michael@0: /*
michael@0:  **********************************************************************
michael@0:  *   Copyright (C) 2005-2012, International Business Machines
michael@0:  *   Corporation and others.  All Rights Reserved.
michael@0:  **********************************************************************
michael@0:  */
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: 
michael@0: #if !UCONFIG_NO_CONVERSION
michael@0: 
michael@0: #include "cstring.h"
michael@0: 
michael@0: #include "csr2022.h"
michael@0: #include "csmatch.h"
michael@0: 
michael@0: U_NAMESPACE_BEGIN
michael@0: 
michael@0: #define ARRAY_SIZE(array) (sizeof array / sizeof array[0])
michael@0: 
michael@0: /**
michael@0:  * Matching function shared among the 2022 detectors JP, CN and KR
michael@0:  * Counts up the number of legal and unrecognized escape sequences in
michael@0:  * the sample of text, and computes a score based on the total number &
michael@0:  * the proportion that fit the encoding.
michael@0:  * 
michael@0:  * 
michael@0:  * @param text the byte buffer containing text to analyse
michael@0:  * @param textLen  the size of the text in the byte.
michael@0:  * @param escapeSequences the byte escape sequences to test for.
michael@0:  * @return match quality, in the range of 0-100.
michael@0:  */
michael@0: int32_t CharsetRecog_2022::match_2022(const uint8_t *text, int32_t textLen, const uint8_t escapeSequences[][5], int32_t escapeSequences_length) const
michael@0: {
michael@0:     int32_t i, j;
michael@0:     int32_t escN;
michael@0:     int32_t hits   = 0;
michael@0:     int32_t misses = 0;
michael@0:     int32_t shifts = 0;
michael@0:     int32_t quality;
michael@0: 
michael@0:     i = 0;
michael@0:     while(i < textLen) {
michael@0:         if(text[i] == 0x1B) {
michael@0:             escN = 0;
michael@0:             while(escN < escapeSequences_length) {
michael@0:                 const uint8_t *seq = escapeSequences[escN];
michael@0:                 int32_t seq_length = (int32_t)uprv_strlen((const char *) seq);
michael@0: 
michael@0:                 if (textLen-i >= seq_length) {
michael@0:                     j = 1;
michael@0:                     while(j < seq_length) {
michael@0:                         if(seq[j] != text[i+j]) {
michael@0:                             goto checkEscapes;
michael@0:                         }
michael@0: 
michael@0:                         j += 1;
michael@0:                     }
michael@0: 
michael@0:                     hits += 1;
michael@0:                     i += seq_length-1;
michael@0:                     goto scanInput;
michael@0:                 }
michael@0:                 // else we ran out of string to compare this time.
michael@0: checkEscapes:
michael@0:                 escN += 1;
michael@0:             }
michael@0: 
michael@0:             misses += 1;
michael@0:         }
michael@0: 
michael@0:         if( text[i]== 0x0e || text[i] == 0x0f){
michael@0:             shifts += 1;
michael@0:         }
michael@0: 
michael@0: scanInput:
michael@0:         i += 1;
michael@0:     }
michael@0: 
michael@0:     if (hits == 0) {
michael@0:         return 0;
michael@0:     }
michael@0: 
michael@0:     //
michael@0:     // Initial quality is based on relative proportion of recongized vs.
michael@0:     //   unrecognized escape sequences. 
michael@0:     //   All good:  quality = 100;
michael@0:     //   half or less good: quality = 0;
michael@0:     //   linear inbetween.
michael@0:     quality = (100*hits - 100*misses) / (hits + misses);
michael@0: 
michael@0:     // Back off quality if there were too few escape sequences seen.
michael@0:     //   Include shifts in this computation, so that KR does not get penalized
michael@0:     //   for having only a single Escape sequence, but many shifts.
michael@0:     if (hits+shifts < 5) {
michael@0:         quality -= (5-(hits+shifts))*10;
michael@0:     }
michael@0: 
michael@0:     if (quality < 0) {
michael@0:         quality = 0;
michael@0:     }
michael@0: 
michael@0:     return quality;
michael@0: }
michael@0: 
michael@0: 
michael@0: static const uint8_t escapeSequences_2022JP[][5] = {
michael@0:     {0x1b, 0x24, 0x28, 0x43, 0x00},   // KS X 1001:1992
michael@0:     {0x1b, 0x24, 0x28, 0x44, 0x00},   // JIS X 212-1990
michael@0:     {0x1b, 0x24, 0x40, 0x00, 0x00},   // JIS C 6226-1978
michael@0:     {0x1b, 0x24, 0x41, 0x00, 0x00},   // GB 2312-80
michael@0:     {0x1b, 0x24, 0x42, 0x00, 0x00},   // JIS X 208-1983
michael@0:     {0x1b, 0x26, 0x40, 0x00, 0x00},   // JIS X 208 1990, 1997
michael@0:     {0x1b, 0x28, 0x42, 0x00, 0x00},   // ASCII
michael@0:     {0x1b, 0x28, 0x48, 0x00, 0x00},   // JIS-Roman
michael@0:     {0x1b, 0x28, 0x49, 0x00, 0x00},   // Half-width katakana
michael@0:     {0x1b, 0x28, 0x4a, 0x00, 0x00},   // JIS-Roman
michael@0:     {0x1b, 0x2e, 0x41, 0x00, 0x00},   // ISO 8859-1
michael@0:     {0x1b, 0x2e, 0x46, 0x00, 0x00}    // ISO 8859-7
michael@0: };
michael@0: 
michael@0: static const uint8_t escapeSequences_2022KR[][5] = {
michael@0:     {0x1b, 0x24, 0x29, 0x43, 0x00}   
michael@0: };
michael@0: 
michael@0: static const uint8_t escapeSequences_2022CN[][5] = {
michael@0:     {0x1b, 0x24, 0x29, 0x41, 0x00},   // GB 2312-80
michael@0:     {0x1b, 0x24, 0x29, 0x47, 0x00},   // CNS 11643-1992 Plane 1
michael@0:     {0x1b, 0x24, 0x2A, 0x48, 0x00},   // CNS 11643-1992 Plane 2
michael@0:     {0x1b, 0x24, 0x29, 0x45, 0x00},   // ISO-IR-165
michael@0:     {0x1b, 0x24, 0x2B, 0x49, 0x00},   // CNS 11643-1992 Plane 3
michael@0:     {0x1b, 0x24, 0x2B, 0x4A, 0x00},   // CNS 11643-1992 Plane 4
michael@0:     {0x1b, 0x24, 0x2B, 0x4B, 0x00},   // CNS 11643-1992 Plane 5
michael@0:     {0x1b, 0x24, 0x2B, 0x4C, 0x00},   // CNS 11643-1992 Plane 6
michael@0:     {0x1b, 0x24, 0x2B, 0x4D, 0x00},   // CNS 11643-1992 Plane 7
michael@0:     {0x1b, 0x4e, 0x00, 0x00, 0x00},   // SS2
michael@0:     {0x1b, 0x4f, 0x00, 0x00, 0x00},   // SS3
michael@0: };
michael@0: 
michael@0: CharsetRecog_2022JP::~CharsetRecog_2022JP() {}
michael@0: 
michael@0: const char *CharsetRecog_2022JP::getName() const {
michael@0:     return "ISO-2022-JP";
michael@0: }
michael@0: 
michael@0: UBool CharsetRecog_2022JP::match(InputText *textIn, CharsetMatch *results) const {
michael@0:     int32_t confidence = match_2022(textIn->fInputBytes, 
michael@0:                                     textIn->fInputLen, 
michael@0:                                     escapeSequences_2022JP, 
michael@0:                                     ARRAY_SIZE(escapeSequences_2022JP));
michael@0:     results->set(textIn, this, confidence);
michael@0:     return (confidence > 0);
michael@0: }
michael@0: 
michael@0: CharsetRecog_2022KR::~CharsetRecog_2022KR() {}
michael@0: 
michael@0: const char *CharsetRecog_2022KR::getName() const {
michael@0:     return "ISO-2022-KR";
michael@0: }
michael@0: 
michael@0: UBool CharsetRecog_2022KR::match(InputText *textIn, CharsetMatch *results) const {
michael@0:     int32_t confidence = match_2022(textIn->fInputBytes, 
michael@0:                                     textIn->fInputLen, 
michael@0:                                     escapeSequences_2022KR, 
michael@0:                                     ARRAY_SIZE(escapeSequences_2022KR));
michael@0:     results->set(textIn, this, confidence);
michael@0:     return (confidence > 0);
michael@0: }
michael@0: 
michael@0: CharsetRecog_2022CN::~CharsetRecog_2022CN() {}
michael@0: 
michael@0: const char *CharsetRecog_2022CN::getName() const {
michael@0:     return "ISO-2022-CN";
michael@0: }
michael@0: 
michael@0: UBool CharsetRecog_2022CN::match(InputText *textIn, CharsetMatch *results) const {
michael@0:     int32_t confidence = match_2022(textIn->fInputBytes,
michael@0:                                     textIn->fInputLen,
michael@0:                                     escapeSequences_2022CN,
michael@0:                                     ARRAY_SIZE(escapeSequences_2022CN));
michael@0:     results->set(textIn, this, confidence);
michael@0:     return (confidence > 0);
michael@0: }
michael@0: 
michael@0: CharsetRecog_2022::~CharsetRecog_2022() {
michael@0:     // nothing to do
michael@0: }
michael@0: 
michael@0: U_NAMESPACE_END
michael@0: #endif