icuSources/i18n/csr2022.cpp

   1 /*
   2  **********************************************************************
   3  *   Copyright (C) 2005-2012, 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 #include "csmatch.h"
  16
  17 U_NAMESPACE_BEGIN
  18
  19 #define ARRAY_SIZE(array) (sizeof array / sizeof array[0])
  20
  21 /**
  22  * Matching function shared among the 2022 detectors JP, CN and KR
  23  * Counts up the number of legal and unrecognized escape sequences in
  24  * the sample of text, and computes a score based on the total number &
  25  * the proportion that fit the encoding.
  26  *
  27  *
  28  * @param text the byte buffer containing text to analyse
  29  * @param textLen  the size of the text in the byte.
  30  * @param escapeSequences the byte escape sequences to test for.
  31  * @return match quality, in the range of 0-100.
  32  */
  33 int32_t CharsetRecog_2022::match_2022(const uint8_t *text, int32_t textLen, const uint8_t escapeSequences[][5], int32_t escapeSequences_length) const
  34 {
  35     int32_t i, j;
  36     int32_t escN;
  37     int32_t hits   = 0;
  38     int32_t misses = 0;
  39     int32_t shifts = 0;
  40     int32_t quality;
  41
  42     i = 0;
  43     while(i < textLen) {
  44         if(text[i] == 0x1B) {
  45             escN = 0;
  46             while(escN < escapeSequences_length) {
  47                 const uint8_t *seq = escapeSequences[escN];
  48                 int32_t seq_length = (int32_t)uprv_strlen((const char *) seq);
  49
  50                 if (textLen-i >= seq_length) {
  51                     j = 1;
  52                     while(j < seq_length) {
  53                         if(seq[j] != text[i+j]) {
  54                             goto checkEscapes;
  55                         }
  56
  57                         j += 1;
  58                     }
  59
  60                     hits += 1;
  61                     i += seq_length-1;
  62                     goto scanInput;
  63                 }
  64                 // else we ran out of string to compare this time.
  65 checkEscapes:
  66                 escN += 1;
  67             }
  68
  69             misses += 1;
  70         }
  71
  72         if( text[i]== 0x0e || text[i] == 0x0f){
  73             shifts += 1;
  74         }
  75
  76 scanInput:
  77         i += 1;
  78     }
  79
  80     if (hits == 0) {
  81         return 0;
  82     }
  83
  84     //
  85     // Initial quality is based on relative proportion of recongized vs.
  86     //   unrecognized escape sequences.
  87     //   All good:  quality = 100;
  88     //   half or less good: quality = 0;
  89     //   linear inbetween.
  90     quality = (100*hits - 100*misses) / (hits + misses);
  91
  92     // Back off quality if there were too few escape sequences seen.
  93     //   Include shifts in this computation, so that KR does not get penalized
  94     //   for having only a single Escape sequence, but many shifts.
  95     if (hits+shifts < 5) {
  96         quality -= (5-(hits+shifts))*10;
  97     }
  98
  99     if (quality < 0) {
 100         quality = 0;
 101     }
 102
 103     return quality;
 104 }
 105
 106
 107 static const uint8_t escapeSequences_2022JP[][5] = {
 108     {0x1b, 0x24, 0x28, 0x43, 0x00},   // KS X 1001:1992
 109     {0x1b, 0x24, 0x28, 0x44, 0x00},   // JIS X 212-1990
 110     {0x1b, 0x24, 0x40, 0x00, 0x00},   // JIS C 6226-1978
 111     {0x1b, 0x24, 0x41, 0x00, 0x00},   // GB 2312-80
 112     {0x1b, 0x24, 0x42, 0x00, 0x00},   // JIS X 208-1983
 113     {0x1b, 0x26, 0x40, 0x00, 0x00},   // JIS X 208 1990, 1997
 114     {0x1b, 0x28, 0x42, 0x00, 0x00},   // ASCII
 115     {0x1b, 0x28, 0x48, 0x00, 0x00},   // JIS-Roman
 116     {0x1b, 0x28, 0x49, 0x00, 0x00},   // Half-width katakana
 117     {0x1b, 0x28, 0x4a, 0x00, 0x00},   // JIS-Roman
 118     {0x1b, 0x2e, 0x41, 0x00, 0x00},   // ISO 8859-1
 119     {0x1b, 0x2e, 0x46, 0x00, 0x00}    // ISO 8859-7
 120 };
 121
 122 static const uint8_t escapeSequences_2022KR[][5] = {
 123     {0x1b, 0x24, 0x29, 0x43, 0x00}
 124 };
 125
 126 static const uint8_t escapeSequences_2022CN[][5] = {
 127     {0x1b, 0x24, 0x29, 0x41, 0x00},   // GB 2312-80
 128     {0x1b, 0x24, 0x29, 0x47, 0x00},   // CNS 11643-1992 Plane 1
 129     {0x1b, 0x24, 0x2A, 0x48, 0x00},   // CNS 11643-1992 Plane 2
 130     {0x1b, 0x24, 0x29, 0x45, 0x00},   // ISO-IR-165
 131     {0x1b, 0x24, 0x2B, 0x49, 0x00},   // CNS 11643-1992 Plane 3
 132     {0x1b, 0x24, 0x2B, 0x4A, 0x00},   // CNS 11643-1992 Plane 4
 133     {0x1b, 0x24, 0x2B, 0x4B, 0x00},   // CNS 11643-1992 Plane 5
 134     {0x1b, 0x24, 0x2B, 0x4C, 0x00},   // CNS 11643-1992 Plane 6
 135     {0x1b, 0x24, 0x2B, 0x4D, 0x00},   // CNS 11643-1992 Plane 7
 136     {0x1b, 0x4e, 0x00, 0x00, 0x00},   // SS2
 137     {0x1b, 0x4f, 0x00, 0x00, 0x00},   // SS3
 138 };
 139
 140 CharsetRecog_2022JP::~CharsetRecog_2022JP() {}
 141
 142 const char *CharsetRecog_2022JP::getName() const {
 143     return "ISO-2022-JP";
 144 }
 145
 146 UBool CharsetRecog_2022JP::match(InputText *textIn, CharsetMatch *results) const {
 147     int32_t confidence = match_2022(textIn->fInputBytes,
 148                                     textIn->fInputLen,
 149                                     escapeSequences_2022JP,
 150                                     ARRAY_SIZE(escapeSequences_2022JP));
 151     results->set(textIn, this, confidence);
 152     return (confidence > 0);
 153 }
 154
 155 CharsetRecog_2022KR::~CharsetRecog_2022KR() {}
 156
 157 const char *CharsetRecog_2022KR::getName() const {
 158     return "ISO-2022-KR";
 159 }
 160
 161 UBool CharsetRecog_2022KR::match(InputText *textIn, CharsetMatch *results) const {
 162     int32_t confidence = match_2022(textIn->fInputBytes,
 163                                     textIn->fInputLen,
 164                                     escapeSequences_2022KR,
 165                                     ARRAY_SIZE(escapeSequences_2022KR));
 166     results->set(textIn, this, confidence);
 167     return (confidence > 0);
 168 }
 169
 170 CharsetRecog_2022CN::~CharsetRecog_2022CN() {}
 171
 172 const char *CharsetRecog_2022CN::getName() const {
 173     return "ISO-2022-CN";
 174 }
 175
 176 UBool CharsetRecog_2022CN::match(InputText *textIn, CharsetMatch *results) const {
 177     int32_t confidence = match_2022(textIn->fInputBytes,
 178                                     textIn->fInputLen,
 179                                     escapeSequences_2022CN,
 180                                     ARRAY_SIZE(escapeSequences_2022CN));
 181     results->set(textIn, this, confidence);
 182     return (confidence > 0);
 183 }
 184
 185 CharsetRecog_2022::~CharsetRecog_2022() {
 186     // nothing to do
 187 }
 188
 189 U_NAMESPACE_END
 190 #endif