OLD | NEW |
1 /* | 1 /* |
2 ********************************************************************** | 2 ********************************************************************** |
3 * Copyright (C) 2005-2013, International Business Machines | 3 * Copyright (C) 2005-2013, International Business Machines |
4 * Corporation and others. All Rights Reserved. | 4 * Corporation and others. All Rights Reserved. |
5 ********************************************************************** | 5 ********************************************************************** |
6 */ | 6 */ |
7 | 7 |
8 #ifndef __CSRSBCS_H | 8 #ifndef __CSRSBCS_H |
9 #define __CSRSBCS_H | 9 #define __CSRSBCS_H |
10 | 10 |
(...skipping 32 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
43 void lookup(int32_t thisNgram); | 43 void lookup(int32_t thisNgram); |
44 | 44 |
45 virtual int32_t nextByte(InputText *det); | 45 virtual int32_t nextByte(InputText *det); |
46 virtual void parseCharacters(InputText *det); | 46 virtual void parseCharacters(InputText *det); |
47 | 47 |
48 public: | 48 public: |
49 int32_t parse(InputText *det); | 49 int32_t parse(InputText *det); |
50 | 50 |
51 }; | 51 }; |
52 | 52 |
53 #if !UCONFIG_NO_NON_HTML5_CONVERSION | |
54 class NGramParser_IBM420 : public NGramParser | 53 class NGramParser_IBM420 : public NGramParser |
55 { | 54 { |
56 private: | 55 private: |
57 int32_t alef; | 56 int32_t alef; |
58 int32_t isLamAlef(int32_t b); | 57 int32_t isLamAlef(int32_t b); |
59 int32_t nextByte(InputText *det); | 58 int32_t nextByte(InputText *det); |
60 void parseCharacters(InputText *det); | 59 void parseCharacters(InputText *det); |
61 | 60 |
62 public: | 61 public: |
63 NGramParser_IBM420(const int32_t *theNgramList, const uint8_t *theCharMap); | 62 NGramParser_IBM420(const int32_t *theNgramList, const uint8_t *theCharMap); |
64 }; | 63 }; |
65 #endif | |
66 | 64 |
67 | 65 |
68 class CharsetRecog_sbcs : public CharsetRecognizer | 66 class CharsetRecog_sbcs : public CharsetRecognizer |
69 { | 67 { |
70 public: | 68 public: |
71 CharsetRecog_sbcs(); | 69 CharsetRecog_sbcs(); |
72 virtual ~CharsetRecog_sbcs(); | 70 virtual ~CharsetRecog_sbcs(); |
73 virtual const char *getName() const = 0; | 71 virtual const char *getName() const = 0; |
74 virtual UBool match(InputText *det, CharsetMatch *results) const = 0; | 72 virtual UBool match(InputText *det, CharsetMatch *results) const = 0; |
75 virtual int32_t match_sbcs(InputText *det, const int32_t ngrams[], const uin
t8_t charMap[]) const; | 73 virtual int32_t match_sbcs(InputText *det, const int32_t ngrams[], const uin
t8_t charMap[]) const; |
(...skipping 148 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
224 public: | 222 public: |
225 virtual ~CharsetRecog_KOI8_R(); | 223 virtual ~CharsetRecog_KOI8_R(); |
226 | 224 |
227 const char *getName() const; | 225 const char *getName() const; |
228 | 226 |
229 const char *getLanguage() const; | 227 const char *getLanguage() const; |
230 | 228 |
231 virtual UBool match(InputText *det, CharsetMatch *results) const; | 229 virtual UBool match(InputText *det, CharsetMatch *results) const; |
232 }; | 230 }; |
233 | 231 |
234 #if !UCONFIG_NO_NON_HTML5_CONVERSION | |
235 class CharsetRecog_IBM424_he : public CharsetRecog_sbcs | 232 class CharsetRecog_IBM424_he : public CharsetRecog_sbcs |
236 { | 233 { |
237 public: | 234 public: |
238 virtual ~CharsetRecog_IBM424_he(); | 235 virtual ~CharsetRecog_IBM424_he(); |
239 | 236 |
240 const char *getLanguage() const; | 237 const char *getLanguage() const; |
241 }; | 238 }; |
242 | 239 |
243 class CharsetRecog_IBM424_he_rtl : public CharsetRecog_IBM424_he { | 240 class CharsetRecog_IBM424_he_rtl : public CharsetRecog_IBM424_he { |
244 public: | 241 public: |
(...skipping 31 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
276 virtual UBool match(InputText *det, CharsetMatch *results) const; | 273 virtual UBool match(InputText *det, CharsetMatch *results) const; |
277 }; | 274 }; |
278 | 275 |
279 class CharsetRecog_IBM420_ar_ltr : public CharsetRecog_IBM420_ar { | 276 class CharsetRecog_IBM420_ar_ltr : public CharsetRecog_IBM420_ar { |
280 virtual ~CharsetRecog_IBM420_ar_ltr(); | 277 virtual ~CharsetRecog_IBM420_ar_ltr(); |
281 | 278 |
282 const char *getName() const; | 279 const char *getName() const; |
283 | 280 |
284 virtual UBool match(InputText *det, CharsetMatch *results) const; | 281 virtual UBool match(InputText *det, CharsetMatch *results) const; |
285 }; | 282 }; |
286 #endif | |
287 | 283 |
288 U_NAMESPACE_END | 284 U_NAMESPACE_END |
289 | 285 |
290 #endif /* !UCONFIG_NO_CONVERSION */ | 286 #endif /* !UCONFIG_NO_CONVERSION */ |
291 #endif /* __CSRSBCS_H */ | 287 #endif /* __CSRSBCS_H */ |
OLD | NEW |