Trace each target column back to its sources in SQL Server T-SQL and Oracle SQL and PL/SQL. Embed GSP’s column-level lineage analysis in your .NET application and export relationships as XML or JSON — without a database connection.
Runs in your .NET process · No JVM · No SQL execution
This T-SQL query creates dbo.customer_revenue. GSP traces all five target
columns through bracketed identifiers, four joined tables and nested expressions.
For example, total_revenue depends on both item quantity and product unit price,
while last_order traces through FORMAT(MAX(...)) to the order date.
SELECT
c.[customer_id],
CONCAT(c.[first_name], ' ', c.[last_name]) AS [customer_name],
STRING_AGG(p.[product_name], ', ') AS [product_list],
SUM(oi.[quantity] * p.[unit_price]) AS [total_revenue],
FORMAT(MAX(o.[order_date]), 'yyyy-MM-dd') AS [last_order]
INTO [dbo].[customer_revenue]
FROM [dbo].[orders] o
JOIN [dbo].[order_items] oi ON oi.[order_id] = o.[order_id]
JOIN [dbo].[products] p ON p.[product_id] = oi.[product_id]
JOIN [dbo].[customers] c ON c.[customer_id] = o.[customer_id]
WHERE o.[status] = 'PAID'
GROUP BY c.[customer_id], c.[first_name], c.[last_name];customer_idcustomers.customer_idcustomer_namecustomers.first_namecustomers.last_nameproduct_listproducts.product_nametotal_revenueorder_items.quantityproducts.unit_pricelast_orderorders.order_dateA readable summary of the generated column relationships. The files preserve object IDs and SQL coordinates.
Both worked examples were compiled and run with public NuGet 4.1.0.18. The output files are actual SDK output, not illustrative JSON or hand-written XML.
Turn the .NET tool’s output into an interactive lineage diagram. Copy the complete XML or SQLFlow JSON, paste it into Gudu SQL Omni for VS Code or SQLFlow online, and explore the source-to-target column connections.
Try the SQL Server example from this page:
Copy a sample, then paste it into either viewer.
RelationRows field is also visible.
Open full-size image →
Use the same C# API with EDbVendor.dbvoracle. In this Oracle view,
customer_totals.customer_id comes from orders.customer_id;
total_revenue comes from order_items.quantity and
order_items.unit_price through SUM.
The current .NET implementation extends analysis into Oracle PL/SQL packages, procedures, functions, triggers, cursors and resolvable dynamic SQL. Request a current evaluation build and validate your own scripts: recent advanced Oracle changes are not all included in the public package used for these examples.
CREATE VIEW customer_totals AS
SELECT
o.customer_id,
SUM(i.quantity * i.unit_price) AS total_revenue
FROM orders o
JOIN order_items i ON i.order_id = o.order_id
WHERE o.status = 'PAID'
GROUP BY o.customer_id;Download the example project and unzip it. With the .NET 10 SDK installed, run these commands from the extracted directory:
dotnet run
dotnet run -- oracle
The first command writes sqlserver.xml and sqlserver.json.
The second writes oracle.xml and oracle.json.
The project pins gudusoft.gsqlparser 4.1.0.18 for reproducibility.
Project file · C# source · NuGet package
Construct DataFlowAnalyzer with SQL text and a dialect, call
generateDataFlow(), inspect diagnostics, then dispose of the analyzer.
The true constructor argument selects simplified lineage, connecting
final targets to their value sources. Use full mode when you need intermediate result sets.
getDataFlow() exposes the model. getOption() exposes analysis
and output options. These API names are Java-compatible; your application still runs entirely on .NET.
using gudusoft.gsqlparser;
using gudusoft.gsqlparser.dlineage;
using gudusoft.gsqlparser.util.json;
// Run from this directory: dotnet run, or dotnet run -- oracle
bool oracle = args.Length > 0 && args[0] == "oracle";
string name = oracle ? "oracle" : "sqlserver";
string sql = File.ReadAllText(name + ".sql");
var vendor = oracle ? EDbVendor.dbvoracle : EDbVendor.dbvmssql;
var analyzer = new DataFlowAnalyzer(sql, vendor, true);
try
{
string xml = analyzer.generateDataFlow();
foreach (var error in analyzer.getErrorMessages())
Console.Error.WriteLine(error.getErrorMessage());
if (analyzer.getErrorMessages().Count > 0)
return 1;
File.WriteAllText(name + ".xml", xml);
var model = DataFlowAnalyzer.getSqlflowJSONModel(analyzer.getDataFlow(), vendor);
File.WriteAllText(name + ".json", JSON.toJSONString(model));
Console.WriteLine("Saved " + name + ".xml and " + name + ".json");
return 0;
}
finally
{
analyzer.dispose();
}Build impact analysis by following source-to-target edges, explain an ETL mapping, or feed a data catalog. Choose an output format for the integration you are building.
<dlineage> model includes objects, columns and <relationship> edges. generateDataFlow() returns this format directly.dlineage demo supports detailed or simplified column mappings and table-level CSV for review and downstream processing./json /graph or /lv output when integrating a visualization into your own application. To explore the standard XML or SQLFlow JSON directly, use Gudu SQL Omni or SQLFlow online.
Clone gsp_demo_dotnet, save a SQL Server query as query.sql, and run the commands below from the repository root.
For Oracle, use /t oracle. Use /csv or /tableLineage for other projections.
See the full dlineage user guide for offline metadata, directory input and output options.
# From the gsp_demo_dotnet repository root
dotnet build src/demos/dlineage/demos.dlineage.csproj -c Release -p:GspVersion=4.1.0.18
dotnet src/demos/dlineage/bin/Release/net10.0/dlineage.dll /f query.sql /t mssql /s /json /o lineage.jsonThe explicit package version enables the formal lineage API; the demo repository’s older default package pin does not.
Reference gudusoft.gsqlparser, choose the database dialect, create a DataFlowAnalyzer and call generateDataFlow(). Read the resulting source-to-target relationships as XML or use the model API. The downloadable C# project on this page also exports SQLFlow JSON. No database connection or JVM is required.
Yes. The example on this page traces five SELECT INTO target columns through SQL Server joins, CONCAT, STRING_AGG, SUM, MAX and FORMAT. It was compiled and run with public NuGet 4.1.0.18. For larger batches, stored procedures and dynamic SQL, evaluate representative scripts with the build you plan to deploy.
Yes. The downloadable Oracle CREATE VIEW example traces customer_id and aggregated revenue to their source columns on public NuGet 4.1.0.18. For advanced PL/SQL packages, procedures, functions, triggers, cursors and dynamic SQL, request a current evaluation build. Recent advanced Oracle changes are not all included in that public package; validate your scripts with the supplied build.
A parser produces a syntax tree and can enumerate tables and columns. A lineage engine resolves references and follows how source values contribute to target columns through expressions, queries and statements. GSP provides both in .NET. This makes it suitable for impact analysis, ETL mappings and data catalog integrations.
Yes. generateDataFlow() returns XML. The model can be converted to SQLFlow JSON, as shown in the downloadable C# project. The dlineage demo also offers detailed and simplified CSV, table-level lineage and graph JSON. Paste the XML or SQLFlow JSON into Gudu SQL Omni in VS Code or the online SQLFlow tool to explore an interactive diagram.
Use Gudu SQL Omni for VS Code or SQLFlow online. In VS Code, paste the complete XML or SQLFlow JSON into an editor and run GSP: Data Lineage (Ctrl+Alt+L, or Cmd+Alt+L on macOS). Online, sign in if prompted, paste the same output into the SQLFlow editor and click Visualize Lineage. Both viewers let you zoom, pan and explore column connections. Use the dlineage XML or SQLFlow JSON generated by the .NET tool, including the sample files on this page.
Yes. GSP analyzes SQL text in your .NET process without executing it or connecting to the database. Supply offline schema metadata when wildcard expansion or ambiguous unqualified columns require catalog context. Dynamic SQL can be analyzed when its statement text can be resolved; SQL text alone cannot reveal arbitrary runtime-generated statements.
No. DataFlowAnalyzer, Option, generateDataFlow() and the model methods use Java-compatible API names, but the implementation runs as C# on .NET. The example targets .NET 10. The package also supplies a netstandard2.0 assembly for compatible hosts, including .NET Framework 4.6.2 and later.
Yes. Start with the public NuGet trial and the examples on this page. The public trial rejects SQL longer than 10,000 characters. Request an evaluation build for longer scripts, advanced procedure analysis and confirmation of your dialect, runtime and redistribution requirements.
Start with the runnable examples, then send a representative query or stored procedure. Tell us you need .NET data lineage, your target runtime and the output you expect. We can recommend an evaluation build and review the SQL with you.
The public NuGet trial has a 10,000-character SQL limit. Request an evaluation build for larger scripts and the latest advanced lineage support.