OLD | NEW |
1 // Copyright 2014 PDFium Authors. All rights reserved. | 1 // Copyright 2014 PDFium Authors. All rights reserved. |
2 // Use of this source code is governed by a BSD-style license that can be | 2 // Use of this source code is governed by a BSD-style license that can be |
3 // found in the LICENSE file. | 3 // found in the LICENSE file. |
4 | 4 |
5 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com | 5 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com |
6 | 6 |
7 #include "../../include/fpdftext/fpdf_text.h" | 7 #include "../../include/fpdftext/fpdf_text.h" |
8 extern const FX_WCHAR g_UnicodeData_Normalization[65536]; | 8 extern const FX_WCHAR g_UnicodeData_Normalization[65536]; |
9 extern const FX_WCHAR g_UnicodeData_Normalization_Map1[5376]; | 9 extern const FX_WCHAR g_UnicodeData_Normalization_Map1[5376]; |
10 extern const FX_WCHAR g_UnicodeData_Normalization_Map2[1734]; | 10 extern const FX_WCHAR g_UnicodeData_Normalization_Map2[1734]; |
11 extern const FX_WCHAR g_UnicodeData_Normalization_Map3[1164]; | 11 extern const FX_WCHAR g_UnicodeData_Normalization_Map3[1164]; |
12 extern const FX_WCHAR g_UnicodeData_Normalization_Map4[488]; | 12 extern const FX_WCHAR g_UnicodeData_Normalization_Map4[488]; |
13 FX_LPCWSTR g_UnicodeData_Normalization_Maps[5] = { | 13 FX_LPCWSTR g_UnicodeData_Normalization_Maps[5] = { |
14 NULL, | 14 NULL, |
15 g_UnicodeData_Normalization_Map1, | 15 g_UnicodeData_Normalization_Map1, |
16 g_UnicodeData_Normalization_Map2, | 16 g_UnicodeData_Normalization_Map2, |
17 g_UnicodeData_Normalization_Map3, | 17 g_UnicodeData_Normalization_Map3, |
18 g_UnicodeData_Normalization_Map4 | 18 g_UnicodeData_Normalization_Map4 |
19 }; | 19 }; |
20 FX_STRSIZE FX_Unicode_GetNormalization(FX_WCHAR wch, FX_LPWSTR pDst) | 20 FX_STRSIZE FX_Unicode_GetNormalization(FX_WCHAR wch, FX_LPWSTR pDst) { |
21 { | 21 wch = wch & 0xFFFF; |
22 wch = wch & 0xFFFF; | 22 FX_WCHAR wFind = g_UnicodeData_Normalization[wch]; |
23 FX_WCHAR wFind = g_UnicodeData_Normalization[wch]; | 23 if (!wFind) { |
24 if (!wFind) { | 24 if (pDst) { |
25 if (pDst) { | 25 *pDst = wch; |
26 *pDst = wch; | |
27 } | |
28 return 1; | |
29 } | 26 } |
30 if(wFind >= 0x8000) { | 27 return 1; |
31 wch = wFind - 0x8000; | 28 } |
32 wFind = 1; | 29 if (wFind >= 0x8000) { |
| 30 wch = wFind - 0x8000; |
| 31 wFind = 1; |
| 32 } else { |
| 33 wch = wFind & 0x0FFF; |
| 34 wFind >>= 12; |
| 35 } |
| 36 FX_LPCWSTR pMap = g_UnicodeData_Normalization_Maps[wFind]; |
| 37 if (pMap == g_UnicodeData_Normalization_Map4) { |
| 38 pMap = g_UnicodeData_Normalization_Map4 + wch; |
| 39 wFind = (FX_WCHAR)(*pMap++); |
| 40 } else { |
| 41 pMap += wch; |
| 42 } |
| 43 if (pDst) { |
| 44 FX_WCHAR n = wFind; |
| 45 while (n--) { |
| 46 *pDst++ = *pMap++; |
| 47 } |
| 48 } |
| 49 return (FX_STRSIZE)wFind; |
| 50 } |
| 51 FX_STRSIZE FX_WideString_GetNormalization(FX_WSTR wsSrc, FX_LPWSTR pDst) { |
| 52 FX_STRSIZE nCount = 0; |
| 53 for (FX_STRSIZE len = 0; len < wsSrc.GetLength(); len++) { |
| 54 FX_WCHAR wch = wsSrc.GetAt(len); |
| 55 if (pDst) { |
| 56 nCount += FX_Unicode_GetNormalization(wch, pDst + nCount); |
33 } else { | 57 } else { |
34 wch = wFind & 0x0FFF; | 58 nCount += FX_Unicode_GetNormalization(wch, pDst); |
35 wFind >>= 12; | |
36 } | 59 } |
37 FX_LPCWSTR pMap = g_UnicodeData_Normalization_Maps[wFind]; | 60 } |
38 if (pMap == g_UnicodeData_Normalization_Map4) { | 61 return nCount; |
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 } | 62 } |
52 FX_STRSIZE FX_WideString_GetNormalization(FX_WSTR wsSrc, FX_LPWSTR pDst) | 63 FX_STRSIZE FX_WideString_GetNormalization(FX_WSTR wsSrc, |
53 { | 64 CFX_WideString& wsDst) { |
54 FX_STRSIZE nCount = 0; | 65 FX_STRSIZE nLen = FX_WideString_GetNormalization(wsSrc, (FX_LPWSTR)NULL); |
55 for (FX_STRSIZE len = 0; len < wsSrc.GetLength(); len ++) { | 66 if (!nLen) { |
56 FX_WCHAR wch = wsSrc.GetAt(len); | 67 return 0; |
57 if(pDst) { | 68 } |
58 nCount += FX_Unicode_GetNormalization(wch, pDst + nCount); | 69 FX_LPWSTR pBuf = wsDst.GetBuffer(nLen); |
59 } else { | 70 FX_WideString_GetNormalization(wsSrc, pBuf); |
60 nCount += FX_Unicode_GetNormalization(wch, pDst); | 71 wsDst.ReleaseBuffer(nLen); |
61 } | 72 return nLen; |
62 } | |
63 return nCount; | |
64 } | 73 } |
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 } | |
OLD | NEW |