| OLD | NEW |
| 1 /* | 1 /* |
| 2 ********************************************************************** | 2 ********************************************************************** |
| 3 * Copyright (C) 2005-2013, International Business Machines | 3 * Copyright (C) 2005-2015, 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 |
| 11 #include "unicode/uobject.h" | 11 #include "unicode/uobject.h" |
| 12 | 12 |
| 13 #if !UCONFIG_NO_CONVERSION | 13 #if !UCONFIG_NO_CONVERSION |
| (...skipping 12 matching lines...) Expand all Loading... |
| 26 int32_t hitCount; | 26 int32_t hitCount; |
| 27 | 27 |
| 28 protected: | 28 protected: |
| 29 int32_t byteIndex; | 29 int32_t byteIndex; |
| 30 const uint8_t *charMap; | 30 const uint8_t *charMap; |
| 31 | 31 |
| 32 void addByte(int32_t b); | 32 void addByte(int32_t b); |
| 33 | 33 |
| 34 public: | 34 public: |
| 35 NGramParser(const int32_t *theNgramList, const uint8_t *theCharMap); | 35 NGramParser(const int32_t *theNgramList, const uint8_t *theCharMap); |
| 36 virtual ~NGramParser(); |
| 36 | 37 |
| 37 private: | 38 private: |
| 38 /* | 39 /* |
| 39 * Binary search for value in table, which must have exactly 64 entries. | 40 * Binary search for value in table, which must have exactly 64 entries. |
| 40 */ | 41 */ |
| 41 int32_t search(const int32_t *table, int32_t value); | 42 int32_t search(const int32_t *table, int32_t value); |
| 42 | 43 |
| 43 void lookup(int32_t thisNgram); | 44 void lookup(int32_t thisNgram); |
| 44 | 45 |
| 45 virtual int32_t nextByte(InputText *det); | 46 virtual int32_t nextByte(InputText *det); |
| 46 virtual void parseCharacters(InputText *det); | 47 virtual void parseCharacters(InputText *det); |
| 47 | 48 |
| 48 public: | 49 public: |
| 49 int32_t parse(InputText *det); | 50 int32_t parse(InputText *det); |
| 50 | 51 |
| 51 }; | 52 }; |
| 52 | 53 |
| 53 #if !UCONFIG_NO_NON_HTML5_CONVERSION | 54 #if !UCONFIG_ONLY_HTML_CONVERSION |
| 54 class NGramParser_IBM420 : public NGramParser | 55 class NGramParser_IBM420 : public NGramParser |
| 55 { | 56 { |
| 56 private: | |
| 57 int32_t alef; | |
| 58 int32_t isLamAlef(int32_t b); | |
| 59 int32_t nextByte(InputText *det); | |
| 60 void parseCharacters(InputText *det); | |
| 61 | |
| 62 public: | 57 public: |
| 63 NGramParser_IBM420(const int32_t *theNgramList, const uint8_t *theCharMap); | 58 NGramParser_IBM420(const int32_t *theNgramList, const uint8_t *theCharMap); |
| 59 ~NGramParser_IBM420(); |
| 60 |
| 61 private: |
| 62 int32_t alef; |
| 63 int32_t isLamAlef(int32_t b); |
| 64 int32_t nextByte(InputText *det); |
| 65 void parseCharacters(InputText *det); |
| 64 }; | 66 }; |
| 65 #endif | 67 #endif |
| 66 | 68 |
| 67 | 69 |
| 68 class CharsetRecog_sbcs : public CharsetRecognizer | 70 class CharsetRecog_sbcs : public CharsetRecognizer |
| 69 { | 71 { |
| 70 public: | 72 public: |
| 71 CharsetRecog_sbcs(); | 73 CharsetRecog_sbcs(); |
| 72 virtual ~CharsetRecog_sbcs(); | 74 virtual ~CharsetRecog_sbcs(); |
| 73 virtual const char *getName() const = 0; | 75 virtual const char *getName() const = 0; |
| (...skipping 150 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 224 public: | 226 public: |
| 225 virtual ~CharsetRecog_KOI8_R(); | 227 virtual ~CharsetRecog_KOI8_R(); |
| 226 | 228 |
| 227 const char *getName() const; | 229 const char *getName() const; |
| 228 | 230 |
| 229 const char *getLanguage() const; | 231 const char *getLanguage() const; |
| 230 | 232 |
| 231 virtual UBool match(InputText *det, CharsetMatch *results) const; | 233 virtual UBool match(InputText *det, CharsetMatch *results) const; |
| 232 }; | 234 }; |
| 233 | 235 |
| 234 #if !UCONFIG_NO_NON_HTML5_CONVERSION | 236 #if !UCONFIG_ONLY_HTML_CONVERSION |
| 235 class CharsetRecog_IBM424_he : public CharsetRecog_sbcs | 237 class CharsetRecog_IBM424_he : public CharsetRecog_sbcs |
| 236 { | 238 { |
| 237 public: | 239 public: |
| 238 virtual ~CharsetRecog_IBM424_he(); | 240 virtual ~CharsetRecog_IBM424_he(); |
| 239 | 241 |
| 240 const char *getLanguage() const; | 242 const char *getLanguage() const; |
| 241 }; | 243 }; |
| 242 | 244 |
| 243 class CharsetRecog_IBM424_he_rtl : public CharsetRecog_IBM424_he { | 245 class CharsetRecog_IBM424_he_rtl : public CharsetRecog_IBM424_he { |
| 244 public: | 246 public: |
| (...skipping 37 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 282 const char *getName() const; | 284 const char *getName() const; |
| 283 | 285 |
| 284 virtual UBool match(InputText *det, CharsetMatch *results) const; | 286 virtual UBool match(InputText *det, CharsetMatch *results) const; |
| 285 }; | 287 }; |
| 286 #endif | 288 #endif |
| 287 | 289 |
| 288 U_NAMESPACE_END | 290 U_NAMESPACE_END |
| 289 | 291 |
| 290 #endif /* !UCONFIG_NO_CONVERSION */ | 292 #endif /* !UCONFIG_NO_CONVERSION */ |
| 291 #endif /* __CSRSBCS_H */ | 293 #endif /* __CSRSBCS_H */ |
| OLD | NEW |