feat: Make filter_parquet_files more general with regards to compaction level

compactor/src/cold.rs

@@ -384,26 +384,33 @@ mod tests {
 
         // ------------------------------------------------
         // Compact
-        let mut candidates = compactor
+        let mut partition_candidates = compactor
             .cold_partitions_to_compact(compactor.config.max_number_partitions_per_shard)
             .await
             .unwrap();
 
-        assert_eq!(candidates.len(), 1);
+        assert_eq!(partition_candidates.len(), 1);
-        let c = candidates.pop().unwrap();
+        let partition = partition_candidates.pop().unwrap();
 
         let parquet_files_for_compaction =
             parquet_file_lookup::ParquetFilesForCompaction::for_partition_with_size_overrides(
                 Arc::clone(&compactor.catalog),
-                Arc::clone(&c),
+                Arc::clone(&partition),
                 &size_overrides,
             )
             .await
             .unwrap();
 
+        let ParquetFilesForCompaction {
+            level_0,
+            level_1,
+            .. // Ignore other levels
+        } = parquet_files_for_compaction;
+
         let to_compact = parquet_file_filtering::filter_parquet_files(
-            c,
+            partition,
-            parquet_files_for_compaction,
+            level_0,
+            level_1,
             compactor.config.memory_budget_bytes,
             &compactor.parquet_file_candidate_gauge,
             &compactor.parquet_file_candidate_bytes,
@@ -567,26 +574,33 @@ mod tests {
 
         // ------------------------------------------------
         // Compact
-        let mut candidates = compactor
+        let mut partition_candidates = compactor
             .cold_partitions_to_compact(compactor.config.max_number_partitions_per_shard)
             .await
             .unwrap();
 
-        assert_eq!(candidates.len(), 1);
+        assert_eq!(partition_candidates.len(), 1);
-        let c = candidates.pop().unwrap();
+        let partition = partition_candidates.pop().unwrap();
 
         let parquet_files_for_compaction =
             parquet_file_lookup::ParquetFilesForCompaction::for_partition_with_size_overrides(
                 Arc::clone(&compactor.catalog),
-                Arc::clone(&c),
+                Arc::clone(&partition),
                 &size_overrides,
             )
             .await
             .unwrap();
 
+        let ParquetFilesForCompaction {
+            level_0,
+            level_1,
+            .. // Ignore other levels
+        } = parquet_files_for_compaction;
+
         let to_compact = parquet_file_filtering::filter_parquet_files(
-            Arc::clone(&c),
+            Arc::clone(&partition),
-            parquet_files_for_compaction,
+            level_0,
+            level_1,
             compactor.config.memory_budget_bytes,
             &compactor.parquet_file_candidate_gauge,
             &compactor.parquet_file_candidate_bytes,
@@ -656,7 +670,7 @@ mod tests {
         );
 
         // Full compaction will now combine the two level 1 files into one level 2 file
-        full_compaction(&compactor, c, &size_overrides)
+        full_compaction(&compactor, partition, &size_overrides)
             .await
             .unwrap();
 

compactor/src/hot.rs

@@ -77,6 +77,7 @@ mod tests {
         handler::CompactorConfig,
         parquet_file_filtering, parquet_file_lookup,
         tests::{test_setup, TestSetup},
+        ParquetFilesForCompaction,
     };
     use arrow_util::assert_batches_sorted_eq;
     use backoff::BackoffConfig;
@@ -485,7 +486,7 @@ mod tests {
 
         // ------------------------------------------------
         // Compact
-        let mut candidates = compactor
+        let mut partition_candidates = compactor
             .hot_partitions_to_compact(
                 compactor.config.max_number_partitions_per_shard,
                 compactor
@@ -495,21 +496,28 @@ mod tests {
             .await
             .unwrap();
 
-        assert_eq!(candidates.len(), 1);
+        assert_eq!(partition_candidates.len(), 1);
-        let c = candidates.pop().unwrap();
+        let partition = partition_candidates.pop().unwrap();
 
         let parquet_files_for_compaction =
             parquet_file_lookup::ParquetFilesForCompaction::for_partition_with_size_overrides(
                 Arc::clone(&compactor.catalog),
-                Arc::clone(&c),
+                Arc::clone(&partition),
                 &size_overrides,
             )
             .await
             .unwrap();
 
+        let ParquetFilesForCompaction {
+            level_0,
+            level_1,
+            .. // Ignore other levels
+        } = parquet_files_for_compaction;
+
         let to_compact = parquet_file_filtering::filter_parquet_files(
-            c,
+            partition,
-            parquet_files_for_compaction,
+            level_0,
+            level_1,
             compactor.config.memory_budget_bytes,
             &compactor.parquet_file_candidate_gauge,
             &compactor.parquet_file_candidate_bytes,

compactor/src/lib.rs

@@ -27,6 +27,7 @@ use crate::{
     compact::{Compactor, PartitionCompactionCandidateWithInfo},
     parquet_file::CompactorParquetFile,
     parquet_file_filtering::{FilterResult, FilteredFiles},
+    parquet_file_lookup::ParquetFilesForCompaction,
 };
 use data_types::CompactionLevel;
 use metric::Attributes;
@@ -107,9 +108,16 @@ async fn compact_candidates_with_memory_budget<C, Fut>(
             Ok(parquet_files_for_compaction) => {
                 // Return only files under the `remaining_budget_bytes` that should be
                 // compacted
+                let ParquetFilesForCompaction {
+                    level_0,
+                    level_1,
+                    .. // Ignore other levels
+                } = parquet_files_for_compaction;
+
                 let to_compact = parquet_file_filtering::filter_parquet_files(
                     Arc::clone(&partition),
-                    parquet_files_for_compaction,
+                    level_0,
+                    level_1,
                     remaining_budget_bytes,
                     &compactor.parquet_file_candidate_gauge,
                     &compactor.parquet_file_candidate_bytes,

data_types/src/lib.rs

@@ -53,11 +53,24 @@ impl TryFrom<i32> for CompactionLevel {
         match value {
             x if x == Self::Initial as i32 => Ok(Self::Initial),
             x if x == Self::FileNonOverlapped as i32 => Ok(Self::FileNonOverlapped),
+            x if x == Self::Final as i32 => Ok(Self::Final),
             _ => Err("invalid compaction level value".into()),
         }
     }
 }
 
+impl CompactionLevel {
+    /// When compacting files of this level, provide the level that the resulting file should be.
+    /// Does not exceed the maximum available level.
+    pub fn next(&self) -> Self {
+        match self {
+            Self::Initial => Self::FileNonOverlapped,
+            Self::FileNonOverlapped => Self::Final,
+            _ => Self::Final,
+        }
+    }
+}
+
 /// Unique ID for a `Namespace`
 #[derive(Debug, Clone, Copy, PartialEq, Eq, PartialOrd, Ord, Hash, sqlx::Type)]
 #[sqlx(transparent)]