, 
    {0x2680, 0x2691}, {0x26a0, 0x26a1}, {0x2701, 0x2704}, {0x2706, 0x2709}, 
    {0x270c, 0x2727}, {0x2729, 0x274b}, {0x274d, 0x274d}, {0x274f, 0x2752}, 
    {0x2756, 0x2756}, {0x2758, 0x275e}, {0x2761, 0x2767}, {0x2794, 0x2794}, 
    {0x2798, 0x27af}, {0x27b1, 0x27be}, {0x27d0, 0x27e5}, {0x27f0, 0x2982}, 
    {0x2999, 0x29d7}, {0x29dc, 0x29fb}, {0x29fe, 0x2b0d}, {0x2e80, 0x2e99}, 
    {0x2e9b, 0x2ef3}, {0x2f00, 0x2fd5}, {0x2ff0, 0x2ffb}, {0x3004, 0x3004}, 
    {0x3012, 0x3013}, {0x3020, 0x3020}, {0x3036, 0x3037}, {0x303e, 0x303f}, 
    {0x309b, 0x309c}, {0x3190, 0x3191}, {0x3196, 0x319f}, {0x3200, 0x321e}, 
    {0x322a, 0x3243}, {0x3250, 0x3250}, {0x3260, 0x327d}, {0x327f, 0x327f}, 
    {0x328a, 0x32b0}, {0x32c0, 0x32fe}, {0x3300, 0x33ff}, {0x4dc0, 0x4dff}, 
    {0xa490, 0xa4c6}, {0xfb29, 0xfb29}, {0xfdfc, 0xfdfd}, {0xfe62, 0xfe62}, 
    {0xfe64, 0xfe66}, {0xfe69, 0xfe69}, {0xff04, 0xff04}, {0xff0b, 0xff0b}, 
    {0xff1c, 0xff1e}, {0xff3e, 0xff3e}, {0xff40, 0xff40}, {0xff5c, 0xff5c}, 
    {0xff5e, 0xff5e}, {0xffe0, 0xffe6}, {0xffe8, 0xffee}, {0xfffc, 0xfffd} };
static xmlChLRange xmlSL[] = {{0x10102, 0x10102}, {0x10137, 0x1013f}, 
    {0x1d000, 0x1d0f5}, {0x1d100, 0x1d126}, {0x1d12a, 0x1d164}, 
    {0x1d16a, 0x1d16c}, {0x1d183, 0x1d184}, {0x1d18c, 0x1d1a9}, 
    {0x1d1ae, 0x1d1dd}, {0x1d300, 0x1d356}, {0x1d6c1, 0x1d6c1}, 
    {0x1d6db, 0x1d6db}, {0x1d6fb, 0x1d6fb}, {0x1d715, 0x1d715}, 
    {0x1d735, 0x1d735}, {0x1d74f, 0x1d74f}, {0x1d76f, 0x1d76f}, 
    {0x1d789, 0x1d789}, {0x1d7a9, 0x1d7a9}, {0x1d7c3, 0x1d7c3} };
static xmlChRangeGroup xmlSG = {133,20,xmlSS,xmlSL};

static xmlChSRange xmlScS[] = {{0x24, 0x24}, {0xa2, 0xa5}, {0x9f2, 0x9f3}, 
    {0xaf1, 0xaf1}, {0xbf9, 0xbf9}, {0xe3f, 0xe3f}, {0x17db, 0x17db}, 
    {0x20a0, 0x20b1}, {0xfdfc, 0xfdfc}, {0xfe69, 0xfe69}, {0xff04, 0xff04}, 
    {0xffe0, 0xffe1}, {0xffe5, 0xffe6} };
static xmlChRangeGroup xmlScG = {13,0,xmlScS,NULL};

static xmlChSRange xmlSkS[] = {{0x5e, 0x5e}, {0x60, 0x60}, {0xa8, 0xa8}, 
    {0xaf, 0xaf}, {0xb4, 0xb4}, {0xb8, 0xb8}, {0x2c2, 0x2c5}, 
    {0x2d2, 0x2df}, {0x2e5, 0x2ed}, {0x2ef, 0x2ff}, {0x374, 0x375}, 
    {0x384, 0x385}, {0x1fbd, 0x1fbd}, {0x1fbf, 0x1fc1}, {0x1fcd, 0x1fcf}, 
    {0x1fdd, 0x1fdf}, {0x1fed, 0x1fef}, {0x1ffd, 0x1ffe}, {0x309b, 0x309c}, 
    {0xff3e, 0xff3e}, {0xff40, 0xff40}, {0xffe3, 0xffe3} };
static xmlChRangeGroup xmlSkG = {22,0,xmlSkS,NULL};

static xmlChSRange xmlSmS[] = {{0x2b, 0x2b}, {0x3c, 0x3e}, {0x7c, 0x7c}, 
    {0x7e, 0x7e}, {0xac, 0xac}, {0xb1, 0xb1}, {0xd7, 0xd7}, {0xf7, 0xf7}, 
    {0x3f6, 0x3f6}, {0x2044, 0x2044}, {0x2052, 0x2052}, {0x207a, 0x207c}, 
    {0x208a, 0x208c}, {0x2140, 0x2144}, {0x214b, 0x214b}, {0x2190, 0x2194}, 
    {0x219a, 0x219b}, {0x21a0, 0x21a0}, {0x21a3, 0x21a3}, {0x21a6, 0x21a6}, 
    {0x21ae, 0x21ae}, {0x21ce, 0x21cf}, {0x21d2, 0x21d2}, {0x21d4, 0x21d4}, 
    {0x21f4, 0x22ff}, {0x2308, 0x230b}, {0x2320, 0x2321}, {0x237c, 0x237c}, 
    {0x239b, 0x23b3}, {0x25b7, 0x25b7}, {0x25c1, 0x25c1}, {0x25f8, 0x25ff}, 
    {0x266f, 0x266f}, {0x27d0, 0x27e5}, {0x27f0, 0x27ff}, {0x2900, 0x2982}, 
    {0x2999, 0x29d7}, {0x29dc, 0x29fb}, {0x29fe, 0x2aff}, {0xfb29, 0xfb29}, 
    {0xfe62, 0xfe62}, {0xfe64, 0xfe66}, {0xff0b, 0xff0b}, {0xff1c, 0xff1e}, 
    {0xff5c, 0xff5c}, {0xff5e, 0xff5e}, {0xffe2, 0xffe2}, {0xffe9, 0xffec} };
static xmlChLRange xmlSmL[] = {{0x1d6c1, 0x1d6c1}, {0x1d6db, 0x1d6db}, 
    {0x1d6fb, 0x1d6fb}, {0x1d715, 0x1d715}, {0x1d735, 0x1d735}, 
    {0x1d74f, 0x1d74f}, {0x1d76f, 0x1d76f}, {0x1d789, 0x1d789}, 
    {0x1d7a9, 0x1d7a9}, {0x1d7c3, 0x1d7c3} };
static xmlChRangeGroup xmlSmG = {48,10,xmlSmS,xmlSmL};

static xmlChSRange xmlSoS[] = {{0xa6, 0xa7}, {0xa9, 0xa9}, {0xae, 0xae}, 
    {0xb0, 0xb0}, {0xb6, 0xb6}, {0x482, 0x482}, {0x60e, 0x60f}, 
    {0x6e9, 0x6e9}, {0x6fd, 0x6fe}, {0x9fa, 0x9fa}, {0xb70, 0xb70}, 
    {0xbf3, 0xbf8}, {0xbfa, 0xbfa}, {0xf01, 0xf03}, {0xf13, 0xf17}, 
    {0xf1a, 0xf1f}, {0xf34, 0xf34}, {0xf36, 0xf36}, {0xf38, 0xf38}, 
    {0xfbe, 0xfc5}, {0xfc7, 0xfcc}, {0xfcf, 0xfcf}, {0x1940, 0x1940}, 
    {0x19e0, 0x19ff}, {0x2100, 0x2101}, {0x2103, 0x2106}, {0x2108, 0x2109}, 
    {0x2114, 0x2114}, {0x2116, 0x2118}, {0x211e, 0x2123}, {0x2125, 0x2125}, 
    {0x2127, 0x2127}, {0x2129, 0x2129}, {0x212e, 0x212e}, {0x2132, 0x2132}, 
    {0x213a, 0x213b}, {0x214a, 0x214a}, {0x2195, 0x2199}, {0x219c, 0x219f}, 
    {0x21a1, 0x21a2}, {0x21a4, 0x21a5}, {0x21a7, 0x21ad}, {0x21af, 0x21cd}, 
    {0x21d0, 0x21d1}, {0x21d3, 0x21d3}, {0x21d5, 0x21f3}, {0x2300, 0x2307}, 
    {0x230c, 0x231f}, {0x2322, 0x2328}, {0x232b, 0x237b}, {0x237d, 0x239a}, 
    {0x23b7, 0x23d0}, {0x2400, 0x2426}, {0x2440, 0x244a}, {0x249c, 0x24e9}, 
    {0x2500, 0x25b6}, {0x25b8, 0x25c0}, {0x25c2, 0x25f7}, {0x2600, 0x2617}, 
    {0x2619, 0x266e}, {0x2670, 0x267d}, {0x2680, 0x2691}, {0x26a0, 0x26a1}, 
    {0x2701, 0x2704}, {0x2706, 0x2709}, {0x270c, 0x2727}, {0x2729, 0x274b}, 
    {0x274d, 0x274d}, {0x274f, 0x2752}, {0x2756, 0x2756}, {0x2758, 0x275e}, 
    {0x2761, 0x2767}, {0x2794, 0x2794}, {0x2798, 0x27af}, {0x27b1, 0x27be}, 
    {0x2800, 0x28ff}, {0x2b00, 0x2b0d}, {0x2e80, 0x2e99}, {0x2e9b, 0x2ef3}, 
    {0x2f00, 0x2fd5}, {0x2ff0, 0x2ffb}, {0x3004, 0x3004}, {0x3012, 0x3013}, 
    {0x3020, 0x3020}, {0x3036, 0x3037}, {0x303e, 0x303f}, {0x3190, 0x3191}, 
    {0x3196, 0x319f}, {0x3200, 0x321e}, {0x322a, 0x3243}, {0x3250, 0x3250}, 
    {0x3260, 0x327d}, {0x327f, 0x327f}, {0x328a, 0x32b0}, {0x32c0, 0x32fe}, 
    {0x3300, 0x33ff}, {0x4dc0, 0x4dff}, {0xa490, 0xa4c6}, {0xfdfd, 0xfdfd}, 
    {0xffe4, 0xffe4}, {0xffe8, 0xffe8}, {0xffed, 0xffee}, {0xfffc, 0xfffd} };
static xmlChLRange xmlSoL[] = {{0x10102, 0x10102}, {0x10137, 0x1013f}, 
    {0x1d000, 0x1d0f5}, {0x1d100, 0x1d126}, {0x1d12a, 0x1d164}, 
    {0x1d16a, 0x1d16c}, {0x1d183, 0x1d184}, {0x1d18c, 0x1d1a9}, 
    {0x1d1ae, 0x1d1dd}, {0x1d300, 0x1d356} };
static xmlChRangeGroup xmlSoG = {103,10,xmlSoS,xmlSoL};

static xmlChSRange xmlZS[] = {{0x20, 0x20}, {0xa0, 0xa0}, {0x1680, 0x1680}, 
    {0x180e, 0x180e}, {0x2000, 0x200b}, {0x2028, 0x2029}, {0x202f, 0x202f}, 
    {0x205f, 0x205f}, {0x3000, 0x3000} };
static xmlChRangeGroup xmlZG = {9,0,xmlZS,NULL};

xmlUnicodeNameTable xmlUnicodeBlockTbl = {xmlUnicodeBlocks, 128};
xmlUnicodeNameTable xmlUnicodeCatTbl = {xmlUnicodeCats, 36};

/**
 * xmlUnicodeLookup:
 * @tptr: pointer to the name table
 * @name: name to be found
 *
 * binary table lookup for user-supplied name
 *
 * Returns pointer to range function if found, otherwise NULL
 */
static xmlIntFunc
*xmlUnicodeLookup(xmlUnicodeNameTable *tptr, const char *tname) {
    int low, high, mid, cmp;
    xmlUnicodeRange *sptr;

    if ((tptr == NULL) || (tname == NULL)) return(NULL);

    low = 0;
    high = tptr->numentries - 1;
    sptr = tptr->table;
    while (low <= high) {
	mid = (low + high) / 2;
	if ((cmp=strcmp(tname, sptr[mid].rangename)) == 0)
	    return (sptr[mid].func);
	if (cmp < 0)
	    high = mid - 1;
	else
	    low = mid + 1;
    }
    return (NULL);    
}

/**
 * xmlUCSIsAegeanNumbers:
 * @code: UCS code point
 *
 * Check whether the character is part of AegeanNumbers UCS Block
 *
 * Returns 1 if true 0 otherwise
 */
int
xmlUCSIsAegeanNumbers(int code) {
    return(((code >= 0x10100) && (code <= 0x1013F)));
}

/**
 * xmlUCSIsAlphabeticPresentationForms:
 * @code: UCS code point
 *
 * Check whether the character is part of AlphabeticPresentationForms UCS Block
 *
 * Returns 1 if true 0 otherwise
 */
int
xmlUCSIsAlphabeticPresentationForms(int code) {
    return(((code >= 0xFB00) && (code <= 0xFB4F)));
}

/**
 * xmlUCSIsArabic:
 * @code: UCS code point
 *
 * Check whether the character is part of Arabic UCS Block
 *
 * Returns 1 if true 0 otherwise
 */
int
xmlUCSIsArabic(int code) {
    return(((code >= 0x0600) && (code <= 0x06FF)));
}

/**
 * xmlUCSIsArabicPresentationFormsA:
 * @code: UCS code point
 *
 * Check whether the character is part of ArabicPresentationForms-A UCS Block
 *
 * Returns 1 if true 0 otherwise
 */
int
xmlUCSIsArabicPresentationFormsA(int code) {
    return(((code >= 0xFB50) && (code <= 0xFDFF)));
}

/**
 * xmlUCSIsArabicPresentationFormsB:
 * @code: UCS code point
 *
 * Check whether the character is part of ArabicPresentationForms-B UCS Block
 *
 * Returns 1 if true 0 otherwise
 */
int
xmlUCSIsArabicPresentationFormsB(int code) {
    return(((code >= 0xFE70) && (code <= 0xFEFF)));
}

/**
 * xmlUCSIsArmenian:
 * @code: UCS code point
 *
 * Check whether the character is part of Armenian UCS Block
 *
 * Returns 1 if true 0 otherwise
 */
int
xmlUCSIsArmenian(int code) {
    return(((code >= 0x0530) && (code <= 0x058F)));
}

/**
 * xmlUCSIsArrows:
 * @code: UCS code point
 *
 * Check whether the character is part of Arrows UCS Block
 *
 * Returns 1 if true 0 otherwise
 */
int
xmlUCSIsArrows(int code) {
    return(((code >= 0x2190) && (code <= 0x21FF)));
}

/**
 * xmlUCSIsBasicLatin:
 * @code: UCS code point