Extend SQL dialect?

Any SQL dialect uses generated + handwritten lexers approach.
So you can hack OracleLexer or even SqlLexer class instead. ;-)

I've added "." as allowed character so your case will be covered.
Expect the fix in the upcoming 12.0.2 EAP later this week.

Thanks Gregory. I extended OracleLexer and added a method similar to lookCustomParameter(). I must say it was a bit challenging without any doc/source code, but I managed to get it working.
Now I hope that since the code is not part of the OpenAPI, it won't change very often, otherwise I'll have to modify my plugin...

Bastien, would you mind sharing your code?  I am looking to do something similar for Anorm which uses named parameters delinated with {}.

Here you go: I have overriden the default OracleLexer to do custom lookaheads for parameters within $, % or #. The content of lookKasperParameter is based on the original code in OracleLexer (haven't tested it in IntelliJ 13 though).

 
package org.intellij.plugins.kasper.parser.sql;

import com.intellij.lexer.Lexer;
import com.intellij.sql.dialects.oracle.OracleLexer;
import com.intellij.sql.psi.SqlTokens;

public class KOracleLexer extends OracleLexer {

    @Override
    protected void lookAhead(Lexer baseLexer) {
        if (!lookKasperParameter(baseLexer)) {
            super.lookAhead(baseLexer);
        }
    }

    protected boolean lookKasperParameter(Lexer baseLexer) {
        CharSequence text = baseLexer.getBufferSequence();
        int tokenStart = baseLexer.getTokenStart();
        int len = baseLexer.getBufferEnd();
        if ((len > tokenStart) && (
                (text.charAt(tokenStart) == '#') || (text.charAt(tokenStart) == '$') || (text.charAt(tokenStart) == '%') || (text.charAt(tokenStart) == '@'))) {
            int stringStart = tokenStart + 1;
            String endDelimiter = String.valueOf(text.subSequence(tokenStart, tokenStart + 1));
            return processCustomQuotation(baseLexer, endDelimiter, stringStart, true, SqlTokens.SQL_CUSTOM_PARAM_LQUOTE, SqlTokens.SQL_CUSTOM_PARAM_RQUOTE, SqlTokens.SQL_IDENT);
        }

        return false;
    }
}

This lexer is used by a custom parser definition:
 
package org.intellij.plugins.kasper.parser.sql;

import com.intellij.lexer.Lexer;
import com.intellij.openapi.project.Project;
import com.intellij.psi.tree.IFileElementType;
import com.intellij.sql.dialects.oracle.OracleParserDefinition;
import org.jetbrains.annotations.NotNull;

public class KOracleSqlParserDefinition extends OracleParserDefinition {
    @NotNull
    @Override
    public Lexer createLexer(Project project) {
        return new KOracleLexer();
    }

    @Override
    public IFileElementType getFileNodeType() {
        return KOracleDialect.ORACLE_SQL_FILE;
    }
}

(KOracleDialect does nothing in particular except returning a custom dialect name, everything else is the same as in OracleDialect.INSTANCE).

Hope it helps.
Bastien

Thanks, Bastien.  I should be able to make it work from your examples, but I can't find the classes you're extending anywhere.  I thought it would be easier, but this plugin may be over my head :8}.

Oh, sorrry, I forgot to mention that I also modified plugin.xml to add extra dependencies to database libraries:

        <depends>com.intellij.modules.lang</depends>
        <depends optional="true">com.intellij.sql</depends>
        <depends optional="true">com.intellij.persistence.database</depends>
        <depends optional="true">com.intellij.javaee</depends>

IIRC IntelliJ will automatically add the corresponding jars to the classpath, and you will be able to override the needed classes. Keep trying, plugins can be hard to write but so much enjoying once they work ;)

Don't I still need the jars from somewhere to be able to make it?