Fix sequence number generation (#15)

Author: SergeiPatiakin

src/iceberg_destination.rs

@@ -203,7 +203,7 @@ pub async fn record_batches_to_iceberg(
         .collect();
 
     let snapshot_id = fastrand::i64(..);
-    let sequence_number = 1;
+    let sequence_number = previous_metadata.last_sequence_number() + 1;
 
     let manifest_file_path = format!("{}/metadata/manifest-{}.avro", target_url, Uuid::new_v4());
     let manifest_file_output = file_io.new_output(manifest_file_path)?;

tests/basic_integration.rs

@@ -1,10 +1,13 @@
+use std::vec;
+
 use arrow::array::{
     Array, BinaryArray, BooleanArray, Date32Array, Decimal128Array, Float32Array, Float64Array,
     Int16Array, Int32Array, Int64Array, Int8Array, StringArray, TimestampMicrosecondArray,
 };
 use arrow::datatypes::DataType;
 use clap::Parser;
 use futures::{StreamExt, TryStreamExt};
+use iceberg::spec::TableMetadata;
 use lakehouse_loader::delta_destination::object_store_keys_from_env;
 use lakehouse_loader::error::DataLoadingError;
 use lakehouse_loader::pg_arrow_source::PgArrowSource;
@@ -57,6 +60,10 @@ async fn test_pg_to_delta_e2e() {
     assert!(paths[2].to_string().ends_with("-c000.snappy.parquet"));
 }
 
+const DATA_FILEPATH_PATTERN: &str = r"^iceberg/data/part-00000-[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}.parquet$";
+const MANIFEST_FILEPATH_PATTERN: &str = r"^iceberg/metadata/manifest-[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}.avro$";
+const MANIFEST_LIST_FILEPATH_PATTERN: &str = r"^iceberg/metadata/manifest-list-[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}.avro$";
+
 #[tokio::test]
 async fn test_pg_to_iceberg() {
     let target_url = "s3://lhl-test-bucket/iceberg";
@@ -78,23 +85,40 @@ async fn test_pg_to_iceberg() {
     let (store, path) =
         object_store::parse_url_opts(&Url::parse(target_url).unwrap(), config).unwrap();
 
+    // THEN iceberg data and metadata files are written
     let mut paths = store
         .list(Some(&path))
         .map_ok(|m| m.location)
         .boxed()
         .try_collect::<Vec<Path>>()
         .await
         .unwrap();
-
     paths.sort();
-
-    // THEN iceberg data and metadata files are written
     assert_eq!(paths.len(), 5);
-    assert!(Regex::new(r"^iceberg/data/part-00000-[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}.parquet$").unwrap().is_match(paths[0].as_ref()));
-    assert!(Regex::new(r"^iceberg/metadata/manifest-[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}.avro$").unwrap().is_match(paths[1].as_ref()));
-    assert!(Regex::new(r"^iceberg/metadata/manifest-list-[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}.avro$").unwrap().is_match(paths[2].as_ref()));
+    assert!(Regex::new(DATA_FILEPATH_PATTERN)
+        .unwrap()
+        .is_match(paths[0].as_ref()));
+    assert!(Regex::new(MANIFEST_FILEPATH_PATTERN)
+        .unwrap()
+        .is_match(paths[1].as_ref()));
+    assert!(Regex::new(MANIFEST_LIST_FILEPATH_PATTERN)
+        .unwrap()
+        .is_match(paths[2].as_ref()));
     assert_eq!(&paths[3].to_string(), "iceberg/metadata/v0.metadata.json");
     assert_eq!(&paths[4].to_string(), "iceberg/metadata/version-hint.text");
+    // THEN iceberg metadata can be parsed
+    let metadata_bytes = store.get(&paths[3]).await.unwrap().bytes().await.unwrap();
+    let metadata_str = core::str::from_utf8(&metadata_bytes).unwrap();
+    let metadata = serde_json::from_str::<TableMetadata>(metadata_str).unwrap();
+    // THEN metadata contains a single snapshot with sequence number 1
+    assert_eq!(metadata.last_sequence_number(), 1);
+    assert_eq!(
+        metadata
+            .snapshots()
+            .map(|s| s.sequence_number())
+            .collect::<Vec<_>>(),
+        vec![1]
+    );
 
     // WHEN we try to write to an existing table without passing the overwrite flag
     // THEN the command errors out
@@ -149,6 +173,50 @@ async fn test_pg_to_iceberg() {
         "--overwrite",
     ];
     assert!(do_main(Cli::parse_from(args.clone())).await.is_ok());
+
+    // THEN iceberg data and metadata files are written
+    let mut paths = store
+        .list(Some(&path))
+        .map_ok(|m| m.location)
+        .boxed()
+        .try_collect::<Vec<Path>>()
+        .await
+        .unwrap();
+    paths.sort();
+    assert_eq!(paths.len(), 9);
+    assert!(Regex::new(DATA_FILEPATH_PATTERN)
+        .unwrap()
+        .is_match(paths[0].as_ref()));
+    assert!(Regex::new(DATA_FILEPATH_PATTERN)
+        .unwrap()
+        .is_match(paths[1].as_ref()));
+    assert!(Regex::new(MANIFEST_FILEPATH_PATTERN)
+        .unwrap()
+        .is_match(paths[2].as_ref()));
+    assert!(Regex::new(MANIFEST_FILEPATH_PATTERN)
+        .unwrap()
+        .is_match(paths[3].as_ref()));
+    assert!(Regex::new(MANIFEST_LIST_FILEPATH_PATTERN)
+        .unwrap()
+        .is_match(paths[4].as_ref()));
+    assert!(Regex::new(MANIFEST_LIST_FILEPATH_PATTERN)
+        .unwrap()
+        .is_match(paths[5].as_ref()));
+    assert_eq!(&paths[6].to_string(), "iceberg/metadata/v0.metadata.json");
+    assert_eq!(&paths[7].to_string(), "iceberg/metadata/v1.metadata.json");
+    assert_eq!(&paths[8].to_string(), "iceberg/metadata/version-hint.text");
+    // THEN iceberg metadata can be parsed
+    let metadata_bytes = store.get(&paths[7]).await.unwrap().bytes().await.unwrap();
+    let metadata_str = core::str::from_utf8(&metadata_bytes).unwrap();
+    let metadata = serde_json::from_str::<TableMetadata>(metadata_str).unwrap();
+    // THEN metadata contains two snapshots with sequence numbers 1 and 2
+    assert_eq!(metadata.last_sequence_number(), 2);
+    let mut snapshot_ids = metadata
+        .snapshots()
+        .map(|s| s.sequence_number())
+        .collect::<Vec<_>>();
+    snapshot_ids.sort();
+    assert_eq!(snapshot_ids, vec![1, 2]);
 }
 
 #[tokio::test]