michael@0: /*
michael@0:  **********************************************************************
michael@0:  *   Copyright (C) 2005-2013, 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 "csrucode.h"
michael@0: #include "csmatch.h"
michael@0: 
michael@0: U_NAMESPACE_BEGIN
michael@0: 
michael@0: CharsetRecog_Unicode::~CharsetRecog_Unicode()
michael@0: {
michael@0:     // nothing to do
michael@0: }
michael@0: 
michael@0: CharsetRecog_UTF_16_BE::~CharsetRecog_UTF_16_BE()
michael@0: {
michael@0:     // nothing to do
michael@0: }
michael@0: 
michael@0: const char *CharsetRecog_UTF_16_BE::getName() const
michael@0: {
michael@0:     return "UTF-16BE";
michael@0: }
michael@0: 
michael@0: UBool CharsetRecog_UTF_16_BE::match(InputText* textIn, CharsetMatch *results) const
michael@0: {
michael@0:     const uint8_t *input = textIn->fRawInput;
michael@0:     int32_t confidence = 0;
michael@0:     int32_t length = textIn->fRawLength;
michael@0: 
michael@0:     if (length >=2 && input[0] == 0xFE && input[1] == 0xFF) {
michael@0:         confidence = 100;
michael@0:     }
michael@0: 
michael@0:     // TODO: Do some statastics to check for unsigned UTF-16BE
michael@0:     results->set(textIn, this, confidence);
michael@0:     return (confidence > 0);
michael@0: }
michael@0: 
michael@0: CharsetRecog_UTF_16_LE::~CharsetRecog_UTF_16_LE()
michael@0: {
michael@0:     // nothing to do
michael@0: }
michael@0: 
michael@0: const char *CharsetRecog_UTF_16_LE::getName() const
michael@0: {
michael@0:     return "UTF-16LE";
michael@0: }
michael@0: 
michael@0: UBool CharsetRecog_UTF_16_LE::match(InputText* textIn, CharsetMatch *results) const
michael@0: {
michael@0:     const uint8_t *input = textIn->fRawInput;
michael@0:     int32_t confidence = 0;
michael@0:     int32_t length = textIn->fRawLength;
michael@0: 
michael@0:     if (length >= 4 && input[0] == 0xFF && input[1] == 0xFE && (input[2] != 0x00 || input[3] != 0x00)) {
michael@0:         confidence = 100;
michael@0:     }
michael@0: 
michael@0:     // TODO: Do some statastics to check for unsigned UTF-16LE
michael@0:     results->set(textIn, this, confidence);
michael@0:     return (confidence > 0);
michael@0: }
michael@0: 
michael@0: CharsetRecog_UTF_32::~CharsetRecog_UTF_32()
michael@0: {
michael@0:     // nothing to do
michael@0: }
michael@0: 
michael@0: UBool CharsetRecog_UTF_32::match(InputText* textIn, CharsetMatch *results) const
michael@0: {
michael@0:     const uint8_t *input = textIn->fRawInput;
michael@0:     int32_t limit = (textIn->fRawLength / 4) * 4;
michael@0:     int32_t numValid = 0;
michael@0:     int32_t numInvalid = 0;
michael@0:     bool hasBOM = FALSE;
michael@0:     int32_t confidence = 0;
michael@0: 
michael@0:     if (limit > 0 && getChar(input, 0) == 0x0000FEFFUL) {
michael@0:         hasBOM = TRUE;
michael@0:     }
michael@0: 
michael@0:     for(int32_t i = 0; i < limit; i += 4) {
michael@0:         int32_t ch = getChar(input, i);
michael@0: 
michael@0:         if (ch < 0 || ch >= 0x10FFFF || (ch >= 0xD800 && ch <= 0xDFFF)) {
michael@0:             numInvalid += 1;
michael@0:         } else {
michael@0:             numValid += 1;
michael@0:         }
michael@0:     }
michael@0: 
michael@0: 
michael@0:     // Cook up some sort of confidence score, based on presense of a BOM
michael@0:     //    and the existence of valid and/or invalid multi-byte sequences.
michael@0:     if (hasBOM && numInvalid==0) {
michael@0:         confidence = 100;
michael@0:     } else if (hasBOM && numValid > numInvalid*10) {
michael@0:         confidence = 80;
michael@0:     } else if (numValid > 3 && numInvalid == 0) {
michael@0:         confidence = 100;            
michael@0:     } else if (numValid > 0 && numInvalid == 0) {
michael@0:         confidence = 80;
michael@0:     } else if (numValid > numInvalid*10) {
michael@0:         // Probably corruput UTF-32BE data.  Valid sequences aren't likely by chance.
michael@0:         confidence = 25;
michael@0:     }
michael@0: 
michael@0:     results->set(textIn, this, confidence);
michael@0:     return (confidence > 0);
michael@0: }
michael@0: 
michael@0: CharsetRecog_UTF_32_BE::~CharsetRecog_UTF_32_BE()
michael@0: {
michael@0:     // nothing to do
michael@0: }
michael@0: 
michael@0: const char *CharsetRecog_UTF_32_BE::getName() const
michael@0: {
michael@0:     return "UTF-32BE";
michael@0: }
michael@0: 
michael@0: int32_t CharsetRecog_UTF_32_BE::getChar(const uint8_t *input, int32_t index) const
michael@0: {
michael@0:     return input[index + 0] << 24 | input[index + 1] << 16 |
michael@0:            input[index + 2] <<  8 | input[index + 3];
michael@0: } 
michael@0: 
michael@0: CharsetRecog_UTF_32_LE::~CharsetRecog_UTF_32_LE()
michael@0: {
michael@0:     // nothing to do
michael@0: }
michael@0: 
michael@0: const char *CharsetRecog_UTF_32_LE::getName() const
michael@0: {
michael@0:     return "UTF-32LE";
michael@0: }
michael@0: 
michael@0: int32_t CharsetRecog_UTF_32_LE::getChar(const uint8_t *input, int32_t index) const
michael@0: {
michael@0:     return input[index + 3] << 24 | input[index + 2] << 16 |
michael@0:            input[index + 1] <<  8 | input[index + 0];
michael@0: }
michael@0: 
michael@0: U_NAMESPACE_END
michael@0: #endif
michael@0: