parallel_processor/buckets/
concurrent.rs1use crate::buckets::bucket_writer::BucketItemSerializer;
2use crate::buckets::{LockFreeBucket, MultiThreadBuckets};
3use crate::memory_data_size::MemoryDataSize;
4use crate::utils::panic_on_drop::PanicOnDrop;
5use std::sync::Arc;
6
7use super::ChunkingStatus;
8
9pub struct BucketsThreadBuffer {
10 buffers: Vec<Vec<u8>>,
11}
12
13impl BucketsThreadBuffer {
14 pub const EMPTY: Self = Self { buffers: vec![] };
15
16 pub fn new(max_buffer_size: MemoryDataSize, buckets_count: usize) -> Self {
17 let mut buffers = Vec::with_capacity(buckets_count);
18 let capacity = max_buffer_size.as_bytes();
19 for _ in 0..buckets_count {
20 buffers.push(Vec::with_capacity(capacity));
21 }
22
23 Self { buffers }
24 }
25}
26
27pub struct BucketsThreadDispatcher<B: LockFreeBucket, S: BucketItemSerializer> {
28 mtb: Arc<MultiThreadBuckets<B>>,
29 thread_data: BucketsThreadBuffer,
30 drop_panic: PanicOnDrop,
31 serializers: Vec<S>,
32}
33
34impl<B: LockFreeBucket, S: BucketItemSerializer> BucketsThreadDispatcher<B, S> {
35 pub fn new(
36 mtb: &Arc<MultiThreadBuckets<B>>,
37 thread_data: BucketsThreadBuffer,
38 deserializer_init_data: S::InitData,
39 ) -> Self
40 where
41 S::InitData: Copy,
42 {
43 assert_eq!(mtb.active_buckets.len(), thread_data.buffers.len());
44 Self {
45 mtb: mtb.clone(),
46 thread_data,
47 drop_panic: PanicOnDrop::new("buckets thread dispatcher not finalized"),
48 serializers: (0..mtb.active_buckets.len())
49 .map(|_| S::new(deserializer_init_data))
50 .collect(),
51 }
52 }
53
54 #[inline]
55 pub fn add_element_extended(
56 &mut self,
57 bucket: u16,
58 extra_data: &S::ExtraData,
59 extra_data_buffer: &S::ExtraDataBuffer,
60 element: &S::InputElementType<'_>,
61 ) -> ChunkingStatus {
62 let bucket_buf = &mut self.thread_data.buffers[bucket as usize];
63 let mut chunking_status = ChunkingStatus::SameChunk;
64 if self.serializers[bucket as usize].get_size(element, extra_data) + bucket_buf.len()
65 > bucket_buf.capacity()
66 && bucket_buf.len() > 0
67 {
68 chunking_status = self.mtb.add_data(bucket, bucket_buf.as_slice());
69 bucket_buf.clear();
70 self.serializers[bucket as usize].reset();
71 }
72 self.serializers[bucket as usize].write_to(
73 element,
74 bucket_buf,
75 extra_data,
76 extra_data_buffer,
77 );
78 chunking_status
79 }
80
81 #[inline]
82 pub fn add_element(
83 &mut self,
84 bucket: u16,
85 extra_data: &S::ExtraData,
86 element: &S::InputElementType<'_>,
87 ) where
88 S: BucketItemSerializer<ExtraDataBuffer = ()>,
89 {
90 self.add_element_extended(bucket, extra_data, &(), element);
91 }
92
93 pub fn finalize(mut self) -> (BucketsThreadBuffer, Arc<MultiThreadBuckets<B>>) {
94 for (index, vec) in self.thread_data.buffers.iter_mut().enumerate() {
95 if vec.len() == 0 {
96 continue;
97 }
98 self.mtb.add_data(index as u16, vec.as_slice());
99 vec.clear();
100 }
101 self.drop_panic.disengage();
102 (self.thread_data, self.mtb)
103 }
104}