Change the project to become a workspace with milli as a default-member

This commit is contained in:
Clément Renault
2021-02-12 16:15:09 +01:00
parent d450b971f9
commit e8639517da
56 changed files with 1053 additions and 2617 deletions

View File

@ -0,0 +1,87 @@
use std::borrow::Cow;
use std::convert::TryInto;
use crate::facet::value_encoding::f64_into_bytes;
use crate::FieldId;
// TODO do not de/serialize right bound when level = 0
pub struct FacetLevelValueF64Codec;
impl<'a> heed::BytesDecode<'a> for FacetLevelValueF64Codec {
type DItem = (FieldId, u8, f64, f64);
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
let (field_id, bytes) = bytes.split_first()?;
let (level, bytes) = bytes.split_first()?;
let (left, right) = if *level != 0 {
let left = bytes[16..24].try_into().ok().map(f64::from_be_bytes)?;
let right = bytes[24..].try_into().ok().map(f64::from_be_bytes)?;
(left, right)
} else {
let left = bytes[8..].try_into().ok().map(f64::from_be_bytes)?;
(left, left)
};
Some((*field_id, *level, left, right))
}
}
impl heed::BytesEncode<'_> for FacetLevelValueF64Codec {
type EItem = (FieldId, u8, f64, f64);
fn bytes_encode((field_id, level, left, right): &Self::EItem) -> Option<Cow<[u8]>> {
let mut buffer = [0u8; 32];
let len = if *level != 0 {
// Write the globally ordered floats.
let bytes = f64_into_bytes(*left)?;
buffer[..8].copy_from_slice(&bytes[..]);
let bytes = f64_into_bytes(*right)?;
buffer[8..16].copy_from_slice(&bytes[..]);
// Then the f64 values just to be able to read them back.
let bytes = left.to_be_bytes();
buffer[16..24].copy_from_slice(&bytes[..]);
let bytes = right.to_be_bytes();
buffer[24..].copy_from_slice(&bytes[..]);
32 // length
} else {
// Write the globally ordered floats.
let bytes = f64_into_bytes(*left)?;
buffer[..8].copy_from_slice(&bytes[..]);
// Then the f64 values just to be able to read them back.
let bytes = left.to_be_bytes();
buffer[8..16].copy_from_slice(&bytes[..]);
16 // length
};
let mut bytes = Vec::with_capacity(len + 2);
bytes.push(*field_id);
bytes.push(*level);
bytes.extend_from_slice(&buffer[..len]);
Some(Cow::Owned(bytes))
}
}
#[cfg(test)]
mod tests {
use heed::{BytesEncode, BytesDecode};
use super::*;
#[test]
fn globally_ordered_f64() {
let bytes = FacetLevelValueF64Codec::bytes_encode(&(3, 0, 32.0, 0.0)).unwrap();
let (name, level, left, right) = FacetLevelValueF64Codec::bytes_decode(&bytes).unwrap();
assert_eq!((name, level, left, right), (3, 0, 32.0, 32.0));
let bytes = FacetLevelValueF64Codec::bytes_encode(&(3, 1, -32.0, 32.0)).unwrap();
let (name, level, left, right) = FacetLevelValueF64Codec::bytes_decode(&bytes).unwrap();
assert_eq!((name, level, left, right), (3, 1, -32.0, 32.0));
}
}

View File

@ -0,0 +1,44 @@
use std::borrow::Cow;
use std::convert::TryInto;
use crate::facet::value_encoding::{i64_from_bytes, i64_into_bytes};
use crate::FieldId;
pub struct FacetLevelValueI64Codec;
impl<'a> heed::BytesDecode<'a> for FacetLevelValueI64Codec {
type DItem = (FieldId, u8, i64, i64);
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
let (field_id, bytes) = bytes.split_first()?;
let (level, bytes) = bytes.split_first()?;
let left = bytes[..8].try_into().map(i64_from_bytes).ok()?;
let right = if *level != 0 {
bytes[8..].try_into().map(i64_from_bytes).ok()?
} else {
left
};
Some((*field_id, *level, left, right))
}
}
impl heed::BytesEncode<'_> for FacetLevelValueI64Codec {
type EItem = (FieldId, u8, i64, i64);
fn bytes_encode((field_id, level, left, right): &Self::EItem) -> Option<Cow<[u8]>> {
let left = i64_into_bytes(*left);
let right = i64_into_bytes(*right);
let mut bytes = Vec::with_capacity(2 + left.len() + right.len());
bytes.push(*field_id);
bytes.push(*level);
bytes.extend_from_slice(&left[..]);
if *level != 0 {
bytes.extend_from_slice(&right[..]);
}
Some(Cow::Owned(bytes))
}
}

View File

@ -0,0 +1,27 @@
use std::borrow::Cow;
use std::str;
use crate::FieldId;
pub struct FacetValueStringCodec;
impl<'a> heed::BytesDecode<'a> for FacetValueStringCodec {
type DItem = (FieldId, &'a str);
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
let (field_id, bytes) = bytes.split_first()?;
let value = str::from_utf8(bytes).ok()?;
Some((*field_id, value))
}
}
impl<'a> heed::BytesEncode<'a> for FacetValueStringCodec {
type EItem = (FieldId, &'a str);
fn bytes_encode((field_id, value): &Self::EItem) -> Option<Cow<[u8]>> {
let mut bytes = Vec::with_capacity(value.len() + 1);
bytes.push(*field_id);
bytes.extend_from_slice(value.as_bytes());
Some(Cow::Owned(bytes))
}
}

View File

@ -0,0 +1,36 @@
use std::borrow::Cow;
use std::convert::TryInto;
use crate::{FieldId, DocumentId};
use crate::facet::value_encoding::f64_into_bytes;
pub struct FieldDocIdFacetF64Codec;
impl<'a> heed::BytesDecode<'a> for FieldDocIdFacetF64Codec {
type DItem = (FieldId, DocumentId, f64);
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
let (field_id, bytes) = bytes.split_first()?;
let (document_id_bytes, bytes) = bytes.split_at(4);
let document_id = document_id_bytes.try_into().map(u32::from_be_bytes).ok()?;
let value = bytes[8..16].try_into().map(f64::from_be_bytes).ok()?;
Some((*field_id, document_id, value))
}
}
impl<'a> heed::BytesEncode<'a> for FieldDocIdFacetF64Codec {
type EItem = (FieldId, DocumentId, f64);
fn bytes_encode((field_id, document_id, value): &Self::EItem) -> Option<Cow<[u8]>> {
let mut bytes = Vec::with_capacity(1 + 4 + 8 + 8);
bytes.push(*field_id);
bytes.extend_from_slice(&document_id.to_be_bytes());
let value_bytes = f64_into_bytes(*value)?;
bytes.extend_from_slice(&value_bytes);
bytes.extend_from_slice(&value.to_be_bytes());
Some(Cow::Owned(bytes))
}
}

View File

@ -0,0 +1,34 @@
use std::borrow::Cow;
use std::convert::TryInto;
use crate::facet::value_encoding::{i64_into_bytes, i64_from_bytes};
use crate::{FieldId, DocumentId};
pub struct FieldDocIdFacetI64Codec;
impl<'a> heed::BytesDecode<'a> for FieldDocIdFacetI64Codec {
type DItem = (FieldId, DocumentId, i64);
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
let (field_id, bytes) = bytes.split_first()?;
let (document_id_bytes, bytes) = bytes.split_at(4);
let document_id = document_id_bytes.try_into().map(u32::from_be_bytes).ok()?;
let value = bytes[..8].try_into().map(i64_from_bytes).ok()?;
Some((*field_id, document_id, value))
}
}
impl<'a> heed::BytesEncode<'a> for FieldDocIdFacetI64Codec {
type EItem = (FieldId, DocumentId, i64);
fn bytes_encode((field_id, document_id, value): &Self::EItem) -> Option<Cow<[u8]>> {
let mut bytes = Vec::with_capacity(1 + 4 + 8);
bytes.push(*field_id);
bytes.extend_from_slice(&document_id.to_be_bytes());
bytes.extend_from_slice(&i64_into_bytes(*value));
Some(Cow::Owned(bytes))
}
}

View File

@ -0,0 +1,31 @@
use std::borrow::Cow;
use std::convert::TryInto;
use std::str;
use crate::{FieldId, DocumentId};
pub struct FieldDocIdFacetStringCodec;
impl<'a> heed::BytesDecode<'a> for FieldDocIdFacetStringCodec {
type DItem = (FieldId, DocumentId, &'a str);
fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
let (field_id, bytes) = bytes.split_first()?;
let (document_id_bytes, bytes) = bytes.split_at(4);
let document_id = document_id_bytes.try_into().map(u32::from_be_bytes).ok()?;
let value = str::from_utf8(bytes).ok()?;
Some((*field_id, document_id, value))
}
}
impl<'a> heed::BytesEncode<'a> for FieldDocIdFacetStringCodec {
type EItem = (FieldId, DocumentId, &'a str);
fn bytes_encode((field_id, document_id, value): &Self::EItem) -> Option<Cow<[u8]>> {
let mut bytes = Vec::with_capacity(1 + 4 + value.len());
bytes.push(*field_id);
bytes.extend_from_slice(&document_id.to_be_bytes());
bytes.extend_from_slice(value.as_bytes());
Some(Cow::Owned(bytes))
}
}

View File

@ -0,0 +1,13 @@
mod facet_level_value_f64_codec;
mod facet_level_value_i64_codec;
mod facet_value_string_codec;
mod field_doc_id_facet_f64_codec;
mod field_doc_id_facet_i64_codec;
mod field_doc_id_facet_string_codec;
pub use self::facet_level_value_f64_codec::FacetLevelValueF64Codec;
pub use self::facet_level_value_i64_codec::FacetLevelValueI64Codec;
pub use self::facet_value_string_codec::FacetValueStringCodec;
pub use self::field_doc_id_facet_f64_codec::FieldDocIdFacetF64Codec;
pub use self::field_doc_id_facet_i64_codec::FieldDocIdFacetI64Codec;
pub use self::field_doc_id_facet_string_codec::FieldDocIdFacetStringCodec;