.NET SQL Parser SDK for C# and VB.NET

Key takeaways

3-minute quick start

Evaluate GSP .NET 4.1.0.6 locally

Every snippet below was compiled against GSP .NET 4.1.0.6 and executed — the outputs in the comments are the program's real output, not illustrations. If you cannot use a public package repository, request the trial package and restore it from a local folder.

<!-- your .csproj -->
<ItemGroup>
  <PackageReference Include="gudusoft.gsqlparser" Version="4.1.0.6" />
</ItemGroup>

using gudusoft.gsqlparser;

string sql = @"select c.customer_id, sum(o.amount) total
               from customers c join orders o on c.id = o.customer_id
               group by c.customer_id";

TGSqlParser parser = new TGSqlParser(EDbVendor.dbvoracle);
parser.sqltext = sql;
int result = parser.parse();

if (result != 0)
{
    Console.WriteLine(parser.Errormessage);
    // e.g. syntax error, state:90(10101) near: from(1,8)
    return;
}

using gudusoft.gsqlparser.nodes;

for (int i = 0; i < parser.sqlstatements.size(); i++)
{
    TCustomSqlStatement stmt = parser.sqlstatements.get(i);
    Console.WriteLine(stmt.sqlstatementtype);            // sstselect

    for (int t = 0; t < stmt.tables.size(); t++)
    {
        TTable table = stmt.tables.getTable(t);
        Console.WriteLine(table.FullName + " " + table.AliasName);
        // customers c
        // orders o
    }

    for (int c = 0; c < stmt.ResultColumnList.size(); c++)
    {
        TResultColumn rc = stmt.ResultColumnList.getResultColumn(c);
        Console.WriteLine(rc.Expr + (rc.AliasClause != null ? " AS " + rc.AliasClause : ""));
        // c.customer_id
        // sum(o.amount) AS total
    }
}

Built in

Format SQL from the same parse tree

The formatter is part of the library, not a separate product — it renders the AST you already parsed, so it never reflows SQL it did not understand.

using gudusoft.gsqlparser.pp.para;
using gudusoft.gsqlparser.pp.stmtformatter;

TGSqlParser parser = new TGSqlParser(EDbVendor.dbvmssql);
parser.sqltext = "select a.id,b.name from t1 a inner join t2 b on a.id=b.id "
               + "where a.status='X' order by 1";
if (parser.parse() != 0) { Console.Error.WriteLine(parser.Errormessage); return; }

GFmtOpt option = GFmtOptFactory.newInstance();
Console.WriteLine(FormatterFactory.pp(parser, option));

SELECT   a.id,
         b.NAME
FROM     t1 a
         INNER JOIN t2 b
         ON a.id = b.id
WHERE    a.status = 'X'
ORDER BY 1

Column-level lineage

Trace a column back through views and aggregates

The dataFlowAnalyzer demo ships as C# source with the SDK. Below is its real output for a CREATE VIEW over a two-table join with a SUM() — note that TOTAL_AMOUNT is resolved to ORDERS.AMOUNT through the aggregate, and the join keys are recorded separately as a dataflow_recordset relation.

<dlineage>
  <table id="1" name="CUSTOMERS" type="table" alias="C">
    <column id="2" name="CUSTOMER_ID" />
    <column id="1" name="ID" />
  </table>
  <table id="2" name="ORDERS" type="table" alias="O">
    <column id="5" name="AMOUNT" />
  </table>
  <view id="4" name="V_CUSTOMER_TOTAL" type="view">
    <column id="9" name="TOTAL_AMOUNT" />
  </view>
  <relation type="dataflow" id="2">
    <target id="7" column="TOTAL_AMOUNT" parent_name="RESULT_OF_SELECT-QUERY" />
    <source id="5" column="AMOUNT"       parent_name="ORDERS" />
  </relation>
  <relation type="dataflow_recordset" id="5">
    <target id="7" column="TOTAL_AMOUNT" parent_name="RESULT_OF_SELECT-QUERY" function="SUM" />
    <source id="1" column="ID"           parent_name="CUSTOMERS" />
  </relation>
</dlineage>

Same engine behind Gudu SQLFlow and the lineage integrations for DataHub and OpenMetadata.

Runtime support

One package, old and new .NET

Looking for the JVM instead? See the Java SQL Parser SDK. The two editions share the same grammar lineage and AST model, with .NET exposing it through properties.

FAQ

Questions .NET evaluators actually ask