projects / apple / icu.git / blame
| Commit | Line | Data |
|---|---|---|
| 73c04bcf A |
1 | /* |
| 2 | ********************************************************************** | |
| 3 | * Copyright (C) 2005-2006, International Business Machines | |
| 4 | * Corporation and others. All Rights Reserved. | |
| 5 | ********************************************************************** | |
| 6 | */ | |
| 7 | ||
| 8 | #ifndef __CSR2022_H | |
| 9 | #define __CSR2022_H | |
| 10 | ||
| 11 | #include "unicode/utypes.h" | |
| 12 | ||
| 13 | #if !UCONFIG_NO_CONVERSION | |
| 14 | ||
| 15 | #include "csrecog.h" | |
| 16 | ||
| 17 | U_NAMESPACE_BEGIN | |
| 18 | ||
| 19 | /** | |
| 20 | * class CharsetRecog_2022 part of the ICU charset detection imlementation. | |
| 21 | * This is a superclass for the individual detectors for | |
| 22 | * each of the detectable members of the ISO 2022 family | |
| 23 | * of encodings. | |
| 24 | * | |
| 25 | * The separate classes are nested within this class. | |
| 26 | * | |
| 27 | * @internal | |
| 28 | */ | |
| 29 | class CharsetRecog_2022 : public CharsetRecognizer | |
| 30 | { | |
| 31 | ||
| 32 | public: | |
| 33 | virtual ~CharsetRecog_2022() = 0; | |
| 34 | ||
| 35 | protected: | |
| 36 | ||
| 37 | /** | |
| 38 | * Matching function shared among the 2022 detectors JP, CN and KR | |
| 39 | * Counts up the number of legal an unrecognized escape sequences in | |
| 40 | * the sample of text, and computes a score based on the total number & | |
| 41 | * the proportion that fit the encoding. | |
| 42 | * | |
| 43 | * | |
| 44 | * @param text the byte buffer containing text to analyse | |
| 45 | * @param textLen the size of the text in the byte. | |
| 46 | * @param escapeSequences the byte escape sequences to test for. | |
| 47 | * @return match quality, in the range of 0-100. | |
| 48 | */ | |
| 49 | int32_t match_2022(const uint8_t *text, int32_t textLen, const uint8_t escapeSequences[][5], int32_t escapeSequences_length); | |
| 50 | ||
| 51 | }; | |
| 52 | ||
| 53 | class CharsetRecog_2022JP :public CharsetRecog_2022 | |
| 54 | { | |
| 55 | public: | |
| 56 | virtual ~CharsetRecog_2022JP() {} | |
| 57 | ||
| 58 | const char *getName() const; | |
| 59 | ||
| 60 | int32_t match(InputText *textIn); | |
| 61 | }; | |
| 62 | ||
| 63 | class CharsetRecog_2022KR :public CharsetRecog_2022 { | |
| 64 | public: | |
| 65 | virtual ~CharsetRecog_2022KR() {} | |
| 66 | ||
| 67 | const char *getName() const; | |
| 68 | ||
| 69 | int32_t match(InputText *textIn); | |
| 70 | ||
| 71 | }; | |
| 72 | ||
| 73 | class CharsetRecog_2022CN :public CharsetRecog_2022 | |
| 74 | { | |
| 75 | public: | |
| 76 | virtual ~CharsetRecog_2022CN() {} | |
| 77 | ||
| 78 | const char* getName() const; | |
| 79 | ||
| 80 | int32_t match(InputText *textIn); | |
| 81 | }; | |
| 82 | ||
| 83 | U_NAMESPACE_END | |
| 84 | ||
| 85 | #endif | |
| 86 | #endif /* __CSR2022_H */ |