blob: 159a285a3b4dadf3dc034bb232037d6e72054aa6 [file] [log] [blame]
/*
* Copyright (C) 2016 Apple Inc. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY APPLE INC. ``AS IS'' AND ANY
* EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
* PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL APPLE INC. OR
* CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
* EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
* PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
* OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*/
#include "config.h"
#include "TextDecoder.h"
#include "HTMLParserIdioms.h"
namespace WebCore {
ExceptionOr<Ref<TextDecoder>> TextDecoder::create(const String& label, Options options)
{
String strippedLabel = stripLeadingAndTrailingHTMLSpaces(label);
const UChar nullCharacter = '\0';
if (strippedLabel.contains(nullCharacter))
return Exception { RangeError };
auto decoder = adoptRef(*new TextDecoder(strippedLabel.utf8().data(), options));
if (!decoder->m_textEncoding.isValid() || !strcmp(decoder->m_textEncoding.name(), "replacement"))
return Exception { RangeError };
return WTFMove(decoder);
}
TextDecoder::TextDecoder(const char* label, Options options)
: m_textEncoding(label)
, m_options(options)
{
}
void TextDecoder::ignoreBOMIfNecessary(const uint8_t*& data, size_t& length)
{
const uint8_t utf8BOMBytes[3] = {0xEF, 0xBB, 0xBF};
const uint8_t utf16BEBOMBytes[2] = {0xFE, 0xFF};
const uint8_t utf16LEBOMBytes[2] = {0xFF, 0xFE};
if (m_textEncoding == UTF8Encoding()
&& length >= sizeof(utf8BOMBytes)
&& data[0] == utf8BOMBytes[0]
&& data[1] == utf8BOMBytes[1]
&& data[2] == utf8BOMBytes[2]) {
data += sizeof(utf8BOMBytes);
length -= sizeof(utf8BOMBytes);
} else if (m_textEncoding == UTF16BigEndianEncoding()
&& length >= sizeof(utf16BEBOMBytes)
&& data[0] == utf16BEBOMBytes[0]
&& data[1] == utf16BEBOMBytes[1]) {
data += sizeof(utf16BEBOMBytes);
length -= sizeof(utf16BEBOMBytes);
} else if (m_textEncoding == UTF16LittleEndianEncoding()
&& length >= sizeof(utf16LEBOMBytes)
&& data[0] == utf16LEBOMBytes[0]
&& data[1] == utf16LEBOMBytes[1]) {
data += sizeof(utf16LEBOMBytes);
length -= sizeof(utf16LEBOMBytes);
}
}
String TextDecoder::prependBOMIfNecessary(const String& decoded)
{
if (m_hasDecoded || !m_options.ignoreBOM)
return decoded;
const UChar utf16BEBOM[2] = {0xFEFF, '\0'};
// FIXME: Make TextCodec::decode take a flag for prepending BOM so we don't need to do this extra allocation and copy.
return makeString(utf16BEBOM, decoded);
}
static size_t codeUnitByteSize(const TextEncoding& encoding)
{
return encoding.isByteBasedEncoding() ? 1 : 2;
}
ExceptionOr<String> TextDecoder::decode(std::optional<BufferSource::VariantType> input, DecodeOptions options)
{
std::optional<BufferSource> inputBuffer;
const uint8_t* data = nullptr;
size_t length = 0;
if (input) {
inputBuffer = BufferSource(WTFMove(input.value()));
data = inputBuffer->data();
length = inputBuffer->length();
}
ignoreBOMIfNecessary(data, length);
if (m_buffer.size()) {
m_buffer.append(data, length);
data = m_buffer.data();
length = m_buffer.size();
}
const bool stopOnError = true;
bool sawError = false;
if (length % codeUnitByteSize(m_textEncoding))
sawError = true;
const char* charData = reinterpret_cast<const char*>(data);
String result;
if (!sawError)
result = prependBOMIfNecessary(m_textEncoding.decode(charData, length, stopOnError, sawError));
if (sawError) {
if (options.stream) {
result = String();
if (!m_buffer.size())
m_buffer.append(data, length);
} else {
if (m_options.fatal)
return Exception { TypeError };
result = prependBOMIfNecessary(m_textEncoding.decode(charData, length));
}
} else
m_buffer.clear();
m_hasDecoded = true;
return WTFMove(result);
}
String TextDecoder::encoding() const
{
return String(m_textEncoding.name()).convertToASCIILowercase();
}
}