1 // Copyright 2014 PDFium Authors. All rights reserved. 2 // Use of this source code is governed by a BSD-style license that can be 3 // found in the LICENSE file. 4 5 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com 6 7 #include "core/include/fxcrt/fx_string.h" 8 #include "unicodenormalizationdata.h" 9 10 const FX_WCHAR* const g_UnicodeData_Normalization_Maps[5] = { 11 nullptr, 12 g_UnicodeData_Normalization_Map1, 13 g_UnicodeData_Normalization_Map2, 14 g_UnicodeData_Normalization_Map3, 15 g_UnicodeData_Normalization_Map4}; 16 17 FX_STRSIZE FX_Unicode_GetNormalization(FX_WCHAR wch, FX_WCHAR* pDst) { 18 wch = wch & 0xFFFF; 19 FX_WCHAR wFind = g_UnicodeData_Normalization[wch]; 20 if (!wFind) { 21 if (pDst) { 22 *pDst = wch; 23 } 24 return 1; 25 } 26 if (wFind >= 0x8000) { 27 wch = wFind - 0x8000; 28 wFind = 1; 29 } else { 30 wch = wFind & 0x0FFF; 31 wFind >>= 12; 32 } 33 const FX_WCHAR* pMap = g_UnicodeData_Normalization_Maps[wFind]; 34 if (pMap == g_UnicodeData_Normalization_Map4) { 35 pMap = g_UnicodeData_Normalization_Map4 + wch; 36 wFind = (FX_WCHAR)(*pMap++); 37 } else { 38 pMap += wch; 39 } 40 if (pDst) { 41 FX_WCHAR n = wFind; 42 while (n--) { 43 *pDst++ = *pMap++; 44 } 45 } 46 return (FX_STRSIZE)wFind; 47 } 48 FX_STRSIZE FX_WideString_GetNormalization(const CFX_WideStringC& wsSrc, 49 FX_WCHAR* pDst) { 50 FX_STRSIZE nCount = 0; 51 for (FX_STRSIZE len = 0; len < wsSrc.GetLength(); len++) { 52 FX_WCHAR wch = wsSrc.GetAt(len); 53 if (pDst) { 54 nCount += FX_Unicode_GetNormalization(wch, pDst + nCount); 55 } else { 56 nCount += FX_Unicode_GetNormalization(wch, pDst); 57 } 58 } 59 return nCount; 60 } 61 FX_STRSIZE FX_WideString_GetNormalization(const CFX_WideStringC& wsSrc, 62 CFX_WideString& wsDst) { 63 FX_STRSIZE nLen = FX_WideString_GetNormalization(wsSrc, (FX_WCHAR*)NULL); 64 if (!nLen) { 65 return 0; 66 } 67 FX_WCHAR* pBuf = wsDst.GetBuffer(nLen); 68 FX_WideString_GetNormalization(wsSrc, pBuf); 69 wsDst.ReleaseBuffer(nLen); 70 return nLen; 71 } 72
