MeiliSearch/src/data/doc_ids.rs

56 lines
1.6 KiB
Rust
Raw Normal View History

use std::slice::from_raw_parts;
use std::sync::Arc;
use std::{io, mem};
use byteorder::{LittleEndian, ReadBytesExt, WriteBytesExt};
use sdset::Set;
use crate::DocumentId;
use crate::data::SharedData;
2018-12-01 18:37:21 +01:00
#[derive(Default, Clone)]
pub struct DocIds {
data: SharedData,
}
impl DocIds {
pub fn from_bytes(vec: Vec<u8>) -> io::Result<Self> {
let len = vec.len();
DocIds::from_shared_bytes(Arc::new(vec), 0, len)
}
pub fn from_shared_bytes(bytes: Arc<Vec<u8>>, offset: usize, len: usize) -> io::Result<Self> {
let data = SharedData { bytes, offset, len };
DocIds::from_data(data)
}
fn from_data(data: SharedData) -> io::Result<Self> {
let len = data.as_ref().read_u64::<LittleEndian>()?;
let data = data.range(mem::size_of::<u64>(), len as usize);
2018-11-23 18:00:24 +01:00
Ok(DocIds { data })
}
pub fn from_raw(vec: Vec<DocumentId>) -> Self {
DocIds::from_bytes(unsafe { mem::transmute(vec) }).unwrap()
}
pub fn write_to_bytes(&self, bytes: &mut Vec<u8>) {
let len = self.data.len() as u64;
bytes.write_u64::<LittleEndian>(len).unwrap();
bytes.extend_from_slice(&self.data);
}
pub fn contains(&self, doc: DocumentId) -> bool {
// FIXME prefer using the sdset::exponential_search function
self.doc_ids().binary_search(&doc).is_ok()
}
pub fn doc_ids(&self) -> &Set<DocumentId> {
2018-11-23 18:00:24 +01:00
let slice = &self.data;
let ptr = slice.as_ptr() as *const DocumentId;
let len = slice.len() / mem::size_of::<DocumentId>();
let slice = unsafe { from_raw_parts(ptr, len) };
Set::new_unchecked(slice)
}
}