feature: Improve string representation of datafusion classes

src/catalog.rs

@@ -81,6 +81,13 @@ impl PyCatalog {
             ))),
         }
     }
+
+    fn __repr__(&self) -> PyResult<String> {
+        Ok(format!(
+            "Catalog(schema_names=[{}])",
+            self.names().join(";")
+        ))
+    }
 }
 
 #[pymethods]
@@ -97,6 +104,13 @@ impl PyDatabase {
         }
     }
 
+    fn __repr__(&self) -> PyResult<String> {
+        Ok(format!(
+            "Database(table_names=[{}])",
+            Vec::from_iter(self.names()).join(";")
+        ))
+    }
+
     // register_table
     // deregister_table
 }
@@ -119,6 +133,11 @@ impl PyTable {
         }
     }
 
+    fn __repr__(&self) -> PyResult<String> {
+        let kind = self.kind();
+        Ok(format!("Table(kind={kind})"))
+    }
+
     // fn scan
     // fn statistics
     // fn has_exact_statistics

src/config.rs

@@ -72,6 +72,14 @@ impl PyConfig {
         }
         Ok(dict.into())
     }
+
+    fn __repr__(&mut self, py: Python) -> PyResult<String> {
+        let dict = self.get_all(py);
+        match dict {
+            Ok(result) => Ok(format!("Config({result})")),
+            Err(err) => Ok(format!("Error: {:?}", err.to_string())),
+        }
+    }
 }
 
 /// Convert a python object to a ScalarValue

src/context.rs

@@ -460,6 +460,14 @@ impl PySessionContext {
         };
         Ok(PyDataFrame::new(df))
     }
+
+    fn __repr__(&self) -> PyResult<String> {
+        let id = self.session_id();
+        match id {
+            Ok(value) => Ok(format!("SessionContext(session_id={value})")),
+            Err(err) => Ok(format!("Error: {:?}", err.to_string())),
+        }
+    }
 }
 
 impl PySessionContext {

src/dataframe.rs

@@ -66,6 +66,16 @@ impl PyDataFrame {
         })
     }
 
+    fn __repr__(&self, py: Python) -> PyResult<String> {
+        let df = self.df.as_ref().clone().limit(0, Some(10))?;
+        let batches = wait_for_future(py, df.collect())?;
+        let batches_as_string = pretty::pretty_format_batches(&batches);
+        match batches_as_string {
+            Ok(batch) => Ok(format!("DataFrame()\n{batch}")),
+            Err(err) => Ok(format!("Error: {:?}", err.to_string())),
+        }
+    }
+
     /// Returns the schema from the logical plan
     fn schema(&self) -> PyArrowType<Schema> {
         PyArrowType(self.df.schema().into())

src/expression.rs

@@ -57,8 +57,8 @@ impl PyExpr {
         expr.into()
     }
 
-    fn __str__(&self) -> PyResult<String> {
-        Ok(format!("{}", self.expr))
+    fn __repr__(&self) -> PyResult<String> {
+        Ok(format!("Expr({})", self.expr))
     }
 
     fn __add__(&self, rhs: PyExpr) -> PyResult<PyExpr> {

src/udaf.rs

@@ -141,4 +141,8 @@ impl PyAggregateUDF {
         let args = args.iter().map(|e| e.expr.clone()).collect();
         Ok(self.function.call(args).into())
     }
+
+    fn __repr__(&self) -> PyResult<String> {
+        Ok(format!("AggregateUDF({})", self.function.name))
+    }
 }

src/udf.rs

@@ -92,4 +92,8 @@ impl PyScalarUDF {
         let args = args.iter().map(|e| e.expr.clone()).collect();
         Ok(self.function.call(args).into())
     }
+
+    fn __repr__(&self) -> PyResult<String> {
+        Ok(format!("ScalarUDF({})", self.function.name))
+    }
 }