fix: make it compile

Signed-off-by: Andrew Lamb <[email protected]>
Signed-off-by: Liam Brannigan <[email protected]>

Author: alamb

crates/core/src/data_catalog/storage/mod.rs

@@ -149,7 +149,7 @@ mod tests {
     use super::*;
     use datafusion::assert_batches_sorted_eq;
     use datafusion::catalog::CatalogProvider;
-    use datafusion::catalog_common::MemoryCatalogProvider;
+    use datafusion::catalog::MemoryCatalogProvider;
     use datafusion::execution::context::SessionContext;
 
     #[test]

crates/core/src/delta_datafusion/mod.rs

@@ -51,8 +51,8 @@ use datafusion::physical_optimizer::pruning::PruningPredicate;
 use datafusion_common::scalar::ScalarValue;
 use datafusion_common::tree_node::{TreeNode, TreeNodeRecursion, TreeNodeVisitor};
 use datafusion_common::{
-    config::ConfigOptions, Column, DFSchema, DataFusionError, Result as DataFusionResult,
-    TableReference, ToDFSchema,
+    config::ConfigOptions, Column, Constraints, DFSchema, DataFusionError,
+    Result as DataFusionResult, TableReference, ToDFSchema,
 };
 use datafusion_expr::execution_props::ExecutionProps;
 use datafusion_expr::logical_plan::CreateExternalTable;
@@ -659,6 +659,7 @@ impl<'a> DeltaScanBuilder<'a> {
             } else {
                 file_groups.into_values().collect()
             },
+            constraints: Constraints::default(),
             statistics: stats,
             projection: self.projection.cloned(),
             limit: self.limit,

crates/core/src/kernel/snapshot/log_data.rs

@@ -638,6 +638,7 @@ mod datafusion {
                 null_count,
                 max_value,
                 min_value,
+                sum_value: Precision::Absent,
                 distinct_count: Precision::Absent,
             })
         }
@@ -653,6 +654,7 @@ mod datafusion {
                 null_count: self.null_count.add(&other.null_count),
                 max_value: self.max_value.max(&other.max_value),
                 min_value: self.min_value.min(&other.min_value),
+                sum_value: Precision::Absent,
                 distinct_count: self.distinct_count.add(&other.distinct_count),
             }
         }

crates/core/src/operations/load_cdf.rs

@@ -10,7 +10,7 @@ use datafusion::datasource::file_format::parquet::ParquetFormat;
 use datafusion::datasource::file_format::FileFormat;
 use datafusion::datasource::physical_plan::FileScanConfig;
 use datafusion::prelude::SessionContext;
-use datafusion_common::{ScalarValue, Statistics};
+use datafusion_common::{Constraints, ScalarValue, Statistics};
 use datafusion_physical_expr::expressions;
 use datafusion_physical_expr::PhysicalExpr;
 use datafusion_physical_plan::projection::ProjectionExec;
@@ -389,6 +389,7 @@ impl CdfLoadBuilder {
                     object_store_url: self.log_store.object_store_url(),
                     file_schema: cdc_file_schema.clone(),
                     file_groups: cdc_file_groups.into_values().collect(),
+                    constraints: Constraints::default(),
                     statistics: Statistics::new_unknown(&cdc_file_schema),
                     projection: None,
                     limit: None,
@@ -406,6 +407,7 @@ impl CdfLoadBuilder {
                     object_store_url: self.log_store.object_store_url(),
                     file_schema: add_remove_file_schema.clone(),
                     file_groups: add_file_groups.into_values().collect(),
+                    constraints: Constraints::default(),
                     statistics: Statistics::new_unknown(&add_remove_file_schema.clone()),
                     projection: None,
                     limit: None,
@@ -423,6 +425,7 @@ impl CdfLoadBuilder {
                     object_store_url: self.log_store.object_store_url(),
                     file_schema: add_remove_file_schema.clone(),
                     file_groups: remove_file_groups.into_values().collect(),
+                    constraints: Constraints::default(),
                     statistics: Statistics::new_unknown(&add_remove_file_schema),
                     projection: None,
                     limit: None,

crates/core/src/writer/json.rs

@@ -349,6 +349,8 @@ impl DeltaWriter<Vec<Value>> for JsonWriter {
                         ParquetError::IndexOutOfBound(u.to_owned(), v.to_owned())
                     }
                     ParquetError::NYI(msg) => ParquetError::NYI(msg.to_owned()),
+                    // ParquetError is non exhaustive, so have a fallback
+                    e => ParquetError::General(e.to_string()),
                 },
                 skipped_values: partial_writes,
             }

python/src/filesystem.rs

@@ -4,6 +4,7 @@ use crate::RawDeltaTable;
 use deltalake::storage::object_store::{MultipartUpload, PutPayloadMut};
 use deltalake::storage::{DynObjectStore, ListResult, ObjectStoreError, Path};
 use deltalake::DeltaTableBuilder;
+use parking_lot::Mutex;
 use pyo3::exceptions::{PyIOError, PyNotImplementedError, PyValueError};
 use pyo3::prelude::*;
 use pyo3::types::{IntoPyDict, PyBytes, PyType};
@@ -519,7 +520,9 @@ impl ObjectInputFile {
 // TODO the C++ implementation track an internal lock on all random access files, DO we need this here?
 #[pyclass(weakref, module = "deltalake._internal")]
 pub struct ObjectOutputStream {
-    upload: Box<dyn MultipartUpload>,
+    // wrap in mutex as rustc says `MultipartUpload` can't be
+    // shared across threads (it isn't sync)
+    upload: Mutex<Box<dyn MultipartUpload>>,
     pos: i64,
     #[pyo3(get)]
     closed: bool,
@@ -537,7 +540,7 @@ impl ObjectOutputStream {
     ) -> Result<Self, ObjectStoreError> {
         let upload = store.put_multipart(&path).await?;
         Ok(Self {
-            upload,
+            upload: Mutex::new(upload),
             pos: 0,
             closed: false,
             mode: "wb".into(),
@@ -555,14 +558,15 @@ impl ObjectOutputStream {
     }
 
     fn abort(&mut self) -> PyResult<()> {
-        rt().block_on(self.upload.abort())
+        rt().block_on(self.upload.lock().abort())
             .map_err(PythonError::from)?;
         Ok(())
     }
 
     fn upload_buffer(&mut self) -> PyResult<()> {
         let payload = std::mem::take(&mut self.buffer).freeze();
-        match rt().block_on(self.upload.put_part(payload)) {
+        let res = rt().block_on(self.upload.lock().put_part(payload));
+        match res {
             Ok(_) => Ok(()),
             Err(err) => {
                 self.abort()?;
@@ -580,7 +584,7 @@ impl ObjectOutputStream {
             if !self.buffer.is_empty() {
                 self.upload_buffer()?;
             }
-            match rt().block_on(self.upload.complete()) {
+            match rt().block_on(self.upload.lock().complete()) {
                 Ok(_) => Ok(()),
                 Err(err) => Err(PyIOError::new_err(err.to_string())),
             }

python/src/lib.rs

@@ -1563,10 +1563,12 @@ impl RawDeltaTable {
         &self,
         py: Python<'py>,
     ) -> PyResult<Bound<'py, PyCapsule>> {
+        // tokio runtime handle?
+        let handle = None;
         let name = CString::new("datafusion_table_provider").unwrap();
 
         let table = self.with_table(|t| Ok(Arc::new(t.clone())))?;
-        let provider = FFI_TableProvider::new(table, false);
+        let provider = FFI_TableProvider::new(table, false, handle);
 
         PyCapsule::new_bound(py, provider, Some(name.clone()))
     }
@@ -1820,7 +1822,7 @@ fn filestats_to_expression_next<'py>(
             })?
             .data_type()
             .clone();
-        let column_type = PyArrowType(column_type).into_py(py);
+        let column_type = PyArrowType(column_type).into_pyobject(py)?;
         pa.call_method1("scalar", (value,))?
             .call_method1("cast", (column_type,))
     };