|
| 1 | +use std::{hash::Hash, ops::Range, sync::Arc}; |
| 2 | + |
| 3 | +use bytes::Bytes; |
| 4 | +use futures_core::future::BoxFuture; |
| 5 | +use futures_util::FutureExt; |
| 6 | +use parquet::{ |
| 7 | + arrow::async_reader::AsyncFileReader, |
| 8 | + errors::{ParquetError, Result}, |
| 9 | + file::metadata::ParquetMetaData, |
| 10 | +}; |
| 11 | +use serde::{Deserialize, Serialize}; |
| 12 | + |
| 13 | +use crate::{Error, LruCache, Options}; |
| 14 | + |
| 15 | +#[derive(Clone)] |
| 16 | +pub struct FoyerCache<K> |
| 17 | +where |
| 18 | + for<'a> K: Send + Sync + Hash + Eq + Serialize + Deserialize<'a> + 'static, |
| 19 | +{ |
| 20 | + inner: Arc<FoyerCacheInner<K>>, |
| 21 | +} |
| 22 | + |
| 23 | +pub struct FoyerCacheInner<K> |
| 24 | +where |
| 25 | + for<'a> K: Send + Sync + Hash + Eq + Serialize + Deserialize<'a> + 'static, |
| 26 | +{ |
| 27 | + meta: foyer::Cache<K, Arc<ParquetMetaData>>, |
| 28 | + data: foyer::HybridCache<(K, Range<usize>), Bytes>, |
| 29 | +} |
| 30 | + |
| 31 | +impl<K> LruCache<K> for FoyerCache<K> |
| 32 | +where |
| 33 | + for<'a> K: Send + Sync + Hash + Eq + Serialize + Deserialize<'a> + Clone + 'static, |
| 34 | +{ |
| 35 | + type LruReader<R: AsyncFileReader + 'static> = FoyerReader<K, R>; |
| 36 | + |
| 37 | + async fn new(options: Options) -> Result<Self, Error> { |
| 38 | + Ok(Self { |
| 39 | + inner: Arc::new(FoyerCacheInner { |
| 40 | + meta: foyer::CacheBuilder::new(options.meta_capacity).build(), |
| 41 | + data: foyer::HybridCacheBuilder::new() |
| 42 | + .memory(options.data_capacity) |
| 43 | + .storage(foyer::Engine::Large) |
| 44 | + .build() |
| 45 | + .await |
| 46 | + .map_err(|e| Error::External(e.into()))?, |
| 47 | + }), |
| 48 | + }) |
| 49 | + } |
| 50 | + |
| 51 | + async fn get_reader<R: AsyncFileReader>(&self, key: K, reader: R) -> FoyerReader<K, R> { |
| 52 | + FoyerReader::new(self.clone(), key, reader) |
| 53 | + } |
| 54 | +} |
| 55 | + |
| 56 | +pub struct FoyerReader<K, R> |
| 57 | +where |
| 58 | + for<'a> K: Send + Sync + Hash + Eq + Serialize + Deserialize<'a> + 'static, |
| 59 | +{ |
| 60 | + cache: FoyerCache<K>, |
| 61 | + key: K, |
| 62 | + reader: R, |
| 63 | +} |
| 64 | + |
| 65 | +impl<K, R> FoyerReader<K, R> |
| 66 | +where |
| 67 | + for<'a> K: Send + Sync + Hash + Eq + Serialize + Deserialize<'a> + 'static, |
| 68 | + R: AsyncFileReader, |
| 69 | +{ |
| 70 | + fn new(cache: FoyerCache<K>, key: K, reader: R) -> Self { |
| 71 | + Self { cache, key, reader } |
| 72 | + } |
| 73 | +} |
| 74 | + |
| 75 | +impl<K, R> AsyncFileReader for FoyerReader<K, R> |
| 76 | +where |
| 77 | + for<'a> K: Send + Sync + Hash + Eq + Serialize + Deserialize<'a> + Clone + 'static, |
| 78 | + R: AsyncFileReader, |
| 79 | +{ |
| 80 | + fn get_bytes(&mut self, range: Range<usize>) -> BoxFuture<'_, Result<Bytes>> { |
| 81 | + async move { |
| 82 | + if let Some(data) = self |
| 83 | + .cache |
| 84 | + .inner |
| 85 | + .data |
| 86 | + .get(&(self.key.clone(), range.clone())) |
| 87 | + .await |
| 88 | + .map_err(|e| ParquetError::External(e.into()))? |
| 89 | + { |
| 90 | + Ok(data.value().clone()) |
| 91 | + } else { |
| 92 | + let data = self.reader.get_bytes(range.clone()).await?; |
| 93 | + self.cache |
| 94 | + .inner |
| 95 | + .data |
| 96 | + .insert((self.key.clone(), range), data.clone()); |
| 97 | + Ok(data) |
| 98 | + } |
| 99 | + } |
| 100 | + .boxed() |
| 101 | + } |
| 102 | + |
| 103 | + fn get_metadata(&mut self) -> BoxFuture<'_, Result<Arc<ParquetMetaData>>> { |
| 104 | + async move { |
| 105 | + if let Some(meta) = self.cache.inner.meta.get(&self.key) { |
| 106 | + Ok(meta.value().clone()) |
| 107 | + } else { |
| 108 | + let meta = self.reader.get_metadata().await?; |
| 109 | + self.cache.inner.meta.insert(self.key.clone(), meta.clone()); |
| 110 | + Ok(meta) |
| 111 | + } |
| 112 | + } |
| 113 | + .boxed() |
| 114 | + } |
| 115 | + |
| 116 | + fn get_byte_ranges(&mut self, ranges: Vec<Range<usize>>) -> BoxFuture<'_, Result<Vec<Bytes>>> { |
| 117 | + async move { |
| 118 | + let mut missed = Vec::with_capacity(ranges.len()); |
| 119 | + let mut results = Vec::with_capacity(ranges.len()); |
| 120 | + for (id, range) in ranges.iter().enumerate() { |
| 121 | + if let Some(data) = self |
| 122 | + .cache |
| 123 | + .inner |
| 124 | + .data |
| 125 | + .get(&(self.key.clone(), range.clone())) |
| 126 | + .await |
| 127 | + .map_err(|e| ParquetError::External(e.into()))? |
| 128 | + { |
| 129 | + results.push((id, data.value().clone())); |
| 130 | + } else { |
| 131 | + missed.push((id, range)); |
| 132 | + } |
| 133 | + } |
| 134 | + if !missed.is_empty() { |
| 135 | + let data = self |
| 136 | + .reader |
| 137 | + .get_byte_ranges(missed.iter().map(|&(_, r)| r.clone()).collect()) |
| 138 | + .await?; |
| 139 | + for (id, range) in missed { |
| 140 | + let data = data[id].clone(); |
| 141 | + self.cache |
| 142 | + .inner |
| 143 | + .data |
| 144 | + .insert((self.key.clone(), range.clone()), data.clone()); |
| 145 | + results.push((id, data)); |
| 146 | + } |
| 147 | + } |
| 148 | + results.sort_by_key(|(id, _)| *id); |
| 149 | + Ok(results.into_iter().map(|(_, data)| data).collect()) |
| 150 | + } |
| 151 | + .boxed() |
| 152 | + } |
| 153 | +} |
0 commit comments