rust/tantivy
1
0
Fork 0
mirror of https://github.com/quickwit-oss/tantivy.git synced 2026-01-05 16:52:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

Backmerging master

Browse Source
This commit is contained in:
Paul Masurel
2018-02-12 11:08:57 +09:00
parent eb84b8a60d a7ffc0e610
commit c4125bda59
67 changed files with 2060 additions and 1249 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
CHANGELOG.md
View File

@@ -1,3 +1,13 @@
Tantivy 0.5
==========================
- Faceting
- RangeQuery
- Configurable tokenization pipeline
- Allowing super large indexes
- 64 bits file address
- Smarter encoding of the `TermInfo` objects
Tantivy 0.4.3
==========================

23
src/collector/count_collector.rs
View File

@@ -20,7 +20,6 @@ impl CountCollector {
}
}
impl Collector for CountCollector {
fn set_segment(&mut self, _: SegmentLocalId, _: &SegmentReader) -> Result<()> {
Ok(())
@@ -34,18 +33,16 @@ impl Collector for CountCollector {
#[cfg(test)]
mod tests {
use super::*;
use test::Bencher;
use collector::Collector;
use collector::{Collector, CountCollector};
#[bench]
fn build_collector(b: &mut Bencher) {
b.iter(|| {
let mut count_collector = CountCollector::default();
for doc in 0..1_000_000 {
count_collector.collect(doc, 1f32);
}
count_collector.count()
});
#[test]
fn test_count_collector() {
let mut count_collector = CountCollector::default();
assert_eq!(count_collector.count(), 0);
count_collector.collect(0u32, 1f32);
assert_eq!(count_collector.count(), 1);
assert_eq!(count_collector.count(), 1);
count_collector.collect(1u32, 1f32);
assert_eq!(count_collector.count(), 2);
}
}

295
src/collector/facet_collector.rs
View File

@@ -13,11 +13,9 @@ use termdict::TermStreamerBuilder;
use std::collections::BTreeSet;
use termdict::TermMerger;
use postings::SkipResult;
use std::{u64, usize};
use schema::FACET_SEP_BYTE;
use std::{usize, u64};
use std::iter::Peekable;
use DocId;
use Result;
use Score;
@@ -50,7 +48,6 @@ impl<'a> Ord for Hit<'a> {
}
}
struct SegmentFacetCounter {
pub facet_reader: FacetReader,
pub facet_ords: Vec<u64>,
@@ -61,16 +58,10 @@ fn facet_depth(facet_bytes: &[u8]) -> usize {
if facet_bytes.is_empty() {
0
} else {
facet_bytes
.iter()
.cloned()
.filter(|b| *b == 0u8)
.count() + 1
facet_bytes.iter().cloned().filter(|b| *b == 0u8).count() + 1
}
}
/// Collector for faceting
///
/// The collector collects all facets. You need to configure it
@@ -93,7 +84,7 @@ fn facet_depth(facet_bytes: &[u8]) -> usize {
/// (e.g. `/category/fiction`, `/category/biography`, `/category/personal_development`).
///
/// Once collection is finished, you can harvest its results in the form
/// of a `FacetCounts` object, and extract your facet counts from it.
/// of a `FacetCounts` object, and extract your face t counts from it.
///
/// This implementation assumes you are working with a number of facets that
/// is much hundreds of time lower than your number of documents.
@@ -204,7 +195,6 @@ fn facet_depth(facet_bytes: &[u8]) -> usize {
/// Ok(())
/// }
/// ```
pub struct FacetCollector {
facet_ords: Vec<u64>,
field: Field,
@@ -218,23 +208,24 @@ pub struct FacetCollector {
// collapse facet_id -> facet_ord
current_collapse_facet_ords: Vec<u64>,
collapse: BTreeSet<Vec<u8>>,
facets: BTreeSet<Facet>,
}
fn skip<'a, I: Iterator<Item=&'a Vec<u8>>>(target: &[u8], collapse_it: &mut Peekable<I>) -> SkipResult {
fn skip<'a, I: Iterator<Item = &'a Facet>>(
target: &[u8],
collapse_it: &mut Peekable<I>,
) -> SkipResult {
loop {
match collapse_it.peek() {
Some(facet_bytes) => {
match facet_bytes[..].cmp(&target) {
Ordering::Less => {}
Ordering::Greater => {
return SkipResult::OverStep;
}
Ordering::Equal => {
return SkipResult::Reached;
}
Some(facet_bytes) => match facet_bytes.encoded_bytes().cmp(target) {
Ordering::Less => {}
Ordering::Greater => {
return SkipResult::OverStep;
}
}
Ordering::Equal => {
return SkipResult::Reached;
}
},
None => {
return SkipResult::End;
}
@@ -244,71 +235,59 @@ fn skip<'a, I: Iterator<Item=&'a Vec<u8>>>(target: &[u8], collapse_it: &mut Peek
}
impl FacetCollector {
/// Create a facet collector to collect the facets
/// from a specific facet `Field`.
///
/// This function does not check whether the field
/// is of the proper type.
pub fn for_field(field: Field) -> FacetCollector {
let mut facet_collector = FacetCollector {
FacetCollector {
facet_ords: Vec::with_capacity(255),
field: field,
ff_reader: None,
segment_counters: Vec::new(),
collapse: BTreeSet::new(),
field,
ff_reader: None,
facets: BTreeSet::new(),
current_segment_collapse_mapping: Vec::new(),
current_collapse_facet_ords: Vec::new(),
current_segment_counts: Vec::new(),
};
facet_collector.add_facet(Facet::from("/"));
facet_collector
}
/// Adds a facet that we want to collect.
///
/// For all of the facets that are one level below the facet
/// given in argument, the collector will collects the count of unique
/// documents containing the facet.
///
/// Calling `add_facet()` on two facets that have a ancestor
/// descendant relationships is not allowed.
pub fn add_facet<T>(&mut self, facet_from: T)
where Facet: From<T> {
let facet = Facet::from(facet_from);
let facet_bytes: &[u8] = facet.encoded_bytes();
self.collapse.remove(&facet_bytes[..0]);
for pos in facet_bytes.iter()
.cloned()
.position(|b| b == FACET_SEP_BYTE) {
self.collapse.remove(&facet_bytes[..pos]);
}
self.collapse.insert(facet_bytes.to_owned());
}
fn finalize_segment(&mut self) {
if self.ff_reader.is_some() {
self.segment_counters.push(
SegmentFacetCounter {
facet_reader: self.ff_reader.take().unwrap().into_inner(),
facet_ords: mem::replace(&mut self.current_collapse_facet_ords, Vec::new()),
facet_counts: mem::replace(&mut self.current_segment_counts, Vec::new()),
}
/// Adds a facet that we want to record counts
///
/// Adding facet `Facet::from("/country")` for instance,
/// will record the counts of all of the direct children of the facet country
/// (e.g. `/country/FR`, `/country/UK`).
///
/// Adding two facets within which one is the prefix of the other is forbidden.
/// If you need the correct number of unique documents for two such facets,
/// just add them in separate `FacetCollector`.
pub fn add_facet<T>(&mut self, facet_from: T)
where
Facet: From<T>,
{
let facet = Facet::from(facet_from);
for old_facet in &self.facets {
assert!(
!old_facet.is_prefix_of(&facet),
"Tried to add a facet which is a descendant of an already added facet."
);
assert!(
!facet.is_prefix_of(old_facet),
"Tried to add a facet which is an ancestor of an already added facet."
);
}
self.facets.insert(facet);
}
fn set_collapse_mapping(&mut self, facet_reader: &FacetReader) {
self.current_segment_collapse_mapping.clear();
self.current_collapse_facet_ords.clear();
self.current_segment_counts.clear();
let mut collapse_facet_it = self.collapse.iter().peekable();
let mut collapse_facet_it = self.facets.iter().peekable();
self.current_collapse_facet_ords.push(0);
let mut facet_streamer = facet_reader
.facet_dict()
.range()
.into_stream();
let mut facet_streamer = facet_reader.facet_dict().range().into_stream();
if !facet_streamer.advance() {
return;
}
@@ -326,9 +305,11 @@ impl FacetCollector {
let depth = facet_depth(facet_streamer.key());
if depth <= collapse_depth {
continue 'outer;
} else if depth == collapse_depth + 1 {
}
if depth == collapse_depth + 1 {
collapsed_id = self.current_collapse_facet_ords.len();
self.current_collapse_facet_ords.push(facet_streamer.term_ord());
self.current_collapse_facet_ords
.push(facet_streamer.term_ord());
self.current_segment_collapse_mapping.push(collapsed_id);
} else {
self.current_segment_collapse_mapping.push(collapsed_id);
@@ -346,6 +327,16 @@ impl FacetCollector {
}
}
fn finalize_segment(&mut self) {
if self.ff_reader.is_some() {
self.segment_counters.push(SegmentFacetCounter {
facet_reader: self.ff_reader.take().unwrap().into_inner(),
facet_ords: mem::replace(&mut self.current_collapse_facet_ords, Vec::new()),
facet_counts: mem::replace(&mut self.current_segment_counts, Vec::new()),
});
}
}
/// Returns the results of the collection.
///
/// This method does not just return the counters,
@@ -362,14 +353,9 @@ impl FacetCollector {
.map(|segment_counter| &segment_counter.facet_counts[..])
.collect();
let facet_streams = self.segment_counters
.iter()
.map(|seg_counts| seg_counts
.facet_reader
.facet_dict()
.range()
.into_stream())
.map(|seg_counts| seg_counts.facet_reader.facet_dict().range().into_stream())
.collect::<Vec<_>>();
let mut facet_merger = TermMerger::new(facet_streams);
@@ -399,51 +385,43 @@ impl FacetCollector {
facet_counts.insert(Facet::from_encoded(bytes), count);
}
}
FacetCounts {
facet_counts: facet_counts
}
FacetCounts { facet_counts }
}
}
impl Collector for FacetCollector {
fn set_segment(&mut self, _: SegmentLocalId, reader: &SegmentReader) -> Result<()> {
self.finalize_segment();
let facet_reader = reader.facet_reader(self.field)?;
self.set_collapse_mapping(&facet_reader);
self.current_segment_counts.resize(self.current_collapse_facet_ords.len(), 0);
self.current_segment_counts
.resize(self.current_collapse_facet_ords.len(), 0);
self.ff_reader = Some(UnsafeCell::new(facet_reader));
Ok(())
}
fn collect(&mut self, doc: DocId, _: Score) {
let facet_reader: &mut FacetReader =
unsafe {
&mut *self.ff_reader
.as_ref()
.expect("collect() was called before set_segment. This should never happen.")
.get()
};
let facet_reader: &mut FacetReader = unsafe {
&mut *self.ff_reader
.as_ref()
.expect("collect() was called before set_segment. This should never happen.")
.get()
};
facet_reader.facet_ords(doc, &mut self.facet_ords);
let mut previous_collapsed_ord: usize = usize::MAX;
for &facet_ord in &self.facet_ords {
let collapsed_ord = self.current_segment_collapse_mapping[facet_ord as usize];
self.current_segment_counts[collapsed_ord] +=
if collapsed_ord == previous_collapsed_ord {
0
} else {
1
};
self.current_segment_counts[collapsed_ord] += if collapsed_ord == previous_collapsed_ord
{
0
} else {
1
};
previous_collapsed_ord = collapsed_ord;
}
}
}
/// Intermediary result of the `FacetCollector` that stores
/// the facet counts for all the segments.
pub struct FacetCounts {
@@ -451,20 +429,21 @@ pub struct FacetCounts {
}
impl FacetCounts {
pub fn get<'a, T>(&'a self, facet_from: T) -> impl Iterator<Item=(&'a Facet, u64)>
where Facet: From<T> {
#[allow(needless_lifetimes)] //< compiler fails if we remove the lifetime
pub fn get<'a, T>(&'a self, facet_from: T) -> impl Iterator<Item = (&'a Facet, u64)>
where
Facet: From<T>,
{
let facet = Facet::from(facet_from);
let left_bound = Bound::Excluded(facet.clone());
let right_bound =
if facet.is_root() {
Bound::Unbounded
} else {
let mut facet_after_bytes = facet.encoded_bytes().to_owned();
facet_after_bytes.push(1u8);
let facet_after = Facet::from_encoded(facet_after_bytes);
Bound::Excluded(facet_after)
};
let right_bound = if facet.is_root() {
Bound::Unbounded
} else {
let mut facet_after_bytes = facet.encoded_bytes().to_owned();
facet_after_bytes.push(1u8);
let facet_after = Facet::from_encoded(facet_after_bytes);
Bound::Excluded(facet_after)
};
self.facet_counts
.range((left_bound, right_bound))
@@ -472,29 +451,22 @@ impl FacetCounts {
}
pub fn top_k<T>(&self, facet: T, k: usize) -> Vec<(&Facet, u64)>
where Facet: From<T> {
where
Facet: From<T>,
{
let mut heap = BinaryHeap::with_capacity(k);
let mut it = self.get(facet);
for (ref facet, count) in (&mut it).take(k) {
heap.push(Hit {
count: count,
facet: facet
});
for (facet, count) in (&mut it).take(k) {
heap.push(Hit { count, facet });
}
let mut lowest_count: u64 = heap.peek()
.map(|hit| hit.count)
.unwrap_or(u64::MIN);
let mut lowest_count: u64 = heap.peek().map(|hit| hit.count).unwrap_or(u64::MIN);
for (facet, count) in it {
if count > lowest_count {
lowest_count = count;
if let Some(mut head) = heap.peek_mut() {
*head = Hit {
count: count,
facet: facet
};
*head = Hit { count, facet };
}
}
}
@@ -503,19 +475,17 @@ impl FacetCounts {
.map(|hit| (hit.facet, hit.count))
.collect::<Vec<_>>()
}
}
#[cfg(test)]
mod tests {
use test::Bencher;
use core::Index;
use schema::{SchemaBuilder, Document, Facet};
use schema::{Document, Facet, SchemaBuilder};
use query::AllQuery;
use super::{FacetCollector, FacetCounts};
use std::iter;
use schema::Field;
use rand::{thread_rng, Rng};
#[test]
@@ -556,19 +526,36 @@ mod tests {
.get("/top1")
.map(|(facet, count)| (facet.to_string(), count))
.collect();
assert_eq!(facets, [
("/top1/mid0", 50),
("/top1/mid1", 50),
("/top1/mid2", 50),
("/top1/mid3", 50),
].iter()
.map(|&(facet_str, count)| {
(String::from(facet_str), count)
})
.collect::<Vec<_>>());
assert_eq!(
facets,
[
("/top1/mid0", 50),
("/top1/mid1", 50),
("/top1/mid2", 50),
("/top1/mid3", 50),
].iter()
.map(|&(facet_str, count)| (String::from(facet_str), count))
.collect::<Vec<_>>()
);
}
}
#[test]
#[should_panic(expected = "Tried to add a facet which is a descendant of \
an already added facet.")]
fn test_misused_facet_collector() {
let mut facet_collector = FacetCollector::for_field(Field(0));
facet_collector.add_facet(Facet::from("/country"));
facet_collector.add_facet(Facet::from("/country/europe"));
}
#[test]
fn test_non_used_facet_collector() {
let mut facet_collector = FacetCollector::for_field(Field(0));
facet_collector.add_facet(Facet::from("/country"));
facet_collector.add_facet(Facet::from("/countryeurope"));
}
#[test]
fn test_facet_collector_topk() {
let mut schema_builder = SchemaBuilder::new();
@@ -576,18 +563,14 @@ mod tests {
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut docs: Vec<Document> = vec![
("a", 10),
("b", 100),
("c", 7),
("d", 12),
("e", 21)
].into_iter()
.flat_map(|(c, count)| {
let facet = Facet::from(&format!("/facet_{}", c));
let doc = doc!(facet_field => facet);
iter::repeat(doc).take(count)
}).collect();
let mut docs: Vec<Document> = vec![("a", 10), ("b", 100), ("c", 7), ("d", 12), ("e", 21)]
.into_iter()
.flat_map(|(c, count)| {
let facet = Facet::from(&format!("/facet_{}", c));
let doc = doc!(facet_field => facet);
iter::repeat(doc).take(count)
})
.collect();
thread_rng().shuffle(&mut docs[..]);
let mut index_writer = index.writer(3_000_000).unwrap();
@@ -611,8 +594,9 @@ mod tests {
vec![
(&Facet::from("/facet_b"), 100),
(&Facet::from("/facet_e"), 21),
(&Facet::from("/facet_d"), 12)
]);
(&Facet::from("/facet_d"), 12),
]
);
}
}
@@ -623,12 +607,12 @@ mod tests {
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut docs = vec!();
let mut docs = vec![];
for val in 0..50 {
let facet = Facet::from(&format!("/facet_{}", val));
for _ in 0..val*val {
docs.push(doc!(facet_field=>facet.clone()));
}
for _ in 0..val * val {
docs.push(doc!(facet_field=>facet.clone()));
}
}
// 40425 docs
thread_rng().shuffle(&mut docs[..]);
@@ -647,4 +631,3 @@ mod tests {
});
}
}

87
src/common/bitpacker.rs
View File

@@ -4,64 +4,35 @@ use common::serialize::BinarySerializable;
use std::mem;
use std::ops::Deref;
/// Computes the number of bits that will be used for bitpacking.
///
/// In general the target is the minimum number of bits
/// required to express the amplitude given in argument.
///
/// e.g. If the amplitude is 10, we can store all ints on simply 4bits.
///
/// The logic is slightly more convoluted here as for optimization
/// reasons, we want to ensure that a value spawns over at most 8 bytes
/// of aligns bytes.
///
/// Spanning over 9 bytes is possible for instance, if we do
/// bitpacking with an amplitude of 63 bits.
/// In this case, the second int will start on bit
/// 63 (which belongs to byte 7) and ends at byte 15;
/// Hence 9 bytes (from byte 7 to byte 15 included).
///
/// To avoid this, we force the number of bits to 64bits
/// when the result is greater than `64-8 = 56 bits`.
///
/// Note that this only affects rare use cases spawning over
/// a very large range of values. Even in this case, it results
/// in an extra cost of at most 12% compared to the optimal
/// number of bits.
pub fn compute_num_bits(amplitude: u64) -> u8 {
let amplitude = (64u32 - amplitude.leading_zeros()) as u8;
if amplitude <= 64 - 8 {
amplitude
} else {
64
}
}
pub struct BitPacker {
pub(crate) struct BitPacker {
mini_buffer: u64,
mini_buffer_written: usize,
num_bits: usize,
}
impl BitPacker {
pub fn new(num_bits: usize) -> BitPacker {
pub fn new() -> BitPacker {
BitPacker {
mini_buffer: 0u64,
mini_buffer_written: 0,
num_bits,
}
}
pub fn write<TWrite: Write>(&mut self, val: u64, output: &mut TWrite) -> io::Result<()> {
pub fn write<TWrite: Write>(
&mut self,
val: u64,
num_bits: u8,
output: &mut TWrite,
) -> io::Result<()> {
let val_u64 = val as u64;
if self.mini_buffer_written + self.num_bits > 64 {
let num_bits = num_bits as usize;
if self.mini_buffer_written + num_bits > 64 {
self.mini_buffer |= val_u64.wrapping_shl(self.mini_buffer_written as u32);
self.mini_buffer.serialize(output)?;
self.mini_buffer = val_u64.wrapping_shr((64 - self.mini_buffer_written) as u32);
self.mini_buffer_written = self.mini_buffer_written + (self.num_bits as usize) - 64;
self.mini_buffer_written = self.mini_buffer_written + num_bits - 64;
} else {
self.mini_buffer |= val_u64 << self.mini_buffer_written;
self.mini_buffer_written += self.num_bits;
self.mini_buffer_written += num_bits;
if self.mini_buffer_written == 64 {
self.mini_buffer.serialize(output)?;
self.mini_buffer_written = 0;
@@ -71,7 +42,7 @@ impl BitPacker {
Ok(())
}
pub(crate) fn flush<TWrite: Write>(&mut self, output: &mut TWrite) -> io::Result<()> {
pub fn flush<TWrite: Write>(&mut self, output: &mut TWrite) -> io::Result<()> {
if self.mini_buffer_written > 0 {
let num_bytes = (self.mini_buffer_written + 7) / 8;
let arr: [u8; 8] = unsafe { mem::transmute::<u64, [u8; 8]>(self.mini_buffer) };
@@ -103,14 +74,14 @@ impl<Data> BitUnpacker<Data>
where
Data: Deref<Target = [u8]>,
{
pub fn new(data: Data, num_bits: usize) -> BitUnpacker<Data> {
pub fn new(data: Data, num_bits: u8) -> BitUnpacker<Data> {
let mask: u64 = if num_bits == 64 {
!0u64
} else {
(1u64 << num_bits) - 1u64
};
BitUnpacker {
num_bits,
num_bits: num_bits as usize,
mask,
data,
}
@@ -147,7 +118,7 @@ where
}
unsafe { *(buffer[..].as_ptr() as *const u64) }
};
let val_shifted = (val_unshifted_unmasked >> bit_shift) as u64;
let val_shifted = val_unshifted_unmasked >> (bit_shift as u64);
(val_shifted & mask)
}
}
@@ -176,37 +147,25 @@ where
#[cfg(test)]
mod test {
use super::{compute_num_bits, BitPacker, BitUnpacker};
use super::{BitPacker, BitUnpacker};
#[test]
fn test_compute_num_bits() {
assert_eq!(compute_num_bits(1), 1u8);
assert_eq!(compute_num_bits(0), 0u8);
assert_eq!(compute_num_bits(2), 2u8);
assert_eq!(compute_num_bits(3), 2u8);
assert_eq!(compute_num_bits(4), 3u8);
assert_eq!(compute_num_bits(255), 8u8);
assert_eq!(compute_num_bits(256), 9u8);
assert_eq!(compute_num_bits(5_000_000_000), 33u8);
}
fn create_fastfield_bitpacker(len: usize, num_bits: usize) -> (BitUnpacker<Vec<u8>>, Vec<u64>) {
fn create_fastfield_bitpacker(len: usize, num_bits: u8) -> (BitUnpacker<Vec<u8>>, Vec<u64>) {
let mut data = Vec::new();
let mut bitpacker = BitPacker::new(num_bits);
let max_val: u64 = (1 << num_bits) - 1;
let mut bitpacker = BitPacker::new();
let max_val: u64 = (1u64 << num_bits as u64) - 1u64;
let vals: Vec<u64> = (0u64..len as u64)
.map(|i| if max_val == 0 { 0 } else { i % max_val })
.collect();
for &val in &vals {
bitpacker.write(val, &mut data).unwrap();
bitpacker.write(val, num_bits, &mut data).unwrap();
}
bitpacker.close(&mut data).unwrap();
assert_eq!(data.len(), (num_bits * len + 7) / 8 + 7);
assert_eq!(data.len(), ((num_bits as usize) * len + 7) / 8 + 7);
let bitunpacker = BitUnpacker::new(data, num_bits);
(bitunpacker, vals)
}
fn test_bitpacker_util(len: usize, num_bits: usize) {
fn test_bitpacker_util(len: usize, num_bits: u8) {
let (bitunpacker, vals) = create_fastfield_bitpacker(len, num_bits);
for (i, val) in vals.iter().enumerate() {
assert_eq!(bitunpacker.get(i), *val);

402
src/common/bitset.rs
View File

@@ -1,190 +1,260 @@
use DocId;
use std::fmt;
pub trait TinySet {
fn insert(&mut self, b: u32);
fn is_empty(&self) -> bool;
fn pop_lowest(&mut self) -> Option<u32>;
fn remove(&mut self, b: u32);
fn lowest(&mut self) -> Option<u32>;
#[derive(Clone, Copy, Eq, PartialEq)]
pub(crate) struct TinySet(u64);
/// Update self to represent the
/// intersection of its elements and the other
/// set given in arguments.
fn intersect(&mut self, other: Self);
/// Returns a `TinySet` than contains all values up
/// to limit excluded.
///
/// The limit is assumed to be strictly lower than 64.
fn range_lower(limit: u32) -> u64;
/// Returns a `TinySet` that contains all values greater
/// or equal to the given limit, included. (and up to 63)
///
/// The limit is assumed to be strictly lower than 64.
fn range_greater_or_equal(from_included: u32) -> u64 {
assert!(from_included < 64);
0 ^ Self::range_lower(from_included)
impl fmt::Debug for TinySet {
fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
self.into_iter().collect::<Vec<u32>>().fmt(f)
}
}
impl TinySet for u64 {
fn range_lower(from_included: u32) -> u64 {
assert!(from_included < 64);
(1u64 << (from_included as u64)) - 1u64
pub struct TinySetIterator(TinySet);
impl Iterator for TinySetIterator {
type Item = u32;
fn next(&mut self) -> Option<Self::Item> {
self.0.pop_lowest()
}
}
impl IntoIterator for TinySet {
type Item = u32;
type IntoIter = TinySetIterator;
fn into_iter(self) -> Self::IntoIter {
TinySetIterator(self)
}
}
impl TinySet {
/// Returns an empty `TinySet`.
pub fn empty() -> TinySet {
TinySet(0u64)
}
fn intersect(&mut self, filter_mask: u64) {
*self &= filter_mask;
/// Returns the complement of the set in `[0, 64[`.
fn complement(&self) -> TinySet {
TinySet(!self.0)
}
/// Returns true iff the `TinySet` contains the element `el`.
pub fn contains(&self, el: u32) -> bool {
!self.intersect(TinySet::singleton(el)).is_empty()
}
/// Returns the intersection of `self` and `other`
pub fn intersect(&self, other: TinySet) -> TinySet {
TinySet(self.0 & other.0)
}
/// Creates a new `TinySet` containing only one element
/// within `[0; 64[`
#[inline(always)]
fn insert(&mut self, b: u32) {
*self |= 1u64 << (b as u64);
pub fn singleton(el: u32) -> TinySet {
TinySet(1u64 << (el as u64))
}
/// Insert a new element within [0..64[
#[inline(always)]
fn is_empty(&self) -> bool {
*self == 0u64
pub fn insert(self, el: u32) -> TinySet {
self.union(TinySet::singleton(el))
}
/// Insert a new element within [0..64[
#[inline(always)]
fn pop_lowest(&mut self) -> Option<u32> {
pub fn insert_mut(&mut self, el: u32) -> bool {
let old = *self;
*self = old.insert(el);
old != *self
}
/// Returns the union of two tinysets
#[inline(always)]
pub fn union(self, other: TinySet) -> TinySet {
TinySet(self.0 | other.0)
}
/// Returns true iff the `TinySet` is empty.
#[inline(always)]
pub fn is_empty(&self) -> bool {
self.0 == 0u64
}
/// Returns the lowest element in the `TinySet`
/// and removes it.
#[inline(always)]
pub fn pop_lowest(&mut self) -> Option<u32> {
if let Some(lowest) = self.lowest() {
self.remove(lowest);
self.0 ^= TinySet::singleton(lowest).0;
Some(lowest)
} else {
None
}
}
/// Returns the lowest element in the `TinySet`
/// (or None if the set is empty).
#[inline(always)]
fn remove(&mut self, b: u32) {
*self ^= 1 << (b as u64);
}
#[inline(always)]
fn lowest(&mut self) -> Option<u32> {
pub fn lowest(&mut self) -> Option<u32> {
if self.is_empty() {
None
} else {
let least_significant_bit = self.trailing_zeros() as u32;
let least_significant_bit = self.0.trailing_zeros() as u32;
Some(least_significant_bit)
}
}
/// Returns a `TinySet` than contains all values up
/// to limit excluded.
///
/// The limit is assumed to be strictly lower than 64.
pub fn range_lower(upper_bound: u32) -> TinySet {
TinySet((1u64 << ((upper_bound % 64u32) as u64)) - 1u64)
}
/// Returns a `TinySet` that contains all values greater
/// or equal to the given limit, included. (and up to 63)
///
/// The limit is assumed to be strictly lower than 64.
pub fn range_greater_or_equal(from_included: u32) -> TinySet {
TinySet::range_lower(from_included).complement()
}
}
pub struct DocBitSet {
tinybitsets: Box<[u64]>,
size_hint: usize, //< Technically it should be u32, but we
// count multiple inserts.
// `usize` guards us from overflow.
max_doc: DocId
#[derive(Clone)]
pub struct BitSet {
tinysets: Box<[TinySet]>,
len: usize, //< Technically it should be u32, but we
// count multiple inserts.
// `usize` guards us from overflow.
max_value: u32,
}
impl DocBitSet {
pub fn with_maxdoc(max_doc: DocId) -> DocBitSet {
let num_buckets = (max_doc + 63) / 64;
DocBitSet {
tinybitsets: vec![0u64; num_buckets as usize].into_boxed_slice(),
size_hint: 0,
max_doc
fn num_buckets(max_val: u32) -> u32 {
(max_val + 63u32) / 64u32
}
impl BitSet {
/// Create a new `BitSet` that may contain elements
/// within `[0, max_val[`.
pub fn with_max_value(max_value: u32) -> BitSet {
let num_buckets = num_buckets(max_value);
let tinybisets = vec![TinySet::empty(); num_buckets as usize].into_boxed_slice();
BitSet {
tinysets: tinybisets,
len: 0,
max_value,
}
}
pub fn size_hint(&self) -> u32 {
if self.max_doc as usize > self.size_hint {
self.size_hint as u32
} else {
self.max_doc
/// Removes all elements from the `BitSet`.
pub fn clear(&mut self) {
for tinyset in self.tinysets.iter_mut() {
*tinyset = TinySet::empty();
}
}
pub fn insert(&mut self, doc: DocId) {
/// Returns the number of elements in the `BitSet`.
pub fn len(&self) -> usize {
self.len
}
/// Inserts an element in the `BitSet`
pub fn insert(&mut self, el: u32) {
// we do not check saturated els.
self.size_hint += 1;
let bucket = (doc / 64u32) as usize;
self.tinybitsets[bucket].insert(doc % 64u32);
let higher = el / 64u32;
let lower = el % 64u32;
self.len += if self.tinysets[higher as usize].insert_mut(lower) {
1
} else {
0
};
}
pub fn contains(&self, doc: DocId) -> bool {
let tiny_bitset = self.tiny_bitset((doc / 64u32) as usize);
let lower = doc % 64;
let mask = 1u64 << (lower as u64);
(tiny_bitset & mask) != 0u64
/// Returns true iff the elements is in the `BitSet`.
pub fn contains(&self, el: u32) -> bool {
self.tinyset(el / 64u32).contains(el % 64)
}
pub fn max_doc(&self) -> DocId {
self.max_doc
/// Returns the first non-empty `TinySet` associated to a bucket lower
/// or greater than bucket.
///
/// Reminder: the tiny set with the bucket `bucket`, represents the
/// elements from `bucket * 64` to `(bucket+1) * 64`.
pub(crate) fn first_non_empty_bucket(&self, bucket: u32) -> Option<u32> {
self.tinysets[bucket as usize..]
.iter()
.cloned()
.position(|tinyset| !tinyset.is_empty())
.map(|delta_bucket| bucket + delta_bucket as u32)
}
pub fn num_tiny_bitsets(&self) -> usize {
self.tinybitsets.len()
pub fn max_value(&self) -> u32 {
self.max_value
}
pub fn tiny_bitset(&self, bucket: usize) -> u64 {
self.tinybitsets[bucket]
/// Returns the tiny bitset representing the
/// the set restricted to the number range from
/// `bucket * 64` to `(bucket + 1) * 64`.
pub(crate) fn tinyset(&self, bucket: u32) -> TinySet {
self.tinysets[bucket as usize]
}
}
#[cfg(test)]
mod tests {
extern crate test;
use tests;
use std::collections::HashSet;
use DocId;
use super::BitSet;
use super::TinySet;
use super::DocBitSet;
use tests::generate_nonunique_unsorted;
use std::collections::BTreeSet;
use query::BitSetDocSet;
use DocSet;
#[test]
fn test_tiny_set() {
assert!(0u64.is_empty());
assert!(TinySet::empty().is_empty());
{
let mut u = 0u64;
u.insert(1u32);
let mut u = TinySet::empty().insert(1u32);
assert_eq!(u.pop_lowest(), Some(1u32));
assert!(u.pop_lowest().is_none())
}
{
let mut u = 0u64;
u.insert(1u32);
u.insert(1u32);
let mut u = TinySet::empty().insert(1u32).insert(1u32);
assert_eq!(u.pop_lowest(), Some(1u32));
assert!(u.pop_lowest().is_none())
}
{
let mut u = 0u64;
u.insert(2u32);
let mut u = TinySet::empty().insert(2u32);
assert_eq!(u.pop_lowest(), Some(2u32));
u.insert(1u32);
u.insert_mut(1u32);
assert_eq!(u.pop_lowest(), Some(1u32));
assert!(u.pop_lowest().is_none());
}
{
let mut u = 0u64;
u.insert(63u32);
let mut u = TinySet::empty().insert(63u32);
assert_eq!(u.pop_lowest(), Some(63u32));
assert!(u.pop_lowest().is_none());
}
}
#[test]
fn test_docbitset() {
// docs are assumed to be lower than 100.
let test_against_hashset = |docs: &[DocId], max_doc: u32| {
let mut hashset: HashSet<DocId> = HashSet::new();
let mut docbitset = DocBitSet::with_maxdoc(max_doc);
for &doc in docs {
assert!(doc < max_doc);
hashset.insert(doc);
docbitset.insert(doc);
fn test_bitset() {
let test_against_hashset = |els: &[u32], max_value: u32| {
let mut hashset: HashSet<u32> = HashSet::new();
let mut bitset = BitSet::with_max_value(max_value);
for &el in els {
assert!(el < max_value);
hashset.insert(el);
bitset.insert(el);
}
for doc in 0..max_doc {
assert_eq!(
hashset.contains(&doc),
docbitset.contains(doc)
);
for el in 0..max_value {
assert_eq!(hashset.contains(&el), bitset.contains(el));
}
assert_eq!(docbitset.max_doc(), max_doc);
assert_eq!(bitset.max_value(), max_value);
};
test_against_hashset(&[], 0);
@@ -194,17 +264,119 @@ mod tests {
test_against_hashset(&[1u32, 2u32], 4);
test_against_hashset(&[99u32], 100);
test_against_hashset(&[63u32], 64);
test_against_hashset(&[62u32,63u32], 64);
test_against_hashset(&[62u32, 63u32], 64);
}
#[test]
fn test_docbitset_num_buckets() {
assert_eq!(DocBitSet::with_maxdoc(0u32).num_tiny_bitsets(), 0);
assert_eq!(DocBitSet::with_maxdoc(1u32).num_tiny_bitsets(), 1);
assert_eq!(DocBitSet::with_maxdoc(64u32).num_tiny_bitsets(), 1);
assert_eq!(DocBitSet::with_maxdoc(65u32).num_tiny_bitsets(), 2);
assert_eq!(DocBitSet::with_maxdoc(128u32).num_tiny_bitsets(), 2);
assert_eq!(DocBitSet::with_maxdoc(129u32).num_tiny_bitsets(), 3);
fn test_bitset_large() {
let arr = generate_nonunique_unsorted(1_000_000, 50_000);
let mut btreeset: BTreeSet<u32> = BTreeSet::new();
let mut bitset = BitSet::with_max_value(1_000_000);
for el in arr {
btreeset.insert(el);
bitset.insert(el);
}
for i in 0..1_000_000 {
assert_eq!(btreeset.contains(&i), bitset.contains(i));
}
assert_eq!(btreeset.len(), bitset.len());
let mut bitset_docset = BitSetDocSet::from(bitset);
for el in btreeset.into_iter() {
bitset_docset.advance();
assert_eq!(bitset_docset.doc(), el);
}
assert!(!bitset_docset.advance());
}
#[test]
fn test_bitset_num_buckets() {
use super::num_buckets;
assert_eq!(num_buckets(0u32), 0);
assert_eq!(num_buckets(1u32), 1);
assert_eq!(num_buckets(64u32), 1);
assert_eq!(num_buckets(65u32), 2);
assert_eq!(num_buckets(128u32), 2);
assert_eq!(num_buckets(129u32), 3);
}
#[test]
fn test_tinyset_range() {
assert_eq!(
TinySet::range_lower(3).into_iter().collect::<Vec<u32>>(),
[0, 1, 2]
);
assert!(TinySet::range_lower(0).is_empty());
assert_eq!(
TinySet::range_lower(63).into_iter().collect::<Vec<u32>>(),
(0u32..63u32).collect::<Vec<_>>()
);
assert_eq!(
TinySet::range_lower(1).into_iter().collect::<Vec<u32>>(),
[0]
);
assert_eq!(
TinySet::range_lower(2).into_iter().collect::<Vec<u32>>(),
[0, 1]
);
assert_eq!(
TinySet::range_greater_or_equal(3)
.into_iter()
.collect::<Vec<u32>>(),
(3u32..64u32).collect::<Vec<_>>()
);
}
#[test]
fn test_bitset_len() {
let mut bitset = BitSet::with_max_value(1_000);
assert_eq!(bitset.len(), 0);
bitset.insert(3u32);
assert_eq!(bitset.len(), 1);
bitset.insert(103u32);
assert_eq!(bitset.len(), 2);
bitset.insert(3u32);
assert_eq!(bitset.len(), 2);
bitset.insert(103u32);
assert_eq!(bitset.len(), 2);
bitset.insert(104u32);
assert_eq!(bitset.len(), 3);
}
#[test]
fn test_bitset_clear() {
let mut bitset = BitSet::with_max_value(1_000);
let els = tests::sample(1_000, 0.01f32);
for &el in &els {
bitset.insert(el);
}
assert!(els.iter().all(|el| bitset.contains(*el)));
bitset.clear();
for el in 0u32..1000u32 {
assert!(!bitset.contains(el));
}
}
#[bench]
fn bench_tinyset_pop(b: &mut test::Bencher) {
b.iter(|| test::black_box(TinySet::singleton(31u32)).pop_lowest());
}
#[bench]
fn bench_tinyset_sum(b: &mut test::Bencher) {
let tiny_set = TinySet::empty().insert(10u32).insert(14u32).insert(21u32);
b.iter(|| {
assert_eq!(test::black_box(tiny_set).into_iter().sum::<u32>(), 45u32);
});
}
#[bench]
fn bench_tinyarr_sum(b: &mut test::Bencher) {
let v = [10u32, 14u32, 21u32];
b.iter(|| test::black_box(v).iter().cloned().sum::<u32>());
}
#[bench]
fn bench_bitset_initialize(b: &mut test::Bencher) {
b.iter(|| BitSet::with_max_value(1_000_000));
}
}

26
src/common/composite_file.rs
View File

@@ -8,7 +8,6 @@ use std::io::{self, Read};
use directory::ReadOnlySource;
use common::BinarySerializable;
#[derive(Eq, PartialEq, Hash, Copy, Ord, PartialOrd, Clone, Debug)]
pub struct FileAddr {
field: Field,
@@ -19,7 +18,7 @@ impl FileAddr {
fn new(field: Field, idx: usize) -> FileAddr {
FileAddr {
field: field,
idx: idx
idx: idx,
}
}
}
@@ -36,7 +35,7 @@ impl BinarySerializable for FileAddr {
let idx = VInt::deserialize(reader)?.0 as usize;
Ok(FileAddr {
field: field,
idx: idx
idx: idx,
})
}
}
@@ -59,7 +58,7 @@ impl<W: Write> CompositeWrite<W> {
/// Start writing a new field.
pub fn for_field(&mut self, field: Field) -> &mut CountingWriter<W> {
self.for_field_with_idx(field, 0)
self.for_field_with_idx(field, 0)
}
/// Start writing a new field.
@@ -71,7 +70,6 @@ impl<W: Write> CompositeWrite<W> {
&mut self.write
}
/// Close the composite file.
///
/// An index of the different field offsets
@@ -89,9 +87,7 @@ impl<W: Write> CompositeWrite<W> {
let mut prev_offset = 0;
for (offset, file_addr) in offset_fields {
VInt((offset - prev_offset) as u64).serialize(
&mut self.write,
)?;
VInt((offset - prev_offset) as u64).serialize(&mut self.write)?;
file_addr.serialize(&mut self.write)?;
prev_offset = offset;
}
@@ -103,7 +99,6 @@ impl<W: Write> CompositeWrite<W> {
}
}
/// A composite file is an abstraction to store a
/// file partitioned by field.
///
@@ -174,20 +169,20 @@ impl CompositeFile {
/// to a given `Field` and stored in a `CompositeFile`.
pub fn open_read_with_idx(&self, field: Field, idx: usize) -> Option<ReadOnlySource> {
self.offsets_index
.get(&FileAddr {field: field, idx: idx})
.map(|&(from, to)| {
self.data.slice(from, to)
.get(&FileAddr {
field: field,
idx: idx,
})
.map(|&(from, to)| self.data.slice(from, to))
}
}
#[cfg(test)]
mod test {
use std::io::Write;
use super::{CompositeWrite, CompositeFile};
use directory::{RAMDirectory, Directory};
use super::{CompositeFile, CompositeWrite};
use directory::{Directory, RAMDirectory};
use schema::Field;
use common::VInt;
use common::BinarySerializable;
@@ -231,7 +226,6 @@ mod test {
assert_eq!(payload_4, 2u64);
}
}
}
}

62
src/common/mod.rs
View File

@@ -7,18 +7,57 @@ pub mod bitpacker;
mod bitset;
pub(crate) use self::composite_file::{CompositeFile, CompositeWrite};
pub use self::serialize::BinarySerializable;
pub use self::serialize::{BinarySerializable, FixedSize};
pub use self::timer::Timing;
pub use self::timer::TimerTree;
pub use self::timer::OpenTimer;
pub use self::vint::VInt;
pub use self::counting_writer::CountingWriter;
pub use self::bitset::{TinySet, DocBitSet};
pub use self::bitset::BitSet;
pub(crate) use self::bitset::TinySet;
pub use byteorder::LittleEndian as Endianness;
use std::io;
/// Computes the number of bits that will be used for bitpacking.
///
/// In general the target is the minimum number of bits
/// required to express the amplitude given in argument.
///
/// e.g. If the amplitude is 10, we can store all ints on simply 4bits.
///
/// The logic is slightly more convoluted here as for optimization
/// reasons, we want to ensure that a value spawns over at most 8 bytes
/// of aligns bytes.
///
/// Spanning over 9 bytes is possible for instance, if we do
/// bitpacking with an amplitude of 63 bits.
/// In this case, the second int will start on bit
/// 63 (which belongs to byte 7) and ends at byte 15;
/// Hence 9 bytes (from byte 7 to byte 15 included).
///
/// To avoid this, we force the number of bits to 64bits
/// when the result is greater than `64-8 = 56 bits`.
///
/// Note that this only affects rare use cases spawning over
/// a very large range of values. Even in this case, it results
/// in an extra cost of at most 12% compared to the optimal
/// number of bits.
pub(crate) fn compute_num_bits(n: u64) -> u8 {
let amplitude = (64u32 - n.leading_zeros()) as u8;
if amplitude <= 64 - 8 {
amplitude
} else {
64
}
}
pub(crate) fn is_power_of_2(n: usize) -> bool {
(n > 0) && (n & (n - 1) == 0)
}
/// Create a default io error given a string.
pub fn make_io_err(msg: String) -> io::Error {
pub(crate) fn make_io_err(msg: String) -> io::Error {
io::Error::new(io::ErrorKind::Other, msg)
}
@@ -67,9 +106,10 @@ pub fn u64_to_i64(val: u64) -> i64 {
}
#[cfg(test)]
mod test {
pub(crate) mod test {
use super::{i64_to_u64, u64_to_i64};
use super::{compute_num_bits, i64_to_u64, u64_to_i64};
pub use super::serialize::test::fixed_size_test;
fn test_i64_converter_helper(val: i64) {
assert_eq!(u64_to_i64(i64_to_u64(val)), val);
@@ -86,4 +126,16 @@ mod test {
test_i64_converter_helper(i);
}
}
#[test]
fn test_compute_num_bits() {
assert_eq!(compute_num_bits(1), 1u8);
assert_eq!(compute_num_bits(0), 0u8);
assert_eq!(compute_num_bits(2), 2u8);
assert_eq!(compute_num_bits(3), 2u8);
assert_eq!(compute_num_bits(4), 3u8);
assert_eq!(compute_num_bits(255), 8u8);
assert_eq!(compute_num_bits(256), 9u8);
assert_eq!(compute_num_bits(5_000_000_000), 33u8);
}
}

102
src/common/serialize.rs
View File

@@ -1,16 +1,25 @@
use byteorder::{ReadBytesExt, WriteBytesExt};
use byteorder::LittleEndian as Endianness;
use common::Endianness;
use std::fmt;
use std::io::Write;
use std::io::Read;
use std::io;
use common::VInt;
/// Trait for a simple binary serialization.
pub trait BinarySerializable: fmt::Debug + Sized {
/// Serialize
fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()>;
/// Deserialize
fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self>;
}
/// `FixedSize` marks a `BinarySerializable` as
/// always serializing to the same size.
pub trait FixedSize: BinarySerializable {
const SIZE_IN_BYTES: usize;
}
impl BinarySerializable for () {
fn serialize<W: Write>(&self, _: &mut W) -> io::Result<()> {
Ok(())
@@ -20,6 +29,10 @@ impl BinarySerializable for () {
}
}
impl FixedSize for () {
const SIZE_IN_BYTES: usize = 0;
}
impl<T: BinarySerializable> BinarySerializable for Vec<T> {
fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
VInt(self.len() as u64).serialize(writer)?;
@@ -59,6 +72,10 @@ impl BinarySerializable for u32 {
}
}
impl FixedSize for u32 {
const SIZE_IN_BYTES: usize = 4;
}
impl BinarySerializable for u64 {
fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
writer.write_u64::<Endianness>(*self)
@@ -68,6 +85,10 @@ impl BinarySerializable for u64 {
}
}
impl FixedSize for u64 {
const SIZE_IN_BYTES: usize = 8;
}
impl BinarySerializable for i64 {
fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
writer.write_i64::<Endianness>(*self)
@@ -77,6 +98,10 @@ impl BinarySerializable for i64 {
}
}
impl FixedSize for i64 {
const SIZE_IN_BYTES: usize = 8;
}
impl BinarySerializable for u8 {
fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
writer.write_u8(*self)
@@ -86,6 +111,10 @@ impl BinarySerializable for u8 {
}
}
impl FixedSize for u8 {
const SIZE_IN_BYTES: usize = 1;
}
impl BinarySerializable for String {
fn serialize<W: Write>(&self, writer: &mut W) -> io::Result<()> {
let data: &[u8] = self.as_bytes();
@@ -104,63 +133,78 @@ impl BinarySerializable for String {
}
#[cfg(test)]
mod test {
pub mod test {
use common::VInt;
use super::*;
fn serialize_test<T: BinarySerializable + Eq>(v: T, num_bytes: usize) {
pub fn fixed_size_test<O: BinarySerializable + FixedSize + Default>() {
let mut buffer = Vec::new();
O::default().serialize(&mut buffer).unwrap();
assert_eq!(buffer.len(), O::SIZE_IN_BYTES);
}
fn serialize_test<T: BinarySerializable + Eq>(v: T) -> usize {
let mut buffer: Vec<u8> = Vec::new();
if num_bytes != 0 {
v.serialize(&mut buffer).unwrap();
assert_eq!(buffer.len(), num_bytes);
} else {
v.serialize(&mut buffer).unwrap();
}
v.serialize(&mut buffer).unwrap();
let num_bytes = buffer.len();
let mut cursor = &buffer[..];
let deser = T::deserialize(&mut cursor).unwrap();
assert_eq!(deser, v);
num_bytes
}
#[test]
fn test_serialize_u8() {
serialize_test(3u8, 1);
serialize_test(5u8, 1);
fixed_size_test::<u8>();
}
#[test]
fn test_serialize_u32() {
serialize_test(3u32, 4);
serialize_test(5u32, 4);
serialize_test(u32::max_value(), 4);
fixed_size_test::<u32>();
assert_eq!(4, serialize_test(3u32));
assert_eq!(4, serialize_test(5u32));
assert_eq!(4, serialize_test(u32::max_value()));
}
#[test]
fn test_serialize_i64() {
fixed_size_test::<i64>();
}
#[test]
fn test_serialize_u64() {
fixed_size_test::<u64>();
}
#[test]
fn test_serialize_string() {
serialize_test(String::from(""), 1);
serialize_test(String::from("ぽよぽよ"), 1 + 3 * 4);
serialize_test(String::from("富士さん見える。"), 1 + 3 * 8);
assert_eq!(serialize_test(String::from("")), 1);
assert_eq!(serialize_test(String::from("ぽよぽよ")), 1 + 3 * 4);
assert_eq!(
serialize_test(String::from("富士さん見える。")),
1 + 3 * 8
);
}
#[test]
fn test_serialize_vec() {
let v: Vec<u8> = Vec::new();
serialize_test(v, 1);
serialize_test(vec![1u32, 3u32], 1 + 4 * 2);
assert_eq!(serialize_test(Vec::<u8>::new()), 1);
assert_eq!(serialize_test(vec![1u32, 3u32]), 1 + 4 * 2);
}
#[test]
fn test_serialize_vint() {
for i in 0..10_000 {
serialize_test(VInt(i as u64), 0);
serialize_test(VInt(i as u64));
}
serialize_test(VInt(7u64), 1);
serialize_test(VInt(127u64), 1);
serialize_test(VInt(128u64), 2);
serialize_test(VInt(129u64), 2);
serialize_test(VInt(1234u64), 2);
serialize_test(VInt(16_383), 2);
serialize_test(VInt(16_384), 3);
serialize_test(VInt(u64::max_value()), 10);
assert_eq!(serialize_test(VInt(7u64)), 1);
assert_eq!(serialize_test(VInt(127u64)), 1);
assert_eq!(serialize_test(VInt(128u64)), 2);
assert_eq!(serialize_test(VInt(129u64)), 2);
assert_eq!(serialize_test(VInt(1234u64)), 2);
assert_eq!(serialize_test(VInt(16_383u64)), 2);
assert_eq!(serialize_test(VInt(16_384u64)), 3);
assert_eq!(serialize_test(VInt(u64::max_value())), 10);
}
}

4
src/common/vint.rs
View File

@@ -11,6 +11,10 @@ impl VInt {
pub fn val(&self) -> u64 {
self.0
}
pub fn deserialize_u64<R: Read>(reader: &mut R) -> io::Result<u64> {
VInt::deserialize(reader).map(|vint| vint.0)
}
}
impl BinarySerializable for VInt {

14
src/compression/pack/compression_pack_nosimd.rs
View File

@@ -23,9 +23,11 @@ pub fn compress_sorted(vals: &mut [u32], output: &mut [u8], offset: u32) -> usiz
let num_bits = compute_num_bits(max_delta as u64);
counting_writer.write_all(&[num_bits]).unwrap();
let mut bit_packer = BitPacker::new(num_bits as usize);
let mut bit_packer = BitPacker::new();
for val in vals {
bit_packer.write(*val as u64, &mut counting_writer).unwrap();
bit_packer
.write(*val as u64, num_bits, &mut counting_writer)
.unwrap();
}
counting_writer.written_bytes()
}
@@ -61,13 +63,15 @@ impl BlockEncoder {
let num_bits = compute_num_bits(max as u64);
let mut counting_writer = CountingWriter::wrap(output);
counting_writer.write_all(&[num_bits]).unwrap();
let mut bit_packer = BitPacker::new(num_bits as usize);
let mut bit_packer = BitPacker::new();
for val in vals {
bit_packer.write(*val as u64, &mut counting_writer).unwrap();
bit_packer
.write(*val as u64, num_bits, &mut counting_writer)
.unwrap();
}
for _ in vals.len()..COMPRESSION_BLOCK_SIZE {
bit_packer
.write(vals[0] as u64, &mut counting_writer)
.write(vals[0] as u64, num_bits, &mut counting_writer)
.unwrap();
}
bit_packer.flush(&mut counting_writer).expect(

3
src/core/index_meta.rs
View File

@@ -14,7 +14,8 @@ pub struct IndexMeta {
pub segments: Vec<SegmentMeta>,
pub schema: Schema,
pub opstamp: u64,
#[serde(skip_serializing_if = "Option::is_none")] pub payload: Option<String>,
#[serde(skip_serializing_if = "Option::is_none")]
pub payload: Option<String>,
}
impl IndexMeta {

22
src/core/segment_reader.rs
View File

@@ -108,17 +108,21 @@ impl SegmentReader {
pub fn facet_reader(&self, field: Field) -> Result<FacetReader> {
let field_entry = self.schema.get_field_entry(field);
if field_entry.field_type() != &FieldType::HierarchicalFacet {
return Err(ErrorKind::InvalidArgument(format!("The field {:?} is not a \
hierarchical facet.", field_entry)).into())
return Err(ErrorKind::InvalidArgument(format!(
"The field {:?} is not a \
hierarchical facet.",
field_entry
)).into());
}
let term_ords_reader = self.multi_value_reader(field)?;
let termdict_source = self.termdict_composite
.open_read(field)
.ok_or_else(|| {
ErrorKind::InvalidArgument(format!("The field \"{}\" is a hierarchical \
but this segment does not seem to have the field term \
dictionary.", field_entry.name()))
})?;
let termdict_source = self.termdict_composite.open_read(field).ok_or_else(|| {
ErrorKind::InvalidArgument(format!(
"The field \"{}\" is a hierarchical \
but this segment does not seem to have the field term \
dictionary.",
field_entry.name()
))
})?;
let termdict = TermDictionaryImpl::from_source(termdict_source);
let facet_reader = FacetReader::new(term_ords_reader, termdict);
Ok(facet_reader)

41
src/datastruct/skip/mod.rs
View File

@@ -9,12 +9,12 @@ pub use self::skiplist::SkipList;
#[cfg(test)]
mod tests {
use super::*;
use super::{SkipList, SkipListBuilder};
#[test]
fn test_skiplist() {
let mut output: Vec<u8> = Vec::new();
let mut skip_list_builder: SkipListBuilder<u32> = SkipListBuilder::new(10);
let mut skip_list_builder: SkipListBuilder<u32> = SkipListBuilder::new(8);
skip_list_builder.insert(2, &3).unwrap();
skip_list_builder.write::<Vec<u8>>(&mut output).unwrap();
let mut skip_list: SkipList<u32> = SkipList::from(output.as_slice());
@@ -24,7 +24,7 @@ mod tests {
#[test]
fn test_skiplist2() {
let mut output: Vec<u8> = Vec::new();
let skip_list_builder: SkipListBuilder<u32> = SkipListBuilder::new(10);
let skip_list_builder: SkipListBuilder<u32> = SkipListBuilder::new(8);
skip_list_builder.write::<Vec<u8>>(&mut output).unwrap();
let mut skip_list: SkipList<u32> = SkipList::from(output.as_slice());
assert_eq!(skip_list.next(), None);
@@ -71,7 +71,7 @@ mod tests {
#[test]
fn test_skiplist5() {
let mut output: Vec<u8> = Vec::new();
let mut skip_list_builder: SkipListBuilder<()> = SkipListBuilder::new(3);
let mut skip_list_builder: SkipListBuilder<()> = SkipListBuilder::new(4);
skip_list_builder.insert(2, &()).unwrap();
skip_list_builder.insert(3, &()).unwrap();
skip_list_builder.insert(5, &()).unwrap();
@@ -103,7 +103,7 @@ mod tests {
#[test]
fn test_skiplist7() {
let mut output: Vec<u8> = Vec::new();
let mut skip_list_builder: SkipListBuilder<()> = SkipListBuilder::new(3);
let mut skip_list_builder: SkipListBuilder<()> = SkipListBuilder::new(4);
for i in 0..1000 {
skip_list_builder.insert(i, &()).unwrap();
}
@@ -121,35 +121,48 @@ mod tests {
#[test]
fn test_skiplist8() {
let mut output: Vec<u8> = Vec::new();
let mut skip_list_builder: SkipListBuilder<u32> = SkipListBuilder::new(10);
let mut skip_list_builder: SkipListBuilder<u64> = SkipListBuilder::new(8);
skip_list_builder.insert(2, &3).unwrap();
skip_list_builder.write::<Vec<u8>>(&mut output).unwrap();
assert_eq!(output.len(), 13);
assert_eq!(output.len(), 11);
assert_eq!(output[0], 1u8 + 128u8);
}
#[test]
fn test_skiplist9() {
let mut output: Vec<u8> = Vec::new();
let mut skip_list_builder: SkipListBuilder<u32> = SkipListBuilder::new(3);
for i in 0..9 {
let mut skip_list_builder: SkipListBuilder<u64> = SkipListBuilder::new(4);
for i in 0..4 * 4 * 4 {
skip_list_builder.insert(i, &i).unwrap();
}
skip_list_builder.write::<Vec<u8>>(&mut output).unwrap();
assert_eq!(output.len(), 117);
assert_eq!(output[0], 3u8 + 128u8);
assert_eq!(output.len(), 774);
assert_eq!(output[0], 4u8 + 128u8);
}
#[test]
fn test_skiplist10() {
// checking that void gets serialized to nothing.
let mut output: Vec<u8> = Vec::new();
let mut skip_list_builder: SkipListBuilder<()> = SkipListBuilder::new(3);
for i in 0..9 {
let mut skip_list_builder: SkipListBuilder<()> = SkipListBuilder::new(4);
for i in 0..((4 * 4 * 4) - 1) {
skip_list_builder.insert(i, &()).unwrap();
}
skip_list_builder.write::<Vec<u8>>(&mut output).unwrap();
assert_eq!(output.len(), 81);
assert_eq!(output.len(), 230);
assert_eq!(output[0], 128u8 + 3u8);
}
#[test]
fn test_skiplist11() {
// checking that void gets serialized to nothing.
let mut output: Vec<u8> = Vec::new();
let mut skip_list_builder: SkipListBuilder<()> = SkipListBuilder::new(4);
for i in 0..(4 * 4) {
skip_list_builder.insert(i, &()).unwrap();
}
skip_list_builder.write::<Vec<u8>>(&mut output).unwrap();
assert_eq!(output.len(), 65);
assert_eq!(output[0], 128u8 + 3u8);
}

67
src/datastruct/skip/skiplist.rs
View File

@@ -1,6 +1,5 @@
use common::BinarySerializable;
use common::{BinarySerializable, VInt};
use std::marker::PhantomData;
use DocId;
use std::cmp::max;
static EMPTY: [u8; 0] = [];
@@ -8,21 +7,20 @@ static EMPTY: [u8; 0] = [];
struct Layer<'a, T> {
data: &'a [u8],
cursor: &'a [u8],
next_id: DocId,
next_id: Option<u64>,
_phantom_: PhantomData<T>,
}
impl<'a, T: BinarySerializable> Iterator for Layer<'a, T> {
type Item = (DocId, T);
type Item = (u64, T);
fn next(&mut self) -> Option<(DocId, T)> {
if self.next_id == u32::max_value() {
None
} else {
fn next(&mut self) -> Option<(u64, T)> {
if let Some(cur_id) = self.next_id {
let cur_val = T::deserialize(&mut self.cursor).unwrap();
let cur_id = self.next_id;
self.next_id = u32::deserialize(&mut self.cursor).unwrap_or(u32::max_value());
self.next_id = VInt::deserialize_u64(&mut self.cursor).ok();
Some((cur_id, cur_val))
} else {
None
}
}
}
@@ -30,7 +28,7 @@ impl<'a, T: BinarySerializable> Iterator for Layer<'a, T> {
impl<'a, T: BinarySerializable> From<&'a [u8]> for Layer<'a, T> {
fn from(data: &'a [u8]) -> Layer<'a, T> {
let mut cursor = data;
let next_id = u32::deserialize(&mut cursor).unwrap_or(u32::max_value());
let next_id = VInt::deserialize_u64(&mut cursor).ok();
Layer {
data,
cursor,
@@ -45,14 +43,14 @@ impl<'a, T: BinarySerializable> Layer<'a, T> {
Layer {
data: &EMPTY,
cursor: &EMPTY,
next_id: DocId::max_value(),
next_id: None,
_phantom_: PhantomData,
}
}
fn seek_offset(&mut self, offset: usize) {
self.cursor = &self.data[offset..];
self.next_id = u32::deserialize(&mut self.cursor).unwrap_or(u32::max_value());
self.next_id = VInt::deserialize_u64(&mut self.cursor).ok();
}
// Returns the last element (key, val)
@@ -60,54 +58,61 @@ impl<'a, T: BinarySerializable> Layer<'a, T> {
//
// If there is no such element anymore,
// returns None.
fn seek(&mut self, doc_id: DocId) -> Option<(DocId, T)> {
let mut val = None;
while self.next_id < doc_id {
match self.next() {
None => {
break;
}
v => {
val = v;
//
// If the element exists, it will be returned
// at the next call to `.next()`.
fn seek(&mut self, key: u64) -> Option<(u64, T)> {
let mut result: Option<(u64, T)> = None;
loop {
if let Some(next_id) = self.next_id {
if next_id < key {
if let Some(v) = self.next() {
result = Some(v);
continue;
}
}
}
return result;
}
val
}
}
pub struct SkipList<'a, T: BinarySerializable> {
data_layer: Layer<'a, T>,
skip_layers: Vec<Layer<'a, u32>>,
skip_layers: Vec<Layer<'a, u64>>,
}
impl<'a, T: BinarySerializable> Iterator for SkipList<'a, T> {
type Item = (DocId, T);
type Item = (u64, T);
fn next(&mut self) -> Option<(DocId, T)> {
fn next(&mut self) -> Option<(u64, T)> {
self.data_layer.next()
}
}
impl<'a, T: BinarySerializable> SkipList<'a, T> {
pub fn seek(&mut self, doc_id: DocId) -> Option<(DocId, T)> {
let mut next_layer_skip: Option<(DocId, u32)> = None;
pub fn seek(&mut self, key: u64) -> Option<(u64, T)> {
let mut next_layer_skip: Option<(u64, u64)> = None;
for skip_layer in &mut self.skip_layers {
if let Some((_, offset)) = next_layer_skip {
skip_layer.seek_offset(offset as usize);
}
next_layer_skip = skip_layer.seek(doc_id);
next_layer_skip = skip_layer.seek(key);
}
if let Some((_, offset)) = next_layer_skip {
self.data_layer.seek_offset(offset as usize);
}
self.data_layer.seek(doc_id)
self.data_layer.seek(key)
}
}
impl<'a, T: BinarySerializable> From<&'a [u8]> for SkipList<'a, T> {
fn from(mut data: &'a [u8]) -> SkipList<'a, T> {
let offsets: Vec<u32> = Vec::deserialize(&mut data).unwrap();
let offsets: Vec<u64> = Vec::<VInt>::deserialize(&mut data)
.unwrap()
.into_iter()
.map(|el| el.0)
.collect();
let num_layers = offsets.len();
let layers_data: &[u8] = data;
let data_layer: Layer<'a, T> = if num_layers == 0 {

45
src/datastruct/skip/skiplist_builder.rs
View File

@@ -1,13 +1,11 @@
use std::io::Write;
use common::BinarySerializable;
use common::{BinarySerializable, VInt, is_power_of_2};
use std::marker::PhantomData;
use DocId;
use std::io;
struct LayerBuilder<T: BinarySerializable> {
period: usize,
period_mask: usize,
buffer: Vec<u8>,
remaining: usize,
len: usize,
_phantom_: PhantomData<T>,
}
@@ -23,34 +21,33 @@ impl<T: BinarySerializable> LayerBuilder<T> {
}
fn with_period(period: usize) -> LayerBuilder<T> {
assert!(is_power_of_2(period), "The period has to be a power of 2.");
LayerBuilder {
period,
period_mask: (period - 1),
buffer: Vec::new(),
remaining: period,
len: 0,
_phantom_: PhantomData,
}
}
fn insert(&mut self, doc_id: DocId, value: &T) -> io::Result<Option<(DocId, u32)>> {
self.remaining -= 1;
fn insert(&mut self, key: u64, value: &T) -> io::Result<Option<(u64, u64)>> {
self.len += 1;
let offset = self.written_size() as u32;
doc_id.serialize(&mut self.buffer)?;
let offset = self.written_size() as u64;
VInt(key).serialize(&mut self.buffer)?;
value.serialize(&mut self.buffer)?;
Ok(if self.remaining == 0 {
self.remaining = self.period;
Some((doc_id, offset))
let emit_skip_info = (self.period_mask & self.len) == 0;
if emit_skip_info {
Ok(Some((key, offset)))
} else {
None
})
Ok(None)
}
}
}
pub struct SkipListBuilder<T: BinarySerializable> {
period: usize,
data_layer: LayerBuilder<T>,
skip_layers: Vec<LayerBuilder<u32>>,
skip_layers: Vec<LayerBuilder<u64>>,
}
impl<T: BinarySerializable> SkipListBuilder<T> {
@@ -62,7 +59,7 @@ impl<T: BinarySerializable> SkipListBuilder<T> {
}
}
fn get_skip_layer(&mut self, layer_id: usize) -> &mut LayerBuilder<u32> {
fn get_skip_layer(&mut self, layer_id: usize) -> &mut LayerBuilder<u64> {
if layer_id == self.skip_layers.len() {
let layer_builder = LayerBuilder::with_period(self.period);
self.skip_layers.push(layer_builder);
@@ -70,9 +67,9 @@ impl<T: BinarySerializable> SkipListBuilder<T> {
&mut self.skip_layers[layer_id]
}
pub fn insert(&mut self, doc_id: DocId, dest: &T) -> io::Result<()> {
pub fn insert(&mut self, key: u64, dest: &T) -> io::Result<()> {
let mut layer_id = 0;
let mut skip_pointer = self.data_layer.insert(doc_id, dest)?;
let mut skip_pointer = self.data_layer.insert(key, dest)?;
loop {
skip_pointer = match skip_pointer {
Some((skip_doc_id, skip_offset)) => self.get_skip_layer(layer_id)
@@ -86,13 +83,11 @@ impl<T: BinarySerializable> SkipListBuilder<T> {
}
pub fn write<W: Write>(self, output: &mut W) -> io::Result<()> {
let mut size: u32 = 0;
let mut layer_sizes: Vec<u32> = Vec::new();
size += self.data_layer.buffer.len() as u32;
layer_sizes.push(size);
let mut size: u64 = self.data_layer.buffer.len() as u64;
let mut layer_sizes = vec![VInt(size)];
for layer in self.skip_layers.iter().rev() {
size += layer.buffer.len() as u32;
layer_sizes.push(size);
size += layer.buffer.len() as u64;
layer_sizes.push(VInt(size));
}
layer_sizes.serialize(output)?;
self.data_layer.write(output)?;

54
src/datastruct/stacker/hashmap.rs
View File

@@ -1,7 +1,7 @@
use std::iter;
use std::mem;
use postings::UnorderedTermId;
use super::heap::{Heap, HeapAllocable, BytesRef};
use super::heap::{BytesRef, Heap, HeapAllocable};
mod murmurhash2 {
@@ -10,7 +10,7 @@ mod murmurhash2 {
#[inline(always)]
pub fn murmurhash2(key: &[u8]) -> u32 {
let mut key_ptr: *const u32 = key.as_ptr() as *const u32;
let m: u32 = 0x5bd1e995;
let m: u32 = 0x5bd1_e995;
let r = 24;
let len = key.len() as u32;
@@ -31,18 +31,18 @@ mod murmurhash2 {
let key_ptr_u8: *const u8 = key_ptr as *const u8;
match remaining {
3 => {
h ^= unsafe { *key_ptr_u8.wrapping_offset(2) as u32 } << 16;
h ^= unsafe { *key_ptr_u8.wrapping_offset(1) as u32 } << 8;
h ^= unsafe { *key_ptr_u8 as u32 };
h ^= unsafe { u32::from(*key_ptr_u8.wrapping_offset(2)) } << 16;
h ^= unsafe { u32::from(*key_ptr_u8.wrapping_offset(1)) } << 8;
h ^= unsafe { u32::from(*key_ptr_u8) };
h = h.wrapping_mul(m);
}
2 => {
h ^= unsafe { *key_ptr_u8.wrapping_offset(1) as u32 } << 8;
h ^= unsafe { *key_ptr_u8 as u32 };
h ^= unsafe { u32::from(*key_ptr_u8.wrapping_offset(1)) } << 8;
h ^= unsafe { u32::from(*key_ptr_u8) };
h = h.wrapping_mul(m);
}
1 => {
h ^= unsafe { *key_ptr_u8 as u32 };
h ^= unsafe { u32::from(*key_ptr_u8) };
h = h.wrapping_mul(m);
}
_ => {}
@@ -53,9 +53,6 @@ mod murmurhash2 {
}
}
/// Split the thread memory budget into
/// - the heap size
/// - the hash table "table" itself.
@@ -63,14 +60,10 @@ mod murmurhash2 {
/// Returns (the heap size in bytes, the hash table size in number of bits)
pub(crate) fn split_memory(per_thread_memory_budget: usize) -> (usize, usize) {
let table_size_limit: usize = per_thread_memory_budget / 3;
let compute_table_size = |num_bits: usize| {
(1 << num_bits) * mem::size_of::<KeyValue>()
};
let compute_table_size = |num_bits: usize| (1 << num_bits) * mem::size_of::<KeyValue>();
let table_num_bits: usize = (1..)
.into_iter()
.take_while(|num_bits: &usize| {
compute_table_size(*num_bits) < table_size_limit
})
.take_while(|num_bits: &usize| compute_table_size(*num_bits) < table_size_limit)
.last()
.expect(&format!(
"Per thread memory is too small: {}",
@@ -81,7 +74,6 @@ pub(crate) fn split_memory(per_thread_memory_budget: usize) -> (usize, usize) {
(heap_size, table_num_bits)
}
/// `KeyValue` is the item stored in the hash table.
/// The key is actually a `BytesRef` object stored in an external heap.
/// The `value_addr` also points to an address in the heap.
@@ -101,7 +93,6 @@ impl KeyValue {
}
}
/// Customized `HashMap` with string keys
///
/// This `HashMap` takes String as keys. Keys are
@@ -118,7 +109,6 @@ pub struct TermHashMap<'a> {
occupied: Vec<usize>,
}
struct QuadraticProbing {
hash: usize,
i: usize,
@@ -141,7 +131,6 @@ impl QuadraticProbing {
}
}
impl<'a> TermHashMap<'a> {
pub fn new(num_bucket_power_of_2: usize, heap: &'a Heap) -> TermHashMap<'a> {
let table_size = 1 << num_bucket_power_of_2;
@@ -178,18 +167,17 @@ impl<'a> TermHashMap<'a> {
}
pub fn iter<'b: 'a>(&'b self) -> impl Iterator<Item = (&'a [u8], u32, UnorderedTermId)> + 'b {
self.occupied
.iter()
.cloned()
.map(move |bucket: usize| {
let kv = self.table[bucket];
let (key, offset) = self.get_key_value(kv.key_value_addr);
(key, offset, bucket as UnorderedTermId)
})
self.occupied.iter().cloned().map(move |bucket: usize| {
let kv = self.table[bucket];
let (key, offset) = self.get_key_value(kv.key_value_addr);
(key, offset, bucket as UnorderedTermId)
})
}
pub fn get_or_create<S: AsRef<[u8]>, V: HeapAllocable>(&mut self, key: S) -> (UnorderedTermId, &mut V) {
pub fn get_or_create<S: AsRef<[u8]>, V: HeapAllocable>(
&mut self,
key: S,
) -> (UnorderedTermId, &mut V) {
let key_bytes: &[u8] = key.as_ref();
let hash = murmurhash2::murmurhash2(key.as_ref());
let mut probe = self.probe(hash);
@@ -212,7 +200,6 @@ impl<'a> TermHashMap<'a> {
}
}
#[cfg(test)]
mod tests {
@@ -223,7 +210,6 @@ mod tests {
use std::collections::HashSet;
use super::split_memory;
struct TestValue {
val: u32,
_addr: u32,
@@ -245,7 +231,6 @@ mod tests {
assert_eq!(split_memory(10_000_000), (7902848, 18));
}
#[test]
fn test_hash_map() {
let heap = Heap::with_capacity(2_000_000);
@@ -319,5 +304,4 @@ mod tests {
});
}
}

1
src/datastruct/stacker/mod.rs
View File

@@ -39,6 +39,5 @@ fn test_unrolled_linked_list() {
assert!(!it.next().is_some());
}
}
}
}

85
src/directory/mmap_directory.rs
View File

@@ -20,19 +20,17 @@ use std::sync::Arc;
use std::sync::RwLock;
use tempdir::TempDir;
/// Returns None iff the file exists, can be read, but is empty (and hence
/// cannot be mmapped).
///
fn open_mmap(full_path: &PathBuf) -> result::Result<Option<MmapReadOnly>, OpenReadError> {
let file = File::open(&full_path)
.map_err(|e| {
if e.kind() == io::ErrorKind::NotFound {
OpenReadError::FileDoesNotExist(full_path.clone())
} else {
OpenReadError::IOError(IOError::with_path(full_path.to_owned(), e))
}
})?;
fn open_mmap(full_path: &Path) -> result::Result<Option<MmapReadOnly>, OpenReadError> {
let file = File::open(full_path).map_err(|e| {
if e.kind() == io::ErrorKind::NotFound {
OpenReadError::FileDoesNotExist(full_path.to_owned())
} else {
OpenReadError::IOError(IOError::with_path(full_path.to_owned(), e))
}
})?;
let meta_data = file.metadata()
.map_err(|e| IOError::with_path(full_path.to_owned(), e))?;
@@ -44,9 +42,7 @@ fn open_mmap(full_path: &PathBuf) -> result::Result<Option<MmapReadOnly>, OpenRe
}
MmapReadOnly::open(&file)
.map(Some)
.map_err(|e| {
From::from(IOError::with_path(full_path.to_owned(), e))
})
.map_err(|e| From::from(IOError::with_path(full_path.to_owned(), e)))
}
#[derive(Default, Clone, Debug, Serialize, Deserialize)]
@@ -79,7 +75,6 @@ impl Default for MmapCache {
}
impl MmapCache {
/// Removes a `MmapReadOnly` entry from the mmap cache.
fn discard_from_cache(&mut self, full_path: &Path) -> bool {
self.cache.remove(full_path).is_some()
@@ -93,23 +88,23 @@ impl MmapCache {
}
}
fn get_mmap(&mut self, full_path: PathBuf) -> Result<Option<MmapReadOnly>, OpenReadError> {
Ok(match self.cache.entry(full_path.clone()) {
HashMapEntry::Occupied(occupied_entry) => {
let mmap = occupied_entry.get();
self.counters.hit += 1;
Some(mmap.clone())
}
HashMapEntry::Vacant(vacant_entry) => {
self.counters.miss += 1;
if let Some(mmap) = open_mmap(&full_path)? {
vacant_entry.insert(mmap.clone());
Some(mmap)
} else {
None
}
}
})
fn get_mmap(&mut self, full_path: &Path) -> Result<Option<MmapReadOnly>, OpenReadError> {
Ok(match self.cache.entry(full_path.to_owned()) {
HashMapEntry::Occupied(occupied_entry) => {
let mmap = occupied_entry.get();
self.counters.hit += 1;
Some(mmap.clone())
}
HashMapEntry::Vacant(vacant_entry) => {
self.counters.miss += 1;
if let Some(mmap) = open_mmap(full_path)? {
vacant_entry.insert(mmap.clone());
Some(mmap)
} else {
None
}
}
})
}
}
@@ -257,9 +252,9 @@ impl Directory for MmapDirectory {
})?;
Ok(mmap_cache
.get_mmap(full_path)?
.map(ReadOnlySource::Mmap)
.unwrap_or_else(|| ReadOnlySource::Anonymous(SharedVecSlice::empty())))
.get_mmap(&full_path)?
.map(ReadOnlySource::Mmap)
.unwrap_or_else(|| ReadOnlySource::Anonymous(SharedVecSlice::empty())))
}
fn open_write(&mut self, path: &Path) -> Result<WritePtr, OpenWriteError> {
@@ -292,20 +287,19 @@ impl Directory for MmapDirectory {
Ok(BufWriter::new(Box::new(writer)))
}
/// Any entry associated to the path in the mmap will be
/// removed before the file is deleted.
fn delete(&self, path: &Path) -> result::Result<(), DeleteError> {
debug!("Deleting file {:?}", path);
let full_path = self.resolve_path(path);
let mut mmap_cache = self.mmap_cache
.write()
.map_err(|_| {
let msg = format!("Failed to acquired write lock \
on mmap cache while deleting {:?}",
path);
IOError::with_path(path.to_owned(), make_io_err(msg))
})?;
let mut mmap_cache = self.mmap_cache.write().map_err(|_| {
let msg = format!(
"Failed to acquired write lock \
on mmap cache while deleting {:?}",
path
);
IOError::with_path(path.to_owned(), make_io_err(msg))
})?;
mmap_cache.discard_from_cache(path);
// Removing the entry in the MMap cache.
@@ -415,7 +409,10 @@ mod tests {
}
for (i, path) in paths.iter().enumerate() {
mmap_directory.delete(path).unwrap();
assert_eq!(mmap_directory.get_cache_info().mmapped.len(), num_paths - i - 1);
assert_eq!(
mmap_directory.get_cache_info().mmapped.len(),
num_paths - i - 1
);
}
}
assert_eq!(mmap_directory.get_cache_info().counters.hit, 10);

2
src/directory/read_only_source.rs
View File

@@ -4,7 +4,7 @@ use super::shared_vec_slice::SharedVecSlice;
use common::HasLen;
use std::slice;
use std::io::{self, Read};
use stable_deref_trait::{StableDeref, CloneStableDeref};
use stable_deref_trait::{CloneStableDeref, StableDeref};
/// Read object that represents files in tantivy.
///

13
src/fastfield/facet_reader.rs
View File

@@ -4,7 +4,6 @@ use termdict::TermOrdinal;
use schema::Facet;
use termdict::{TermDictionary, TermDictionaryImpl};
/// The facet reader makes it possible to access the list of
/// facets associated to a given document in a specific
/// segment.
@@ -24,7 +23,6 @@ pub struct FacetReader {
}
impl FacetReader {
/// Creates a new `FacetReader`.
///
/// A facet reader just wraps :
@@ -32,8 +30,10 @@ impl FacetReader {
/// access the list of facet ords for a given document.
/// - a `TermDictionaryImpl` that helps associating a facet to
/// an ordinal and vice versa.
pub fn new(term_ords: MultiValueIntFastFieldReader,
term_dict: TermDictionaryImpl) -> FacetReader {
pub fn new(
term_ords: MultiValueIntFastFieldReader,
term_dict: TermDictionaryImpl,
) -> FacetReader {
FacetReader {
term_ords: term_ords,
term_dict: term_dict,
@@ -56,7 +56,8 @@ impl FacetReader {
/// Given a term ordinal returns the term associated to it.
pub fn facet_from_ord(&self, facet_ord: TermOrdinal, output: &mut Facet) {
let found_term = self.term_dict.ord_to_term(facet_ord as u64, output.inner_buffer_mut());
let found_term = self.term_dict
.ord_to_term(facet_ord as u64, output.inner_buffer_mut());
assert!(found_term, "Term ordinal {} no found.", facet_ord);
}
@@ -64,4 +65,4 @@ impl FacetReader {
pub fn facet_ords(&mut self, doc: DocId, output: &mut Vec<u64>) {
self.term_ords.get_vals(doc, output);
}
}
}

36
src/fastfield/mod.rs
View File

@@ -95,7 +95,9 @@ mod tests {
add_single_field_doc(&mut fast_field_writers, *FIELD, 13u64);
add_single_field_doc(&mut fast_field_writers, *FIELD, 14u64);
add_single_field_doc(&mut fast_field_writers, *FIELD, 2u64);
fast_field_writers.serialize(&mut serializer, HashMap::new()).unwrap();
fast_field_writers
.serialize(&mut serializer, &HashMap::new())
.unwrap();
serializer.close().unwrap();
}
let source = directory.open_read(&path).unwrap();
@@ -129,7 +131,9 @@ mod tests {
add_single_field_doc(&mut fast_field_writers, *FIELD, 1_002u64);
add_single_field_doc(&mut fast_field_writers, *FIELD, 1_501u64);
add_single_field_doc(&mut fast_field_writers, *FIELD, 215u64);
fast_field_writers.serialize(&mut serializer, HashMap::new()).unwrap();
fast_field_writers
.serialize(&mut serializer, &HashMap::new())
.unwrap();
serializer.close().unwrap();
}
let source = directory.open_read(&path).unwrap();
@@ -164,7 +168,9 @@ mod tests {
for _ in 0..10_000 {
add_single_field_doc(&mut fast_field_writers, *FIELD, 100_000u64);
}
fast_field_writers.serialize(&mut serializer, HashMap::new()).unwrap();
fast_field_writers
.serialize(&mut serializer, &HashMap::new())
.unwrap();
serializer.close().unwrap();
}
let source = directory.open_read(&path).unwrap();
@@ -199,7 +205,9 @@ mod tests {
5_000_000_000_000_000_000u64 + i,
);
}
fast_field_writers.serialize(&mut serializer, HashMap::new()).unwrap();
fast_field_writers
.serialize(&mut serializer, &HashMap::new())
.unwrap();
serializer.close().unwrap();
}
let source = directory.open_read(&path).unwrap();
@@ -238,7 +246,9 @@ mod tests {
doc.add_i64(i64_field, i);
fast_field_writers.add_document(&doc);
}
fast_field_writers.serialize(&mut serializer, HashMap::new()).unwrap();
fast_field_writers
.serialize(&mut serializer, &HashMap::new())
.unwrap();
serializer.close().unwrap();
}
let source = directory.open_read(&path).unwrap();
@@ -277,7 +287,9 @@ mod tests {
let mut fast_field_writers = FastFieldsWriter::from_schema(&schema);
let doc = Document::default();
fast_field_writers.add_document(&doc);
fast_field_writers.serialize(&mut serializer, HashMap::new()).unwrap();
fast_field_writers
.serialize(&mut serializer, &HashMap::new())
.unwrap();
serializer.close().unwrap();
}
@@ -311,7 +323,9 @@ mod tests {
for x in &permutation {
add_single_field_doc(&mut fast_field_writers, *FIELD, *x);
}
fast_field_writers.serialize(&mut serializer, HashMap::new()).unwrap();
fast_field_writers
.serialize(&mut serializer, &HashMap::new())
.unwrap();
serializer.close().unwrap();
}
let source = directory.open_read(&path).unwrap();
@@ -366,7 +380,9 @@ mod tests {
for x in &permutation {
add_single_field_doc(&mut fast_field_writers, *FIELD, *x);
}
fast_field_writers.serialize(&mut serializer, HashMap::new()).unwrap();
fast_field_writers
.serialize(&mut serializer, &HashMap::new())
.unwrap();
serializer.close().unwrap();
}
let source = directory.open_read(&path).unwrap();
@@ -398,7 +414,9 @@ mod tests {
for x in &permutation {
add_single_field_doc(&mut fast_field_writers, *FIELD, *x);
}
fast_field_writers.serialize(&mut serializer, HashMap::new()).unwrap();
fast_field_writers
.serialize(&mut serializer, &HashMap::new())
.unwrap();
serializer.close().unwrap();
}
let source = directory.open_read(&path).unwrap();

2
src/fastfield/multivalued/mod.rs
View File

@@ -2,4 +2,4 @@ mod writer;
mod reader;
pub use self::writer::MultiValueIntFastFieldWriter;
pub use self::reader::MultiValueIntFastFieldReader;
pub use self::reader::MultiValueIntFastFieldReader;

21
src/fastfield/multivalued/reader.rs
View File

@@ -18,8 +18,10 @@ pub struct MultiValueIntFastFieldReader {
}
impl MultiValueIntFastFieldReader {
pub(crate) fn open(idx_reader: U64FastFieldReader, vals_reader: U64FastFieldReader) -> MultiValueIntFastFieldReader {
pub(crate) fn open(
idx_reader: U64FastFieldReader,
vals_reader: U64FastFieldReader,
) -> MultiValueIntFastFieldReader {
MultiValueIntFastFieldReader {
idx_reader: idx_reader,
vals_reader: vals_reader,
@@ -38,12 +40,11 @@ impl MultiValueIntFastFieldReader {
}
}
#[cfg(test)]
mod tests {
use core::Index;
use schema::{Facet, Document, SchemaBuilder};
use schema::{Document, Facet, SchemaBuilder};
#[test]
fn test_multifastfield_reader() {
@@ -51,7 +52,9 @@ mod tests {
let facet_field = schema_builder.add_facet_field("facets");
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
let mut index_writer = index.writer_with_num_threads(1, 30_000_000).expect("Failed to create index writer.");
let mut index_writer = index
.writer_with_num_threads(1, 30_000_000)
.expect("Failed to create index writer.");
{
let mut doc = Document::new();
doc.add_facet(facet_field, "/category/cat2");
@@ -72,9 +75,7 @@ mod tests {
index.load_searchers().expect("Reloading searchers");
let searcher = index.searcher();
let segment_reader = searcher.segment_reader(0);
let mut facet_reader = segment_reader
.facet_reader(facet_field)
.unwrap();
let mut facet_reader = segment_reader.facet_reader(facet_field).unwrap();
let mut facet = Facet::root();
{
@@ -108,7 +109,5 @@ mod tests {
facet_reader.facet_ords(2, &mut vals);
assert_eq!(&vals[..], &[4]);
}
}
}
}

13
src/fastfield/multivalued/writer.rs
View File

@@ -38,10 +38,15 @@ impl MultiValueIntFastFieldWriter {
}
/// Push the fast fields value to the `FastFieldWriter`.
pub fn serialize(&self, serializer: &mut FastFieldSerializer, mapping: &HashMap<UnorderedTermId, usize>) -> io::Result<()> {
pub fn serialize(
&self,
serializer: &mut FastFieldSerializer,
mapping: &HashMap<UnorderedTermId, usize>,
) -> io::Result<()> {
{
// writing the offset index
let mut doc_index_serializer = serializer.new_u64_fast_field_with_idx(self.field, 0, self.vals.len() as u64, 0)?;
let mut doc_index_serializer =
serializer.new_u64_fast_field_with_idx(self.field, 0, self.vals.len() as u64, 0)?;
for &offset in &self.doc_index {
doc_index_serializer.add_val(offset)?;
}
@@ -50,13 +55,13 @@ impl MultiValueIntFastFieldWriter {
}
{
// writing the values themselves.
let mut value_serializer = serializer.new_u64_fast_field_with_idx(self.field, 0u64, mapping.len() as u64, 1)?;
let mut value_serializer =
serializer.new_u64_fast_field_with_idx(self.field, 0u64, mapping.len() as u64, 1)?;
for val in &self.vals {
value_serializer.add_val(*mapping.get(val).expect("Missing term ordinal") as u64)?;
}
value_serializer.close_field()?;
}
Ok(())
}
}

48
src/fastfield/reader.rs
View File

@@ -1,11 +1,12 @@
use directory::ReadOnlySource;
use common::{self, BinarySerializable};
use common::bitpacker::{compute_num_bits, BitUnpacker};
use common::compute_num_bits;
use common::bitpacker::BitUnpacker;
use DocId;
use schema::SchemaBuilder;
use std::path::Path;
use schema::FAST;
use directory::{WritePtr, RAMDirectory, Directory};
use directory::{Directory, RAMDirectory, WritePtr};
use fastfield::{FastFieldSerializer, FastFieldsWriter};
use schema::FieldType;
use std::mem;
@@ -88,7 +89,7 @@ impl FastFieldReader for U64FastFieldReader {
fn is_enabled(field_type: &FieldType) -> bool {
match *field_type {
FieldType::U64(ref integer_options) => integer_options.is_fast(),
FieldType::HierarchicalFacet => { true },
FieldType::HierarchicalFacet => true,
_ => false,
}
}
@@ -113,21 +114,19 @@ impl FastFieldReader for U64FastFieldReader {
u64::deserialize(&mut cursor).expect("Failed to read the min_value of fast field.");
amplitude =
u64::deserialize(&mut cursor).expect("Failed to read the amplitude of fast field.");
}
let max_value = min_value + amplitude;
let num_bits = compute_num_bits(amplitude);
let owning_ref = OwningRef::new(data).map(|data| &data[16..]);
let bit_unpacker = BitUnpacker::new(owning_ref, num_bits as usize);
let bit_unpacker = BitUnpacker::new(owning_ref, num_bits);
U64FastFieldReader {
min_value: min_value,
max_value: max_value,
bit_unpacker: bit_unpacker,
min_value,
max_value,
bit_unpacker,
}
}
}
impl From<Vec<u64>> for U64FastFieldReader {
fn from(vals: Vec<u64>) -> U64FastFieldReader {
let mut schema_builder = SchemaBuilder::default();
@@ -136,22 +135,23 @@ impl From<Vec<u64>> for U64FastFieldReader {
let path = Path::new("__dummy__");
let mut directory: RAMDirectory = RAMDirectory::create();
{
let write: WritePtr = directory.open_write(path).expect(
"With a RAMDirectory, this should never fail.",
);
let mut serializer = FastFieldSerializer::from_write(write).expect(
"With a RAMDirectory, this should never fail.",
);
let write: WritePtr = directory
.open_write(path)
.expect("With a RAMDirectory, this should never fail.");
let mut serializer = FastFieldSerializer::from_write(write)
.expect("With a RAMDirectory, this should never fail.");
let mut fast_field_writers = FastFieldsWriter::from_schema(&schema);
{
let fast_field_writer = fast_field_writers.get_field_writer(field).expect(
"With a RAMDirectory, this should never fail.",
);
let fast_field_writer = fast_field_writers
.get_field_writer(field)
.expect("With a RAMDirectory, this should never fail.");
for val in vals {
fast_field_writer.add_val(val);
}
}
fast_field_writers.serialize(&mut serializer, HashMap::new()).unwrap();
fast_field_writers
.serialize(&mut serializer, &HashMap::new())
.unwrap();
serializer.close().unwrap();
}
@@ -159,9 +159,9 @@ impl From<Vec<u64>> for U64FastFieldReader {
let composite_file =
CompositeFile::open(&source).expect("Failed to read the composite file");
let field_source = composite_file.open_read(field).expect(
"File component not found",
);
let field_source = composite_file
.open_read(field)
.expect("File component not found");
U64FastFieldReader::open(field_source)
}
}
@@ -222,7 +222,9 @@ impl FastFieldReader for I64FastFieldReader {
/// # Panics
/// Panics if the data is corrupted.
fn open(data: ReadOnlySource) -> I64FastFieldReader {
I64FastFieldReader { underlying: U64FastFieldReader::open(data) }
I64FastFieldReader {
underlying: U64FastFieldReader::open(data),
}
}
fn is_enabled(field_type: &FieldType) -> bool {

24
src/fastfield/serializer.rs
View File

@@ -1,7 +1,8 @@
use common::BinarySerializable;
use directory::WritePtr;
use schema::Field;
use common::bitpacker::{compute_num_bits, BitPacker};
use common::bitpacker::BitPacker;
use common::compute_num_bits;
use common::CountingWriter;
use common::CompositeWrite;
use std::io::{self, Write};
@@ -35,7 +36,9 @@ impl FastFieldSerializer {
pub fn from_write(write: WritePtr) -> io::Result<FastFieldSerializer> {
// just making room for the pointer to header.
let composite_write = CompositeWrite::wrap(write);
Ok(FastFieldSerializer { composite_write: composite_write })
Ok(FastFieldSerializer {
composite_write: composite_write,
})
}
/// Start serializing a new u64 fast field
@@ -54,12 +57,12 @@ impl FastFieldSerializer {
field: Field,
min_value: u64,
max_value: u64,
idx: usize) -> io::Result<FastSingleFieldSerializer<CountingWriter<WritePtr>>> {
idx: usize,
) -> io::Result<FastSingleFieldSerializer<CountingWriter<WritePtr>>> {
let field_write = self.composite_write.for_field_with_idx(field, idx);
FastSingleFieldSerializer::open(field_write, min_value, max_value)
}
/// Closes the serializer
///
/// After this call the data must be persistently save on disk.
@@ -72,6 +75,7 @@ pub struct FastSingleFieldSerializer<'a, W: Write + 'a> {
bit_packer: BitPacker,
write: &'a mut W,
min_value: u64,
num_bits: u8,
}
impl<'a, W: Write> FastSingleFieldSerializer<'a, W> {
@@ -84,18 +88,20 @@ impl<'a, W: Write> FastSingleFieldSerializer<'a, W> {
let amplitude = max_value - min_value;
amplitude.serialize(write)?;
let num_bits = compute_num_bits(amplitude);
let bit_packer = BitPacker::new(num_bits as usize);
let bit_packer = BitPacker::new();
Ok(FastSingleFieldSerializer {
write: write,
bit_packer: bit_packer,
min_value: min_value,
write,
bit_packer,
min_value,
num_bits,
})
}
/// Pushes a new value to the currently open u64 fast field.
pub fn add_val(&mut self, val: u64) -> io::Result<()> {
let val_to_write: u64 = val - self.min_value;
self.bit_packer.write(val_to_write, &mut self.write)?;
self.bit_packer
.write(val_to_write, self.num_bits, &mut self.write)?;
Ok(())
}

57
src/fastfield/writer.rs
View File

@@ -1,4 +1,4 @@
use schema::{Schema, Field, Document, Cardinality};
use schema::{Cardinality, Document, Field, Schema};
use fastfield::FastFieldSerializer;
use std::io;
use schema::Value;
@@ -25,12 +25,11 @@ impl FastFieldsWriter {
for (field_id, field_entry) in schema.fields().iter().enumerate() {
let field = Field(field_id as u32);
let default_value =
if let FieldType::I64(_) = *field_entry.field_type() {
common::i64_to_u64(0i64)
} else {
0u64
};
let default_value = if let FieldType::I64(_) = *field_entry.field_type() {
common::i64_to_u64(0i64)
} else {
0u64
};
match *field_entry.field_type() {
FieldType::I64(ref int_options) | FieldType::U64(ref int_options) => {
match int_options.get_fastfield_cardinality() {
@@ -50,7 +49,7 @@ impl FastFieldsWriter {
let fast_field_writer = MultiValueIntFastFieldWriter::new(field);
multi_values_writers.push(fast_field_writer);
}
_ => {},
_ => {}
}
}
FastFieldsWriter {
@@ -64,7 +63,7 @@ impl FastFieldsWriter {
pub(crate) fn new(fields: Vec<Field>) -> FastFieldsWriter {
FastFieldsWriter {
single_value_writers: fields.into_iter().map(IntFastFieldWriter::new).collect(),
multi_values_writers: vec!(),
multi_values_writers: vec![],
}
}
@@ -73,23 +72,22 @@ impl FastFieldsWriter {
// TODO optimize
self.single_value_writers
.iter_mut()
.find(|field_writer| {
field_writer.field() == field
})
.find(|field_writer| field_writer.field() == field)
}
/// Returns the fast field multi-value writer for the given field.
///
/// Returns None if the field does not exist, or is not
/// configured as a multivalued fastfield in the schema.
pub(crate) fn get_multivalue_writer(&mut self, field: Field) -> Option<&mut MultiValueIntFastFieldWriter> {
pub(crate) fn get_multivalue_writer(
&mut self,
field: Field,
) -> Option<&mut MultiValueIntFastFieldWriter> {
// TODO optimize
// TODO expose for users
self.multi_values_writers
.iter_mut()
.find(|multivalue_writer| {
multivalue_writer.field() == field
})
.find(|multivalue_writer| multivalue_writer.field() == field)
}
/// Indexes all of the fastfields of a new document.
@@ -104,9 +102,11 @@ impl FastFieldsWriter {
/// Serializes all of the `FastFieldWriter`s by pushing them in
/// order to the fast field serializer.
pub fn serialize(&self,
serializer: &mut FastFieldSerializer,
mapping: HashMap<Field, HashMap<UnorderedTermId, usize>>) -> io::Result<()> {
pub fn serialize(
&self,
serializer: &mut FastFieldSerializer,
mapping: &HashMap<Field, HashMap<UnorderedTermId, usize>>,
) -> io::Result<()> {
for field_writer in &self.single_value_writers {
field_writer.serialize(serializer)?;
}
@@ -201,9 +201,9 @@ impl IntFastFieldWriter {
/// associated to the document with the `DocId` n.
/// (Well, `n-1` actually because of 0-indexing)
pub fn add_val(&mut self, val: u64) {
VInt(val).serialize(&mut self.vals).expect(
"unable to serialize VInt to Vec",
);
VInt(val)
.serialize(&mut self.vals)
.expect("unable to serialize VInt to Vec");
if val > self.val_max {
self.val_max = val;
@@ -215,7 +215,6 @@ impl IntFastFieldWriter {
self.val_count += 1;
}
/// Extract the value associated to the fast field for
/// this document.
///
@@ -228,13 +227,11 @@ impl IntFastFieldWriter {
/// only the first one is taken in account.
fn extract_val(&self, doc: &Document) -> u64 {
match doc.get_first(self.field) {
Some(v) => {
match *v {
Value::U64(ref val) => *val,
Value::I64(ref val) => common::i64_to_u64(*val),
_ => panic!("Expected a u64field, got {:?} ", v),
}
}
Some(v) => match *v {
Value::U64(ref val) => *val,
Value::I64(ref val) => common::i64_to_u64(*val),
_ => panic!("Expected a u64field, got {:?} ", v),
},
None => self.val_if_missing,
}
}

10
src/indexer/index_writer.rs
View File

@@ -23,7 +23,6 @@ use indexer::SegmentWriter;
use postings::DocSet;
use schema::IndexRecordOption;
use schema::Document;
use schema::Schema;
use schema::Term;
use std::mem;
use std::mem::swap;
@@ -250,17 +249,18 @@ fn index_documents(
heap: &mut Heap,
table_size: usize,
segment: &Segment,
schema: &Schema,
generation: usize,
document_iterator: &mut Iterator<Item = AddOperation>,
segment_updater: &mut SegmentUpdater,
mut delete_cursor: DeleteCursor,
) -> Result<bool> {
heap.clear();
let schema = segment.schema();
let segment_id = segment.id();
let mut segment_writer = SegmentWriter::for_segment(heap, table_size, segment.clone(), schema)?;
let mut segment_writer =
SegmentWriter::for_segment(heap, table_size, segment.clone(), &schema)?;
for doc in document_iterator {
segment_writer.add_document(doc, schema)?;
segment_writer.add_document(doc, &schema)?;
// There is two possible conditions to close the segment.
// One is the memory arena dedicated to the segment is
// getting full.
@@ -368,7 +368,6 @@ impl IndexWriter {
/// The thread consumes documents from the pipeline.
///
fn add_indexing_worker(&mut self) -> Result<()> {
let schema = self.index.schema();
let document_receiver_clone = self.document_receiver.clone();
let mut segment_updater = self.segment_updater.clone();
let (heap_size, table_size) = split_memory(self.heap_size_in_bytes_per_thread);
@@ -409,7 +408,6 @@ impl IndexWriter {
&mut heap,
table_size,
&segment,
&schema,
generation,
&mut document_iterator,
&mut segment_updater,

2
src/indexer/segment_updater.rs
View File

@@ -69,7 +69,7 @@ pub fn save_metas(
segments: segment_metas,
schema,
opstamp,
payload: payload.clone(),
payload,
};
let mut buffer = serde_json::to_vec_pretty(&metas)?;
write!(&mut buffer, "\n")?;

81
src/indexer/segment_writer.rs
View File

@@ -17,7 +17,7 @@ use super::operation::AddOperation;
use postings::MultiFieldPostingsWriter;
use tokenizer::BoxedTokenizer;
use tokenizer::FacetTokenizer;
use tokenizer::{Tokenizer, TokenStream};
use tokenizer::{TokenStream, Tokenizer};
use schema::Value;
/// A `SegmentWriter` is in charge of creating segment index from a
@@ -126,11 +126,7 @@ impl<'a> SegmentWriter<'a> {
/// Indexes a new document
///
/// As a user, you should rather use `IndexWriter`'s add_document.
pub fn add_document(
&mut self,
add_operation: AddOperation,
schema: &Schema,
) -> io::Result<()> {
pub fn add_document(&mut self, add_operation: AddOperation, schema: &Schema) -> io::Result<()> {
let doc_id = self.max_doc;
let mut doc = add_operation.document;
self.doc_opstamps.push(add_operation.opstamp);
@@ -144,28 +140,26 @@ impl<'a> SegmentWriter<'a> {
}
match *field_options.field_type() {
FieldType::HierarchicalFacet => {
let facets: Vec<&[u8]> = field_values.iter()
.flat_map(|field_value| {
match field_value.value() {
&Value::Facet(ref facet) => Some(facet.encoded_bytes()),
_ => {
panic!("Expected hierarchical facet");
}
let facets: Vec<&[u8]> = field_values
.iter()
.flat_map(|field_value| match *field_value.value() {
Value::Facet(ref facet) => Some(facet.encoded_bytes()),
_ => {
panic!("Expected hierarchical facet");
}
})
.collect();
let mut term = unsafe {Term::with_capacity(100)};
let mut term = unsafe { Term::with_capacity(100) };
term.set_field(field);
for facet_bytes in facets {
let mut unordered_term_id_opt = None;
let fake_str = unsafe { str::from_utf8_unchecked(facet_bytes) };
FacetTokenizer
.token_stream(&fake_str)
.process(&mut |ref token| {
term.set_text(&token.text);
let unordered_term_id = self.multifield_postings.subscribe(doc_id, &term);
unordered_term_id_opt = Some(unordered_term_id);
});
FacetTokenizer.token_stream(fake_str).process(&mut |token| {
term.set_text(&token.text);
let unordered_term_id =
self.multifield_postings.subscribe(doc_id, &term);
unordered_term_id_opt = Some(unordered_term_id);
});
if let Some(unordered_term_id) = unordered_term_id_opt {
self.fast_field_writers
@@ -176,25 +170,26 @@ impl<'a> SegmentWriter<'a> {
}
}
FieldType::Str(_) => {
let num_tokens =
if let Some(ref mut tokenizer) = self.tokenizers[field.0 as usize] {
let texts: Vec<&str> = field_values
.iter()
.flat_map(|field_value| match *field_value.value() {
Value::Str(ref text) => Some(text.as_str()),
_ => None,
})
.collect();
if texts.is_empty() {
0
} else {
let mut token_stream = tokenizer.token_stream_texts(&texts[..]);
self.multifield_postings
.index_text(doc_id, field, &mut token_stream)
}
} else {
let num_tokens = if let Some(ref mut tokenizer) =
self.tokenizers[field.0 as usize]
{
let texts: Vec<&str> = field_values
.iter()
.flat_map(|field_value| match *field_value.value() {
Value::Str(ref text) => Some(text.as_str()),
_ => None,
})
.collect();
if texts.is_empty() {
0
};
} else {
let mut token_stream = tokenizer.token_stream_texts(&texts[..]);
self.multifield_postings
.index_text(doc_id, field, &mut token_stream)
}
} else {
0
};
self.fieldnorms_writer
.get_field_writer(field)
.map(|field_norms_writer| {
@@ -226,9 +221,7 @@ impl<'a> SegmentWriter<'a> {
}
}
self.fieldnorms_writer.fill_val_up_to(doc_id);
doc.filter_fields(|field| {
schema.get_field_entry(field).is_stored()
});
doc.filter_fields(|field| schema.get_field_entry(field).is_stored());
let doc_writer = self.segment_serializer.get_store_writer();
doc_writer.store(&doc)?;
self.max_doc += 1;
@@ -264,8 +257,8 @@ fn write(
mut serializer: SegmentSerializer,
) -> Result<()> {
let term_ord_map = multifield_postings.serialize(serializer.get_postings_serializer())?;
fast_field_writers.serialize(serializer.get_fast_field_serializer(), term_ord_map)?;
fieldnorms_writer.serialize(serializer.get_fieldnorms_serializer(), HashMap::new())?;
fast_field_writers.serialize(serializer.get_fast_field_serializer(), &term_ord_map)?;
fieldnorms_writer.serialize(serializer.get_fieldnorms_serializer(), &HashMap::new())?;
serializer.close()?;
Ok(())

107
src/lib.rs
View File

@@ -4,6 +4,7 @@
#![feature(box_syntax)]
#![feature(optin_builtin_traits)]
#![feature(conservative_impl_trait)]
#![feature(collections_range)]
#![feature(integer_atomics)]
#![feature(drain_filter)]
#![cfg_attr(test, feature(test))]
@@ -19,6 +20,98 @@
//! Tantivy is a search engine library.
//! Think `Lucene`, but in Rust.
//!
//! ```rust
//! # extern crate tempdir;
//! #
//! #[macro_use]
//! extern crate tantivy;
//!
//! // ...
//!
//! # use std::path::Path;
//! # use tempdir::TempDir;
//! # use tantivy::Index;
//! # use tantivy::schema::*;
//! # use tantivy::collector::TopCollector;
//! # use tantivy::query::QueryParser;
//! #
//! # fn main() {
//! # // Let's create a temporary directory for the
//! # // sake of this example
//! # if let Ok(dir) = TempDir::new("tantivy_example_dir") {
//! # run_example(dir.path()).unwrap();
//! # dir.close().unwrap();
//! # }
//! # }
//! #
//! # fn run_example(index_path: &Path) -> tantivy::Result<()> {
//! // First we need to define a schema ...
//!
//! // `TEXT` means the field should be tokenized and indexed,
//! // along with its term frequency and term positions.
//! //
//! // `STORED` means that the field will also be saved
//! // in a compressed, row-oriented key-value store.
//! // This store is useful to reconstruct the
//! // documents that were selected during the search phase.
//! let mut schema_builder = SchemaBuilder::default();
//! let title = schema_builder.add_text_field("title", TEXT | STORED);
//! let body = schema_builder.add_text_field("body", TEXT);
//! let schema = schema_builder.build();
//!
//! // Indexing documents
//!
//! let index = Index::create(index_path, schema.clone())?;
//!
//! // Here we use a buffer of 100MB that will be split
//! // between indexing threads.
//! let mut index_writer = index.writer(100_000_000)?;
//!
//! // Let's index one documents!
//! index_writer.add_document(doc!(
//! title => "The Old Man and the Sea",
//! body => "He was an old man who fished alone in a skiff in \
//! the Gulf Stream and he had gone eighty-four days \
//! now without taking a fish."
//! ));
//!
//! // We need to call .commit() explicitly to force the
//! // index_writer to finish processing the documents in the queue,
//! // flush the current index to the disk, and advertise
//! // the existence of new documents.
//! index_writer.commit()?;
//!
//! // # Searching
//!
//! index.load_searchers()?;
//!
//! let searcher = index.searcher();
//!
//! let query_parser = QueryParser::for_index(&index, vec![title, body]);
//!
//! // QueryParser may fail if the query is not in the right
//! // format. For user facing applications, this can be a problem.
//! // A ticket has been opened regarding this problem.
//! let query = query_parser.parse_query("sea whale")?;
//!
//! let mut top_collector = TopCollector::with_limit(10);
//! searcher.search(&*query, &mut top_collector)?;
//!
//! // Our top collector now contains the 10
//! // most relevant doc ids...
//! let doc_addresses = top_collector.docs();
//! for doc_address in doc_addresses {
//! let retrieved_doc = searcher.doc(&doc_address)?;
//! println!("{}", schema.to_json(&retrieved_doc));
//! }
//!
//! # Ok(())
//! # }
//! ```
//!
//!
//!
//! A good place for you to get started is to check out
//! the example code (
//! [literate programming](http://fulmicoton.com/tantivy-examples/simple_search.html) /
@@ -36,8 +129,6 @@ extern crate log;
#[macro_use]
extern crate error_chain;
extern crate regex;
extern crate tempfile;
extern crate atomicwrites;
extern crate bit_set;
extern crate byteorder;
@@ -51,11 +142,13 @@ extern crate itertools;
extern crate lz4;
extern crate num_cpus;
extern crate owning_ref;
extern crate regex;
extern crate rust_stemmers;
extern crate serde;
extern crate serde_json;
extern crate stable_deref_trait;
extern crate tempdir;
extern crate tempfile;
extern crate time;
extern crate uuid;
@@ -192,6 +285,7 @@ mod tests {
use fastfield::{FastFieldReader, I64FastFieldReader, U64FastFieldReader};
use Postings;
use rand::{Rng, SeedableRng, XorShiftRng};
use rand::distributions::{IndependentSample, Range};
fn generate_array_with_seed(n: usize, ratio: f32, seed_val: u32) -> Vec<u32> {
let seed: &[u32; 4] = &[1, 2, 3, seed_val];
@@ -202,6 +296,15 @@ mod tests {
.collect()
}
pub fn generate_nonunique_unsorted(max_value: u32, n_elems: usize) -> Vec<u32> {
let seed: &[u32; 4] = &[1, 2, 3, 4];
let mut rng: XorShiftRng = XorShiftRng::from_seed(*seed);
let between = Range::new(0u32, max_value);
(0..n_elems)
.map(|_| between.ind_sample(&mut rng))
.collect::<Vec<u32>>()
}
pub fn generate_array(n: usize, ratio: f32) -> Vec<u32> {
generate_array_with_seed(n, ratio, 4)
}

33
src/postings/docset.rs
View File

@@ -2,7 +2,7 @@ use DocId;
use std::borrow::Borrow;
use std::borrow::BorrowMut;
use std::cmp::Ordering;
use common::DocBitSet;
use common::BitSet;
/// Expresses the outcome of a call to `DocSet`'s `.skip_next(...)`.
#[derive(PartialEq, Eq, Debug)]
@@ -95,13 +95,11 @@ pub trait DocSet {
/// length of the docset.
fn size_hint(&self) -> u32;
fn to_doc_bitset(&mut self, max_doc: DocId) -> DocBitSet {
let mut docs = DocBitSet::with_maxdoc(max_doc);
/// Appends all docs to a `bitset`.
fn append_to_bitset(&mut self, bitset: &mut BitSet) {
while self.advance() {
let doc = self.doc();
docs.insert(doc);
bitset.insert(self.doc());
}
docs
}
}
@@ -125,26 +123,9 @@ impl<TDocSet: DocSet + ?Sized> DocSet for Box<TDocSet> {
let unboxed: &TDocSet = self.borrow();
unboxed.size_hint()
}
}
impl<'a, TDocSet: DocSet> DocSet for &'a mut TDocSet {
fn advance(&mut self) -> bool {
let unref: &mut TDocSet = *self;
unref.advance()
}
fn skip_next(&mut self, target: DocId) -> SkipResult {
let unref: &mut TDocSet = *self;
unref.skip_next(target)
}
fn doc(&self) -> DocId {
let unref: &TDocSet = *self;
unref.doc()
}
fn size_hint(&self) -> u32 {
let unref: &TDocSet = *self;
unref.size_hint()
fn append_to_bitset(&mut self, bitset: &mut BitSet) {
let unboxed: &mut TDocSet = self.borrow_mut();
unboxed.append_to_bitset(bitset);
}
}

37
src/postings/mod.rs
View File

@@ -97,39 +97,52 @@ mod tests {
index_writer.commit().unwrap();
index.load_searchers().unwrap();
let searcher = index.searcher();
let query = TermQuery::new(Term::from_field_text(title, "abc"), IndexRecordOption::WithFreqsAndPositions);
let query = TermQuery::new(
Term::from_field_text(title, "abc"),
IndexRecordOption::WithFreqsAndPositions,
);
let weight = query.specialized_weight(&*searcher);
{
let mut scorer = weight.specialized_scorer(searcher.segment_reader(0u32)).unwrap();
let mut scorer = weight
.specialized_scorer(searcher.segment_reader(0u32))
.unwrap();
scorer.advance();
assert_eq!(&[0,1,2], scorer.postings().positions());
assert_eq!(&[0, 1, 2], scorer.postings().positions());
scorer.advance();
assert_eq!(&[0,5], scorer.postings().positions());
assert_eq!(&[0, 5], scorer.postings().positions());
}
{
let mut scorer = weight.specialized_scorer(searcher.segment_reader(0u32)).unwrap();
let mut scorer = weight
.specialized_scorer(searcher.segment_reader(0u32))
.unwrap();
scorer.advance();
scorer.advance();
assert_eq!(&[0,5], scorer.postings().positions());
assert_eq!(&[0, 5], scorer.postings().positions());
}
{
let mut scorer = weight.specialized_scorer(searcher.segment_reader(0u32)).unwrap();
let mut scorer = weight
.specialized_scorer(searcher.segment_reader(0u32))
.unwrap();
assert_eq!(scorer.skip_next(1), SkipResult::Reached);
assert_eq!(scorer.doc(), 1);
assert_eq!(&[0,5], scorer.postings().positions());
assert_eq!(&[0, 5], scorer.postings().positions());
}
{
let mut scorer = weight.specialized_scorer(searcher.segment_reader(0u32)).unwrap();
let mut scorer = weight
.specialized_scorer(searcher.segment_reader(0u32))
.unwrap();
assert_eq!(scorer.skip_next(1002), SkipResult::Reached);
assert_eq!(scorer.doc(), 1002);
assert_eq!(&[0,5], scorer.postings().positions());
assert_eq!(&[0, 5], scorer.postings().positions());
}
{
let mut scorer = weight.specialized_scorer(searcher.segment_reader(0u32)).unwrap();
let mut scorer = weight
.specialized_scorer(searcher.segment_reader(0u32))
.unwrap();
assert_eq!(scorer.skip_next(100), SkipResult::Reached);
assert_eq!(scorer.skip_next(1002), SkipResult::Reached);
assert_eq!(scorer.doc(), 1002);
assert_eq!(&[0,5], scorer.postings().positions());
assert_eq!(&[0, 5], scorer.postings().positions());
}
}

24
src/postings/postings.rs
View File

@@ -30,15 +30,15 @@ impl<TPostings: Postings> Postings for Box<TPostings> {
unboxed.positions()
}
}
impl<'a, TPostings: Postings> Postings for &'a mut TPostings {
fn term_freq(&self) -> u32 {
let unref: &TPostings = *self;
unref.term_freq()
}
fn positions(&self) -> &[u32] {
let unref: &TPostings = *self;
unref.positions()
}
}
//
//impl<'a, TPostings: Postings> Postings for &'a mut TPostings {
// fn term_freq(&self) -> u32 {
// let unref: &TPostings = *self;
// unref.term_freq()
// }
//
// fn positions(&self) -> &[u32] {
// let unref: &TPostings = *self;
// unref.positions()
// }
//}

86
src/postings/postings_writer.rs
View File

@@ -1,15 +1,15 @@
use DocId;
use schema::Term;
use postings::{InvertedIndexSerializer, FieldSerializer};
use postings::{FieldSerializer, InvertedIndexSerializer};
use std::io;
use std::collections::HashMap;
use postings::Recorder;
use Result;
use schema::{Schema, Field};
use schema::{Field, Schema};
use std::marker::PhantomData;
use std::ops::DerefMut;
use datastruct::stacker::{TermHashMap, Heap};
use postings::{NothingRecorder, TermFrequencyRecorder, TFAndPositionRecorder};
use datastruct::stacker::{Heap, TermHashMap};
use postings::{NothingRecorder, TFAndPositionRecorder, TermFrequencyRecorder};
use schema::FieldEntry;
use schema::FieldType;
use tokenizer::Token;
@@ -17,39 +17,31 @@ use tokenizer::TokenStream;
use schema::IndexRecordOption;
use postings::UnorderedTermId;
fn posting_from_field_entry<'a>(
field_entry: &FieldEntry,
heap: &'a Heap,
) -> Box<PostingsWriter + 'a> {
match *field_entry.field_type() {
FieldType::Str(ref text_options) => {
text_options
FieldType::Str(ref text_options) => text_options
.get_indexing_options()
.map(|indexing_options| {
match indexing_options.index_option() {
IndexRecordOption::Basic => {
SpecializedPostingsWriter::<NothingRecorder>::new_boxed(heap)
}
IndexRecordOption::WithFreqs => {
SpecializedPostingsWriter::<TermFrequencyRecorder>::new_boxed(heap)
}
IndexRecordOption::WithFreqsAndPositions => {
SpecializedPostingsWriter::<TFAndPositionRecorder>::new_boxed(heap)
}
.map(|indexing_options| match indexing_options.index_option() {
IndexRecordOption::Basic => {
SpecializedPostingsWriter::<NothingRecorder>::new_boxed(heap)
}
IndexRecordOption::WithFreqs => {
SpecializedPostingsWriter::<TermFrequencyRecorder>::new_boxed(heap)
}
IndexRecordOption::WithFreqsAndPositions => {
SpecializedPostingsWriter::<TFAndPositionRecorder>::new_boxed(heap)
}
})
.unwrap_or_else(|| {
SpecializedPostingsWriter::<NothingRecorder>::new_boxed(heap)
})
.unwrap_or_else(|| SpecializedPostingsWriter::<NothingRecorder>::new_boxed(heap)),
FieldType::U64(_) | FieldType::I64(_) | FieldType::HierarchicalFacet => {
SpecializedPostingsWriter::<NothingRecorder>::new_boxed(heap)
}
FieldType::U64(_) |
FieldType::I64(_) |
FieldType::HierarchicalFacet => SpecializedPostingsWriter::<NothingRecorder>::new_boxed(heap),
}
}
pub struct MultiFieldPostingsWriter<'a> {
heap: &'a Heap,
term_index: TermHashMap<'a>,
@@ -88,7 +80,10 @@ impl<'a> MultiFieldPostingsWriter<'a> {
/// It pushes all term, one field at a time, towards the
/// postings serializer.
#[allow(needless_range_loop)]
pub fn serialize(&self, serializer: &mut InvertedIndexSerializer) -> Result<HashMap<Field, HashMap<UnorderedTermId, usize>>> {
pub fn serialize(
&self,
serializer: &mut InvertedIndexSerializer,
) -> Result<HashMap<Field, HashMap<UnorderedTermId, usize>>> {
let mut term_offsets: Vec<(&[u8], u32, UnorderedTermId)> = self.term_index.iter().collect();
term_offsets.sort_by_key(|&(k, _, _)| k);
@@ -99,8 +94,8 @@ impl<'a> MultiFieldPostingsWriter<'a> {
.map(|(key, _, _)| Term::wrap(key).field())
.enumerate();
let mut unordered_term_mappings: HashMap<Field, HashMap<UnorderedTermId, usize>> = HashMap::new();
let mut unordered_term_mappings: HashMap<Field, HashMap<UnorderedTermId, usize>> =
HashMap::new();
let mut prev_field = Field(u32::max_value());
for (offset, field) in term_offsets_it {
@@ -120,8 +115,9 @@ impl<'a> MultiFieldPostingsWriter<'a> {
let mut mapping = HashMap::new();
for (term_ord, term_unord_id) in term_offsets[start..stop]
.iter()
.map(|&(_,_,bucket)| bucket)
.enumerate() {
.map(|&(_, _, bucket)| bucket)
.enumerate()
{
mapping.insert(term_unord_id, term_ord);
}
unordered_term_mappings.insert(field, mapping);
@@ -144,7 +140,6 @@ impl<'a> MultiFieldPostingsWriter<'a> {
}
}
/// The `PostingsWriter` is in charge of receiving documenting
/// and building a `Segment` in anonymous memory.
///
@@ -168,20 +163,22 @@ pub trait PostingsWriter {
/// Serializes the postings on disk.
/// The actual serialization format is handled by the `PostingsSerializer`.
fn serialize(&self,
term_addrs: &[(&[u8], u32, UnorderedTermId)],
serializer: &mut FieldSerializer,
heap: &Heap)
-> io::Result<()>;
fn serialize(
&self,
term_addrs: &[(&[u8], u32, UnorderedTermId)],
serializer: &mut FieldSerializer,
heap: &Heap,
) -> io::Result<()>;
/// Tokenize a text and subscribe all of its token.
fn index_text<'a>(&mut self,
term_index: &mut TermHashMap,
doc_id: DocId,
field: Field,
token_stream: &mut TokenStream,
heap: &Heap)
-> u32 {
fn index_text(
&mut self,
term_index: &mut TermHashMap,
doc_id: DocId,
field: Field,
token_stream: &mut TokenStream,
heap: &Heap,
) -> u32 {
let mut term = unsafe { Term::with_capacity(100) };
term.set_field(field);
let mut sink = |token: &Token| {
@@ -215,7 +212,6 @@ impl<'a, Rec: Recorder + 'static> SpecializedPostingsWriter<'a, Rec> {
}
impl<'a, Rec: Recorder + 'static> PostingsWriter for SpecializedPostingsWriter<'a, Rec> {
fn subscribe(
&mut self,
term_index: &mut TermHashMap,
@@ -237,8 +233,6 @@ impl<'a, Rec: Recorder + 'static> PostingsWriter for SpecializedPostingsWriter<'
term_ord
}
fn serialize(
&self,
term_addrs: &[(&[u8], u32, UnorderedTermId)],

12
src/postings/segment_postings.rs
View File

@@ -1,6 +1,6 @@
use compression::{BlockDecoder, CompressedIntStream, VIntDecoder, COMPRESSION_BLOCK_SIZE};
use DocId;
use common::DocBitSet;
use common::BitSet;
use postings::{DocSet, HasLen, Postings, SkipResult};
use std::cmp;
use fst::Streamer;
@@ -250,22 +250,20 @@ impl DocSet for SegmentPostings {
);
docs[self.cur]
}
fn to_doc_bitset(&mut self, max_doc: DocId) -> DocBitSet {
fn append_to_bitset(&mut self, bitset: &mut BitSet) {
// finish the current block
let mut docs = DocBitSet::with_maxdoc(max_doc);
if self.advance() {
for &doc in &self.block_cursor.docs()[self.cur..] {
docs.insert(doc);
bitset.insert(doc);
}
// ... iterate through the remaining blocks.
while self.block_cursor.advance() {
for &doc in self.block_cursor.docs() {
docs.insert(doc);
bitset.insert(doc);
}
}
}
return docs;
}
}

25
src/postings/term_info.rs
View File

@@ -1,4 +1,4 @@
use common::BinarySerializable;
use common::{BinarySerializable, FixedSize};
use std::io;
/// `TermInfo` contains all of the information
@@ -23,10 +23,13 @@ pub struct TermInfo {
pub positions_inner_offset: u8,
}
impl TermInfo {
/// Size required to encode the `TermInfo`.
// TODO make this smaller when positions are unused for instance.
pub(crate) const SIZE_IN_BYTES: usize = 4 + 8 + 8 + 1;
impl FixedSize for TermInfo {
/// Size required for the binary serialization of `TermInfo`.
/// This is large, but in practise, all `TermInfo` but the first one
/// of the block are bitpacked.
///
/// See `TermInfoStore`.
const SIZE_IN_BYTES: usize = u32::SIZE_IN_BYTES + 2 * u64::SIZE_IN_BYTES + u8::SIZE_IN_BYTES;
}
impl BinarySerializable for TermInfo {
@@ -50,3 +53,15 @@ impl BinarySerializable for TermInfo {
})
}
}
#[cfg(test)]
mod tests {
use super::TermInfo;
use common::test::fixed_size_test;
#[test]
fn test_fixed_size() {
fixed_size_test::<TermInfo>();
}
}

26
src/postings/union.rs
View File

@@ -10,7 +10,7 @@ const HORIZON: usize = 64 * HORIZON_NUM_TINYBITSETS;
/// Creates a `DocSet` that iterator through the intersection of two `DocSet`s.
pub struct UnionDocSet<TDocSet: DocSet> {
docsets: Vec<TDocSet>,
bitsets: Box<[u64; HORIZON_NUM_TINYBITSETS]>,
bitsets: Box<[TinySet; HORIZON_NUM_TINYBITSETS]>,
cursor: usize,
offset: DocId,
doc: DocId,
@@ -31,7 +31,7 @@ impl<TDocSet: DocSet> From<Vec<TDocSet>> for UnionDocSet<TDocSet> {
.collect();
UnionDocSet {
docsets: non_empty_docsets,
bitsets: Box::new([0u64; HORIZON_NUM_TINYBITSETS]),
bitsets: Box::new([TinySet::empty(); HORIZON_NUM_TINYBITSETS]),
cursor: HORIZON_NUM_TINYBITSETS,
offset: 0,
doc: 0
@@ -40,10 +40,10 @@ impl<TDocSet: DocSet> From<Vec<TDocSet>> for UnionDocSet<TDocSet> {
}
fn refill<TDocSet: DocSet>(docsets: &mut Vec<TDocSet>, bitsets: &mut [u64; HORIZON_NUM_TINYBITSETS], min_doc: DocId) {
fn refill<TDocSet: DocSet>(docsets: &mut Vec<TDocSet>, bitsets: &mut [TinySet; HORIZON_NUM_TINYBITSETS], min_doc: DocId) {
docsets
.drain_filter(|docset| {
let horizon = min_doc + HORIZON_NUM_TINYBITSETS as u32;
let horizon = min_doc + HORIZON as u32;
loop {
let doc = docset.doc();
if doc >= horizon {
@@ -51,7 +51,7 @@ fn refill<TDocSet: DocSet>(docsets: &mut Vec<TDocSet>, bitsets: &mut [u64; HORIZ
}
// add this document
let delta = doc - min_doc;
bitsets[(delta / 64) as usize] |= 1 << (delta % 64);
bitsets[(delta / 64) as usize].insert_mut(delta % 64u32);
if !docset.advance() {
// remove the docset, it has been entirely consumed.
return true;
@@ -91,14 +91,6 @@ impl<TDocSet: DocSet> DocSet for UnionDocSet<TDocSet> {
self.refill()
}
fn doc(&self) -> DocId {
self.doc
}
fn size_hint(&self) -> u32 {
0u32
}
fn skip_next(&mut self, target: DocId) -> SkipResult {
let mut reached = false;
self.docsets
@@ -122,6 +114,14 @@ impl<TDocSet: DocSet> DocSet for UnionDocSet<TDocSet> {
}
}
}
fn doc(&self) -> DocId {
self.doc
}
fn size_hint(&self) -> u32 {
0u32
}
}

9
src/query/all_query.rs
View File

@@ -9,7 +9,6 @@ use DocId;
use std::any::Any;
use core::Searcher;
/// Query that matches all of the documents.
///
/// All of the document get the score 1f32.
@@ -34,12 +33,11 @@ impl Weight for AllWeight {
Ok(box AllScorer {
started: false,
doc: 0u32,
max_doc: reader.max_doc()
max_doc: reader.max_doc(),
})
}
}
/// Scorer associated to the `AllQuery` query.
pub struct AllScorer {
started: bool,
@@ -51,8 +49,7 @@ impl DocSet for AllScorer {
fn advance(&mut self) -> bool {
if self.started {
self.doc += 1u32;
}
else {
} else {
self.started = true;
}
self.doc < self.max_doc
@@ -71,4 +68,4 @@ impl Scorer for AllScorer {
fn score(&self) -> Score {
1f32
}
}
}

211
src/query/bitset/mod.rs
View File

@@ -1,4 +1,4 @@
use common::{DocBitSet, TinySet};
use common::{BitSet, TinySet};
use DocId;
use postings::DocSet;
use postings::SkipResult;
@@ -8,107 +8,100 @@ use std::cmp::Ordering;
///
/// # Implementation detail
///
/// Skipping is relatively fast here as we can directly point to the
/// Skipping is relatively fast here as we can directly point to the
/// right tiny bitset bucket.
///
/// TODO: Consider implementing a `BitTreeSet` in order to advance faster
/// TODO: Consider implementing a `BitTreeSet` in order to advance faster
/// when the bitset is sparse
pub struct BitSetDocSet {
docs: DocBitSet,
cursor_bucket: usize, //< index associated to the current tiny bitset
cursor_tinybitset: u64,
doc: u32
docs: BitSet,
cursor_bucket: u32, //< index associated to the current tiny bitset
cursor_tinybitset: TinySet,
doc: u32,
}
impl From<DocBitSet> for BitSetDocSet {
fn from(docs: DocBitSet) -> BitSetDocSet {
let first_tiny_bitset =
if docs.num_tiny_bitsets() == 0 {
0u64
} else {
docs.tiny_bitset(0) as u64
};
impl BitSetDocSet {
fn go_to_bucket(&mut self, bucket_addr: u32) {
self.cursor_bucket = bucket_addr;
self.cursor_tinybitset = self.docs.tinyset(bucket_addr);
}
}
impl From<BitSet> for BitSetDocSet {
fn from(docs: BitSet) -> BitSetDocSet {
let first_tiny_bitset = if docs.max_value() == 0 {
TinySet::empty()
} else {
docs.tinyset(0)
};
BitSetDocSet {
docs,
cursor_bucket: 0,
cursor_tinybitset: first_tiny_bitset,
doc: 0u32
doc: 0u32,
}
}
}
impl DocSet for BitSetDocSet {
fn advance(&mut self) -> bool {
loop {
if let Some(lower) = self.cursor_tinybitset.pop_lowest() {
self.doc = (self.cursor_bucket as u32 * 64u32) | lower;
return true;
} else {
if self.cursor_bucket < self.docs.num_tiny_bitsets() - 1 {
self.cursor_bucket += 1;
self.cursor_tinybitset = self.docs.tiny_bitset(self.cursor_bucket);
} else {
return false;
}
}
if let Some(lower) = self.cursor_tinybitset.pop_lowest() {
self.doc = (self.cursor_bucket as u32 * 64u32) | lower;
return true;
}
if let Some(cursor_bucket) = self.docs.first_non_empty_bucket(self.cursor_bucket + 1) {
self.go_to_bucket(cursor_bucket);
let lower = self.cursor_tinybitset.pop_lowest().unwrap();
self.doc = (cursor_bucket * 64u32) | lower;
true
} else {
false
}
}
fn skip_next(&mut self, target: DocId) -> SkipResult {
// skip is required to advance.
if !self.advance() {
return SkipResult::End;
}
let target_bucket = (target / 64u32) as usize;
let target_bucket = target / 64u32;
// Mask for all of the bits greater or equal
// to our target document.
match target_bucket.cmp(&self.cursor_bucket) {
Ordering::Greater => {
self.cursor_bucket = target_bucket;
self.cursor_tinybitset = self.docs.tiny_bitset(target_bucket);
// let greater: u64 = <u64 as TinySet>::range_greater_or_equal(target % 64);
// self.cursor_tinybitset.intersect(greater);
loop {
if !self.advance() {
return SkipResult::End;
self.go_to_bucket(target_bucket);
let greater_filter: TinySet = TinySet::range_greater_or_equal(target);
self.cursor_tinybitset = self.cursor_tinybitset.intersect(greater_filter);
if !self.advance() {
SkipResult::End
} else {
if self.doc() == target {
SkipResult::Reached
} else {
if self.doc() == target {
return SkipResult::Reached;
} else {
// assert!(self.doc() > target);
if self.doc() > target {
return SkipResult::OverStep;
}
}
debug_assert!(self.doc() > target);
SkipResult::OverStep
}
}
}
Ordering::Equal => {
loop {
match self.doc().cmp(&target) {
Ordering::Less => {
if !self.advance() {
return SkipResult::End;
}
}
Ordering::Equal => {
assert!(self.doc() == target);
return SkipResult::Reached;
}
Ordering::Greater => {
assert!(self.doc() > target);
return SkipResult::OverStep;
Ordering::Equal => loop {
match self.doc().cmp(&target) {
Ordering::Less => {
if !self.advance() {
return SkipResult::End;
}
}
Ordering::Equal => {
return SkipResult::Reached;
}
Ordering::Greater => {
debug_assert!(self.doc() > target);
return SkipResult::OverStep;
}
}
}
},
Ordering::Less => {
assert!(self.doc() > target);
debug_assert!(self.doc() > target);
SkipResult::OverStep
}
}
@@ -135,19 +128,20 @@ impl DocSet for BitSetDocSet {
/// but we don't have access to any better
/// value.
fn size_hint(&self) -> u32 {
self.docs.size_hint()
self.docs.len() as u32
}
}
#[cfg(test)]
mod tests {
use DocId;
use common::DocBitSet;
use postings::{SkipResult, DocSet};
use common::BitSet;
use postings::{DocSet, SkipResult};
use super::BitSetDocSet;
extern crate test;
fn create_docbitset(docs: &[DocId], max_doc: DocId) -> BitSetDocSet {
let mut docset = DocBitSet::with_maxdoc(max_doc);
let mut docset = BitSet::with_max_value(max_doc);
for &doc in docs {
docset.insert(doc);
}
@@ -167,10 +161,10 @@ mod tests {
#[test]
fn test_docbitset_sequential() {
test_go_through_sequential(&[]);
test_go_through_sequential(&[1,2,3]);
test_go_through_sequential(&[1,2,3,4,5,63,64,65]);
test_go_through_sequential(&[63,64,65]);
test_go_through_sequential(&[1,2,3,4,95,96,97,98,99]);
test_go_through_sequential(&[1, 2, 3]);
test_go_through_sequential(&[1, 2, 3, 4, 5, 63, 64, 65]);
test_go_through_sequential(&[63, 64, 65]);
test_go_through_sequential(&[1, 2, 3, 4, 95, 96, 97, 98, 99]);
}
#[test]
@@ -206,6 +200,73 @@ mod tests {
assert_eq!(docset.doc(), 5112);
assert!(!docset.advance());
}
{
let mut docset = create_docbitset(&[1, 5, 6, 7, 5112, 5500, 6666], 10_000);
assert_eq!(docset.skip_next(5112), SkipResult::Reached);
assert_eq!(docset.doc(), 5112);
assert!(docset.advance());
assert_eq!(docset.doc(), 5500);
assert!(docset.advance());
assert_eq!(docset.doc(), 6666);
assert!(!docset.advance());
}
{
let mut docset = create_docbitset(&[1, 5, 6, 7, 5112, 5500, 6666], 10_000);
assert_eq!(docset.skip_next(5111), SkipResult::OverStep);
assert_eq!(docset.doc(), 5112);
assert!(docset.advance());
assert_eq!(docset.doc(), 5500);
assert!(docset.advance());
assert_eq!(docset.doc(), 6666);
assert!(!docset.advance());
}
{
let mut docset = create_docbitset(&[1, 5, 6, 7, 5112, 5513, 6666], 10_000);
assert_eq!(docset.skip_next(5111), SkipResult::OverStep);
assert_eq!(docset.doc(), 5112);
assert!(docset.advance());
assert_eq!(docset.doc(), 5513);
assert!(docset.advance());
assert_eq!(docset.doc(), 6666);
assert!(!docset.advance());
}
}
}
#[bench]
fn bench_bitset_1pct_insert(b: &mut test::Bencher) {
use tests;
let els = tests::generate_nonunique_unsorted(1_000_000u32, 10_000);
b.iter(|| {
let mut bitset = BitSet::with_max_value(1_000_000);
for el in els.iter().cloned() {
bitset.insert(el);
}
});
}
#[bench]
fn bench_bitset_1pct_clone(b: &mut test::Bencher) {
use tests;
let els = tests::generate_nonunique_unsorted(1_000_000u32, 10_000);
let mut bitset = BitSet::with_max_value(1_000_000);
for el in els {
bitset.insert(el);
}
b.iter(|| bitset.clone());
}
#[bench]
fn bench_bitset_1pct_clone_iterate(b: &mut test::Bencher) {
use tests;
use DocSet;
let els = tests::generate_nonunique_unsorted(1_000_000u32, 10_000);
let mut bitset = BitSet::with_max_value(1_000_000);
for el in els {
bitset.insert(el);
}
b.iter(|| {
let mut docset = BitSetDocSet::from(bitset.clone());
while docset.advance() {}
});
}
}

14
src/query/boolean_query/boolean_query.rs
View File

@@ -39,6 +39,13 @@ impl Query for BooleanQuery {
self
}
fn disable_scoring(&mut self) {
self.scoring_disabled = true;
for &mut (_, ref mut subquery) in &mut self.subqueries {
subquery.disable_scoring();
}
}
fn weight(&self, searcher: &Searcher) -> Result<Box<Weight>> {
let sub_weights = self.subqueries
.iter()
@@ -48,13 +55,6 @@ impl Query for BooleanQuery {
.collect::<Result<_>>()?;
Ok(box BooleanWeight::new(sub_weights, self.scoring_disabled))
}
fn disable_scoring(&mut self) {
self.scoring_disabled = true;
for &mut (_, ref mut subquery) in &mut self.subqueries {
subquery.disable_scoring();
}
}
}
impl BooleanQuery {

5
src/query/boolean_query/boolean_weight.rs
View File

@@ -24,13 +24,14 @@ impl BooleanWeight {
}
}
impl Weight for BooleanWeight {
fn scorer<'a>(&'a self, reader: &'a SegmentReader) -> Result<Box<Scorer + 'a>> {
if self.weights.is_empty() {
Ok(box EmptyScorer)
} else if self.weights.len() == 1 {
let &(occur, ref weight) = &self.weights[0];
if occur == Occur::MustNot {
let &(occur, ref weight) = &self.weights[0];
if occur == Occur::MustNot {
Ok(box EmptyScorer)
} else {
weight.scorer(reader)

5
src/query/mod.rs
View File

@@ -15,7 +15,6 @@ mod all_query;
mod bitset;
mod range_query;
pub use self::bitset::BitSetDocSet;
pub use self::boolean_query::BooleanQuery;
pub use self::occur_filter::OccurFilter;
@@ -28,7 +27,7 @@ pub use self::scorer::EmptyScorer;
pub use self::scorer::Scorer;
pub use self::term_query::TermQuery;
pub use self::weight::Weight;
pub use self::all_query::{AllQuery, AllWeight, AllScorer};
pub use self::range_query::{RangeQuery,RangeDefinition, RangeWeight};
pub use self::all_query::{AllQuery, AllScorer, AllWeight};
pub use self::range_query::RangeQuery;
pub use self::scorer::ConstScorer;

9
src/query/phrase_query/mod.rs
View File

@@ -6,7 +6,6 @@ pub use self::phrase_query::PhraseQuery;
pub use self::phrase_weight::PhraseWeight;
pub use self::phrase_scorer::PhraseScorer;
#[cfg(test)]
mod tests {
@@ -75,8 +74,6 @@ mod tests {
assert_eq!(test_query(vec!["g", "a"]), empty_vec);
}
#[test] // motivated by #234
pub fn test_phrase_query_docfreq_order() {
let mut schema_builder = SchemaBuilder::default();
@@ -90,11 +87,13 @@ mod tests {
let doc = doc!(text_field=>"b");
index_writer.add_document(doc);
}
{ // 1
{
// 1
let doc = doc!(text_field=>"a b");
index_writer.add_document(doc);
}
{ // 2
{
// 2
let doc = doc!(text_field=>"b a");
index_writer.add_document(doc);
}

9
src/query/phrase_query/phrase_scorer.rs
View File

@@ -1,6 +1,6 @@
use query::Scorer;
use DocId;
use postings::{SkipResult, IntersectionDocSet, DocSet, Postings, SegmentPostings};
use postings::{DocSet, IntersectionDocSet, Postings, SegmentPostings, SkipResult};
struct PostingsWithOffset {
offset: u32,
@@ -11,7 +11,7 @@ impl PostingsWithOffset {
pub fn new(segment_postings: SegmentPostings, offset: u32) -> PostingsWithOffset {
PostingsWithOffset {
offset,
segment_postings
segment_postings,
}
}
}
@@ -49,7 +49,6 @@ pub struct PhraseScorer {
}
impl PhraseScorer {
pub fn new(term_postings: Vec<SegmentPostings>) -> PhraseScorer {
let postings_with_offsets: Vec<_> = term_postings
.into_iter()
@@ -57,12 +56,11 @@ impl PhraseScorer {
.map(|(offset, postings)| PostingsWithOffset::new(postings, offset as u32))
.collect();
PhraseScorer {
intersection_docset: IntersectionDocSet::from(postings_with_offsets)
intersection_docset: IntersectionDocSet::from(postings_with_offsets),
}
}
fn phrase_match(&self) -> bool {
// TODO maybe we could avoid decoding positions lazily for all terms
// when there is > 2 terms.
//
@@ -74,7 +72,6 @@ impl PhraseScorer {
positions_arr[docset.offset as usize] = docset.positions();
}
let num_postings = positions_arr.len() as u32;
let mut ord = 1u32;

3
src/query/phrase_query/phrase_weight.rs
View File

@@ -23,7 +23,8 @@ impl Weight for PhraseWeight {
for term in &self.phrase_terms {
if let Some(postings) = reader
.inverted_index(term.field())
.read_postings(term, IndexRecordOption::WithFreqsAndPositions) {
.read_postings(term, IndexRecordOption::WithFreqsAndPositions)
{
term_postings_list.push(postings);
} else {
return Ok(box EmptyScorer);

305
src/query/range_query.rs
View File

@@ -1,91 +1,129 @@
use schema::{Field, Term, IndexRecordOption};
use query::{Query, Weight, Scorer};
use schema::{Field, IndexRecordOption, Term};
use query::{Query, Scorer, Weight};
use termdict::{TermDictionary, TermStreamer, TermStreamerBuilder};
use core::SegmentReader;
use common::DocBitSet;
use common::BitSet;
use Result;
use std::any::Any;
use core::Searcher;
use query::BitSetDocSet;
use query::ConstScorer;
use std::collections::Bound;
use std::collections::range::RangeArgument;
#[derive(Clone, Debug)]
enum Boundary {
Included(Vec<u8>),
Excluded(Vec<u8>),
Unbounded,
}
#[derive(Clone, Debug)]
pub struct RangeDefinition {
field: Field,
left_bound: Boundary,
right_bound: Boundary
}
impl RangeDefinition {
pub fn for_field(field: Field) -> RangeDefinition{
RangeDefinition {
field,
left_bound: Boundary::Unbounded,
right_bound: Boundary::Unbounded
}
}
pub fn left_included(mut self, left: Term) -> RangeDefinition {
assert_eq!(left.field(), self.field);
self.left_bound = Boundary::Included(left.value_bytes().to_owned());
self
}
pub fn left_excluded(mut self, left: Term) -> RangeDefinition {
assert_eq!(left.field(), self.field);
self.left_bound = Boundary::Excluded(left.value_bytes().to_owned());
self
}
pub fn right_included(mut self, right: Term) -> RangeDefinition {
assert_eq!(right.field(), self.field);
self.right_bound = Boundary::Included(right.value_bytes().to_owned());
self
}
pub fn right_excluded(mut self, right: Term) -> RangeDefinition {
assert_eq!(right.field(), self.field);
self.right_bound = Boundary::Excluded(right.value_bytes().to_owned());
self
}
pub fn term_range<'a, T>(&self, term_dict: &'a T) -> T::Streamer
where T: TermDictionary<'a> + 'a
{
use self::Boundary::*;
let mut term_stream_builder = term_dict.range();
term_stream_builder =
match &self.left_bound {
&Included(ref term_val) => term_stream_builder.ge(term_val),
&Excluded(ref term_val) => term_stream_builder.gt(term_val),
&Unbounded => term_stream_builder
};
term_stream_builder =
match &self.right_bound {
&Included(ref term_val) => term_stream_builder.le(term_val),
&Excluded(ref term_val) => term_stream_builder.lt(term_val),
&Unbounded => term_stream_builder
};
term_stream_builder.into_stream()
fn map_bound<TFrom, Transform: Fn(TFrom) -> Vec<u8>>(
bound: Bound<TFrom>,
transform: &Transform,
) -> Bound<Vec<u8>> {
use self::Bound::*;
match bound {
Excluded(from_val) => Excluded(transform(from_val)),
Included(from_val) => Included(transform(from_val)),
Unbounded => Unbounded,
}
}
/// `RangeQuery` match all documents that have at least one term within a defined range.
///
/// Matched document will all get a constant `Score` of one.
///
/// # Implementation
///
/// The current implement will iterate over the terms within the range
/// and append all of the document cross into a `BitSet`.
///
/// # Example
///
/// ```rust
///
/// # #[macro_use]
/// # extern crate tantivy;
/// # use tantivy::Index;
/// # use tantivy::schema::{SchemaBuilder, INT_INDEXED};
/// # use tantivy::collector::CountCollector;
/// # use tantivy::query::Query;
/// # use tantivy::Result;
/// # use tantivy::query::RangeQuery;
/// #
/// # fn run() -> Result<()> {
/// # let mut schema_builder = SchemaBuilder::new();
/// # let year_field = schema_builder.add_u64_field("year", INT_INDEXED);
/// # let schema = schema_builder.build();
/// #
/// # let index = Index::create_in_ram(schema);
/// # {
/// # let mut index_writer = index.writer_with_num_threads(1, 6_000_000).unwrap();
/// # for year in 1950u64..2017u64 {
/// # let num_docs_within_year = 10 + (year - 1950) * (year - 1950);
/// # for _ in 0..num_docs_within_year {
/// # index_writer.add_document(doc!(year_field => year));
/// # }
/// # }
/// # index_writer.commit().unwrap();
/// # }
/// # index.load_searchers()?;
/// let searcher = index.searcher();
///
/// let docs_in_the_sixties = RangeQuery::new_u64(year_field, 1960..1970);
///
/// // ... or `1960..=1969` if inclusive range is enabled.
/// let mut count_collector = CountCollector::default();
/// docs_in_the_sixties.search(&*searcher, &mut count_collector)?;
///
/// let num_60s_books = count_collector.count();
///
/// # assert_eq!(num_60s_books, 2285);
/// # Ok(())
/// # }
/// #
/// # fn main() {
/// # run().unwrap()
/// # }
/// ```
#[derive(Debug)]
pub struct RangeQuery {
range_definition: RangeDefinition
field: Field,
left_bound: Bound<Vec<u8>>,
right_bound: Bound<Vec<u8>>,
}
impl RangeQuery {
pub fn new(range_definition: RangeDefinition) -> RangeQuery {
/// Create a new `RangeQuery` over a `i64` field.
pub fn new_i64<TRangeArgument: RangeArgument<i64>>(
field: Field,
range: TRangeArgument,
) -> RangeQuery {
let make_term_val = |val: &i64| Term::from_field_i64(field, *val).value_bytes().to_owned();
RangeQuery {
range_definition
field,
left_bound: map_bound(range.start(), &make_term_val),
right_bound: map_bound(range.end(), &make_term_val),
}
}
/// Create a new `RangeQuery` over a `u64` field.
pub fn new_u64<TRangeArgument: RangeArgument<u64>>(
field: Field,
range: TRangeArgument,
) -> RangeQuery {
let make_term_val = |val: &u64| Term::from_field_u64(field, *val).value_bytes().to_owned();
RangeQuery {
field,
left_bound: map_bound(range.start(), &make_term_val),
right_bound: map_bound(range.end(), &make_term_val),
}
}
/// Create a new `RangeQuery` over a `Str` field.
pub fn new_str<'b, TRangeArgument: RangeArgument<&'b str>>(
field: Field,
range: TRangeArgument,
) -> RangeQuery {
let make_term_val = |val: &&str| val.as_bytes().to_vec();
RangeQuery {
field,
left_bound: map_bound(range.start(), &make_term_val),
right_bound: map_bound(range.end(), &make_term_val),
}
}
}
@@ -97,27 +135,52 @@ impl Query for RangeQuery {
fn weight(&self, _searcher: &Searcher) -> Result<Box<Weight>> {
Ok(box RangeWeight {
range_definition: self.range_definition.clone()
field: self.field,
left_bound: self.left_bound.clone(),
right_bound: self.right_bound.clone(),
})
}
}
pub struct RangeWeight {
range_definition: RangeDefinition
field: Field,
left_bound: Bound<Vec<u8>>,
right_bound: Bound<Vec<u8>>,
}
impl RangeWeight {
fn term_range<'a, T>(&self, term_dict: &'a T) -> T::Streamer
where
T: TermDictionary<'a> + 'a,
{
use std::collections::Bound::*;
let mut term_stream_builder = term_dict.range();
term_stream_builder = match &self.left_bound {
&Included(ref term_val) => term_stream_builder.ge(term_val),
&Excluded(ref term_val) => term_stream_builder.gt(term_val),
&Unbounded => term_stream_builder,
};
term_stream_builder = match &self.right_bound {
&Included(ref term_val) => term_stream_builder.le(term_val),
&Excluded(ref term_val) => term_stream_builder.lt(term_val),
&Unbounded => term_stream_builder,
};
term_stream_builder.into_stream()
}
}
impl Weight for RangeWeight {
fn scorer<'a>(&'a self, reader: &'a SegmentReader) -> Result<Box<Scorer + 'a>> {
let max_doc = reader.max_doc();
let mut doc_bitset = DocBitSet::with_maxdoc(max_doc);
let mut doc_bitset = BitSet::with_max_value(max_doc);
let inverted_index = reader.inverted_index(self.range_definition.field);
let inverted_index = reader.inverted_index(self.field);
let term_dict = inverted_index.terms();
let mut term_range = self.range_definition.term_range(term_dict);
let mut term_range = self.term_range(term_dict);
while term_range.advance() {
let term_info = term_range.value();
let mut block_segment_postings = inverted_index.read_block_postings_from_terminfo(term_info,IndexRecordOption::Basic);
let mut block_segment_postings = inverted_index
.read_block_postings_from_terminfo(term_info, IndexRecordOption::Basic);
while block_segment_postings.advance() {
for &doc in block_segment_postings.docs() {
doc_bitset.insert(doc);
@@ -133,8 +196,45 @@ impl Weight for RangeWeight {
mod tests {
use Index;
use schema::{SchemaBuilder, Field, Document, INT_INDEXED};
use schema::{Document, Field, SchemaBuilder, INT_INDEXED};
use collector::CountCollector;
use std::collections::Bound;
use query::Query;
use Result;
use super::RangeQuery;
#[test]
fn test_range_query_simple() {
fn run() -> Result<()> {
let mut schema_builder = SchemaBuilder::new();
let year_field = schema_builder.add_u64_field("year", INT_INDEXED);
let schema = schema_builder.build();
let index = Index::create_in_ram(schema);
{
let mut index_writer = index.writer_with_num_threads(1, 6_000_000).unwrap();
for year in 1950u64..2017u64 {
let num_docs_within_year = 10 + (year - 1950) * (year - 1950);
for _ in 0..num_docs_within_year {
index_writer.add_document(doc!(year_field => year));
}
}
index_writer.commit().unwrap();
}
index.load_searchers().unwrap();
let searcher = index.searcher();
let docs_in_the_sixties = RangeQuery::new_u64(year_field, 1960u64..1970u64);
// ... or `1960..=1969` if inclusive range is enabled.
let mut count_collector = CountCollector::default();
docs_in_the_sixties.search(&*searcher, &mut count_collector)?;
assert_eq!(count_collector.count(), 2285);
Ok(())
}
run().unwrap();
}
#[test]
fn test_range_query() {
@@ -163,43 +263,30 @@ mod tests {
}
index.load_searchers().unwrap();
let searcher = index.searcher();
use collector::CountCollector;
use schema::Term;
use query::Query;
use super::{RangeQuery, RangeDefinition};
let count_multiples = |range: RangeDefinition| {
let count_multiples = |range_query: RangeQuery| {
let mut count_collector = CountCollector::default();
let range_query = RangeQuery::new(range);
range_query.search(&*searcher, &mut count_collector).unwrap();
range_query
.search(&*searcher, &mut count_collector)
.unwrap();
count_collector.count()
};
assert_eq!(count_multiples(RangeQuery::new_i64(int_field, 10..11)), 9);
assert_eq!(
count_multiples(RangeDefinition::for_field(int_field)
.left_included(Term::from_field_i64(int_field, 10))
.right_excluded(Term::from_field_i64(int_field, 11)))
, 9
count_multiples(RangeQuery::new_i64(
int_field,
(Bound::Included(10), Bound::Included(11))
)),
18
);
assert_eq!(
count_multiples(RangeDefinition::for_field(int_field)
.left_included(Term::from_field_i64(int_field, 10))
.right_included(Term::from_field_i64(int_field, 11)))
, 18
count_multiples(RangeQuery::new_i64(
int_field,
(Bound::Excluded(9), Bound::Included(10))
)),
9
);
assert_eq!(
count_multiples(RangeDefinition::for_field(int_field)
.left_excluded(Term::from_field_i64(int_field, 9))
.right_included(Term::from_field_i64(int_field, 10)))
, 9
);
assert_eq!(
count_multiples(RangeDefinition::for_field(int_field)
.left_excluded(Term::from_field_i64(int_field, 9)))
, 90
);
assert_eq!(count_multiples(RangeQuery::new_i64(int_field, 9..)), 91);
}
}
}

19
src/query/scorer.rs
View File

@@ -3,7 +3,7 @@ use DocId;
use Score;
use collector::Collector;
use postings::SkipResult;
use common::DocBitSet;
use common::BitSet;
use std::ops::{Deref, DerefMut};
/// Scored set of documents matching a query within a specific segment.
@@ -62,19 +62,27 @@ impl Scorer for EmptyScorer {
}
}
/// Wraps a `DocSet` and simply returns a constant `Scorer`.
/// The `ConstScorer` is useful if you have a `DocSet` where
/// you needed a scorer.
///
/// The `ConstScorer`'s constant score can be set
/// by calling `.set_score(...)`.
pub struct ConstScorer<TDocSet: DocSet> {
docset: TDocSet,
score: Score
score: Score,
}
impl<TDocSet: DocSet> ConstScorer<TDocSet> {
/// Creates a new `ConstScorer`.
pub fn new(docset: TDocSet) -> ConstScorer<TDocSet> {
ConstScorer {
docset,
score: 1f32
score: 1f32,
}
}
/// Sets the constant score to a different value.
pub fn set_score(&mut self, score: Score) {
self.score = score;
}
@@ -101,12 +109,11 @@ impl<TDocSet: DocSet> DocSet for ConstScorer<TDocSet> {
self.docset.size_hint()
}
fn to_doc_bitset(&mut self, max_doc: DocId) -> DocBitSet {
self.docset.to_doc_bitset(max_doc)
fn append_to_bitset(&mut self, bitset: &mut BitSet) {
self.docset.append_to_bitset(bitset);
}
}
impl<TDocSet: DocSet> Scorer for ConstScorer<TDocSet> {
fn score(&self) -> Score {
1f32

15
src/schema/document.rs
View File

@@ -21,9 +21,7 @@ pub struct Document {
impl From<Vec<FieldValue>> for Document {
fn from(field_values: Vec<FieldValue>) -> Self {
Document {
field_values
}
Document { field_values }
}
}
@@ -38,7 +36,6 @@ impl PartialEq for Document {
}
}
impl Eq for Document {}
impl Document {
@@ -59,14 +56,16 @@ impl Document {
/// Retain only the field that are matching the
/// predicate given in argument.
pub fn filter_fields<P: Fn(Field)->bool>(&mut self, predicate: P) {
pub fn filter_fields<P: Fn(Field) -> bool>(&mut self, predicate: P) {
self.field_values
.retain(|field_value| predicate(field_value.field()));
}
/// Adding a facet to the document.
pub fn add_facet<F>(&mut self, field: Field, path: F)
where Facet: From<F> {
where
Facet: From<F>,
{
let facet = Facet::from(path);
let value = Value::Facet(facet);
self.add(FieldValue::new(field, value));
@@ -144,9 +143,7 @@ impl BinarySerializable for Document {
fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
let num_field_values = VInt::deserialize(reader)?.val() as usize;
let field_values = (0..num_field_values)
.map(|_| {
FieldValue::deserialize(reader)
})
.map(|_| FieldValue::deserialize(reader))
.collect::<io::Result<Vec<FieldValue>>>()?;
Ok(Document::from(field_values))
}

62
src/schema/facet.rs
View File

@@ -1,14 +1,14 @@
use std::fmt::{self, Display, Debug, Formatter};
use std::fmt::{self, Debug, Display, Formatter};
use std::str;
use std::io::{self, Read, Write};
use regex::Regex;
use std::borrow::Borrow;
use serde::{Deserialize, Deserializer, Serialize, Serializer};
use std::borrow::Cow;
use common::BinarySerializable;
const SLASH_BYTE: u8 = '/' as u8;
const ESCAPE_BYTE: u8 = '\\' as u8;
const SLASH_BYTE: u8 = b'/';
const ESCAPE_BYTE: u8 = b'\\';
/// BYTE used as a level separation in the binary
/// representation of facets.
@@ -29,7 +29,6 @@ pub const FACET_SEP_BYTE: u8 = 0u8;
pub struct Facet(Vec<u8>);
impl Facet {
/// Returns a new instance of the "root facet"
/// Equivalent to `/`.
pub fn root() -> Facet {
@@ -64,8 +63,10 @@ impl Facet {
/// It is conceptually, if one of the steps of this path
/// contains a `/` or a `\`, it should be escaped
/// using an anti-slash `/`.
pub fn from_text<'a, T>(path: &'a T) -> Facet
where T: ?Sized + AsRef<str> {
pub fn from_text<T>(path: &T) -> Facet
where
T: ?Sized + AsRef<str>,
{
From::from(path)
}
@@ -74,9 +75,10 @@ impl Facet {
///
/// The steps are expected to be unescaped.
pub fn from_path<Path>(path: Path) -> Facet
where
Path: IntoIterator,
Path::Item: ToString {
where
Path: IntoIterator,
Path::Item: ToString,
{
let mut facet_bytes: Vec<u8> = Vec::with_capacity(100);
let mut step_it = path.into_iter();
if let Some(step) = step_it.next() {
@@ -93,11 +95,28 @@ impl Facet {
pub(crate) fn inner_buffer_mut(&mut self) -> &mut Vec<u8> {
&mut self.0
}
/// Returns `true` iff other is a subfacet of `self`.
#[allow(collapsible_if)]
pub fn is_prefix_of(&self, other: &Facet) -> bool {
let self_bytes: &[u8] = self.encoded_bytes();
let other_bytes: &[u8] = other.encoded_bytes();
if self_bytes.len() < other_bytes.len() {
if other_bytes.starts_with(self_bytes) {
return other_bytes[self_bytes.len()] == 0u8;
}
}
false
}
}
impl Borrow<[u8]> for Facet {
fn borrow(&self) -> &[u8] {
self.encoded_bytes()
}
}
impl<'a, T: ?Sized + AsRef<str>> From<&'a T> for Facet {
fn from(path_asref: &'a T) -> Facet {
#[derive(Copy, Clone)]
enum State {
@@ -110,9 +129,7 @@ impl<'a, T: ?Sized + AsRef<str>> From<&'a T> for Facet {
let path_bytes = path.as_bytes();
for &c in &path_bytes[1..] {
match (state, c) {
(State::Idle, ESCAPE_BYTE) => {
state = State::Escaped
}
(State::Idle, ESCAPE_BYTE) => state = State::Escaped,
(State::Idle, SLASH_BYTE) => {
facet_encoded.push(FACET_SEP_BYTE);
}
@@ -160,16 +177,19 @@ fn escape_slashes(s: &str) -> Cow<str> {
impl Serialize for Facet {
fn serialize<S>(&self, serializer: S) -> Result<S::Ok, S::Error>
where S: Serializer {
where
S: Serializer,
{
serializer.serialize_str(&self.to_string())
}
}
impl<'de> Deserialize<'de> for Facet {
fn deserialize<D>(deserializer: D) -> Result<Self, D::Error> where
D: Deserializer<'de> {
<&'de str as Deserialize<'de>>::deserialize(deserializer)
.map(Facet::from)
fn deserialize<D>(deserializer: D) -> Result<Self, D::Error>
where
D: Deserializer<'de>,
{
<&'de str as Deserialize<'de>>::deserialize(deserializer).map(Facet::from)
}
}
@@ -180,7 +200,6 @@ impl Debug for Facet {
}
}
#[cfg(test)]
mod tests {
@@ -207,7 +226,6 @@ mod tests {
}
}
#[test]
fn test_facet_debug() {
let v = ["first", "second", "third"];
@@ -215,4 +233,4 @@ mod tests {
assert_eq!(format!("{:?}", facet), "Facet(/first/second/third)");
}
}
}

2
src/schema/field_entry.rs
View File

@@ -71,7 +71,7 @@ impl FieldEntry {
match self.field_type {
FieldType::Str(ref options) => options.get_indexing_options().is_some(),
FieldType::U64(ref options) | FieldType::I64(ref options) => options.is_indexed(),
FieldType::HierarchicalFacet => true
FieldType::HierarchicalFacet => true,
}
}

8
src/schema/field_type.rs
View File

@@ -39,7 +39,7 @@ impl FieldType {
FieldType::U64(ref int_options) | FieldType::I64(ref int_options) => {
int_options.is_indexed()
}
FieldType::HierarchicalFacet => true
FieldType::HierarchicalFacet => true,
}
}
@@ -59,7 +59,7 @@ impl FieldType {
None
}
}
FieldType::HierarchicalFacet => Some(IndexRecordOption::Basic)
FieldType::HierarchicalFacet => Some(IndexRecordOption::Basic),
}
}
@@ -75,9 +75,7 @@ impl FieldType {
FieldType::U64(_) | FieldType::I64(_) => Err(ValueParsingError::TypeError(
format!("Expected an integer, got {:?}", json),
)),
FieldType::HierarchicalFacet => {
Ok(Value::Facet(Facet::from(field_text)))
}
FieldType::HierarchicalFacet => Ok(Value::Facet(Facet::from(field_text))),
},
JsonValue::Number(ref field_val_num) => match *self {
FieldType::I64(_) => {

7
src/schema/int_options.rs
View File

@@ -1,6 +1,5 @@
use std::ops::BitOr;
/// Express whether a field is single-value or multi-valued.
#[derive(Clone, Copy, PartialEq, Eq, Debug, Serialize, Deserialize)]
pub enum Cardinality {
@@ -10,14 +9,14 @@ pub enum Cardinality {
/// The document can have any number of values associated to the document.
/// This is more memory and CPU expensive than the SingleValue solution.
#[serde(rename = "multi")]
MultiValues
MultiValues,
}
/// Define how an int field should be handled by tantivy.
#[derive(Clone, Debug, PartialEq, Eq, Serialize, Deserialize)]
pub struct IntOptions {
indexed: bool,
#[serde(skip_serializing_if="Option::is_none")]
#[serde(skip_serializing_if = "Option::is_none")]
fast: Option<Cardinality>,
stored: bool,
}
@@ -86,7 +85,6 @@ impl Default for IntOptions {
}
}
/// Shortcut for a u64 fast field.
///
/// Such a shortcut can be composed as follows `STORED | FAST | INT_INDEXED`
@@ -114,7 +112,6 @@ pub const INT_STORED: IntOptions = IntOptions {
fast: None,
};
impl BitOr for IntOptions {
type Output = IntOptions;

20
src/schema/schema.rs
View File

@@ -334,8 +334,12 @@ mod tests {
#[test]
pub fn test_schema_serialization() {
let mut schema_builder = SchemaBuilder::default();
let count_options = IntOptions::default().set_stored().set_fast(Cardinality::SingleValue);
let popularity_options = IntOptions::default().set_stored().set_fast(Cardinality::SingleValue);
let count_options = IntOptions::default()
.set_stored()
.set_fast(Cardinality::SingleValue);
let popularity_options = IntOptions::default()
.set_stored()
.set_fast(Cardinality::SingleValue);
schema_builder.add_text_field("title", TEXT);
schema_builder.add_text_field("author", STRING);
schema_builder.add_u64_field("count", count_options);
@@ -399,7 +403,9 @@ mod tests {
#[test]
pub fn test_document_to_json() {
let mut schema_builder = SchemaBuilder::default();
let count_options = IntOptions::default().set_stored().set_fast(Cardinality::SingleValue);
let count_options = IntOptions::default()
.set_stored()
.set_fast(Cardinality::SingleValue);
schema_builder.add_text_field("title", TEXT);
schema_builder.add_text_field("author", STRING);
schema_builder.add_u64_field("count", count_options);
@@ -418,8 +424,12 @@ mod tests {
#[test]
pub fn test_parse_document() {
let mut schema_builder = SchemaBuilder::default();
let count_options = IntOptions::default().set_stored().set_fast(Cardinality::SingleValue);
let popularity_options = IntOptions::default().set_stored().set_fast(Cardinality::SingleValue);
let count_options = IntOptions::default()
.set_stored()
.set_fast(Cardinality::SingleValue);
let popularity_options = IntOptions::default()
.set_stored()
.set_fast(Cardinality::SingleValue);
let title_field = schema_builder.add_text_field("title", TEXT);
let author_field = schema_builder.add_text_field("author", STRING);
let count_field = schema_builder.add_u64_field("count", count_options);

2
src/schema/text_options.rs
View File

@@ -44,7 +44,7 @@ impl Default for TextOptions {
/// Configuration defining indexing for a text field.
/// It wraps:
///
///
/// * record (See [`IndexRecordOption`](./enum.IndexRecordOption.html))
/// * tokenizer
#[derive(Clone, PartialEq, Eq, Debug, Serialize, Deserialize)]

6
src/schema/value.rs
View File

@@ -26,7 +26,7 @@ impl Serialize for Value {
Value::Str(ref v) => serializer.serialize_str(v),
Value::U64(u) => serializer.serialize_u64(u),
Value::I64(u) => serializer.serialize_i64(u),
Value::Facet(ref facet) => facet.serialize(serializer)
Value::Facet(ref facet) => facet.serialize(serializer),
}
}
}
@@ -178,9 +178,7 @@ mod binary_serialize {
let value = i64::deserialize(reader)?;
Ok(Value::I64(value))
}
HIERARCHICAL_FACET_CODE => {
Ok(Value::Facet(Facet::deserialize(reader)?))
}
HIERARCHICAL_FACET_CODE => Ok(Value::Facet(Facet::deserialize(reader)?)),
_ => Err(io::Error::new(
io::ErrorKind::InvalidData,
format!("No field type is associated with code {:?}", type_code),

3
src/store/reader.rs
View File

@@ -40,7 +40,8 @@ impl StoreReader {
fn block_offset(&self, doc_id: DocId) -> (DocId, u64) {
self.block_index()
.seek(doc_id + 1)
.seek(doc_id as u64 + 1)
.map(|(doc, offset)| (doc as DocId, offset))
.unwrap_or((0u32, 0u64))
}

21
src/store/writer.rs
View File

@@ -1,6 +1,6 @@
use directory::WritePtr;
use DocId;
use common::{VInt, BinarySerializable};
use common::{BinarySerializable, VInt};
use std::io::{self, Write};
use super::StoreReader;
use lz4;
@@ -34,7 +34,7 @@ impl StoreWriter {
pub fn new(writer: WritePtr) -> StoreWriter {
StoreWriter {
doc: 0,
offset_index_writer: SkipListBuilder::new(3),
offset_index_writer: SkipListBuilder::new(4),
writer: CountingWriter::wrap(writer),
intermediary_buffer: Vec::new(),
current_block: Vec::new(),
@@ -46,7 +46,7 @@ impl StoreWriter {
/// The document id is implicitely the number of times
/// this method has been called.
///
pub fn store<'a>(&mut self, stored_document: &Document) -> io::Result<()> {
pub fn store(&mut self, stored_document: &Document) -> io::Result<()> {
self.intermediary_buffer.clear();
stored_document.serialize(&mut self.intermediary_buffer)?;
let doc_num_bytes = self.intermediary_buffer.len();
@@ -66,10 +66,8 @@ impl StoreWriter {
pub fn stack(&mut self, store_reader: &StoreReader) -> io::Result<()> {
if !self.current_block.is_empty() {
self.write_and_compress_block()?;
self.offset_index_writer.insert(
self.doc,
&(self.writer.written_bytes() as u64),
)?;
self.offset_index_writer
.insert(self.doc as u64, &(self.writer.written_bytes() as u64))?;
}
let doc_offset = self.doc;
let start_offset = self.writer.written_bytes() as u64;
@@ -80,10 +78,9 @@ impl StoreWriter {
// concatenate the index of the `store_reader`, after translating
// its start doc id and its start file offset.
for (next_doc_id, block_addr) in store_reader.block_index() {
self.doc = doc_offset + next_doc_id;
self.offset_index_writer.insert(
self.doc,
&(start_offset + block_addr))?;
self.doc = doc_offset + next_doc_id as u32;
self.offset_index_writer
.insert(self.doc as u64, &(start_offset + block_addr))?;
}
Ok(())
}
@@ -99,7 +96,7 @@ impl StoreWriter {
(self.intermediary_buffer.len() as u32).serialize(&mut self.writer)?;
self.writer.write_all(&self.intermediary_buffer)?;
self.offset_index_writer
.insert(self.doc, &(self.writer.written_bytes() as u64))?;
.insert(self.doc as u64, &(self.writer.written_bytes() as u64))?;
self.current_block.clear();
Ok(())
}

2
src/termdict/fstdict/mod.rs
View File

@@ -16,8 +16,10 @@ Keys (`&[u8]`) in this datastructure are sorted.
mod termdict;
mod streamer;
mod term_info_store;
pub use self::termdict::TermDictionaryImpl;
pub use self::termdict::TermDictionaryBuilderImpl;
pub use self::term_info_store::{TermInfoStore, TermInfoStoreWriter};
pub use self::streamer::TermStreamerImpl;
pub use self::streamer::TermStreamerBuilderImpl;

6
src/termdict/fstdict/streamer.rs
View File

@@ -1,8 +1,8 @@
use fst::{IntoStreamer, Streamer};
use fst::map::{StreamBuilder, Stream};
use fst::map::{Stream, StreamBuilder};
use postings::TermInfo;
use super::TermDictionaryImpl;
use termdict::{TermOrdinal, TermDictionary, TermStreamerBuilder, TermStreamer};
use termdict::{TermDictionary, TermOrdinal, TermStreamer, TermStreamerBuilder};
/// See [`TermStreamerBuilder`](./trait.TermStreamerBuilder.html)
pub struct TermStreamerBuilderImpl<'a> {
@@ -53,7 +53,6 @@ impl<'a> TermStreamerBuilder for TermStreamerBuilderImpl<'a> {
}
}
/// See [`TermStreamer`](./trait.TermStreamer.html)
pub struct TermStreamerImpl<'a> {
fst_map: &'a TermDictionaryImpl,
@@ -88,4 +87,3 @@ impl<'a> TermStreamer for TermStreamerImpl<'a> {
&self.current_value
}
}

330
src/termdict/fstdict/term_info_store.rs Normal file
View File

@@ -0,0 +1,330 @@
use std::io;
use std::cmp;
use std::io::{Read, Write};
use postings::TermInfo;
use common::{BinarySerializable, FixedSize};
use common::compute_num_bits;
use common::Endianness;
use common::bitpacker::BitPacker;
use directory::ReadOnlySource;
use termdict::TermOrdinal;
use byteorder::ByteOrder;
const BLOCK_LEN: usize = 256;
#[derive(Debug, Eq, PartialEq, Default)]
struct TermInfoBlockMeta {
offset: u64,
ref_term_info: TermInfo,
doc_freq_nbits: u8,
postings_offset_nbits: u8,
positions_offset_nbits: u8,
}
impl BinarySerializable for TermInfoBlockMeta {
fn serialize<W: Write>(&self, write: &mut W) -> io::Result<()> {
self.offset.serialize(write)?;
self.ref_term_info.serialize(write)?;
write.write_all(&[
self.doc_freq_nbits,
self.postings_offset_nbits,
self.positions_offset_nbits,
])?;
Ok(())
}
fn deserialize<R: Read>(reader: &mut R) -> io::Result<Self> {
let offset = u64::deserialize(reader)?;
let ref_term_info = TermInfo::deserialize(reader)?;
let mut buffer = [0u8; 3];
reader.read_exact(&mut buffer)?;
Ok(TermInfoBlockMeta {
offset,
ref_term_info,
doc_freq_nbits: buffer[0],
postings_offset_nbits: buffer[1],
positions_offset_nbits: buffer[2],
})
}
}
impl FixedSize for TermInfoBlockMeta {
const SIZE_IN_BYTES: usize =
u64::SIZE_IN_BYTES + TermInfo::SIZE_IN_BYTES + 3 * u8::SIZE_IN_BYTES;
}
impl TermInfoBlockMeta {
fn num_bits(&self) -> u8 {
self.doc_freq_nbits + self.postings_offset_nbits + self.positions_offset_nbits + 7
}
fn deserialize_term_info(&self, data: &[u8], inner_offset: usize) -> TermInfo {
let num_bits = self.num_bits() as usize;
let mut cursor = num_bits * inner_offset;
let doc_freq = extract_bits(data, cursor, self.doc_freq_nbits) as u32;
cursor += self.doc_freq_nbits as usize;
let postings_offset = extract_bits(data, cursor, self.postings_offset_nbits);
cursor += self.postings_offset_nbits as usize;
let positions_offset = extract_bits(data, cursor, self.positions_offset_nbits);
cursor += self.positions_offset_nbits as usize;
let positions_inner_offset = extract_bits(data, cursor, 7) as u8;
TermInfo {
doc_freq,
postings_offset: postings_offset + self.ref_term_info.postings_offset,
positions_offset: positions_offset + self.ref_term_info.positions_offset,
positions_inner_offset,
}
}
}
pub struct TermInfoStore {
num_terms: usize,
block_meta_source: ReadOnlySource,
term_info_source: ReadOnlySource,
}
fn extract_bits(data: &[u8], addr_bits: usize, num_bits: u8) -> u64 {
assert!(num_bits <= 56);
let addr_byte = addr_bits / 8;
let bit_shift = (addr_bits % 8) as u64;
let val_unshifted_unmasked: u64 = unsafe { *(data[addr_byte..].as_ptr() as *const u64) };
let val_shifted_unmasked = val_unshifted_unmasked >> bit_shift;
let mask = (1u64 << (num_bits as u64)) - 1;
val_shifted_unmasked & mask
}
impl TermInfoStore {
pub fn open(data: ReadOnlySource) -> TermInfoStore {
let buffer = data.as_slice();
let len = Endianness::read_u64(&buffer[0..8]) as usize;
let num_terms = Endianness::read_u64(&buffer[8..16]) as usize;
let block_meta_source = data.slice(16, 16 + len);
let term_info_source = data.slice_from(16 + len);
TermInfoStore {
num_terms,
block_meta_source,
term_info_source,
}
}
pub fn get(&self, term_ord: TermOrdinal) -> TermInfo {
let block_id = (term_ord as usize) / BLOCK_LEN;
let buffer = self.block_meta_source.as_slice();
let mut block_data: &[u8] = &buffer[block_id * TermInfoBlockMeta::SIZE_IN_BYTES..];
let term_info_block_data = TermInfoBlockMeta::deserialize(&mut block_data)
.expect("Failed to deserialize terminfoblockmeta");
let inner_offset = (term_ord as usize) % BLOCK_LEN;
if inner_offset == 0 {
term_info_block_data.ref_term_info
} else {
let term_info_data = self.term_info_source.as_slice();
term_info_block_data.deserialize_term_info(
&term_info_data[term_info_block_data.offset as usize..],
inner_offset - 1,
)
}
}
pub fn num_terms(&self) -> usize {
self.num_terms
}
}
pub struct TermInfoStoreWriter {
buffer_block_metas: Vec<u8>,
buffer_term_infos: Vec<u8>,
term_infos: Vec<TermInfo>,
num_terms: u64,
}
fn bitpack_serialize<W: Write>(
write: &mut W,
bit_packer: &mut BitPacker,
term_info_block_meta: &TermInfoBlockMeta,
term_info: &TermInfo,
) -> io::Result<()> {
bit_packer.write(
term_info.doc_freq as u64,
term_info_block_meta.doc_freq_nbits,
write,
)?;
bit_packer.write(
term_info.postings_offset,
term_info_block_meta.postings_offset_nbits,
write,
)?;
bit_packer.write(
term_info.positions_offset,
term_info_block_meta.positions_offset_nbits,
write,
)?;
bit_packer.write(term_info.positions_inner_offset as u64, 7, write)?;
Ok(())
}
impl TermInfoStoreWriter {
pub fn new() -> TermInfoStoreWriter {
TermInfoStoreWriter {
buffer_block_metas: Vec::new(),
buffer_term_infos: Vec::new(),
term_infos: Vec::with_capacity(BLOCK_LEN),
num_terms: 0u64,
}
}
fn flush_block(&mut self) -> io::Result<()> {
if self.term_infos.is_empty() {
return Ok(());
}
let mut bit_packer = BitPacker::new();
let ref_term_info = self.term_infos[0].clone();
for term_info in &mut self.term_infos[1..] {
term_info.postings_offset -= ref_term_info.postings_offset;
term_info.positions_offset -= ref_term_info.positions_offset;
}
let mut max_doc_freq: u32 = 0u32;
let mut max_postings_offset: u64 = 0u64;
let mut max_positions_offset: u64 = 0u64;
for term_info in &self.term_infos[1..] {
max_doc_freq = cmp::max(max_doc_freq, term_info.doc_freq);
max_postings_offset = cmp::max(max_postings_offset, term_info.postings_offset);
max_positions_offset = cmp::max(max_positions_offset, term_info.positions_offset);
}
let max_doc_freq_nbits: u8 = compute_num_bits(max_doc_freq as u64);
let max_postings_offset_nbits = compute_num_bits(max_postings_offset);
let max_positions_offset_nbits = compute_num_bits(max_positions_offset);
let term_info_block_meta = TermInfoBlockMeta {
offset: self.buffer_term_infos.len() as u64,
ref_term_info,
doc_freq_nbits: max_doc_freq_nbits,
postings_offset_nbits: max_postings_offset_nbits,
positions_offset_nbits: max_positions_offset_nbits,
};
term_info_block_meta.serialize(&mut self.buffer_block_metas)?;
for term_info in self.term_infos[1..].iter().cloned() {
bitpack_serialize(
&mut self.buffer_term_infos,
&mut bit_packer,
&term_info_block_meta,
&term_info,
)?;
}
// Block need end up at the end of a byte.
bit_packer.flush(&mut self.buffer_term_infos)?;
self.term_infos.clear();
Ok(())
}
pub fn write_term_info(&mut self, term_info: &TermInfo) -> io::Result<()> {
self.num_terms += 1u64;
self.term_infos.push(term_info.clone());
if self.term_infos.len() >= BLOCK_LEN {
self.flush_block()?;
}
Ok(())
}
pub fn serialize<W: io::Write>(&mut self, write: &mut W) -> io::Result<()> {
if !self.term_infos.is_empty() {
self.flush_block()?;
}
let len = self.buffer_block_metas.len() as u64;
len.serialize(write)?;
self.num_terms.serialize(write)?;
write.write_all(&self.buffer_block_metas)?;
write.write_all(&self.buffer_term_infos)?;
write.write_all(&[0u8; 7])?;
Ok(())
}
}
#[cfg(test)]
mod tests {
use super::extract_bits;
use common::bitpacker::BitPacker;
use common::BinarySerializable;
use super::TermInfoBlockMeta;
use super::{TermInfoStore, TermInfoStoreWriter};
use directory::ReadOnlySource;
use postings::TermInfo;
use common::compute_num_bits;
use common;
#[test]
fn test_term_info_block() {
common::test::fixed_size_test::<TermInfoBlockMeta>();
}
#[test]
fn test_bitpacked() {
let mut buffer = Vec::new();
let mut bitpack = BitPacker::new();
bitpack.write(321u64, 9, &mut buffer).unwrap();
assert_eq!(compute_num_bits(321u64), 9);
bitpack.write(2u64, 2, &mut buffer).unwrap();
assert_eq!(compute_num_bits(2u64), 2);
bitpack.write(51, 6, &mut buffer).unwrap();
assert_eq!(compute_num_bits(51), 6);
bitpack.close(&mut buffer).unwrap();
assert_eq!(buffer.len(), 3 + 7);
assert_eq!(extract_bits(&buffer[..], 0, 9), 321u64);
assert_eq!(extract_bits(&buffer[..], 9, 2), 2u64);
assert_eq!(extract_bits(&buffer[..], 11, 6), 51u64);
}
#[test]
fn test_term_info_block_meta_serialization() {
let term_info_block_meta = TermInfoBlockMeta {
offset: 2009,
ref_term_info: TermInfo {
doc_freq: 512,
postings_offset: 51,
positions_offset: 3584,
positions_inner_offset: 0,
},
doc_freq_nbits: 10,
postings_offset_nbits: 5,
positions_offset_nbits: 11,
};
let mut buffer: Vec<u8> = Vec::new();
term_info_block_meta.serialize(&mut buffer).unwrap();
let mut cursor: &[u8] = &buffer[..];
let term_info_block_meta_serde = TermInfoBlockMeta::deserialize(&mut cursor).unwrap();
assert_eq!(term_info_block_meta_serde, term_info_block_meta);
}
#[test]
fn test_pack() {
let mut store_writer = TermInfoStoreWriter::new();
let mut term_infos = vec![];
for i in 0..1000 {
let term_info = TermInfo {
doc_freq: i as u32,
postings_offset: (i / 10) as u64,
positions_offset: (i * 7) as u64,
positions_inner_offset: (i % 128) as u8,
};
store_writer.write_term_info(&term_info).unwrap();
term_infos.push(term_info);
}
let mut buffer = Vec::new();
store_writer.serialize(&mut buffer).unwrap();
let term_info_store = TermInfoStore::open(ReadOnlySource::from(buffer));
for i in 0..1000 {
assert_eq!(term_info_store.get(i as u64), term_infos[i]);
}
}
}

70
src/termdict/fstdict/termdict.rs
View File

@@ -3,10 +3,11 @@ use fst;
use fst::raw::Fst;
use directory::ReadOnlySource;
use common::BinarySerializable;
use common::CountingWriter;
use schema::FieldType;
use postings::TermInfo;
use termdict::{TermDictionary, TermDictionaryBuilder, TermOrdinal};
use super::{TermStreamerImpl, TermStreamerBuilderImpl};
use super::{TermInfoStore, TermInfoStoreWriter, TermStreamerBuilderImpl, TermStreamerImpl};
fn convert_fst_error(e: fst::Error) -> io::Error {
io::Error::new(io::ErrorKind::Other, e)
@@ -15,7 +16,7 @@ fn convert_fst_error(e: fst::Error) -> io::Error {
/// See [`TermDictionaryBuilder`](./trait.TermDictionaryBuilder.html)
pub struct TermDictionaryBuilderImpl<W> {
fst_builder: fst::MapBuilder<W>,
data: Vec<u8>,
term_info_store_writer: TermInfoStoreWriter,
term_ord: u64,
}
@@ -41,8 +42,8 @@ where
/// # Warning
///
/// Horribly dangerous internal API. See `.insert_key(...)`.
pub(crate) fn insert_value(&mut self, value: &TermInfo) -> io::Result<()> {
value.serialize(&mut self.data)?;
pub(crate) fn insert_value(&mut self, term_info: &TermInfo) -> io::Result<()> {
self.term_info_store_writer.write_term_info(term_info)?;
Ok(())
}
}
@@ -54,25 +55,28 @@ where
fn new(w: W, _field_type: FieldType) -> io::Result<Self> {
let fst_builder = fst::MapBuilder::new(w).map_err(convert_fst_error)?;
Ok(TermDictionaryBuilderImpl {
fst_builder: fst_builder,
data: Vec::new(),
fst_builder,
term_info_store_writer: TermInfoStoreWriter::new(),
term_ord: 0,
})
}
fn insert<K: AsRef<[u8]>>(&mut self, key_ref: K, value: &TermInfo) -> io::Result<()> {
let key = key_ref.as_ref();
self.insert_key(key.as_ref())?;
self.insert_key(key)?;
self.insert_value(value)?;
Ok(())
}
fn finish(self) -> io::Result<W> {
fn finish(mut self) -> io::Result<W> {
let mut file = self.fst_builder.into_inner().map_err(convert_fst_error)?;
let footer_size = self.data.len() as u32;
file.write_all(&self.data)?;
(footer_size as u32).serialize(&mut file)?;
file.flush()?;
{
let mut counting_writer = CountingWriter::wrap(&mut file);
self.term_info_store_writer.serialize(&mut counting_writer)?;
let footer_size = counting_writer.written_bytes();
(footer_size as u64).serialize(&mut counting_writer)?;
counting_writer.flush()?;
}
Ok(file)
}
}
@@ -92,10 +96,9 @@ fn open_fst_index(source: ReadOnlySource) -> fst::Map {
/// See [`TermDictionary`](./trait.TermDictionary.html)
pub struct TermDictionaryImpl {
fst_index: fst::Map,
values_mmap: ReadOnlySource,
term_info_store: TermInfoStore,
}
impl<'a> TermDictionary<'a> for TermDictionaryImpl {
type Streamer = TermStreamerImpl<'a>;
@@ -103,23 +106,26 @@ impl<'a> TermDictionary<'a> for TermDictionaryImpl {
fn from_source(source: ReadOnlySource) -> Self {
let total_len = source.len();
let length_offset = total_len - 4;
let length_offset = total_len - 8;
let mut split_len_buffer: &[u8] = &source.as_slice()[length_offset..];
let footer_size = u32::deserialize(&mut split_len_buffer).expect(
"Deserializing 4 bytes should always work",
) as usize;
let footer_size = u64::deserialize(&mut split_len_buffer)
.expect("Deserializing 8 bytes should always work") as usize;
let split_len = length_offset - footer_size;
let fst_source = source.slice(0, split_len);
let values_source = source.slice(split_len, length_offset);
let fst_index = open_fst_index(fst_source);
TermDictionaryImpl {
fst_index: fst_index,
values_mmap: values_source,
fst_index,
term_info_store: TermInfoStore::open(values_source),
}
}
fn num_terms(&self) -> usize {
self.values_mmap.len() / TermInfo::SIZE_IN_BYTES
self.term_info_store.num_terms()
}
fn term_ord<K: AsRef<[u8]>>(&self, key: K) -> Option<TermOrdinal> {
self.fst_index.get(key)
}
fn ord_to_term(&self, mut ord: TermOrdinal, bytes: &mut Vec<u8>) -> bool {
@@ -128,32 +134,22 @@ impl<'a> TermDictionary<'a> for TermDictionaryImpl {
let mut node = fst.root();
while ord != 0 || !node.is_final() {
if let Some(transition) = node.transitions()
.take_while(|transition| {
transition.out.value() <= ord
})
.last() {
.take_while(|transition| transition.out.value() <= ord)
.last()
{
ord -= transition.out.value();
bytes.push(transition.inp);
let new_node_addr = transition.addr;
node = fst.node(new_node_addr);
}
else {
} else {
return false;
}
}
return true;
}
fn term_ord<K: AsRef<[u8]>>(&self, key: K) -> Option<TermOrdinal> {
self.fst_index.get(key)
true
}
fn term_info_from_ord(&self, term_ord: TermOrdinal) -> TermInfo {
let buffer = self.values_mmap.as_slice();
let offset = term_ord as usize * TermInfo::SIZE_IN_BYTES;
let mut cursor = &buffer[offset..];
TermInfo::deserialize(&mut cursor)
.expect("The fst is corrupted. Failed to deserialize a value.")
self.term_info_store.get(term_ord)
}
fn get<K: AsRef<[u8]>>(&self, key: K) -> Option<TermInfo> {

79
src/termdict/mod.rs
View File

@@ -5,7 +5,6 @@ that serves as an address in their respective posting list.
The term dictionary API makes it possible to iterate through
a range of keys in a sorted manner.
```
# Implementations
@@ -48,34 +47,30 @@ followed by a streaming through at most `1024` elements in the
term `stream`.
*/
use schema::{Term, Field, FieldType};
use schema::{Field, FieldType, Term};
use directory::ReadOnlySource;
use postings::TermInfo;
/// Position of the term in the sorted list of terms.
pub type TermOrdinal = u64;
pub use self::merger::TermMerger;
#[cfg(not(feature = "streamdict"))]
mod fstdict;
#[cfg(not(feature = "streamdict"))]
pub use self::fstdict::{TermDictionaryImpl, TermDictionaryBuilderImpl, TermStreamerImpl,
TermStreamerBuilderImpl};
pub use self::fstdict::{TermDictionaryBuilderImpl, TermDictionaryImpl, TermStreamerBuilderImpl,
TermStreamerImpl};
#[cfg(feature = "streamdict")]
mod streamdict;
#[cfg(feature = "streamdict")]
pub use self::streamdict::{TermDictionaryImpl, TermDictionaryBuilderImpl, TermStreamerImpl,
TermStreamerBuilderImpl};
pub use self::streamdict::{TermDictionaryBuilderImpl, TermDictionaryImpl, TermStreamerBuilderImpl,
TermStreamerImpl};
mod merger;
use std::io;
/// Dictionary associating sorted `&[u8]` to values
pub trait TermDictionary<'a>
where
@@ -90,6 +85,10 @@ where
/// Opens a `TermDictionary` given a data source.
fn from_source(source: ReadOnlySource) -> Self;
/// Returns the number of terms in the dictionary.
/// Term ordinals range from 0 to `num_terms() - 1`.
fn num_terms(&self) -> usize;
/// Returns the ordinal associated to a given term.
fn term_ord<K: AsRef<[u8]>>(&self, term: K) -> Option<TermOrdinal>;
@@ -107,10 +106,6 @@ where
/// Returns the number of terms in the dictionary.
fn term_info_from_ord(&self, term_ord: TermOrdinal) -> TermInfo;
/// Returns the number of terms in the dictionary.
/// Term ordinals range from 0 to `num_terms() - 1`.
fn num_terms(&self) -> usize;
/// Lookups the value corresponding to the key.
fn get<K: AsRef<[u8]>>(&self, target_key: K) -> Option<TermInfo>;
@@ -154,7 +149,6 @@ where
fn finish(self) -> io::Result<W>;
}
/// `TermStreamer` acts as a cursor over a range of terms of a segment.
/// Terms are guaranteed to be sorted.
pub trait TermStreamer: Sized {
@@ -202,7 +196,6 @@ pub trait TermStreamer: Sized {
}
}
/// `TermStreamerBuilder` is an helper object used to define
/// a range of terms that should be streamed.
pub trait TermStreamerBuilder {
@@ -226,13 +219,12 @@ pub trait TermStreamerBuilder {
fn into_stream(self) -> Self::Streamer;
}
#[cfg(test)]
mod tests {
use super::{TermDictionaryImpl, TermDictionaryBuilderImpl, TermStreamerImpl};
use directory::{RAMDirectory, Directory, ReadOnlySource};
use super::{TermDictionaryBuilderImpl, TermDictionaryImpl, TermStreamerImpl};
use directory::{Directory, RAMDirectory, ReadOnlySource};
use std::path::PathBuf;
use schema::{FieldType, SchemaBuilder, Document, TEXT};
use schema::{Document, FieldType, SchemaBuilder, TEXT};
use core::Index;
use std::str;
use termdict::TermStreamer;
@@ -243,17 +235,15 @@ mod tests {
const BLOCK_SIZE: usize = 1_500;
fn make_term_info(val: u64) -> TermInfo {
TermInfo {
doc_freq: val as u32,
positions_offset: val * 2u64,
positions_offset: val * 2u64,
postings_offset: val * 3u64,
positions_inner_offset: 5u8,
}
}
#[test]
fn test_term_ordinals() {
const COUNTRIES: [&'static str; 7] = [
@@ -263,15 +253,15 @@ mod tests {
"Slovenia",
"Spain",
"Sweden",
"Switzerland"
"Switzerland",
];
let mut directory = RAMDirectory::create();
let path = PathBuf::from("TermDictionary");
{
let write = directory.open_write(&path).unwrap();
let field_type = FieldType::Str(TEXT);
let mut term_dictionary_builder = TermDictionaryBuilderImpl::new(write, field_type)
.unwrap();
let mut term_dictionary_builder =
TermDictionaryBuilderImpl::new(write, field_type).unwrap();
for term in COUNTRIES.iter() {
term_dictionary_builder
.insert(term.as_bytes(), &make_term_info(0u64))
@@ -283,7 +273,7 @@ mod tests {
let term_dict: TermDictionaryImpl = TermDictionaryImpl::from_source(source);
for (term_ord, term) in COUNTRIES.iter().enumerate() {
assert_eq!(term_dict.term_ord(term).unwrap(), term_ord as u64);
let mut bytes = vec!();
let mut bytes = vec![];
assert!(term_dict.ord_to_term(term_ord as u64, &mut bytes));
assert_eq!(bytes, term.as_bytes());
}
@@ -296,8 +286,8 @@ mod tests {
{
let write = directory.open_write(&path).unwrap();
let field_type = FieldType::Str(TEXT);
let mut term_dictionary_builder = TermDictionaryBuilderImpl::new(write, field_type)
.unwrap();
let mut term_dictionary_builder =
TermDictionaryBuilderImpl::new(write, field_type).unwrap();
term_dictionary_builder
.insert("abc".as_bytes(), &make_term_info(34u64))
.unwrap();
@@ -377,7 +367,6 @@ mod tests {
assert_eq!(&*term_string, "abcdef");
}
#[test]
fn test_term_dictionary_stream() {
let ids: Vec<_> = (0u32..10_000u32)
@@ -385,8 +374,8 @@ mod tests {
.collect();
let field_type = FieldType::Str(TEXT);
let buffer: Vec<u8> = {
let mut term_dictionary_builder = TermDictionaryBuilderImpl::new(vec![], field_type)
.unwrap();
let mut term_dictionary_builder =
TermDictionaryBuilderImpl::new(vec![], field_type).unwrap();
for &(ref id, ref i) in &ids {
term_dictionary_builder
.insert(id.as_bytes(), &make_term_info(*i as u64))
@@ -411,13 +400,12 @@ mod tests {
term_dictionary.get(key.as_bytes());
}
#[test]
fn test_stream_high_range_prefix_suffix() {
let field_type = FieldType::Str(TEXT);
let buffer: Vec<u8> = {
let mut term_dictionary_builder = TermDictionaryBuilderImpl::new(vec![], field_type)
.unwrap();
let mut term_dictionary_builder =
TermDictionaryBuilderImpl::new(vec![], field_type).unwrap();
// term requires more than 16bits
term_dictionary_builder
.insert("abcdefghijklmnopqrstuvwxy", &make_term_info(1))
@@ -451,8 +439,8 @@ mod tests {
.collect();
let field_type = FieldType::Str(TEXT);
let buffer: Vec<u8> = {
let mut term_dictionary_builder = TermDictionaryBuilderImpl::new(vec![], field_type)
.unwrap();
let mut term_dictionary_builder =
TermDictionaryBuilderImpl::new(vec![], field_type).unwrap();
for &(ref id, ref i) in &ids {
term_dictionary_builder
.insert(id.as_bytes(), &make_term_info(*i as u64))
@@ -520,14 +508,15 @@ mod tests {
fn test_empty_string() {
let field_type = FieldType::Str(TEXT);
let buffer: Vec<u8> = {
let mut term_dictionary_builder = TermDictionaryBuilderImpl::new(vec![], field_type)
let mut term_dictionary_builder =
TermDictionaryBuilderImpl::new(vec![], field_type).unwrap();
term_dictionary_builder
.insert(&[], &make_term_info(1 as u64))
.unwrap();
term_dictionary_builder
.insert(&[], &make_term_info(1 as u64)).unwrap();
term_dictionary_builder
.insert(&[1u8], &make_term_info(2 as u64)).unwrap();
term_dictionary_builder
.finish().unwrap()
.insert(&[1u8], &make_term_info(2 as u64))
.unwrap();
term_dictionary_builder.finish().unwrap()
};
let source = ReadOnlySource::from(buffer);
let term_dictionary: TermDictionaryImpl = TermDictionaryImpl::from_source(source);
@@ -543,8 +532,8 @@ mod tests {
fn test_stream_range_boundaries() {
let field_type = FieldType::Str(TEXT);
let buffer: Vec<u8> = {
let mut term_dictionary_builder = TermDictionaryBuilderImpl::new(vec![], field_type)
.unwrap();
let mut term_dictionary_builder =
TermDictionaryBuilderImpl::new(vec![], field_type).unwrap();
for i in 0u8..10u8 {
let number_arr = [i; 1];
term_dictionary_builder

24
src/termdict/streamdict/delta_encoder.rs
View File

@@ -49,7 +49,6 @@ impl TermDeltaDecoder {
}
}
// code
// first bit represents whether the prefix / suffix len can be encoded
// on the same byte. (the next one)
@@ -57,18 +56,17 @@ impl TermDeltaDecoder {
#[inline(always)]
pub fn decode<'a>(&mut self, code: u8, mut cursor: &'a [u8]) -> &'a [u8] {
let (prefix_len, suffix_len): (usize, usize) =
if (code & 1u8) == 1u8 {
let b = cursor[0];
cursor = &cursor[1..];
let prefix_len = (b & 15u8) as usize;
let suffix_len = (b >> 4u8) as usize;
(prefix_len, suffix_len)
} else {
let prefix_len = u32::deserialize(&mut cursor).unwrap();
let suffix_len = u32::deserialize(&mut cursor).unwrap();
(prefix_len as usize, suffix_len as usize)
};
let (prefix_len, suffix_len): (usize, usize) = if (code & 1u8) == 1u8 {
let b = cursor[0];
cursor = &cursor[1..];
let prefix_len = (b & 15u8) as usize;
let suffix_len = (b >> 4u8) as usize;
(prefix_len, suffix_len)
} else {
let prefix_len = u32::deserialize(&mut cursor).unwrap();
let suffix_len = u32::deserialize(&mut cursor).unwrap();
(prefix_len as usize, suffix_len as usize)
};
unsafe { self.term.set_len(prefix_len) };
self.term.extend_from_slice(&(*cursor)[..suffix_len]);
&cursor[suffix_len..]

24
src/tokenizer/alphanum_only.rs
View File

@@ -6,32 +6,28 @@ use super::{Token, TokenFilter, TokenStream};
pub struct AlphaNumOnlyFilter;
pub struct AlphaNumOnlyFilterStream<TailTokenStream>
where TailTokenStream: TokenStream
where
TailTokenStream: TokenStream,
{
tail: TailTokenStream,
}
impl<TailTokenStream> AlphaNumOnlyFilterStream<TailTokenStream>
where TailTokenStream: TokenStream
where
TailTokenStream: TokenStream,
{
fn predicate(&self, token: &Token) -> bool {
token.text.chars().all(|c| c.is_ascii_alphanumeric())
}
fn wrap(
tail: TailTokenStream,
) -> AlphaNumOnlyFilterStream<TailTokenStream> {
AlphaNumOnlyFilterStream {
tail
}
fn wrap(tail: TailTokenStream) -> AlphaNumOnlyFilterStream<TailTokenStream> {
AlphaNumOnlyFilterStream { tail }
}
}
impl<TailTokenStream> TokenFilter<TailTokenStream> for AlphaNumOnlyFilter
where
TailTokenStream: TokenStream,
where
TailTokenStream: TokenStream,
{
type ResultTokenStream = AlphaNumOnlyFilterStream<TailTokenStream>;
@@ -41,8 +37,8 @@ impl<TailTokenStream> TokenFilter<TailTokenStream> for AlphaNumOnlyFilter
}
impl<TailTokenStream> TokenStream for AlphaNumOnlyFilterStream<TailTokenStream>
where
TailTokenStream: TokenStream
where
TailTokenStream: TokenStream,
{
fn token(&self) -> &Token {
self.tail.token()

36
src/tokenizer/facet_tokenizer.rs
View File

@@ -1,8 +1,7 @@
use super::{Token, Tokenizer, TokenStream};
use super::{Token, TokenStream, Tokenizer};
use std::str;
use schema::FACET_SEP_BYTE;
/// The `FacetTokenizer` process a `Facet` binary representation
/// and emits a token for all of its parent.
///
@@ -39,27 +38,27 @@ impl<'a> Tokenizer<'a> for FacetTokenizer {
}
}
impl<'a> TokenStream for FacetTokenStream<'a> {
fn advance(&mut self) -> bool {
match self.state {
State::RootFacetNotEmitted => {
self.state =
if self.text.is_empty() {
State::Terminated
} else {
State::UpToPosition(0)
};
self.state = if self.text.is_empty() {
State::Terminated
} else {
State::UpToPosition(0)
};
true
}
State::UpToPosition(cursor) => {
let bytes: &[u8] = self.text.as_bytes();
if let Some(next_sep_pos) = bytes[cursor+1..]
if let Some(next_sep_pos) = bytes[cursor + 1..]
.iter()
.cloned()
.position(|b| b == FACET_SEP_BYTE)
.map(|pos| cursor + 1 + pos) {
let facet_part = unsafe { str::from_utf8_unchecked(&bytes[cursor..next_sep_pos]) };
.map(|pos| cursor + 1 + pos)
{
let facet_part =
unsafe { str::from_utf8_unchecked(&bytes[cursor..next_sep_pos]) };
self.token.text.push_str(facet_part);
self.state = State::UpToPosition(next_sep_pos);
} else {
@@ -69,9 +68,7 @@ impl<'a> TokenStream for FacetTokenStream<'a> {
}
true
}
State::Terminated => {
false
}
State::Terminated => false,
}
}
@@ -87,9 +84,10 @@ impl<'a> TokenStream for FacetTokenStream<'a> {
#[cfg(test)]
mod tests {
use tokenizer::{TokenStream, Token, Tokenizer};
use tokenizer::{Token, TokenStream, Tokenizer};
use super::FacetTokenizer;
use schema::Facet;
use std::str;
#[test]
fn test_facet_tokenizer() {
@@ -101,7 +99,7 @@ mod tests {
tokens.push(format!("{}", facet));
};
FacetTokenizer
.token_stream(unsafe { ::std::str::from_utf8_unchecked(facet.encoded_bytes()) })
.token_stream(unsafe { str::from_utf8_unchecked(facet.encoded_bytes()) })
.process(&mut add_token);
}
assert_eq!(tokens.len(), 4);
@@ -121,10 +119,10 @@ mod tests {
tokens.push(format!("{}", facet));
};
FacetTokenizer
.token_stream(unsafe { ::std::str::from_utf8_unchecked(facet.encoded_bytes()) })
.token_stream(unsafe { str::from_utf8_unchecked(facet.encoded_bytes()) })
.process(&mut add_token);
}
assert_eq!(tokens.len(), 1);
assert_eq!(tokens[0], "/");
}
}
}

1
src/tokenizer/tokenizer.rs
View File

@@ -253,7 +253,6 @@ where
}
}
/// Trait for the pluggable components of `Tokenizer`s.
pub trait TokenFilter<TailTokenStream: TokenStream>: Clone {
/// The resulting `TokenStream` type.