Home | History | Annotate | Download | only in fpdftext 
      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 "../../include/fpdftext/fpdf_text.h"
      8 extern const FX_WCHAR g_UnicodeData_Normalization[65536];
      9 extern const FX_WCHAR g_UnicodeData_Normalization_Map1[5376];
     10 extern const FX_WCHAR g_UnicodeData_Normalization_Map2[1734];
     11 extern const FX_WCHAR g_UnicodeData_Normalization_Map3[1164];
     12 extern const FX_WCHAR g_UnicodeData_Normalization_Map4[488];
     13 FX_LPCWSTR g_UnicodeData_Normalization_Maps[5] = {
     14     NULL,
     15     g_UnicodeData_Normalization_Map1,
     16     g_UnicodeData_Normalization_Map2,
     17     g_UnicodeData_Normalization_Map3,
     18     g_UnicodeData_Normalization_Map4
     19 };
     20 FX_STRSIZE FX_Unicode_GetNormalization(FX_WCHAR wch, FX_LPWSTR pDst)
     21 {
     22     wch = wch & 0xFFFF;
     23     FX_WCHAR wFind = g_UnicodeData_Normalization[wch];
     24     if (!wFind) {
     25         if (pDst) {
     26             *pDst = wch;
     27         }
     28         return 1;
     29     }
     30     if(wFind >= 0x8000) {
     31         wch = wFind - 0x8000;
     32         wFind = 1;
     33     } else {
     34         wch = wFind & 0x0FFF;
     35         wFind >>= 12;
     36     }
     37     FX_LPCWSTR pMap = g_UnicodeData_Normalization_Maps[wFind];
     38     if (pMap == g_UnicodeData_Normalization_Map4) {
     39         pMap = g_UnicodeData_Normalization_Map4 + wch;
     40         wFind = (FX_WCHAR)(*pMap ++);
     41     } else {
     42         pMap += wch;
     43     }
     44     if (pDst) {
     45         FX_WCHAR n = wFind;
     46         while (n --) {
     47             *pDst ++ = *pMap ++;
     48         }
     49     }
     50     return (FX_STRSIZE)wFind;
     51 }
     52 FX_STRSIZE FX_WideString_GetNormalization(FX_WSTR wsSrc, FX_LPWSTR pDst)
     53 {
     54     FX_STRSIZE nCount = 0;
     55     for (FX_STRSIZE len = 0; len < wsSrc.GetLength(); len ++) {
     56         FX_WCHAR wch = wsSrc.GetAt(len);
     57         if(pDst) {
     58             nCount += FX_Unicode_GetNormalization(wch, pDst + nCount);
     59         } else {
     60             nCount += FX_Unicode_GetNormalization(wch, pDst);
     61         }
     62     }
     63     return nCount;
     64 }
     65 FX_STRSIZE FX_WideString_GetNormalization(FX_WSTR wsSrc, CFX_WideString &wsDst)
     66 {
     67     FX_STRSIZE nLen = FX_WideString_GetNormalization(wsSrc, (FX_LPWSTR)NULL);
     68     if (!nLen) {
     69         return 0;
     70     }
     71     FX_LPWSTR pBuf = wsDst.GetBuffer(nLen);
     72     FX_WideString_GetNormalization(wsSrc, pBuf);
     73     wsDst.ReleaseBuffer(nLen);
     74     return nLen;
     75 }
     76