fix: avro serialisation of nested records with implicit namespace (#545)

Author: andrewinci

backend/Cargo.lock

@@ -17,19 +17,33 @@ openssl-src = { version = "111", features = ["force-engine"] }
 serde_json = "1.0"
 dirs = "5.0.1"
 serde = { version = "1.0.171", features = ["derive"] }
-tauri = { version = "1.4", features = ["dialog-open", "dialog-save", "fs-write-file", "os-all", "updater", "window-create", "window-set-focus", "window-set-min-size", "window-set-title"] }
+tauri = { version = "1.4", features = [
+    "dialog-open",
+    "dialog-save",
+    "fs-write-file",
+    "os-all",
+    "updater",
+    "window-create",
+    "window-set-focus",
+    "window-set-min-size",
+    "window-set-title",
+] }
 reqwest = { version = "0.11", features = ["json", "blocking"] }
 url = { version = "2", features = ["serde"] }
 tokio = { version = "1", features = ["full"] }
 futures = { version = "0.3" }
-apache-avro = { git = "https://github.com/apache/avro", rev = "8c3ee165432cf557429a984df6acbc075a848ac9" }
+# todo: replace the fork with the apache repo when https://github.com/apache/avro/pull/2374 is merged
+# apache-avro = { git = "https://github.com/apache/avro", rev = "<todo>" }
+apache-avro = { git = "https://github.com/andrewinci/avro", rev = "c0583d12944567872079c6162e976f9ea9477e8a" }
 log = { version = "0.4" }
 env_logger = { version = "0.10.0" }
 async-trait = "0.1.71"
 num-bigint = "0.4"
 rust_decimal = "1.30"
 rusqlite = { version = "0.28.0", features = ["bundled", "backup", "hooks"] }
-rust-keystore = { git = "https://github.com/andrewinci/rust-keystore", features = ["p12"], tag = "v0.1.2" }
+rust-keystore = { git = "https://github.com/andrewinci/rust-keystore", features = [
+    "p12",
+], tag = "v0.1.2" }
 r2d2 = "0.8.10"
 r2d2_sqlite = "0.21.0"
 toml = "0.7"
@@ -69,6 +83,6 @@ custom-protocol = ["tauri/custom-protocol"]
 integration_tests = []
 
 [profile.release]
-strip = true 
+strip = true
 lto = true
 opt-level = "s"

backend/Cargo.toml

@@ -222,12 +222,42 @@ mod tests {
     use super::parse_decimal;
     use crate::core::avro::avro_schema::AvroSchema;
     use crate::core::avro::avro_schema::RecordField;
+    use crate::core::avro::error::AvroResult;
     use crate::core::avro::AvroError;
+    use crate::core::avro::AvroParser;
+    use crate::core::avro::ResolvedAvroSchema;
+    use crate::core::avro::SchemaProvider;
+    use apache_avro::Schema;
 
     use apache_avro::types::Value as AvroValue;
+    use async_trait::async_trait;
     use serde_json::json;
     use serde_json::Value as JsonValue;
 
+    struct MockSchemaProvider {}
+    #[async_trait]
+    impl SchemaProvider for MockSchemaProvider {
+        async fn get_schema_by_id(&self, id: i32) -> AvroResult<ResolvedAvroSchema> {
+            let json_schema = &get_test_avro_schema();
+            let schema = Schema::parse_str(json_schema).expect("invalid test schema");
+            Ok(ResolvedAvroSchema::from(id, &schema))
+        }
+        async fn get_schema_by_name(&self, _: &str) -> AvroResult<ResolvedAvroSchema> {
+            let json_schema = &get_test_avro_schema();
+            let schema = Schema::parse_str(json_schema).expect("invalid test schema");
+            Ok(ResolvedAvroSchema::from(123, &schema))
+        }
+    }
+
+    #[tokio::test]
+    async fn test_parse_nested_records_with_implicit_namespace() {
+        let mock_provider = MockSchemaProvider {};
+        let sut = AvroParser::new(mock_provider.into());
+        let sample_json = &get_test_avro_message();
+        let res = sut.json_to_avro(sample_json, "sample").await;
+        assert!(res.is_ok())
+    }
+
     #[test]
     fn test_decimal() {
         // happy path
@@ -311,4 +341,49 @@ mod tests {
             schema: schema,
         }
     }
+
+    fn get_test_avro_message() -> String {
+        r#"{
+            "outer_field_1": {
+                "middle_field_1": {
+                    "inner_field_1": 1.7
+                },
+                "middle_field_2": {
+                    "inner_field_1": 1.8
+                }
+            }
+          }"#
+        .to_string()
+    }
+    fn get_test_avro_schema() -> String {
+        r#"{
+            "name": "record_name",
+            "namespace": "space",
+            "type": "record",
+            "fields": [
+              {
+                "name": "outer_field_1",
+                "type": {
+                  "type": "record",
+                  "name": "middle_record_name",
+                  "namespace": "middle_namespace",
+                  "fields": [
+                    {
+                      "name": "middle_field_1",
+                      "type": {
+                        "type": "record",
+                        "name": "inner_record_name",
+                        "fields": [
+                          { "name": "inner_field_1", "type": "double" }
+                        ]
+                      }
+                    },
+                    { "name": "middle_field_2", "type": "inner_record_name" }
+                  ]
+                }
+              }
+            ]
+          }"#
+        .to_string()
+    }
 }