icuSources/i18n/csr2022.cpp

   1 /*
   2  **********************************************************************
   3  *   Copyright (C) 2005-2016, 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 "cmemory.h"
  13 #include "cstring.h"
  14
  15 #include "csr2022.h"
  16 #include "csmatch.h"
  17
  18 U_NAMESPACE_BEGIN
  19
  20 /**
  21  * Matching function shared among the 2022 detectors JP, CN and KR
  22  * Counts up the number of legal and 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) const
  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                 const uint8_t *seq = escapeSequences[escN];
  47                 int32_t seq_length = (int32_t)uprv_strlen((const char *) seq);
  48
  49                 if (textLen-i >= seq_length) {
  50                     j = 1;
  51                     while(j < seq_length) {
  52                         if(seq[j] != text[i+j]) {
  53                             goto checkEscapes;
  54                         }
  55
  56                         j += 1;
  57                     }
  58
  59                     hits += 1;
  60                     i += seq_length-1;
  61                     goto scanInput;
  62                 }
  63                 // else we ran out of string to compare this time.
  64 checkEscapes:
  65                 escN += 1;
  66             }
  67
  68             misses += 1;
  69         }
  70
  71         if( text[i]== 0x0e || text[i] == 0x0f){
  72             shifts += 1;
  73         }
  74
  75 scanInput:
  76         i += 1;
  77     }
  78
  79     if (hits == 0) {
  80         return 0;
  81     }
  82
  83     //
  84     // Initial quality is based on relative proportion of recongized vs.
  85     //   unrecognized escape sequences.
  86     //   All good:  quality = 100;
  87     //   half or less good: quality = 0;
  88     //   linear inbetween.
  89     quality = (100*hits - 100*misses) / (hits + misses);
  90
  91     // Back off quality if there were too few escape sequences seen.
  92     //   Include shifts in this computation, so that KR does not get penalized
  93     //   for having only a single Escape sequence, but many shifts.
  94     if (hits+shifts < 5) {
  95         quality -= (5-(hits+shifts))*10;
  96     }
  97
  98     if (quality < 0) {
  99         quality = 0;
 100     }
 101
 102     return quality;
 103 }
 104
 105
 106 static const uint8_t escapeSequences_2022JP[][5] = {
 107     {0x1b, 0x24, 0x28, 0x43, 0x00},   // KS X 1001:1992
 108     {0x1b, 0x24, 0x28, 0x44, 0x00},   // JIS X 212-1990
 109     {0x1b, 0x24, 0x40, 0x00, 0x00},   // JIS C 6226-1978
 110     {0x1b, 0x24, 0x41, 0x00, 0x00},   // GB 2312-80
 111     {0x1b, 0x24, 0x42, 0x00, 0x00},   // JIS X 208-1983
 112     {0x1b, 0x26, 0x40, 0x00, 0x00},   // JIS X 208 1990, 1997
 113     {0x1b, 0x28, 0x42, 0x00, 0x00},   // ASCII
 114     {0x1b, 0x28, 0x48, 0x00, 0x00},   // JIS-Roman
 115     {0x1b, 0x28, 0x49, 0x00, 0x00},   // Half-width katakana
 116     {0x1b, 0x28, 0x4a, 0x00, 0x00},   // JIS-Roman
 117     {0x1b, 0x2e, 0x41, 0x00, 0x00},   // ISO 8859-1
 118     {0x1b, 0x2e, 0x46, 0x00, 0x00}    // ISO 8859-7
 119 };
 120
 121 #if !UCONFIG_ONLY_HTML_CONVERSION
 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 #endif
 140
 141 CharsetRecog_2022JP::~CharsetRecog_2022JP() {}
 142
 143 const char *CharsetRecog_2022JP::getName() const {
 144     return "ISO-2022-JP";
 145 }
 146
 147 UBool CharsetRecog_2022JP::match(InputText *textIn, CharsetMatch *results) const {
 148     int32_t confidence = match_2022(textIn->fInputBytes,
 149                                     textIn->fInputLen,
 150                                     escapeSequences_2022JP,
 151                                     UPRV_LENGTHOF(escapeSequences_2022JP));
 152     results->set(textIn, this, confidence);
 153     return (confidence > 0);
 154 }
 155
 156 #if !UCONFIG_ONLY_HTML_CONVERSION
 157 CharsetRecog_2022KR::~CharsetRecog_2022KR() {}
 158
 159 const char *CharsetRecog_2022KR::getName() const {
 160     return "ISO-2022-KR";
 161 }
 162
 163 UBool CharsetRecog_2022KR::match(InputText *textIn, CharsetMatch *results) const {
 164     int32_t confidence = match_2022(textIn->fInputBytes,
 165                                     textIn->fInputLen,
 166                                     escapeSequences_2022KR,
 167                                     UPRV_LENGTHOF(escapeSequences_2022KR));
 168     results->set(textIn, this, confidence);
 169     return (confidence > 0);
 170 }
 171
 172 CharsetRecog_2022CN::~CharsetRecog_2022CN() {}
 173
 174 const char *CharsetRecog_2022CN::getName() const {
 175     return "ISO-2022-CN";
 176 }
 177
 178 UBool CharsetRecog_2022CN::match(InputText *textIn, CharsetMatch *results) const {
 179     int32_t confidence = match_2022(textIn->fInputBytes,
 180                                     textIn->fInputLen,
 181                                     escapeSequences_2022CN,
 182                                     UPRV_LENGTHOF(escapeSequences_2022CN));
 183     results->set(textIn, this, confidence);
 184     return (confidence > 0);
 185 }
 186 #endif
 187
 188 CharsetRecog_2022::~CharsetRecog_2022() {
 189     // nothing to do
 190 }
 191
 192 U_NAMESPACE_END
 193 #endif