Rick, sorry it took me so long to respond. I was just starting to look at this preferred way when I noticed that you have included the line "create table t( a int );" In your sample instructions.
Does this imply that it will need to bind all of the SQL statements to a current data dictionary in order for it to parse the SQL? Unfortunately, I will not have tables associated with the SQL I parse. That was one of the advantages of the previous method - it was not required. It stopped immediately after parsing and did not require a valid data dictionary. -Kent -----Original Message----- From: Rick Hillegas [mailto:rick.hille...@oracle.com] Sent: Tuesday, October 14, 2014 8:41 AM To: Kenton Garner; Derby Discussion Subject: Re: issues.apache.org/jira/browse/DERBY-3946 Hi Kent, I'm moving this discussion to the derby-user@db.apache.org mailing list so that it will be archived in case anyone else is interested later on. The work on DERBY-3946 was superseded by the work on DERBY-4415. I've resolved DERBY-3946. Hopefully that will help point people at DERBY-4415 so that they don't spend too much time wrestling with the old approach. The solution on DERBY-4415 works with production (non-debug) engines. I have just verified that it still works with the latest 10.11.1.1 release. What I recommend is this: Write your own custom version of XmlASTPrinter. Then follow DERBY-4415's instructions for printing an xml-formatted tree: connect 'jdbc:derby:memory:dummy;create=true'; create table t( a int ); call syscs_util.syscs_register_tool( 'customTool', true, 'ASTInspector', 'XmlASTPrinter' ); select * from t; call syscs_util.syscs_register_tool( 'customTool', false, 'ASTInspector' ); - now look for the output in derbyAST.xml Feel free to ask more questions or make more suggestions, either on this email thread or on DERBY-4415. Hope this helps, -Rick On 10/13/14 3:36 PM, Kenton Garner wrote: > > Rick, > > I have been looking for a SQL Parser for a while now with no success ( > various ANTLR/ JAVACC examples came close - but no joy ). > > Today I came across a discussion about the support you added to Derby > to allow it to be used to generate Abstract Syntax Trees. This is great! > > I have to be able to take SQL statements parsed from a log for basic > DML ( INSERT, UPDATE, DELETE ) and be able to convert the SQL to a > custom XML format and potentially re-apply the SQL programmatically to > another database. > > I know your work for this was done a while ago, but I would love to > make use of it. This is what I have done so far. > > 1.Download the debug version of the Derby libs from db.apache.org > (Derby distributed with Java 7 JDK does not work here). > > 2.Update your sample ASTParser.java and TreeWalker.java to use the > "org.apache.derby.shared.common.sanity.SanityManager". > > 3.Read your discussions on the Derby jira page in regards to the > TreeWalker/Visitor and class relationships. > > 4.Ran some tests with basic INSERT SQL Statement. > > It appears that since this is using an in memory database that I do > not have to have a data dictionary or other details pre-existing in > the database for the basic SQL Parsing to work. - Great News! > > Note: I would be glad to ask my questions in another forum if you > prefer, but I do not know how to add questions to the Derby jira page. > > My question is basically... > > Since both the ASTParser and the TreeWalker make use of package level > calls that I have no access to; And knowing the node class from the > TreeWalker does not provide me any greater access to the information > that I need. How would you recommend I make best use of this code? > > My first though was that I would have to modify the Derby code to get > access to package, protected and private methods/variables - but I > would obviously rather not alter the Derby code. Do you have any > suggestions? Perhaps I am missing the obvious. > > Thank you so much. > > -Kent >