parallel_processor/buckets/
concurrent.rs

1use 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(mtb: &Arc<MultiThreadBuckets<B>>, thread_data: BucketsThreadBuffer) -> Self {
36        assert_eq!(mtb.active_buckets.len(), thread_data.buffers.len());
37        Self {
38            mtb: mtb.clone(),
39            thread_data,
40            drop_panic: PanicOnDrop::new("buckets thread dispatcher not finalized"),
41            serializers: (0..mtb.active_buckets.len()).map(|_| S::new()).collect(),
42        }
43    }
44
45    #[inline]
46    pub fn add_element_extended(
47        &mut self,
48        bucket: u16,
49        extra_data: &S::ExtraData,
50        extra_data_buffer: &S::ExtraDataBuffer,
51        element: &S::InputElementType<'_>,
52    ) -> ChunkingStatus {
53        let bucket_buf = &mut self.thread_data.buffers[bucket as usize];
54        let mut chunking_status = ChunkingStatus::SameChunk;
55        if self.serializers[bucket as usize].get_size(element, extra_data) + bucket_buf.len()
56            > bucket_buf.capacity()
57            && bucket_buf.len() > 0
58        {
59            chunking_status = self.mtb.add_data(bucket, bucket_buf.as_slice());
60            bucket_buf.clear();
61            self.serializers[bucket as usize].reset();
62        }
63        self.serializers[bucket as usize].write_to(
64            element,
65            bucket_buf,
66            extra_data,
67            extra_data_buffer,
68        );
69        chunking_status
70    }
71
72    #[inline]
73    pub fn add_element(
74        &mut self,
75        bucket: u16,
76        extra_data: &S::ExtraData,
77        element: &S::InputElementType<'_>,
78    ) where
79        S: BucketItemSerializer<ExtraDataBuffer = ()>,
80    {
81        self.add_element_extended(bucket, extra_data, &(), element);
82    }
83
84    pub fn finalize(mut self) -> (BucketsThreadBuffer, Arc<MultiThreadBuckets<B>>) {
85        for (index, vec) in self.thread_data.buffers.iter_mut().enumerate() {
86            if vec.len() == 0 {
87                continue;
88            }
89            self.mtb.add_data(index as u16, vec.as_slice());
90            vec.clear();
91        }
92        self.drop_panic.disengage();
93        (self.thread_data, self.mtb)
94    }
95}
parallel_processor/buckets/concurrent.rs

parallel_processor/buckets/
concurrent.rs
