PLSQL parsing using Oracle.VsDevTools.SqlAndPlsqlParser - oracle

Question

I downloaded Oracle Developer Tools for Visual studio and explored the libraries (.dll). I need to parse the PLSQL scripts to get the SQL Statements in that and using the visitors to visit the nodes. I found Oracle.VsDevTools.SqlAndPlsqlParser library and explored further and tried to parse the PLSQL scripts. Please refer the below code.
string plSQLScript="CREATE TABLE Student(Age int);";
List<LexerToken> tokens= LexerToken.Parse(plSQLScript, false, true);
But the above code gives tokens only, I need PLSQL parser.
Is there any possible way to parser the PLSQL scripts and get the sql statements. Also I need to explicit visit the sql statements like CreateTableVisitor, CreateProcedureVisitor, etc.
I already created the parser for TSQL files by exploring Microsoft SQL library called Microsoft.SqlServer.TransactSql.ScriptDom and using the below mentioned code for parsing.
TSql100Parser parser = new TSql100Parser(true);
IList<ParseError> error;
using (TextReader tr = new StreamReader(filePath))
{
TSqlFragment fragment = parser.Parse(tr, out error);
tr.Close();
}
And using TSqlFragmentVisitor to visit the sql statements. Please refer the below mentioned code.
public override void ExplicitVisit(CreateTableStatement node)
{
//----Coding----
}
I want the same for PLSQL by using Oracle.VsDevTools.SqlAndPlsqlParser library.
Please let me know if there any possibility for this.
Thanks,Sivaprakash.

Answer 1

With some reflection used it's possible. The parser, nodes and grammar classes are internal. I made small example for you:
private static Assembly odacDevAssembly = Assembly.Load("Oracle.VsDevTools.14.0");
private static Type parseNodeType = odacDevAssembly.GetType("Oracle.VsDevTools.SqlAndPlsqlParser.ParseNode");
void Main()
{
var parserType = odacDevAssembly.GetType("Oracle.VsDevTools.SqlAndPlsqlParser.OracleSqlEarley");
var parser = Activator.CreateInstance(parserType);
const string sqlScriptText = "DECLARE x NUMBER; BEGIN SELECT DUMMY INTO :DUMMY FROM DUAL T1; x := 1; SELECT DUMMY INTO :DUMMY FROM DUAL T2; x := 2; INSERT INTO T3 (C) VALUES (1); x := 3; END;";
var tokens = LexerToken.Parse(sqlScriptText);
var parserGrammar = parserType.GetProperty("EarleyGrammar").GetValue(parser);
var allSymbols = (string[])parserGrammar.GetType().GetField("m_vAllSymbols", BindingFlags.Public | BindingFlags.Instance).GetValue(parserGrammar);
var sqlStatementIndex = -1;
for (var i = 0; i < allSymbols.Length; i++)
{
if (allSymbols[i] == "unlabeled_nonblock_stmt")
{
sqlStatementIndex = i;
break;
}
}
var parseTree = parserType.GetMethod("Parse").Invoke(parser, new object[] { sqlScriptText, tokens });
//parseTree.GetType().GetMethod("PrintTree", new Type[] { parserGrammar.GetType() }).Invoke(parseTree, new object[] { parserGrammar });
var commandNodes = GetDescendants(parseTree).Where(d => GetPayloadIn(d) == sqlStatementIndex);
foreach (var commandNode in commandNodes)
{
var commandText = GetCommandText(commandNode, sqlScriptText, tokens);
Console.WriteLine(commandText);
}
}
private static string GetCommandText(object parseNode, string sqlScriptText, List<LexerToken> tokens)
{
var from = (int)parseNodeType.GetProperty("From").GetValue(parseNode);
var to = (int)parseNodeType.GetProperty("To").GetValue(parseNode) - 1;
var begin = tokens[from].m_vBegin;
return sqlScriptText.Substring(begin, tokens[to].m_vEnd - begin);
}
private static IEnumerable<object> GetDescendants(object parseNode)
{
yield return parseNode;
foreach (var child in GetChildren(parseNode))
foreach (var descendant in GetDescendants(child))
yield return descendant;
}
private static int GetPayloadIn(object parseNode)
{
return (int)parseNodeType.GetProperty("PayloadIn").GetValue(parseNode);
}
private static IEnumerable<object> GetChildren(object parseNode)
{
return (IEnumerable<object>)parseNodeType.GetMethod("Children").Invoke(parseNode, null);
}
UPDATE:
I'm not exactly sure what do you want to achieve but I hope it will be at least somehow helpful:
private static Assembly odacDevAssembly = Assembly.Load("Oracle.VsDevTools.14.0");
private static Type parseNodeType = odacDevAssembly.GetType("Oracle.VsDevTools.SqlAndPlsqlParser.ParseNode");
void Main()
{
var parserType = odacDevAssembly.GetType("Oracle.VsDevTools.SqlAndPlsqlParser.OracleSqlEarley");
var parser = Activator.CreateInstance(parserType);
const string sqlScriptText = "DECLARE x NUMBER; BEGIN SELECT DUMMY INTO :DUMMY FROM DUAL T1; x := 1; SELECT DUMMY INTO :DUMMY FROM DUAL T2; x := 2; INSERT INTO T3 (C) VALUES (1); x := 3; END;";
var tokens = LexerToken.Parse(sqlScriptText);
var parserGrammar = parserType.GetProperty("EarleyGrammar").GetValue(parser);
var allSymbols = (string[])parserGrammar.GetType().GetField("m_vAllSymbols", BindingFlags.Public | BindingFlags.Instance).GetValue(parserGrammar);
const int sqlStatementIndex = 4453; // unlabeled_nonblock_stmt
var parseTree = parserType.GetMethod("Parse").Invoke(parser, new object[] { sqlScriptText, tokens });
//parseTree.GetType().GetMethod("PrintTree", new Type[] { parserGrammar.GetType() }).Invoke(parseTree, new object[] { parserGrammar });
var commandNodes = GetDescendants(parseTree)
.Select(n => ParseNodeFactory.CreateNode(n, sqlScriptText, tokens))
.Where(n => n != null);
var visitor = new GrammarNodeVisitor();
foreach (var commandNode in commandNodes)
{
commandNode.Accept(visitor);
}
}
public static class ParseNodeFactory
{
private const int queryBlockIndex = 3849; // query_block
private const int insertIndex = 3136; // insert
public static IParseNode CreateNode(object parseNode, string sqlScriptText, List<LexerToken> tokens)
{
var symbolIndex = GetPayloadIn(parseNode);
var parseData = new ParseData { ParseNode = parseNode, SqlScriptText = sqlScriptText, Tokens = tokens };
switch (symbolIndex)
{
case queryBlockIndex:
return new QueryBlock { ParseData = parseData };
case insertIndex:
return new Insert { ParseData = parseData };
default:
return null;
}
}
}
public class GrammarNodeVisitor : IParseNodeVisitor
{
public void VisitQueryBlock(QueryBlock queryBlock)
{
Console.WriteLine($"Visited query block: {GetCommandText(queryBlock.ParseData.ParseNode, queryBlock.ParseData.SqlScriptText, queryBlock.ParseData.Tokens)}");
}
public void VisitInsert(Insert insert)
{
Console.WriteLine($"Visited insert command: {GetCommandText(insert.ParseData.ParseNode, insert.ParseData.SqlScriptText, insert.ParseData.Tokens)}");
}
}
public interface IParseNodeVisitor
{
void VisitQueryBlock(QueryBlock queryBlock);
void VisitInsert(Insert insert);
}
public interface IParseNode
{
ParseData ParseData { get; }
void Accept(IParseNodeVisitor visitor);
}
public class ParseData
{
public object ParseNode { get; set; }
public string SqlScriptText { get; set; }
public List<LexerToken> Tokens { get; set; }
}
public class QueryBlock : IParseNode
{
public ParseData ParseData { get; set; }
public void Accept(IParseNodeVisitor visitor)
{
visitor.VisitQueryBlock(this);
}
}
public class Insert : IParseNode
{
public ParseData ParseData { get; set; }
public void Accept(IParseNodeVisitor visitor)
{
visitor.VisitInsert(this);
}
}
private static string GetCommandText(object parseNode, string sqlScriptText, List<LexerToken> tokens)
{
var from = (int)parseNodeType.GetProperty("From").GetValue(parseNode);
var to = (int)parseNodeType.GetProperty("To").GetValue(parseNode) - 1;
var begin = tokens[from].m_vBegin;
return sqlScriptText.Substring(begin, tokens[to].m_vEnd - begin);
}
private static IEnumerable<object> GetDescendants(object parseNode)
{
yield return parseNode;
foreach (var child in GetChildren(parseNode))
foreach (var descendant in GetDescendants(child))
yield return descendant;
}
private static int GetPayloadIn(object parseNode)
{
return (int)parseNodeType.GetProperty("PayloadIn").GetValue(parseNode);
}
private static IEnumerable<object> GetChildren(object parseNode)
{
return (IEnumerable<object>)parseNodeType.GetMethod("Children").Invoke(parseNode, null);
}