fix: workaround for Add actions being read slightly differently out of parquet files

Cargo.toml

@@ -27,7 +27,7 @@ debug = "line-tables-only"
 
 [workspace.dependencies]
 delta_kernel = { version = "0.4.1", features = ["sync-engine"] }
-# delta_kernel = { path = "../delta-kernel-rs/kernel", version = "0.3.0" }
+#delta_kernel = { path = "../delta-kernel-rs/kernel", features = ["sync-engine"]  }
 
 # arrow
 arrow = { version = "53" }

crates/core/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "deltalake-core"
-version = "0.22.0"
+version = "0.22.1"
 authors.workspace = true
 keywords.workspace = true
 readme.workspace = true

crates/core/src/kernel/models/fields.rs

@@ -258,11 +258,11 @@ fn deletion_vector_field() -> StructField {
     StructField::new(
         "deletionVector",
         DataType::Struct(Box::new(StructType::new(vec![
-            StructField::new("storageType", DataType::STRING, true),
-            StructField::new("pathOrInlineDv", DataType::STRING, true),
+            StructField::new("storageType", DataType::STRING, false),
+            StructField::new("pathOrInlineDv", DataType::STRING, false),
             StructField::new("offset", DataType::INTEGER, true),
-            StructField::new("sizeInBytes", DataType::INTEGER, true),
-            StructField::new("cardinality", DataType::LONG, true),
+            StructField::new("sizeInBytes", DataType::INTEGER, false),
+            StructField::new("cardinality", DataType::LONG, false),
         ]))),
         true,
     )

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

@@ -245,14 +245,23 @@ impl LogicalFile<'_> {
 
     /// Defines a deletion vector
     pub fn deletion_vector(&self) -> Option<DeletionVectorView<'_>> {
-        self.deletion_vector.as_ref().and_then(|arr| {
-            arr.storage_type
-                .is_valid(self.index)
-                .then_some(DeletionVectorView {
+        if let Some(arr) = self.deletion_vector.as_ref() {
+            // With v0.22 and the upgrade to a more recent arrow. Reading nullable structs with
+            // non-nullable entries back out of parquet is resulting in the DeletionVector having
+            // an empty string rather than a null. The addition check on the value ensures that a
+            // [DeletionVectorView] is not created in this scenario
+            //
+            // <https://github.com/delta-io/delta-rs/issues/3030>
+            if arr.storage_type.is_valid(self.index)
+                && !arr.storage_type.value(self.index).is_empty()
+            {
+                return Some(DeletionVectorView {
                     data: arr,
                     index: self.index,
-                })
-        })
+                });
+            }
+        }
+        None
     }
 
     /// The number of records stored in the data file.
@@ -509,7 +518,7 @@ mod datafusion {
         fn collect_count(&self, name: &str) -> Precision<usize> {
             let num_records = extract_and_cast_opt::<Int64Array>(self.stats, name);
             if let Some(num_records) = num_records {
-                if num_records.len() == 0 {
+                if num_records.is_empty() {
                     Precision::Exact(0)
                 } else if let Some(null_count_mulls) = num_records.nulls() {
                     if null_count_mulls.null_count() > 0 {

crates/core/src/protocol/checkpoints.rs

@@ -567,6 +567,7 @@ mod tests {
     use crate::operations::DeltaOps;
     use crate::protocol::Metadata;
     use crate::writer::test_utils::get_delta_schema;
+    use crate::DeltaResult;
 
     #[tokio::test]
     async fn test_create_checkpoint_for() {
@@ -1102,7 +1103,7 @@ mod tests {
 
     #[ignore = "This test is only useful if the batch size has been made small"]
     #[tokio::test]
-    async fn test_checkpoint_large_table() -> crate::DeltaResult<()> {
+    async fn test_checkpoint_large_table() -> DeltaResult<()> {
         use crate::writer::test_utils::get_arrow_schema;
 
         let table_schema = get_delta_schema();
@@ -1160,4 +1161,44 @@ mod tests {
         );
         Ok(())
     }
+
+    /// <https://github.com/delta-io/delta-rs/issues/3030>
+    #[tokio::test]
+    async fn test_create_checkpoint_overwrite() -> DeltaResult<()> {
+        use crate::protocol::SaveMode;
+        use crate::writer::test_utils::get_arrow_schema;
+
+        let batch = RecordBatch::try_new(
+            Arc::clone(&get_arrow_schema(&None)),
+            vec![
+                Arc::new(arrow::array::StringArray::from(vec!["C"])),
+                Arc::new(arrow::array::Int32Array::from(vec![30])),
+                Arc::new(arrow::array::StringArray::from(vec!["2021-02-03"])),
+            ],
+        )
+        .unwrap();
+        let table = DeltaOps::try_from_uri_with_storage_options("memory://", HashMap::default())
+            .await?
+            .write(vec![batch])
+            .await?;
+        assert_eq!(table.version(), 0);
+
+        create_checkpoint_for(0, table.snapshot().unwrap(), table.log_store.as_ref()).await?;
+
+        let batch = RecordBatch::try_new(
+            Arc::clone(&get_arrow_schema(&None)),
+            vec![
+                Arc::new(arrow::array::StringArray::from(vec!["A"])),
+                Arc::new(arrow::array::Int32Array::from(vec![0])),
+                Arc::new(arrow::array::StringArray::from(vec!["2021-02-02"])),
+            ],
+        )
+        .unwrap();
+        let table = DeltaOps(table)
+            .write(vec![batch])
+            .with_save_mode(SaveMode::Overwrite)
+            .await?;
+        assert_eq!(table.version(), 1);
+        Ok(())
+    }
 }

crates/deltalake/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "deltalake"
-version = "0.22.0"
+version = "0.22.1"
 authors.workspace = true
 keywords.workspace = true
 readme.workspace = true

python/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "deltalake-python"
-version = "0.22.0"
+version = "0.22.1"
 authors = ["Qingping Hou <[email protected]>", "Will Jones <[email protected]>"]
 homepage = "https://github.com/delta-io/delta-rs"
 license = "Apache-2.0"

python/tests/test_checkpoint.py

@@ -468,3 +468,33 @@ def test_checkpoint_with_nullable_false(tmp_path: pathlib.Path):
     assert checkpoint_path.exists()
 
     assert DeltaTable(str(tmp_table_path)).to_pyarrow_table() == data
+
+
+@pytest.mark.pandas
+def test_checkpoint_with_multiple_writes(tmp_path: pathlib.Path):
+    import pandas as pd
+
+    write_deltalake(
+        tmp_path,
+        pd.DataFrame(
+            {
+                "a": ["a"],
+                "b": [3],
+            }
+        ),
+    )
+    DeltaTable(tmp_path).create_checkpoint()
+
+    dt = DeltaTable(tmp_path)
+    print(dt.to_pandas())
+
+    write_deltalake(
+        tmp_path,
+        pd.DataFrame(
+            {
+                "a": ["a"],
+                "b": [100],
+            }
+        ),
+        mode="overwrite",
+    )