#ifndef HTMLTokenizer_h
#define HTMLTokenizer_h
#include "HTMLToken.h"
#include "MarkupTokenizerBase.h"
#include "SegmentedString.h"
namespace WebCore {
class Frame;
class HTMLTokenizerState {
public:
enum State {
DataState,
CharacterReferenceInDataState,
RCDATAState,
CharacterReferenceInRCDATAState,
RAWTEXTState,
ScriptDataState,
PLAINTEXTState,
TagOpenState,
EndTagOpenState,
TagNameState,
RCDATALessThanSignState,
RCDATAEndTagOpenState,
RCDATAEndTagNameState,
RAWTEXTLessThanSignState,
RAWTEXTEndTagOpenState,
RAWTEXTEndTagNameState,
ScriptDataLessThanSignState,
ScriptDataEndTagOpenState,
ScriptDataEndTagNameState,
ScriptDataEscapeStartState,
ScriptDataEscapeStartDashState,
ScriptDataEscapedState,
ScriptDataEscapedDashState,
ScriptDataEscapedDashDashState,
ScriptDataEscapedLessThanSignState,
ScriptDataEscapedEndTagOpenState,
ScriptDataEscapedEndTagNameState,
ScriptDataDoubleEscapeStartState,
ScriptDataDoubleEscapedState,
ScriptDataDoubleEscapedDashState,
ScriptDataDoubleEscapedDashDashState,
ScriptDataDoubleEscapedLessThanSignState,
ScriptDataDoubleEscapeEndState,
BeforeAttributeNameState,
AttributeNameState,
AfterAttributeNameState,
BeforeAttributeValueState,
AttributeValueDoubleQuotedState,
AttributeValueSingleQuotedState,
AttributeValueUnquotedState,
CharacterReferenceInAttributeValueState,
AfterAttributeValueQuotedState,
SelfClosingStartTagState,
BogusCommentState,
ContinueBogusCommentState,
MarkupDeclarationOpenState,
CommentStartState,
CommentStartDashState,
CommentState,
CommentEndDashState,
CommentEndState,
CommentEndBangState,
DOCTYPEState,
BeforeDOCTYPENameState,
DOCTYPENameState,
AfterDOCTYPENameState,
AfterDOCTYPEPublicKeywordState,
BeforeDOCTYPEPublicIdentifierState,
DOCTYPEPublicIdentifierDoubleQuotedState,
DOCTYPEPublicIdentifierSingleQuotedState,
AfterDOCTYPEPublicIdentifierState,
BetweenDOCTYPEPublicAndSystemIdentifiersState,
AfterDOCTYPESystemKeywordState,
BeforeDOCTYPESystemIdentifierState,
DOCTYPESystemIdentifierDoubleQuotedState,
DOCTYPESystemIdentifierSingleQuotedState,
AfterDOCTYPESystemIdentifierState,
BogusDOCTYPEState,
CDATASectionState,
CDATASectionRightSquareBracketState,
CDATASectionDoubleRightSquareBracketState,
};
};
class HTMLTokenizer : public MarkupTokenizerBase<HTMLToken, HTMLTokenizerState> {
WTF_MAKE_NONCOPYABLE(HTMLTokenizer);
WTF_MAKE_FAST_ALLOCATED;
public:
static PassOwnPtr<HTMLTokenizer> create(bool usePreHTML5ParserQuirks) { return adoptPtr(new HTMLTokenizer(usePreHTML5ParserQuirks)); }
~HTMLTokenizer();
void reset();
bool nextToken(SegmentedString&, HTMLToken&);
String bufferedCharacters() const;
size_t numberOfBufferedCharacters() const
{
return m_temporaryBuffer.size() ? m_temporaryBuffer.size() + 2 : 0;
}
void updateStateFor(const AtomicString& tagName, Frame*);
bool forceNullCharacterReplacement() const { return m_forceNullCharacterReplacement; }
void setForceNullCharacterReplacement(bool value) { m_forceNullCharacterReplacement = value; }
bool shouldAllowCDATA() const { return m_shouldAllowCDATA; }
void setShouldAllowCDATA(bool value) { m_shouldAllowCDATA = value; }
private:
explicit HTMLTokenizer(bool usePreHTML5ParserQuirks);
inline bool processEntity(SegmentedString&);
inline void parseError();
inline bool emitAndResumeIn(SegmentedString& source, HTMLTokenizerState::State state)
{
saveEndTagNameIfNeeded();
return MarkupTokenizerBase<HTMLToken, HTMLTokenizerState>::emitAndResumeIn(source, state);
}
inline bool emitAndReconsumeIn(SegmentedString& source, HTMLTokenizerState::State state)
{
saveEndTagNameIfNeeded();
return MarkupTokenizerBase<HTMLToken, HTMLTokenizerState>::emitAndReconsumeIn(source, state);
}
inline bool flushEmitAndResumeIn(SegmentedString&, HTMLTokenizerState::State);
inline bool flushBufferedEndTag(SegmentedString&);
inline bool temporaryBufferIs(const String&);
inline void addToPossibleEndTag(UChar cc);
inline void saveEndTagNameIfNeeded()
{
ASSERT(m_token->type() != HTMLTokenTypes::Uninitialized);
if (m_token->type() == HTMLTokenTypes::StartTag)
m_appropriateEndTagName = m_token->name();
}
inline bool isAppropriateEndTag();
Vector<UChar, 32> m_appropriateEndTagName;
bool m_shouldAllowCDATA;
Vector<UChar, 32> m_temporaryBuffer;
Vector<UChar, 32> m_bufferedEndTagName;
bool m_usePreHTML5ParserQuirks;
};
}
#endif