icuSources/i18n/csr2022.cpp

   1 /*
   2  **********************************************************************
   3  *   Copyright (C) 2005-2006, International Business Machines
   4  *   Corporation and others.  All Rights Reserved.
   5  **********************************************************************
   6  */
   7
   8 #include "unicode/utypes.h"
   9
  10 #if !UCONFIG_NO_CONVERSION
  11
  12 #include "cstring.h"
  13
  14 #include "csr2022.h"
  15
  16 U_NAMESPACE_BEGIN
  17
  18 #define ARRAY_SIZE(array) (sizeof array / sizeof array[0])
  19
  20 /**
  21  * Matching function shared among the 2022 detectors JP, CN and KR
  22  * Counts up the number of legal an unrecognized escape sequences in
  23  * the sample of text, and computes a score based on the total number &
  24  * the proportion that fit the encoding.
  25  *
  26  *
  27  * @param text the byte buffer containing text to analyse
  28  * @param textLen  the size of the text in the byte.
  29  * @param escapeSequences the byte escape sequences to test for.
  30  * @return match quality, in the range of 0-100.
  31  */
  32 int32_t CharsetRecog_2022::match_2022(const uint8_t *text, int32_t textLen, const uint8_t escapeSequences[][5], int32_t escapeSequences_length)
  33 {
  34     int32_t i, j;
  35     int32_t escN;
  36     int32_t hits   = 0;
  37     int32_t misses = 0;
  38     int32_t shifts = 0;
  39     int32_t quality;
  40
  41     i = 0;
  42     while(i < textLen) {
  43         if(text[i] == 0x1B) {
  44             escN = 0;
  45             while(escN < escapeSequences_length) {
  46                 int32_t seq_length = uprv_strlen((const char *) escapeSequences[escN]);
  47                 const uint8_t *seq = escapeSequences[escN];
  48
  49                 j = 1;
  50                 while(j < seq_length) {
  51                     if(seq[j] != text[i+j]) {
  52                         goto checkEscapes;
  53                     }
  54
  55                     j += 1;
  56                 }
  57
  58                 hits += 1;
  59                 i += seq_length-1;
  60                 goto scanInput;
  61 checkEscapes:
  62                 escN += 1;
  63             }
  64
  65             misses += 1;
  66         }
  67
  68         if( text[i]== 0x0e || text[i] == 0x0f){
  69             shifts += 1;
  70         }
  71
  72 scanInput:
  73         i += 1;
  74     }
  75
  76     if (hits == 0) {
  77         return 0;
  78     }
  79
  80     //
  81     // Initial quality is based on relative proportion of recongized vs.
  82     //   unrecognized escape sequences.
  83     //   All good:  quality = 100;
  84     //   half or less good: quality = 0;
  85     //   linear inbetween.
  86     quality = (100*hits - 100*misses) / (hits + misses);
  87
  88     // Back off quality if there were too few escape sequences seen.
  89     //   Include shifts in this computation, so that KR does not get penalized
  90     //   for having only a single Escape sequence, but many shifts.
  91     if (hits+shifts < 5) {
  92         quality -= (5-(hits+shifts))*10;
  93     }
  94
  95     if (quality < 0) {
  96         quality = 0;
  97     }
  98
  99     return quality;
 100 }
 101
 102
 103 static const uint8_t escapeSequences_2022JP[][5] = {
 104     {0x1b, 0x24, 0x28, 0x43, 0x00},   // KS X 1001:1992
 105     {0x1b, 0x24, 0x28, 0x44, 0x00},   // JIS X 212-1990
 106     {0x1b, 0x24, 0x40, 0x00, 0x00},   // JIS C 6226-1978
 107     {0x1b, 0x24, 0x41, 0x00, 0x00},   // GB 2312-80
 108     {0x1b, 0x24, 0x42, 0x00, 0x00},   // JIS X 208-1983
 109     {0x1b, 0x26, 0x40, 0x00, 0x00},   // JIS X 208 1990, 1997
 110     {0x1b, 0x28, 0x42, 0x00, 0x00},   // ASCII
 111     {0x1b, 0x28, 0x48, 0x00, 0x00},   // JIS-Roman
 112     {0x1b, 0x28, 0x49, 0x00, 0x00},   // Half-width katakana
 113     {0x1b, 0x28, 0x4a, 0x00, 0x00},   // JIS-Roman
 114     {0x1b, 0x2e, 0x41, 0x00, 0x00},   // ISO 8859-1
 115     {0x1b, 0x2e, 0x46, 0x00, 0x00}    // ISO 8859-7
 116 };
 117
 118 static const uint8_t escapeSequences_2022KR[][5] = {
 119     {0x1b, 0x24, 0x29, 0x43, 0x00}
 120 };
 121
 122 static const uint8_t escapeSequences_2022CN[][5] = {
 123     {0x1b, 0x24, 0x29, 0x41, 0x00},   // GB 2312-80
 124     {0x1b, 0x24, 0x29, 0x47, 0x00},   // CNS 11643-1992 Plane 1
 125     {0x1b, 0x24, 0x2A, 0x48, 0x00},   // CNS 11643-1992 Plane 2
 126     {0x1b, 0x24, 0x29, 0x45, 0x00},   // ISO-IR-165
 127     {0x1b, 0x24, 0x2B, 0x49, 0x00},   // CNS 11643-1992 Plane 3
 128     {0x1b, 0x24, 0x2B, 0x4A, 0x00},   // CNS 11643-1992 Plane 4
 129     {0x1b, 0x24, 0x2B, 0x4B, 0x00},   // CNS 11643-1992 Plane 5
 130     {0x1b, 0x24, 0x2B, 0x4C, 0x00},   // CNS 11643-1992 Plane 6
 131     {0x1b, 0x24, 0x2B, 0x4D, 0x00},   // CNS 11643-1992 Plane 7
 132     {0x1b, 0x4e, 0x00, 0x00, 0x00},   // SS2
 133     {0x1b, 0x4f, 0x00, 0x00, 0x00},   // SS3
 134 };
 135
 136 const char *CharsetRecog_2022JP::getName() const
 137 {
 138     return "ISO-2022-JP";
 139 }
 140
 141 int32_t CharsetRecog_2022JP::match(InputText *textIn)
 142 {
 143     return match_2022(textIn->fInputBytes, textIn->fInputLen, escapeSequences_2022JP, ARRAY_SIZE(escapeSequences_2022JP));
 144 }
 145
 146 const char *CharsetRecog_2022KR::getName() const
 147 {
 148     return "ISO-2022-KR";
 149 }
 150
 151 int32_t CharsetRecog_2022KR::match(InputText *textIn)
 152 {
 153     return match_2022(textIn->fInputBytes, textIn->fInputLen, escapeSequences_2022KR, ARRAY_SIZE(escapeSequences_2022KR));
 154 }
 155
 156 const char *CharsetRecog_2022CN::getName() const
 157 {
 158     return "ISO-2022-CN";
 159 }
 160
 161 int32_t CharsetRecog_2022CN::match(InputText *textIn)
 162 {
 163     return match_2022(textIn->fInputBytes, textIn->fInputLen, escapeSequences_2022CN, ARRAY_SIZE(escapeSequences_2022CN));
 164 }
 165
 166 CharsetRecog_2022::~CharsetRecog_2022()
 167 {
 168     // nothing to do
 169 }
 170
 171 U_NAMESPACE_END
 172 #endif