| 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 |
| 53 class NGramParser_IBM420 : public NGramParser | 54 class NGramParser_IBM420 : public NGramParser |
| 54 { | 55 { |
| 55 private: | 56 private: |
| 56 int32_t alef; | 57 int32_t alef; |
| 57 int32_t isLamAlef(int32_t b); | 58 int32_t isLamAlef(int32_t b); |
| 58 int32_t nextByte(InputText *det); | 59 int32_t nextByte(InputText *det); |
| 59 void parseCharacters(InputText *det); | 60 void parseCharacters(InputText *det); |
| 60 | 61 |
| 61 public: | 62 public: |
| 62 NGramParser_IBM420(const int32_t *theNgramList, const uint8_t *theCharMap); | 63 NGramParser_IBM420(const int32_t *theNgramList, const uint8_t *theCharMap); |
| 63 }; | 64 }; |
| 65 #endif |
| 64 | 66 |
| 65 | 67 |
| 66 class CharsetRecog_sbcs : public CharsetRecognizer | 68 class CharsetRecog_sbcs : public CharsetRecognizer |
| 67 { | 69 { |
| 68 public: | 70 public: |
| 69 CharsetRecog_sbcs(); | 71 CharsetRecog_sbcs(); |
| 70 virtual ~CharsetRecog_sbcs(); | 72 virtual ~CharsetRecog_sbcs(); |
| 71 virtual const char *getName() const = 0; | 73 virtual const char *getName() const = 0; |
| 72 virtual UBool match(InputText *det, CharsetMatch *results) const = 0; | 74 virtual UBool match(InputText *det, CharsetMatch *results) const = 0; |
| 73 virtual int32_t match_sbcs(InputText *det, const int32_t ngrams[], const uin
t8_t charMap[]) const; | 75 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... |
| 222 public: | 224 public: |
| 223 virtual ~CharsetRecog_KOI8_R(); | 225 virtual ~CharsetRecog_KOI8_R(); |
| 224 | 226 |
| 225 const char *getName() const; | 227 const char *getName() const; |
| 226 | 228 |
| 227 const char *getLanguage() const; | 229 const char *getLanguage() const; |
| 228 | 230 |
| 229 virtual UBool match(InputText *det, CharsetMatch *results) const; | 231 virtual UBool match(InputText *det, CharsetMatch *results) const; |
| 230 }; | 232 }; |
| 231 | 233 |
| 234 #if !UCONFIG_NO_NON_HTML5_CONVERSION |
| 232 class CharsetRecog_IBM424_he : public CharsetRecog_sbcs | 235 class CharsetRecog_IBM424_he : public CharsetRecog_sbcs |
| 233 { | 236 { |
| 234 public: | 237 public: |
| 235 virtual ~CharsetRecog_IBM424_he(); | 238 virtual ~CharsetRecog_IBM424_he(); |
| 236 | 239 |
| 237 const char *getLanguage() const; | 240 const char *getLanguage() const; |
| 238 }; | 241 }; |
| 239 | 242 |
| 240 class CharsetRecog_IBM424_he_rtl : public CharsetRecog_IBM424_he { | 243 class CharsetRecog_IBM424_he_rtl : public CharsetRecog_IBM424_he { |
| 241 public: | 244 public: |
| (...skipping 31 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 273 virtual UBool match(InputText *det, CharsetMatch *results) const; | 276 virtual UBool match(InputText *det, CharsetMatch *results) const; |
| 274 }; | 277 }; |
| 275 | 278 |
| 276 class CharsetRecog_IBM420_ar_ltr : public CharsetRecog_IBM420_ar { | 279 class CharsetRecog_IBM420_ar_ltr : public CharsetRecog_IBM420_ar { |
| 277 virtual ~CharsetRecog_IBM420_ar_ltr(); | 280 virtual ~CharsetRecog_IBM420_ar_ltr(); |
| 278 | 281 |
| 279 const char *getName() const; | 282 const char *getName() const; |
| 280 | 283 |
| 281 virtual UBool match(InputText *det, CharsetMatch *results) const; | 284 virtual UBool match(InputText *det, CharsetMatch *results) const; |
| 282 }; | 285 }; |
| 286 #endif |
| 283 | 287 |
| 284 U_NAMESPACE_END | 288 U_NAMESPACE_END |
| 285 | 289 |
| 286 #endif /* !UCONFIG_NO_CONVERSION */ | 290 #endif /* !UCONFIG_NO_CONVERSION */ |
| 287 #endif /* __CSRSBCS_H */ | 291 #endif /* __CSRSBCS_H */ |
| OLD | NEW |