xercesc中文转换和gdb与utf-8互转

#include "strnormalize.h" #include <stdlib.h> #define SWAPBYTE(x) (((unsigned short)(x) >> 8) | (unsigned short)(x) << 8) #define COMPBYTE(x, y) ((unsigned char)(x) << 8 | (unsigned char)(y)) static const unsigned short *_pTrad2Simp_gbk; static const unsigned short *_pTrad2Simp_utf16; static const unsigned short *_pPlain_gbk; static const unsigned short *_pPlain_utf16; static const char *_pUpper2Lower; static const char *_pLower2Upper; static const char *_pPlain; static const unsigned short *_pGbk2Utf16; static const unsigned short *_pUtf162Gbk; static const unsigned short *_pTns; static unsigned short _tns_size; static const unsigned short *_pGbk2utf16_2; static unsigned short _gbk2utf16_2_size; static const unsigned short *_pGbk2utf16_3; static unsigned short _gbk2utf16_3_size; const unsigned short *_initTns(); const unsigned short *_initGbk2utf16_2(); const unsigned short *_initGbk2utf16_3(); const unsigned short *_initTrad2Simp_utf16(); const unsigned short *_initTrad2Simp_gbk(); const unsigned short *_initPlain_utf16(); const unsigned short *_initPlain_gbk(); const char *_initUpper2Lower(); const char *_initLower2Upper(); const char *_initPlain(); const unsigned short *_initGbk2Utf16(); const unsigned short *_initUtf162Gbk(); void str_normalize_init() { _pTns = _initTns(); _pGbk2utf16_2 = _initGbk2utf16_2(); _pGbk2utf16_3 = _initGbk2utf16_3(); _pTrad2Simp_gbk = _initTrad2Simp_gbk(); _pTrad2Simp_utf16 = _initTrad2Simp_utf16(); _pPlain_gbk = _initPlain_gbk(); _pPlain_utf16 = _initPlain_utf16(); _pUpper2Lower = _initUpper2Lower(); _pLower2Upper = _initLower2Upper(); _pPlain = _initPlain(); _pGbk2Utf16 = _initGbk2Utf16(); _pUtf162Gbk = _initUtf162Gbk(); } const unsigned short *_initTns() { static const unsigned short _tns[] = { 0x8147 /* 丟 */, 0xB6AA /* 丢 */, // {{{ Simplified-Traditional table 0x814B /* 並 */, 0xB2A2 /* 并 */, 0x8179 /* 亂 */, 0xC2D2 /* 乱 */, 0x8183 /* 亙 */, 0xD8A8 /* 亘 */, 0x8186 /* 亞 */, 0xD1C7 /* 亚 */, 0x81B2 /* 伈 */, 0xD0C4 /* 心 */, 0x81B8 /* 伕 */, 0xB7F2 /* 夫 */, 0x81D0 /* 佇 */, 0xD8F9 /* 伫 */, 0x81D1 /* 佈 */, 0xB2BC /* 布 */, 0x81D7 /* 佔 */, 0xD5BC /* 占 */, 0x81E3 /* 併 */, 0xB2A2 /* 并 */, 0x81ED /* 來 */, 0xC0B4 /* 来 */, 0x81F6 /* 侖 */, 0xC2D8 /* 仑 */, 0x81F9 /* 侚 */, 0xE1DF /* 徇 */, 0x8248 /* 侶 */, 0xC2C2 /* 侣 */, 0x8249 /* 侷 */, 0xBED6 /* 局 */, 0x8252 /* 俁 */, 0xD9B6 /* 俣 */, 0x8253 /* 係 */, 0xCFB5 /* 系 */, 0x8262 /* 俠 */, 0xCFC0 /* 侠 */, 0x826C /* 俵 */, 0xB1ED /* 表 */, 0x8274 /* 倀 */, 0xD8F6 /* 伥 */, 0x827A /* 倆 */, 0xC1A9 /* 俩 */, 0x827D /* 倉 */, 0xB2D6 /* 仓 */, 0x8280 /* 個 */, 0xB8F6 /* 个 */, 0x8283 /* 們 */, 0xC3C7 /* 们 */, 0x8286 /* 倖 */, 0xD0D2 /* 幸 */, 0x828D /* 倣 */, 0xB7C2 /* 仿 */, 0x8290 /* 倫 */, 0xC2D7 /* 伦 */, 0x82A5 /* 偉 */, 0xCEB0 /* 伟 */, 0x82BF /* 偪 */, 0xB1C6 /* 逼 */, 0x82C8 /* 側 */, 0xB2E0 /* 侧 */, 0x82C9 /* 偵 */, 0xD5EC /* 侦 */, 0x82CA /* 偸 */, 0xCDB5 /* 偷 */, 0x82CC /* 偺 */, 0xD4DB /* 咱 */, 0x82CE /* 偽 */, 0xCEB1 /* 伪 */, 0x82DC /* 傑 */, 0xBDDC /* 杰 */, 0x82E1 /* 傖 */, 0xD8F7 /* 伧 */, 0x82E3 /* 傘 */, 0xC9A1 /* 伞 */, 0x82E4 /* 備 */, 0xB1B8 /* 备 */, 0x82E5 /* 傚 */, 0xD0A7 /* 效 */, 0x82ED /* 傢 */, 0xBCD2 /* 家 */, 0x82F2 /* 傭 */, 0xD3B6 /* 佣 */, 0x82F4 /* 傯 */, 0xD9CC /* 偬 */, 0x82F7 /* 傳 */, 0xB4AB /* 传 */, 0x82F8 /* 傴 */, 0xD8F1 /* 伛 */, 0x82F9 /* 債 */, 0xD5AE /* 债 */, 0x82FB /* 傷 */, 0xC9CB /* 伤 */, 0x8341 /* 傾 */, 0xC7E3 /* 倾 */, 0x8345 /* 僂 */, 0xD9CD /* 偻 */, 0x8348 /* 僅 */, 0xBDF6 /* 仅 */, 0x834C /* 僉 */, 0xD9DD /* 佥 */, 0x834D /* 僊 */, 0xCFC9 /* 仙 */, 0x8353 /* 僑 */, 0xC7C8 /* 侨 */, 0x8357 /* 僕 */, 0xC6CD /* 仆 */, 0x835E /* 僞 */, 0xCEB1 /* 伪 */, 0x8365 /* 僥 */, 0xBDC4 /* 侥 */, 0x8366 /* 僨 */, 0xD9C7 /* 偾 */, 0x836C /* 僱 */, 0xB9CD /* 雇 */, 0x8372 /* 價 */, 0xBCDB /* 价 */, 0x8378 /* 儀 */, 0xD2C7 /* 仪 */, 0x837A /* 儂 */, 0xD9AF /* 侬 */, 0x837C /* 億 */, 0xD2DA /* 亿 */, 0x837D /* 儅 */, 0xB5B1 /* 当 */, 0x837E /* 儈 */, 0xBFEB /* 侩 */, 0x8380 /* 儉 */, 0xBCF3 /* 俭 */, 0x8386 /* 儐 */, 0xD9CF /* 傧 */, 0x8389 /* 儔 */, 0xD9B1 /* 俦 */, 0x838A /* 儕 */, 0xD9AD /* 侪 */, 0x838D /* 儘 */, 0xBEA1 /* 尽 */, 0x8394 /* 償 */, 0xB3A5 /* 偿 */, 0x839E /* 優 */, 0xD3C5 /* 优 */, 0x83A6 /* 儲 */, 0xB4A2 /* 储 */, 0x83A7 /* 儳 */, 0xB2F6 /* 馋 */, 0x83AB /* 儷 */, 0xD9B3 /* 俪 */, 0x83AC /* 儸 */, 0xC2DE /* 罗 */, 0x83AD /* 儹 */, 0xD4DC /* 攒 */, 0x83AE /* 儺 */, 0xD9D0 /* 傩 */, 0x83AF /* 儻 */, 0xD9CE /* 傥 */, 0x83B0 /* 儼 */, 0xD9B2 /* 俨 */, 0x83B4 /* 兇 */, 0xD0D7 /* 凶 */, 0x83B6 /* 兌 */, 0xB6D2 /* 兑 */, 0x83BA /* 兒 */, 0xB6F9 /* 儿 */, 0x83BC /* 兗 */, 0xD9F0 /* 兖 */, 0x83C8 /* 內 */, 0xC4DA /* 内 */, 0x83C9 /* 兩 */, 0xC1BD /* 两 */, 0x83D4 /* 冊 */, 0xB2E1 /* 册 */, 0x83D9 /* 冑 */, 0xEBD0 /* 胄 */, 0x83E7 /* 冪 */, 0xC3DD /* 幂 */, 0x83F2 /* 凃 */, 0xCDBF /* 涂 */, 0x83F3 /* 凅 */, 0xB9CC /* 固 */, 0x83F4 /* 凈 */, 0xBEBB /* 净 */, 0x83F6 /* 凍 */, 0xB6B3 /* 冻 */, 0x8443 /* 凜 */, 0xC1DD /* 凛 */, 0x8445 /* 凟 */, 0xE4C2 /* 渎 */, 0x8449 /* 処 */, 0xB4A6 /* 处 */, 0x8450 /* 凱 */, 0xBFAD /* 凯 */, 0x8452 /* 凴 */, 0xC6BE /* 凭 */, 0x8465 /* 別 */, 0xB1F0 /* 别 */, 0x8468 /* 刪 */, 0xC9BE /* 删 */, 0x8471 /* 剄 */, 0xD8D9 /* 刭 */, 0x8474 /* 則 */, 0xD4F2 /* 则 */, 0x8476 /* 剉 */, 0xEFB1 /* 锉 */, 0x8477 /* 剋 */, 0xBFCB /* 克 */, 0x8478 /* 剎 */, 0xC9B2 /* 刹 */, 0x847D /* 剗 */, 0x8469 /* 刬 */, 0x8482 /* 剛 */, 0xB8D5 /* 刚 */, 0x8483 /* 剝 */, 0xB0FE /* 剥 */, 0x848E /* 剮 */, 0xB9D0 /* 剐 */, 0x8492 /* 剴 */, 0xD8DC /* 剀 */, 0x8493 /* 創 */, 0xB4B4 /* 创 */, 0x8495 /* 剷 */, 0xB2F9 /* 铲 */, 0x849D /* 劃 */, 0xBBAE /* 划 */, 0x849E /* 劄 */, 0xD4FD /* 札 */, 0x84A1 /* 劇 */, 0xBEE7 /* 剧 */, 0x84A2 /* 劉 */, 0xC1F5 /* 刘 */, 0x84A3 /* 劊 */, 0xB9F4 /* 刽 */, 0x84A5 /* 劌 */, 0xD8DB /* 刿 */, 0x84A6 /* 劍 */, 0xBDA3 /* 剑 */, 0x84A9 /* 劑 */, 0xBCC1 /* 剂 */, 0x84AA /* 劒 */, 0xBDA3 /* 剑 */, 0x84C5 /* 勁 */, 0xBEA2 /* 劲 */, 0x84CD /* 勍 */, 0xC7E6 /* 擎 */, 0x84D3 /* 動 */, 0xB6AF /* 动 */, 0x84D4 /* 勗 */, 0xDBC3 /* 勖 */, 0x84D5 /* 務 */, 0xCEF1 /* 务 */, 0x84D7 /* 勛 */, 0xD1AB /* 勋 */, 0x84D9 /* 勝 */, 0xCAA4 /* 胜 */, 0x84DA /* 勞 */, 0xC0CD /* 劳 */, 0x84DD /* 勢 */, 0xCAC6 /* 势 */, 0x84DE /* 勣 */, 0xBCA8 /* 绩 */, 0x84E0 /* 勦 */, 0xBDCB /* 剿 */, 0x84E3 /* 勩 */, 0x84D6 /* 勚 */, 0x84EA /* 勱 */, 0xDBBD /* 劢 */, 0x84EC /* 勳 */, 0xD1AB /* 勋 */, 0x84EE /* 勵 */, 0xC0F8 /* 励 */, 0x84F1 /* 勸 */, 0xC8B0 /* 劝 */, 0x84F2 /* 勻 */, 0xD4C8 /* 匀 */, 0x84FC /* 匋 */, 0xCCD5 /* 陶 */, 0x8551 /* 匭 */, 0xD8D0 /* 匦 */, 0x8552 /* 匯 */, 0xBBE3 /* 汇 */, 0x8554 /* 匱 */, 0xD8D1 /* 匮 */, 0x8555 /* 匲 */, 0xDEC6 /* 奁 */, 0x8556 /* 匳 */, 0xDEC6 /* 奁 */, 0x8558 /* 匵 */, 0xE8FC /* 椟 */, 0x855E /* 區 */, 0xC7F8 /* 区 */, 0x8564 /* 卍 */, 0xCDF2 /* 万 */, 0x8566 /* 協 */, 0xD0AD /* 协 */, 0x856E /* 卬 */, 0xB0BA /* 昂 */, 0x8572 /* 卹 */, 0xD0F4 /* 恤 */, 0x8573 /* 卻 */, 0xC8B4 /* 却 */, 0x8587 /* 厙 */, 0xD8C7 /* 厍 */, 0x8588 /* 厛 */, 0xCCFC /* 厅 */, 0x858B /* 厠 */, 0xB2DE /* 厕 */, 0x858D /* 厤 */, 0xC0FA /* 历 */, 0x8592 /* 厭 */, 0xD1E1 /* 厌 */, 0x8594 /* 厰 */, 0xB3A7 /* 厂 */, 0x8596 /* 厲 */, 0xC0F7 /* 厉 */, 0x8598 /* 厴 */, 0xD8C9 /* 厣 */, 0x85A2 /* 參 */, 0xB2CE /* 参 */, 0x85B2 /* 叢 */, 0xB4D4 /* 丛 */, 0x85BC /* 吋 */, 0x