ClassAd *ClassAdXMLParser::
ParseClassAd(ClassAd *classad_in)
{
bool in_classad;
ClassAd *classad = NULL;
ClassAd *local_ad = NULL;
XMLLexer::Token token;
classad = NULL;
in_classad = false;
while (lexer.PeekToken(&token)) {
if (!in_classad) {
lexer.ConsumeToken(NULL);
if ( token.token_type == XMLLexer::tokenType_Tag
&& token.tag_id == XMLLexer::tagID_ClassAd) {
// We have a ClassAd tag
if (token.tag_type == XMLLexer::tagType_Start) {
in_classad = true;
if ( classad_in ) {
classad_in->Clear();
classad = classad_in;
} else {
local_ad = new ClassAd();
classad = local_ad;
}
classad->DisableDirtyTracking();
} else {
// We're done, return the ClassAd we got, if any.
in_classad = false;
break;
}
}
} else {
if (token.token_type == XMLLexer::tokenType_Tag) {
if (token.tag_id == XMLLexer::tagID_Attribute) {
if (token.tag_type == XMLLexer::tagType_Invalid) {
delete local_ad;
return NULL;
} else if( token.tag_type == XMLLexer::tagType_Start) {
string attribute_name;
ExprTree *tree;
tree = ParseAttribute(attribute_name);
if (tree != NULL) {
classad->Insert(attribute_name, tree);
}
else {
delete local_ad;
return NULL;
}
} else {
lexer.ConsumeToken(NULL);
}
} else if (token.tag_id == XMLLexer::tagID_ClassAd) {
lexer.ConsumeToken(NULL);
if (token.tag_type == XMLLexer::tagType_End) {
in_classad = false;
break;
} else {
// This is invalid, but we'll just ignore it.
}
} else if ( token.tag_id != XMLLexer::tagID_XML
&& token.tag_id != XMLLexer::tagID_XMLStylesheet
&& token.tag_id != XMLLexer::tagID_Doctype
&& token.tag_id != XMLLexer::tagID_ClassAds) {
// We got a non-attribute, non-xml thingy within a
// ClassAd. That must be an error, but we'll just skip
// it in the hopes of recovering.
lexer.ConsumeToken(NULL);
break;
}
} else {
lexer.ConsumeToken(NULL);
}
}
}
if (classad != NULL) {
classad->EnableDirtyTracking();
}
return classad;
}
// ClassAd ::= '[' AttributeList ']'
// AttributeList ::= (epsilon)
// | Attribute ';' AttributeList
// Attribute ::= Identifier '=' Expression
bool ClassAdParser::
parseClassAd( ClassAd &ad , bool full )
{
Lexer::TokenType tt;
Lexer::TokenValue tv;
ExprTree *tree = NULL;
string s;
ad.Clear( );
ad.DisableDirtyTracking();
if( ( tt = lexer.ConsumeToken() ) != Lexer::LEX_OPEN_BOX ) return false;
tt = lexer.PeekToken();
while( tt != Lexer::LEX_CLOSE_BOX ) {
// Get the name of the expression
tt = lexer.ConsumeToken( &tv );
if( tt == Lexer::LEX_SEMICOLON ) {
// We allow empty expressions, so if someone give a double semicolon, it doesn't
// hurt. Technically it's not right, but we shouldn't make users pay the price for
// a meaningless mistake. See condor-support #1881 for a user that was bitten by this.
continue;
}
if( tt != Lexer::LEX_IDENTIFIER ) {
CondorErrno = ERR_PARSE_ERROR;
CondorErrMsg = "while parsing classad: expected LEX_IDENTIFIER "
" but got " + string( Lexer::strLexToken( tt ) );
return false;
}
// consume the intermediate '='
if( ( tt = lexer.ConsumeToken() ) != Lexer::LEX_BOUND_TO ) {
CondorErrno = ERR_PARSE_ERROR;
CondorErrMsg = "while parsing classad: expected LEX_BOUND_TO "
" but got " + string( Lexer::strLexToken( tt ) );
return false;
}
// parse the expression
parseExpression( tree );
if( tree == NULL ) {
return false;
}
// insert the attribute into the classad
tv.GetStringValue( s );
if( !ad.Insert( s, tree ) ) {
delete tree;
return false;
}
// the next token must be a ';' or a ']'
tt = lexer.PeekToken();
if( tt != Lexer::LEX_SEMICOLON && tt != Lexer::LEX_CLOSE_BOX ) {
CondorErrno = ERR_PARSE_ERROR;
CondorErrMsg = "while parsing classad: expected LEX_SEMICOLON or "
"LEX_CLOSE_BOX but got " + string( Lexer::strLexToken( tt ) );
return( false );
}
// Slurp up any extra semicolons. This does not duplicate the work at the top of the loop
// because it accounts for the case where the last expression has extra semicolons,
// while the first case accounts for optional beginning semicolons.
while( tt == Lexer::LEX_SEMICOLON ) {
lexer.ConsumeToken();
tt = lexer.PeekToken();
}
}
lexer.ConsumeToken();
// if a full parse was requested, ensure that input is exhausted
if( full && ( lexer.ConsumeToken() != Lexer::LEX_END_OF_INPUT ) ) {
CondorErrno = ERR_PARSE_ERROR;
CondorErrMsg = "while parsing classad: expected LEX_END_OF_INPUT for "
"full parse but got " + string( Lexer::strLexToken( tt ) );
return false;
}
ad.EnableDirtyTracking();
return true;
}