CSSParserToken CSSTokenizer::dollarSign(UChar cc)
{
ASSERT(cc == '$');
if (consumeIfNext('='))
return CSSParserToken(SuffixMatchToken);
return CSSParserToken(DelimiterToken, '$');
}
CSSParserToken CSSTokenizer::consumeUnicodeRange()
{
ASSERT(isASCIIHexDigit(m_input.nextInputChar()) || m_input.nextInputChar() == '?');
int lengthRemaining = 6;
UChar32 start = 0;
while (lengthRemaining && isASCIIHexDigit(m_input.nextInputChar())) {
start = start * 16 + toASCIIHexValue(consume());
--lengthRemaining;
}
if (lengthRemaining && consumeIfNext('?')) {
UChar32 end = start;
do {
start *= 16;
end = end * 16 + 0xF;
--lengthRemaining;
} while (lengthRemaining && consumeIfNext('?'));
return CSSParserToken(UnicodeRangeToken, start, end);
}
if (m_input.nextInputChar() == '-' && isASCIIHexDigit(m_input.peek(1))) {
consume();
lengthRemaining = 6;
UChar32 end = 0;
do {
end = end * 16 + toASCIIHexValue(consume());
--lengthRemaining;
} while (lengthRemaining && isASCIIHexDigit(m_input.nextInputChar()));
return CSSParserToken(UnicodeRangeToken, start, end);
}
return CSSParserToken(UnicodeRangeToken, start, start);
}
CSSParserToken CSSTokenizer::tilde(UChar cc)
{
ASSERT(cc == '~');
if (consumeIfNext('='))
return CSSParserToken(IncludeMatchToken);
return CSSParserToken(DelimiterToken, '~');
}
// http://dev.w3.org/csswg/css-syntax/#consume-a-string-token
CSSParserToken CSSTokenizer::consumeStringTokenUntil(UChar endingCodePoint)
{
StringBuilder output;
while (true) {
UChar cc = consume();
if (cc == endingCodePoint || cc == kEndOfFileMarker) {
// The "reconsume" here deviates from the spec, but is required to avoid consuming past the EOF
if (cc == kEndOfFileMarker)
reconsume(cc);
return CSSParserToken(StringToken, output.toString());
}
if (isNewLine(cc)) {
reconsume(cc);
return CSSParserToken(BadStringToken);
}
if (cc == '\\') {
if (m_input.nextInputChar() == kEndOfFileMarker)
continue;
if (isNewLine(m_input.nextInputChar()))
consumeSingleWhitespaceIfNext(); // This handles \r\n for us
else
output.append(consumeEscape());
} else {
output.append(cc);
}
}
}
CSSParserToken CSSTokenizer::commercialAt(UChar cc)
{
ASSERT(cc == '@');
if (nextCharsAreIdentifier())
return CSSParserToken(AtKeywordToken, consumeName());
return CSSParserToken(DelimiterToken, '@');
}
CSSParserToken CSSTokenizer::asterisk(UChar cc)
{
ASSERT(cc == '*');
if (consumeIfNext('='))
return CSSParserToken(SubstringMatchToken);
return CSSParserToken(DelimiterToken, '*');
}
CSSParserToken CSSTokenizer::circumflexAccent(UChar cc)
{
ASSERT(cc == '^');
if (consumeIfNext('='))
return CSSParserToken(PrefixMatchToken);
return CSSParserToken(DelimiterToken, '^');
}
CSSParserToken CSSTokenizer::solidus(UChar cc)
{
if (consumeIfNext('*')) {
// We're intentionally deviating from the spec here, by creating tokens for CSS comments.
return consumeUntilCommentEndFound()? CSSParserToken(CommentToken): CSSParserToken(EOFToken);
}
return CSSParserToken(DelimiterToken, cc);
}
CSSParserToken CSSTokenizer::lessThan(UChar cc)
{
ASSERT(cc == '<');
if (m_input.peek(0) == '!' && m_input.peek(1) == '-' && m_input.peek(2) == '-') {
consume(3);
return CSSParserToken(CDOToken);
}
return CSSParserToken(DelimiterToken, '<');
}
CSSParserToken CSSTokenizer::verticalLine(UChar cc)
{
ASSERT(cc == '|');
if (consumeIfNext('='))
return CSSParserToken(DashMatchToken);
if (consumeIfNext('|'))
return CSSParserToken(ColumnToken);
return CSSParserToken(DelimiterToken, '|');
}
CSSParserToken CSSTokenizer::hash(UChar cc)
{
UChar nextChar = m_input.nextInputChar();
if (isNameChar(nextChar) || twoCharsAreValidEscape(nextChar, m_input.peek(1))) {
HashTokenType type = nextCharsAreIdentifier() ? HashTokenId : HashTokenUnrestricted;
return CSSParserToken(type, consumeName());
}
return CSSParserToken(DelimiterToken, cc);
}
CSSParserToken CSSTokenizer::solidus(UChar cc)
{
if (consumeIfNext('*')) {
// These get ignored, but we need a value to return.
consumeUntilCommentEndFound();
return CSSParserToken(CommentToken);
}
return CSSParserToken(DelimiterToken, cc);
}
CSSParserToken CSSTokenizer::whiteSpace(UChar cc)
{
// CSS Tokenization is currently lossy, but we could record
// the exact whitespace instead of discarding it here.
consumeUntilNonWhitespace();
return CSSParserToken(WhitespaceToken);
}
CSSParserToken CSSTokenizer::hyphenMinus(UChar cc)
{
if (nextCharsAreNumber(cc)) {
reconsume(cc);
return consumeNumericToken();
}
if (m_input.peek(0) == '-' && m_input.peek(1) == '>') {
consume(2);
return CSSParserToken(CDCToken);
}
if (nextCharsAreIdentifier(cc)) {
reconsume(cc);
return consumeIdentLikeToken();
}
return CSSParserToken(DelimiterToken, cc);
}
CSSParserToken CSSTokenizer::reverseSolidus(UChar cc)
{
if (twoCharsAreValidEscape(cc, m_input.nextInputChar())) {
reconsume(cc);
return consumeIdentLikeToken();
}
return CSSParserToken(DelimiterToken, cc);
}
// http://dev.w3.org/csswg/css-syntax/#consume-url-token
CSSParserToken CSSTokenizer::consumeUrlToken()
{
consumeUntilNonWhitespace();
// URL tokens without escapes get handled without allocations
for (unsigned size = 0; ; size++) {
UChar cc = m_input.peekWithoutReplacement(size);
if (cc == ')') {
unsigned startOffset = m_input.offset();
m_input.advance(size + 1);
return CSSParserToken(UrlToken, m_input.rangeAsCSSParserString(startOffset, size));
}
if (cc <= ' ' || cc == '\\' || cc == '"' || cc == '\'' || cc == '(' || cc == '\x7f')
break;
}
StringBuilder result;
while (true) {
UChar cc = consume();
if (cc == ')' || cc == kEndOfFileMarker)
return CSSParserToken(UrlToken, registerString(result.toString()));
if (isHTMLSpace(cc)) {
consumeUntilNonWhitespace();
if (consumeIfNext(')') || m_input.nextInputChar() == kEndOfFileMarker)
return CSSParserToken(UrlToken, registerString(result.toString()));
break;
}
if (cc == '"' || cc == '\'' || cc == '(' || isNonPrintableCodePoint(cc))
break;
if (cc == '\\') {
if (twoCharsAreValidEscape(cc, m_input.nextInputChar())) {
result.append(consumeEscape());
continue;
}
break;
}
result.append(cc);
}
consumeBadUrlRemnants();
return CSSParserToken(BadUrlToken);
}
CSSParserToken CSSTokenizer::plusOrFullStop(UChar cc)
{
if (nextCharsAreNumber(cc)) {
reconsume(cc);
return consumeNumericToken();
}
return CSSParserToken(DelimiterToken, cc);
}
CSSParserToken CSSTokenizer::hyphenMinus(UChar cc)
{
if (nextCharsAreNumber(cc)) {
reconsume(cc);
return consumeNumericToken();
}
if (nextCharsAreIdentifier(cc)) {
reconsume(cc);
return consumeIdentLikeToken();
}
return CSSParserToken(DelimiterToken, cc);
}
// http://dev.w3.org/csswg/css-syntax/#consume-a-string-token
CSSParserToken CSSTokenizer::consumeStringTokenUntil(UChar endingCodePoint)
{
// Strings without escapes get handled without allocations
for (unsigned size = 0; ; size++) {
UChar cc = m_input.peekWithoutReplacement(size);
if (cc == endingCodePoint) {
unsigned startOffset = m_input.offset();
m_input.advance(size + 1);
return CSSParserToken(StringToken, m_input.rangeAsCSSParserString(startOffset, size));
}
if (isNewLine(cc)) {
m_input.advance(size);
return CSSParserToken(BadStringToken);
}
if (cc == '\0' || cc == '\\')
break;
}
StringBuilder output;
while (true) {
UChar cc = consume();
if (cc == endingCodePoint || cc == kEndOfFileMarker)
return CSSParserToken(StringToken, registerString(output.toString()));
if (isNewLine(cc)) {
reconsume(cc);
return CSSParserToken(BadStringToken);
}
if (cc == '\\') {
if (m_input.nextInputChar() == kEndOfFileMarker)
continue;
if (isNewLine(m_input.nextInputChar()))
consumeSingleWhitespaceIfNext(); // This handles \r\n for us
else
output.append(consumeEscape());
} else {
output.append(cc);
}
}
}
// http://dev.w3.org/csswg/css-syntax/#consume-url-token
CSSParserToken CSSTokenizer::consumeUrlToken()
{
consumeUntilNonWhitespace();
StringBuilder result;
while (true) {
UChar cc = consume();
if (cc == ')' || cc == kEndOfFileMarker) {
// The "reconsume" here deviates from the spec, but is required to avoid consuming past the EOF
if (cc == kEndOfFileMarker)
reconsume(cc);
return CSSParserToken(UrlToken, result.toString());
}
if (isHTMLSpace(cc)) {
consumeUntilNonWhitespace();
if (consumeIfNext(')') || m_input.nextInputChar() == kEndOfFileMarker)
return CSSParserToken(UrlToken, result.toString());
break;
}
if (cc == '"' || cc == '\'' || cc == '(' || isNonPrintableCodePoint(cc))
break;
if (cc == '\\') {
if (twoCharsAreValidEscape(cc, m_input.nextInputChar())) {
result.append(consumeEscape());
continue;
}
break;
}
result.append(cc);
}
consumeBadUrlRemnants();
return CSSParserToken(BadUrlToken);
}
// http://dev.w3.org/csswg/css-syntax/#consume-ident-like-token
CSSParserToken CSSTokenizer::consumeIdentLikeToken()
{
CSSParserString name = consumeName();
if (consumeIfNext('(')) {
if (name.equalIgnoringASCIICase("url")) {
// The spec is slightly different so as to avoid dropping whitespace
// tokens, but they wouldn't be used and this is easier.
consumeUntilNonWhitespace();
UChar next = m_input.nextInputChar();
if (next != '"' && next != '\'')
return consumeUrlToken();
}
return blockStart(LeftParenthesisToken, FunctionToken, name);
}
return CSSParserToken(IdentToken, name);
}
// The state machine loop
PassRefPtrWillBeRawPtr<MediaQuerySet> MediaQueryParser::parseImpl(CSSParserTokenRange range)
{
while (!range.atEnd())
processToken(range.consume());
// FIXME: Can we get rid of this special case?
if (m_parserType == MediaQuerySetParser)
processToken(CSSParserToken(EOFToken));
if (m_state != ReadAnd && m_state != ReadRestrictor && m_state != Done && m_state != ReadMediaNot)
m_querySet->addMediaQuery(MediaQuery::createNotAll());
else if (m_mediaQueryData.currentMediaQueryChanged())
m_querySet->addMediaQuery(m_mediaQueryData.takeMediaQuery());
return m_querySet;
}
// The state machine loop
RefPtr<MediaQuerySet> MediaQueryParser::parseInternal(CSSParserTokenRange range)
{
while (!range.atEnd())
processToken(range.consume());
// FIXME: Can we get rid of this special case?
if (m_parserType == MediaQuerySetParser)
processToken(CSSParserToken(EOFToken));
if (m_state != ReadAnd && m_state != ReadRestrictor && m_state != Done && m_state != ReadMediaNot) {
MediaQuery query = MediaQuery(MediaQuery::Not, "all", Vector<MediaQueryExpression>());
m_querySet->addMediaQuery(WTFMove(query));
} else if (m_mediaQueryData.currentMediaQueryChanged())
commitMediaQuery();
return m_querySet;
}
// This method merges the following spec sections for efficiency
// http://www.w3.org/TR/css3-syntax/#consume-a-number
// http://www.w3.org/TR/css3-syntax/#convert-a-string-to-a-number
CSSParserToken CSSTokenizer::consumeNumber()
{
ASSERT(nextCharsAreNumber());
NumericValueType type = IntegerValueType;
double value = 0;
unsigned offset = 0;
int exponentSign = 1;
int sign = getSign(m_input, offset);
unsigned long long integerPart = getInteger(m_input, offset);
unsigned integerPartEndOffset = offset;
double fractionPart = getFraction(m_input, offset);
unsigned long long exponentPart = getExponent(m_input, offset, exponentSign);
double exponent = pow(10, (float)exponentSign * (double)exponentPart);
value = (double)sign * ((double)integerPart + fractionPart) * exponent;
m_input.advance(offset);
if (offset != integerPartEndOffset)
type = NumberValueType;
return CSSParserToken(NumberToken, value, type);
}
CSSParserToken CSSTokenizer::nextToken()
{
// Unlike the HTMLTokenizer, the CSS Syntax spec is written
// as a stateless, (fixed-size) look-ahead tokenizer.
// We could move to the stateful model and instead create
// states for all the "next 3 codepoints are X" cases.
// State-machine tokenizers are easier to write to handle
// incremental tokenization of partial sources.
// However, for now we follow the spec exactly.
UChar cc = consume();
CodePoint codePointFunc = 0;
if (isASCII(cc)) {
ASSERT_WITH_SECURITY_IMPLICATION(cc < codePointsNumber);
codePointFunc = codePoints[cc];
} else {
codePointFunc = &CSSTokenizer::nameStart;
}
if (codePointFunc)
return ((this)->*(codePointFunc))(cc);
return CSSParserToken(DelimiterToken, cc);
}
static CSSParserToken unicodeRange(UChar32 start, UChar32 end)
{
return CSSParserToken(UnicodeRangeToken, start, end);
}
static CSSParserToken delim(char c) { return CSSParserToken(DelimiterToken, c); }
static CSSParserToken hash(const String& string, HashTokenType type) { return CSSParserToken(type, toParserString(string)); }
static CSSParserToken url(const String& string) { return CSSParserToken(UrlToken, toParserString(string)); }
static CSSParserToken function(const String& string) { return CSSParserToken(FunctionToken, toParserString(string)); }
