added initial default logic

Author: rambleraptor

crates/iceberg/src/avro/schema.rs

@@ -43,6 +43,41 @@ const MAP_LOGICAL_TYPE: &str = "map";
 // This const may better to maintain in avro-rs.
 const LOGICAL_TYPE: &str = "logicalType";
 
+fn literal_to_json(literal: &crate::spec::Literal) -> Result<Value> {
+    match literal {
+        crate::spec::Literal::Primitive(p) => match p {
+            crate::spec::PrimitiveLiteral::Boolean(b) => Ok(Value::Bool(*b)),
+            crate::spec::PrimitiveLiteral::Int(i) => Ok(Value::Number(Number::from(*i))),
+            crate::spec::PrimitiveLiteral::Long(l) => Ok(Value::Number(Number::from(*l))),
+            crate::spec::PrimitiveLiteral::Float(f) => Ok(Value::Number(
+                Number::from_f64(f.0 as f64).ok_or_else(|| {
+                    Error::new(
+                        ErrorKind::DataInvalid,
+                        "Failed to convert float to json number",
+                    )
+                })?,
+            )),
+            crate::spec::PrimitiveLiteral::Double(d) => Ok(Value::Number(
+                Number::from_f64(d.0).ok_or_else(|| {
+                    Error::new(
+                        ErrorKind::DataInvalid,
+                        "Failed to convert double to json number",
+                    )
+                })?,
+            )),
+            crate::spec::PrimitiveLiteral::String(s) => Ok(Value::String(s.clone())),
+            _ => Err(Error::new(
+                ErrorKind::FeatureUnsupported,
+                "Unsupported literal type to convert to json",
+            )),
+        },
+        _ => Err(Error::new(
+            ErrorKind::FeatureUnsupported,
+            "Unsupported literal type to convert to json",
+        )),
+    }
+}
+
 struct SchemaToAvroSchema {
     schema: String,
 }
@@ -92,9 +127,12 @@ impl SchemaVisitor for SchemaToAvroSchema {
             custom_attributes: Default::default(),
         };
 
-        if !field.required {
+        if let Some(default) = &field.initial_default {
+            avro_record_field.default = Some(literal_to_json(default)?);
+        } else if !field.required {
             avro_record_field.default = Some(Value::Null);
         }
+
         avro_record_field.custom_attributes.insert(
             FILED_ID_PROP.to_string(),
             Value::Number(Number::from(field.id)),

crates/iceberg/src/spec/manifest/_serde.rs

@@ -99,7 +99,7 @@ impl ManifestEntryV1 {
 #[serde_as]
 #[derive(Serialize, Deserialize)]
 pub(super) struct DataFileSerde {
-    #[serde(default)]
+    #[serde(default = "default_to_zero")]
     content: i32,
     file_path: String,
     file_format: String,
@@ -125,6 +125,10 @@ pub(super) struct DataFileSerde {
     content_size_in_bytes: Option<i64>,
 }
 
+fn default_to_zero() -> i32 {
+    0
+}
+
 impl DataFileSerde {
     pub fn try_from(
         value: super::DataFile,
@@ -400,7 +404,7 @@ mod tests {
     }
 
     #[tokio::test]
-    async fn test_data_file_serialize_deserialize_v1() {
+    async fn test_data_file_serialize_deserialize_v1_data_on_v2_reader() {
         let schema = Arc::new(
             Schema::builder()
                 .with_fields(vec![
@@ -461,7 +465,7 @@ mod tests {
             &schema,
             0,
             &StructType::new(vec![]),
-            FormatVersion::V1,
+            FormatVersion::V2,
         )
         .unwrap();
 

crates/iceberg/src/spec/manifest/entry.rs

@@ -234,7 +234,7 @@ static CONTENT: Lazy<NestedFieldRef> = {
     Lazy::new(|| {
         Arc::new(
             NestedField::required(134, "content", Type::Primitive(PrimitiveType::Int))
-                .with_initial_default(Literal::Primitive(PrimitiveLiteral::Int(1))),
+                .with_initial_default(Literal::Primitive(PrimitiveLiteral::Int(0))),
         )
     })
 };