1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/i18n/csrutf8.cpp Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,113 @@
1.4 +/*
1.5 + **********************************************************************
1.6 + * Copyright (C) 2005-2012, International Business Machines
1.7 + * Corporation and others. All Rights Reserved.
1.8 + **********************************************************************
1.9 + */
1.10 +
1.11 +#include "unicode/utypes.h"
1.12 +
1.13 +#if !UCONFIG_NO_CONVERSION
1.14 +
1.15 +#include "csrutf8.h"
1.16 +#include "csmatch.h"
1.17 +
1.18 +U_NAMESPACE_BEGIN
1.19 +
1.20 +CharsetRecog_UTF8::~CharsetRecog_UTF8()
1.21 +{
1.22 + // nothing to do
1.23 +}
1.24 +
1.25 +const char *CharsetRecog_UTF8::getName() const
1.26 +{
1.27 + return "UTF-8";
1.28 +}
1.29 +
1.30 +UBool CharsetRecog_UTF8::match(InputText* input, CharsetMatch *results) const {
1.31 + bool hasBOM = FALSE;
1.32 + int32_t numValid = 0;
1.33 + int32_t numInvalid = 0;
1.34 + const uint8_t *inputBytes = input->fRawInput;
1.35 + int32_t i;
1.36 + int32_t trailBytes = 0;
1.37 + int32_t confidence;
1.38 +
1.39 + if (input->fRawLength >= 3 &&
1.40 + inputBytes[0] == 0xEF && inputBytes[1] == 0xBB && inputBytes[2] == 0xBF) {
1.41 + hasBOM = TRUE;
1.42 + }
1.43 +
1.44 + // Scan for multi-byte sequences
1.45 + for (i=0; i < input->fRawLength; i += 1) {
1.46 + int32_t b = inputBytes[i];
1.47 +
1.48 + if ((b & 0x80) == 0) {
1.49 + continue; // ASCII
1.50 + }
1.51 +
1.52 + // Hi bit on char found. Figure out how long the sequence should be
1.53 + if ((b & 0x0E0) == 0x0C0) {
1.54 + trailBytes = 1;
1.55 + } else if ((b & 0x0F0) == 0x0E0) {
1.56 + trailBytes = 2;
1.57 + } else if ((b & 0x0F8) == 0xF0) {
1.58 + trailBytes = 3;
1.59 + } else {
1.60 + numInvalid += 1;
1.61 +
1.62 + if (numInvalid > 5) {
1.63 + break;
1.64 + }
1.65 +
1.66 + trailBytes = 0;
1.67 + }
1.68 +
1.69 + // Verify that we've got the right number of trail bytes in the sequence
1.70 + for (;;) {
1.71 + i += 1;
1.72 +
1.73 + if (i >= input->fRawLength) {
1.74 + break;
1.75 + }
1.76 +
1.77 + b = inputBytes[i];
1.78 +
1.79 + if ((b & 0xC0) != 0x080) {
1.80 + numInvalid += 1;
1.81 + break;
1.82 + }
1.83 +
1.84 + if (--trailBytes == 0) {
1.85 + numValid += 1;
1.86 + break;
1.87 + }
1.88 + }
1.89 +
1.90 + }
1.91 +
1.92 + // Cook up some sort of confidence score, based on presense of a BOM
1.93 + // and the existence of valid and/or invalid multi-byte sequences.
1.94 + confidence = 0;
1.95 + if (hasBOM && numInvalid == 0) {
1.96 + confidence = 100;
1.97 + } else if (hasBOM && numValid > numInvalid*10) {
1.98 + confidence = 80;
1.99 + } else if (numValid > 3 && numInvalid == 0) {
1.100 + confidence = 100;
1.101 + } else if (numValid > 0 && numInvalid == 0) {
1.102 + confidence = 80;
1.103 + } else if (numValid == 0 && numInvalid == 0) {
1.104 + // Plain ASCII.
1.105 + confidence = 10;
1.106 + } else if (numValid > numInvalid*10) {
1.107 + // Probably corruput utf-8 data. Valid sequences aren't likely by chance.
1.108 + confidence = 25;
1.109 + }
1.110 +
1.111 + results->set(input, this, confidence);
1.112 + return (confidence > 0);
1.113 +}
1.114 +
1.115 +U_NAMESPACE_END
1.116 +#endif
