SML0127 commented on code in PR #307: URL: https://github.com/apache/paimon-rust/pull/307#discussion_r3523501631
########## crates/paimon/src/table/partition_stat.rs: ########## @@ -0,0 +1,171 @@ +// Licensed to the Apache Software Foundation (ASF) under one +// or more contributor license agreements. See the NOTICE file +// distributed with this work for additional information +// regarding copyright ownership. The ASF licenses this file +// to you under the Apache License, Version 2.0 (the +// "License"); you may not use this file except in compliance +// with the License. You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, +// software distributed under the License is distributed on an +// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +// KIND, either express or implied. See the License for the +// specific language governing permissions and limitations +// under the License. + +//! Per-partition statistics computed by scanning the latest snapshot's manifest entries. +//! +//! Mirrors what Java Paimon exposes via the `$partitions` system table for runtime introspection. + +use std::collections::HashMap; + +use crate::io::FileIO; +use crate::spec::{ + avro::from_avro_bytes_fast, merge_active_entries, BinaryRow, CoreOptions, ManifestEntry, + ManifestFileMeta, PartitionComputer, Snapshot, +}; +use crate::table::SnapshotManager; +use crate::table::Table; + +const MANIFEST_DIR: &str = "manifest"; + +/// Per-partition aggregated statistics. +#[derive(Debug, Clone, PartialEq, Eq)] +pub struct PartitionStat { + /// Partition key/value mapping (e.g. `{"dt": "2024-01-01", "hr": "10"}`). + pub partition: HashMap<String, String>, + /// Total physical row count across all live data files in this partition. + pub record_count: i64, + /// Number of live data files in this partition. + pub file_count: u64, + /// Total bytes of live data files in this partition. + pub total_size_bytes: u64, +} + +#[derive(Default)] +struct Accum { + record_count: i64, + file_count: usize, + total_size_bytes: i64, +} + +impl Table { + /// Compute per-partition statistics from the latest snapshot. + /// + /// **Warning:** This method reads all manifest lists and entries from the latest snapshot. + /// For tables with a large number of manifests, this operation can be expensive. + /// + /// Returns an empty Vec when the table has no snapshots yet. + pub async fn partition_stats(&self) -> crate::Result<Vec<PartitionStat>> { + let sm = SnapshotManager::new(self.file_io().clone(), self.location().to_string()); + let snapshot = match sm.get_latest_snapshot().await? { + Some(s) => s, + None => return Ok(Vec::new()), + }; + + let entries = read_all_manifest_entries(self.file_io(), self.location(), &snapshot).await?; + + let schema = self.schema(); + let core = CoreOptions::new(schema.options()); + let computer = PartitionComputer::new( + schema.partition_keys(), + schema.fields(), + core.partition_default_name(), + core.legacy_partition_name(), + )?; + + aggregate_partition_stats(entries, &computer) + } + + /// List all partition values present in the latest snapshot. + /// + /// **Warning:** This method computes partition statistics which reads all manifest lists + /// and entries. For large tables, this operation can be expensive. + /// + /// Returns an empty Vec when the table has no snapshots yet. + pub async fn list_partitions(&self) -> crate::Result<Vec<HashMap<String, String>>> { + Ok(self + .partition_stats() + .await? + .into_iter() + .map(|s| s.partition) + .collect()) + } +} + +async fn read_manifest_list( + file_io: &FileIO, + table_path: &str, + list_name: &str, +) -> crate::Result<Vec<ManifestFileMeta>> { + if list_name.is_empty() { + return Ok(Vec::new()); + } + let path = format!( + "{}/{}/{}", + table_path.trim_end_matches('/'), + MANIFEST_DIR, + list_name + ); + let input = file_io.new_input(&path)?; + let bytes = input.read().await?; + from_avro_bytes_fast::<ManifestFileMeta>(&bytes) +} + +async fn read_all_manifest_entries( + file_io: &FileIO, + table_path: &str, + snapshot: &Snapshot, +) -> crate::Result<Vec<ManifestEntry>> { + let mut metas = read_manifest_list(file_io, table_path, snapshot.base_manifest_list()).await?; + metas.extend(read_manifest_list(file_io, table_path, snapshot.delta_manifest_list()).await?); + + let manifest_dir = format!("{}/{}", table_path.trim_end_matches('/'), MANIFEST_DIR); + let mut all_entries = Vec::new(); + for meta in metas { + let path = format!("{}/{}", manifest_dir, meta.file_name()); + let input = file_io.new_input(&path)?; + let bytes = input.read().await?; + let entries = from_avro_bytes_fast::<ManifestEntry>(&bytes)?; + all_entries.extend(entries); + } + Ok(all_entries) +} + +fn aggregate_partition_stats( + entries: Vec<ManifestEntry>, + computer: &PartitionComputer, +) -> crate::Result<Vec<PartitionStat>> { + // Mirror catalog::partition_listing::list_partitions_from_file_system: + // drop DELETE-shadowed entries first so stats are based on the active + // (live) file set, consistent with scan and catalog partition semantics. + let active = merge_active_entries(entries); + + let mut grouped: HashMap<Vec<u8>, Accum> = HashMap::new(); Review Comment: Thanks for the great feedback! I used btreemap for deterministic partition ordering. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
