Add DuckDB-style Table Functions for Catalog Metadata

README.md

@@ -16,6 +16,8 @@ A DataFusion extension for querying [DuckLake](https://ducklake.select). DuckLak
 - Parquet footer size hints for optimized I/O
 - Filter pushdown to Parquet for row group pruning and page-level filtering
 - Dynamic metadata lookup (no upfront catalog caching)
+- SQL-queryable `information_schema` for catalog metadata (snapshots, schemas, tables, columns, files)
+- DuckDB-style table functions: `ducklake_snapshots()`, `ducklake_table_info()`, `ducklake_list_files()`
 
 ## Known Limitations
 

examples/basic_query.rs

@@ -13,7 +13,7 @@
 
 use datafusion::execution::runtime_env::RuntimeEnv;
 use datafusion::prelude::*;
-use datafusion_ducklake::{DuckLakeCatalog, DuckdbMetadataProvider};
+use datafusion_ducklake::{DuckLakeCatalog, DuckdbMetadataProvider, register_ducklake_functions};
 use object_store::ObjectStore;
 use object_store::aws::AmazonS3Builder;
 use std::env;
@@ -66,10 +66,19 @@ async fn main() -> Result<(), Box<dyn std::error::Error>> {
     // Create DataFusion session context
     let ctx = SessionContext::new_with_config_rt(config, runtime.clone());
 
-    // Register the DuckLake catalog
+    // Get the provider before moving the catalog
+    let provider = ducklake_catalog.provider();
+
+    // Register the DuckLake catalog (standard DataFusion pattern)
     ctx.register_catalog("ducklake", Arc::new(ducklake_catalog));
 
+    // Register table functions (ducklake_snapshots, ducklake_table_info, ducklake_list_files)
+    register_ducklake_functions(&ctx, provider);
+
     println!("✓ Registered DuckLake catalog with DataFusion");
+    println!(
+        "✓ Registered table functions (ducklake_snapshots, ducklake_table_info, ducklake_list_files)"
+    );
 
     // List available schemas
     let catalogs = ctx.catalog_names();

src/catalog.rs

@@ -4,6 +4,7 @@ use std::any::Any;
 use std::sync::Arc;
 
 use crate::Result;
+use crate::information_schema::InformationSchemaProvider;
 use crate::metadata_provider::MetadataProvider;
 use crate::path_resolver::parse_object_store_url;
 use crate::schema::DuckLakeSchema;
@@ -38,6 +39,13 @@ impl DuckLakeCatalog {
         })
     }
 
+    /// Get the metadata provider for this catalog
+    ///
+    /// This is useful when you need to register table functions separately.
+    pub fn provider(&self) -> Arc<dyn MetadataProvider> {
+        self.provider.clone()
+    }
+
     fn get_current_snapshot_id(&self) -> Result<i64> {
         self.provider
             .get_current_snapshot()
@@ -53,25 +61,38 @@ impl CatalogProvider for DuckLakeCatalog {
     fn schema_names(&self) -> Vec<String> {
         let snapshot_id = match self.get_current_snapshot_id() {
             Ok(id) => id,
-            Err(_) => return Vec::new(),
+            Err(_) => return vec!["information_schema".to_string()],
         };
 
-        // Query database with snapshot_id
-        self.provider
+        // Start with information_schema
+        let mut names = vec!["information_schema".to_string()];
+
+        // Add data schemas from catalog
+        let data_schemas = self
+            .provider
             .list_schemas(snapshot_id)
             .unwrap_or_default()
             .into_iter()
-            .map(|s| s.schema_name)
-            .collect()
+            .map(|s| s.schema_name);
+
+        names.extend(data_schemas);
+        names
     }
 
     fn schema(&self, name: &str) -> Option<Arc<dyn SchemaProvider>> {
+        // Handle information_schema specially
+        if name == "information_schema" {
+            return Some(Arc::new(InformationSchemaProvider::new(Arc::clone(
+                &self.provider,
+            ))));
+        }
+
         let snapshot_id = match self.get_current_snapshot_id() {
             Ok(id) => id,
             Err(_) => return None,
         };
 
-        // Query database with snapshot_id
+        // Query database with snapshot_id for data schemas
         match self.provider.get_schema_by_name(name, snapshot_id) {
             Ok(Some(meta)) => {
                 // Resolve schema path hierarchically