/**
* This method is repeatedly called by the tokenizer.
* Each time, we determine the kind of token we're about to
* read, and then we call the appropriate method to handle
* that token type.
*
* @param aScanner The source of our input.
* @param aFlushTokens An OUT parameter to tell the caller whether it should
* process our queued tokens up to now (e.g., when we
* reach a <script>).
* @return Success or error
*/
nsresult
nsHTMLTokenizer::ConsumeToken(nsScanner& aScanner, bool& aFlushTokens)
{
PRUnichar theChar;
CToken* theToken = nullptr;
nsresult result = aScanner.Peek(theChar);
switch(result) {
case kEOF:
// Tell our caller that'we finished.
return result;
case NS_OK:
default:
if (!(mFlags & NS_IPARSER_FLAG_PLAIN_TEXT)) {
if (kLessThan == theChar) {
return ConsumeTag(theChar, theToken, aScanner, aFlushTokens);
} else if (kAmpersand == theChar) {
return ConsumeEntity(theChar, theToken, aScanner);
}
}
if (kCR == theChar || kLF == theChar) {
return ConsumeNewline(theChar, theToken, aScanner);
} else {
if (!nsCRT::IsAsciiSpace(theChar)) {
if (theChar != '\0') {
result = ConsumeText(theToken, aScanner);
} else {
// Skip the embedded null char. Fix bug 64098.
aScanner.GetChar(theChar);
}
break;
}
result = ConsumeWhitespace(theChar, theToken, aScanner);
}
break;
}
return result;
}
BOOL
XMLInternalParser::ReadTextDecl (BOOL xmldecl)
{
unsigned start_index = index;
textdecl_encoding_start = textdecl_encoding_length = 0;
expecting_eof = TRUE;
BOOL has_start = Match (UNI_L ("<?xml"), 5), unproblematic;
expecting_eof = FALSE;
ParseError error = xmldecl ? PARSE_ERROR_Invalid_XMLDecl : PARSE_ERROR_Invalid_TextDecl;
if (has_start && ConsumeWhitespace ())
{
textdecl_version = XMLVERSION_1_0;
textdecl_standalone = XMLSTANDALONE_NONE;
textdecl_encoding = 0;
textdecl_encoding_length = 0;
if (Match (UNI_L ("version"), 7))
{
ConsumeWhitespace ();
if (!Match (UNI_L ("="), 1))
HandleError (error);
ConsumeWhitespace ();
if (!ReadQuotedLiteral (unproblematic) || index == length || literal_length != 3 || literal[0] != '1' || literal[1] != '.' || literal[2] != '0' && literal[2] != '1')
HandleError ((ParseError) (error + 1));
if (literal[2] == '0')
textdecl_version = XMLVERSION_1_0;
else
textdecl_version = XMLVERSION_1_1;
if (current_context != PARSE_CONTEXT_DOCUMENT && version < textdecl_version)
HandleError (WELL_FORMEDNESS_ERROR_ExternalEntityHasLaterVersion);
if (!ConsumeWhitespace ())
if (!xmldecl)
HandleError (error);
else
goto end;
}
else if (xmldecl)
HandleError (error);
if (Match (UNI_L ("encoding"), 8))
{
ConsumeWhitespace ();
if (!Match (UNI_L ("="), 1))
HandleError (error);
ConsumeWhitespace ();
if (!ReadQuotedLiteral (unproblematic) || index == length || literal_length == 0)
HandleError (error);
uni_char ch = literal[0];
if ((ch < 'A' || ch > 'Z') && (ch < 'a' || ch > 'z'))
HandleError ((ParseError) (error + 2));
for (const uni_char *ptr = literal, *ptr_end = ptr + literal_length; ptr != ptr_end; ++ptr)
{
uni_char ch = *ptr;
if ((ch < 'A' || ch > 'Z') && (ch < 'a' || ch > 'z') && (ch < '0' || ch > '9') && ch != '.' && ch != '_' && ch != '-')
HandleError ((ParseError) (error + 2));
}
textdecl_encoding_start = literal_start;
textdecl_encoding_length = literal_length;
if (!ConsumeWhitespace ())
goto end;
}
else if (!xmldecl)
HandleError (error);
unsigned index_before_sddecl = index;
if (Match (UNI_L ("standalone"), 10))
{
ConsumeWhitespace ();
if (!Match (UNI_L ("="), 1))
HandleError (error);
ConsumeWhitespace ();
if (!ReadQuotedLiteral (unproblematic) || index == length)
HandleError (error);
if (!xmldecl)
HandleError (error, index_before_sddecl, index - index_before_sddecl);
if (literal_length == 2 && literal[0] == 'n' && literal[1] == 'o')
textdecl_standalone = XMLSTANDALONE_NO;
else if (literal_length == 3 && literal[0] == 'y' && literal[1] == 'e' && literal[2] == 's')
textdecl_standalone = XMLSTANDALONE_YES;
else
HandleError ((ParseError) (error + 3));
ConsumeWhitespace ();
}
}
else
{
index = start_index;
return FALSE;
}
end:
if (!Match (UNI_L ("?>"), 2))
HandleError (error);
textdecl_encoding = textdecl_encoding_start != 0 ? buffer + textdecl_encoding_start : 0;
return TRUE;
}
