README

The classes represented on the diagram are the infrastructure for ANTLR parser usage.

Working with ANTLR generated parser starts with extending SQLStandardParser with STMParserOverrides.

We also override AST Nodes to extend their functionality. We have three types of nodes:

• terminal node (STMTreeTermNode)
• non-terminal node (STMTreeRuleNode)
• error node (STMTreeTermErrorNode)

All of them have the common parent STMTreeNode which extends Tree class of ANTLR library. So, when we parse some text by the grammar, we get an object of type STMTreeNode.

The key interface is LSMAnalyzer. It encapsulates parser and lexer creation and parsing process. To parse sql query you need to call LSMAnalyzer:parseSQLQueryTree method which has two arguments. The first argument is STMSource. It's responsible for providing the source text to the parser. The second argument is of type STMErrorListener and used to define the behaviour on errors while parsing. Use STMLoggingErrorListener to log errors and STMSkippingErrorListener to just skip them and do nothing.

The concrete implementation of LSMAnalyzer used according to SQLDialect implementation with LSMDialectRegistry. LSMDialectRegistry contains a map with analyzer to dialect correspondence, which comes from extension point declared in plugin.xml. SQLStandardAnalyzer is a common analyzer with parser and lexer by the common grammar for most SQL databases without any specific settings. It's being instantiated with the SQLStandardAnalyzerFactory implementation of the LSMAnalyzerFactory interface via extension point in org.jkiss.dbeaver.model.sql.plugin.xml.

<extension point="org.jkiss.dbeaver.lsm.dialectSyntax">
    <lsmDialect analyzerFactoryClass="org.jkiss.dbeaver.model.lsm.sql.dialect.SQLStandardAnalyzerFactory">
        <appliesTo dialectClass="org.jkiss.dbeaver.model.impl.sql.BasicSQLDialect"/>
    </lsmDialect>
</extension>

If we need to use a common grammar, but with a specific parser configuration, or want to use analyzer with database-specific parser and lexer, then we need to describe it accordingly using extension point. The following listing gives an example of configuring parser for specific dialect to use square brackets quotation for identifiers.

public class SQLiteSQLAnalyzer extends SQLStandardAnalyzer {
    public SQLStandardAnalyzer(@NotNull SQLDialect dialect) {
        super(dialect); 
    }
    @NotNull
    @Override
    protected SQLStandardParser prepareParser(@NotNull STMSource source, @Nullable STMErrorListener errorListener) {
        SQLStandardParser parser = super.prepareParser(source, errorListener);
        parser.setIsSupportSquareBracketQuotation(true);
        return parser;
    }
}

Analyzer accepts SQLDialect carrying non-syntax and non-semantic information about certain SQL language context with respect to the connection session, so it is required to create analyzer factory and associate it with corresponding dialect's family.

public class SQLiteAnalyzerFactory implements LSMAnalyzerFactory {
    @Override
    public LSMAnalyzer createAnalyzer(SQLDialect dialect) {
        return new SQLiteSQLAnalyzer(dialect);
    }
}

Usage of the corresponding extension point for this case is the following:

    <extension point="org.jkiss.dbeaver.lsm.dialectSyntax">
        <lsmDialect analyzerFactoryClass="org.jkiss.dbeaver.ext.sqlite.model.SQLiteSQLAnalyzer">
            <appliesTo dialectClass="org.jkiss.dbeaver.ext.sqlite.model.SQLiteSQLDialect"/>
        </lsmDialect>
    </extension>