icuSources/i18n/csr2022.h

   1 /*
   2  **********************************************************************
   3  *   Copyright (C) 2005-2012, 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 class CharsetMatch;
  20
  21 /**
  22  *  class CharsetRecog_2022  part of the ICU charset detection imlementation.
  23  *                           This is a superclass for the individual detectors for
  24  *                           each of the detectable members of the ISO 2022 family
  25  *                           of encodings.
  26  *
  27  *                           The separate classes are nested within this class.
  28  *
  29  * @internal
  30  */
  31 class CharsetRecog_2022 : public CharsetRecognizer
  32 {
  33
  34 public:
  35     virtual ~CharsetRecog_2022() = 0;
  36
  37 protected:
  38
  39     /**
  40      * Matching function shared among the 2022 detectors JP, CN and KR
  41      * Counts up the number of legal an unrecognized escape sequences in
  42      * the sample of text, and computes a score based on the total number &
  43      * the proportion that fit the encoding.
  44      *
  45      *
  46      * @param text the byte buffer containing text to analyse
  47      * @param textLen  the size of the text in the byte.
  48      * @param escapeSequences the byte escape sequences to test for.
  49      * @return match quality, in the range of 0-100.
  50      */
  51     int32_t match_2022(const uint8_t *text,
  52                        int32_t textLen,
  53                        const uint8_t escapeSequences[][5],
  54                        int32_t escapeSequences_length) const;
  55
  56 };
  57
  58 class CharsetRecog_2022JP :public CharsetRecog_2022
  59 {
  60 public:
  61     virtual ~CharsetRecog_2022JP();
  62
  63     const char *getName() const;
  64
  65     UBool match(InputText *textIn, CharsetMatch *results) const;
  66 };
  67
  68 class CharsetRecog_2022KR :public CharsetRecog_2022 {
  69 public:
  70     virtual ~CharsetRecog_2022KR();
  71
  72     const char *getName() const;
  73
  74     UBool match(InputText *textIn, CharsetMatch *results) const;
  75
  76 };
  77
  78 class CharsetRecog_2022CN :public CharsetRecog_2022
  79 {
  80 public:
  81     virtual ~CharsetRecog_2022CN();
  82
  83     const char* getName() const;
  84
  85     UBool match(InputText *textIn, CharsetMatch *results) const;
  86 };
  87
  88 U_NAMESPACE_END
  89
  90 #endif
  91 #endif /* __CSR2022_H */