Merge pull request #255 from RoaringBitmap/deserialize-run-containers

Deserialize Run Containers

README.md

@@ -49,7 +49,8 @@ Unless you explicitly state otherwise, any contribution intentionally submitted
 for inclusion in the work by you shall be dual licensed as above, without any
 additional terms or conditions.
 
-[github-actions-badge]: https://img.shields.io/github/workflow/status/RoaringBitmap/roaring-rs/Continuous%20integration.svg?style=flat-square
+[github-actions-badge]:
+https://github.com/RoaringBitmap/roaring-rs/actions/workflows/test.yml/badge.svg
 [github-actions]: https://github.com/RoaringBitmap/roaring-rs/actions
 [release-badge]: https://img.shields.io/github/release/RoaringBitmap/roaring-rs.svg?style=flat-square
 [cargo]: https://crates.io/crates/roaring

src/bitmap/container.rs

@@ -6,7 +6,7 @@ use std::ops::{
 use super::store::{self, Store};
 use super::util;
 
-const ARRAY_LIMIT: u64 = 4096;
+pub const ARRAY_LIMIT: u64 = 4096;
 
 #[derive(PartialEq, Clone)]
 pub struct Container {

src/bitmap/serialization.rs

@@ -3,15 +3,19 @@ use byteorder::{LittleEndian, ReadBytesExt, WriteBytesExt};
 use std::convert::{Infallible, TryFrom};
 use std::error::Error;
 use std::io;
+use std::ops::RangeInclusive;
 
-use super::container::Container;
-use crate::bitmap::store::{ArrayStore, BitmapStore, Store};
+use crate::bitmap::container::{Container, ARRAY_LIMIT};
+use crate::bitmap::store::{ArrayStore, BitmapStore, Store, BITMAP_LENGTH};
 use crate::RoaringBitmap;
 
 const SERIAL_COOKIE_NO_RUNCONTAINER: u32 = 12346;
 const SERIAL_COOKIE: u16 = 12347;
-// TODO: Need this once run containers are supported
-// const NO_OFFSET_THRESHOLD: u8 = 4;
+const NO_OFFSET_THRESHOLD: usize = 4;
+
+// Sizes of header structures
+const DESCRIPTION_BYTES: usize = 4;
+const OFFSET_BYTES: usize = 4;
 
 impl RoaringBitmap {
     /// Return the size in bytes of the serialized output.
@@ -163,49 +167,81 @@ impl RoaringBitmap {
         B: Fn(u64, Box<[u64; 1024]>) -> Result<BitmapStore, BErr>,
         BErr: Error + Send + Sync + 'static,
     {
-        let (size, has_offsets) = {
+        // First read the cookie to determine which version of the format we are reading
+        let (size, has_offsets, has_run_containers) = {
             let cookie = reader.read_u32::<LittleEndian>()?;
             if cookie == SERIAL_COOKIE_NO_RUNCONTAINER {
-                (reader.read_u32::<LittleEndian>()? as usize, true)
+                (reader.read_u32::<LittleEndian>()? as usize, true, false)
             } else if (cookie as u16) == SERIAL_COOKIE {
-                return Err(io::Error::new(io::ErrorKind::Other, "run containers are unsupported"));
+                let size = ((cookie >> 16) + 1) as usize;
+                (size, size >= NO_OFFSET_THRESHOLD, true)
             } else {
                 return Err(io::Error::new(io::ErrorKind::Other, "unknown cookie value"));
             }
         };
 
+        // Read the run container bitmap if necessary
+        let run_container_bitmap = if has_run_containers {
+            let mut bitmap = vec![0u8; (size + 7) / 8];
+            reader.read_exact(&mut bitmap)?;
+            Some(bitmap)
+        } else {
+            None
+        };
+
         if size > u16::MAX as usize + 1 {
             return Err(io::Error::new(io::ErrorKind::Other, "size is greater than supported"));
         }
 
-        let mut description_bytes = vec![0u8; size * 4];
+        // Read the container descriptions
+        let mut description_bytes = vec![0u8; size * DESCRIPTION_BYTES];
         reader.read_exact(&mut description_bytes)?;
         let mut description_bytes = &description_bytes[..];
 
         if has_offsets {
-            let mut offsets = vec![0u8; size * 4];
+            let mut offsets = vec![0u8; size * OFFSET_BYTES];
             reader.read_exact(&mut offsets)?;
             drop(offsets); // Not useful when deserializing into memory
         }
 
         let mut containers = Vec::with_capacity(size);
 
-        for _ in 0..size {
+        // Read each container
+        for i in 0..size {
             let key = description_bytes.read_u16::<LittleEndian>()?;
-            let len = u64::from(description_bytes.read_u16::<LittleEndian>()?) + 1;
+            let cardinality = u64::from(description_bytes.read_u16::<LittleEndian>()?) + 1;
+
+            // If the run container bitmap is present, check if this container is a run container
+            let is_run_container =
+                run_container_bitmap.as_ref().map_or(false, |bm| bm[i / 8] & (1 << (i % 8)) != 0);
+
+            let store = if is_run_container {
+                let runs = reader.read_u16::<LittleEndian>()?;
+                let mut intervals = vec![[0, 0]; runs as usize];
+                reader.read_exact(cast_slice_mut(&mut intervals))?;
+                intervals.iter_mut().for_each(|[s, len]| {
+                    *s = u16::from_le(*s);
+                    *len = u16::from_le(*len);
+                });
 
-            let store = if len <= 4096 {
-                let mut values = vec![0; len as usize];
+                let cardinality = intervals.iter().map(|[_, len]| *len as usize).sum();
+                let mut store = Store::with_capacity(cardinality);
+                intervals.into_iter().for_each(|[s, len]| {
+                    store.insert_range(RangeInclusive::new(s, s + len));
+                });
+                store
+            } else if cardinality <= ARRAY_LIMIT {
+                let mut values = vec![0; cardinality as usize];
                 reader.read_exact(cast_slice_mut(&mut values))?;
                 values.iter_mut().for_each(|n| *n = u16::from_le(*n));
                 let array = a(values).map_err(|e| io::Error::new(io::ErrorKind::InvalidData, e))?;
                 Store::Array(array)
             } else {
-                let mut values = Box::new([0; 1024]);
+                let mut values = Box::new([0; BITMAP_LENGTH]);
                 reader.read_exact(cast_slice_mut(&mut values[..]))?;
                 values.iter_mut().for_each(|n| *n = u64::from_le(*n));
-                let bitmap =
-                    b(len, values).map_err(|e| io::Error::new(io::ErrorKind::InvalidData, e))?;
+                let bitmap = b(cardinality, values)
+                    .map_err(|e| io::Error::new(io::ErrorKind::InvalidData, e))?;
                 Store::Bitmap(bitmap)
             };
 

src/bitmap/store/array_store/mod.rs

@@ -21,6 +21,10 @@ impl ArrayStore {
         ArrayStore { vec: vec![] }
     }
 
+    pub fn with_capacity(capacity: usize) -> ArrayStore {
+        ArrayStore { vec: Vec::with_capacity(capacity) }
+    }
+
     ///
     /// Create a new SortedU16Vec from a given vec
     /// It is up to the caller to ensure the vec is sorted and deduplicated

src/bitmap/store/mod.rs

@@ -7,12 +7,14 @@ use std::ops::{
 };
 use std::{slice, vec};
 
-use self::bitmap_store::BITMAP_LENGTH;
+pub use self::bitmap_store::BITMAP_LENGTH;
 use self::Store::{Array, Bitmap};
 
 pub use self::array_store::ArrayStore;
 pub use self::bitmap_store::{BitmapIter, BitmapStore};
 
+use crate::bitmap::container::ARRAY_LIMIT;
+
 #[derive(Clone)]
 pub enum Store {
     Array(ArrayStore),
@@ -31,6 +33,14 @@ impl Store {
         Store::Array(ArrayStore::new())
     }
 
+    pub fn with_capacity(capacity: usize) -> Store {
+        if capacity <= ARRAY_LIMIT as usize {
+            Store::Array(ArrayStore::with_capacity(capacity))
+        } else {
+            Store::Bitmap(BitmapStore::new())
+        }
+    }
+
     pub fn full() -> Store {
         Store::Bitmap(BitmapStore::full())
     }

tests/serialization.rs

@@ -4,6 +4,7 @@ use roaring::RoaringBitmap;
 
 // Test data from https://github.com/RoaringBitmap/RoaringFormatSpec/tree/master/testdata
 static BITMAP_WITHOUT_RUNS: &[u8] = include_bytes!("bitmapwithoutruns.bin");
+static BITMAP_WITH_RUNS: &[u8] = include_bytes!("bitmapwithruns.bin");
 
 fn test_data_bitmap() -> RoaringBitmap {
     (0..100)
@@ -21,10 +22,18 @@ fn serialize_and_deserialize(bitmap: &RoaringBitmap) -> RoaringBitmap {
 }
 
 #[test]
-fn test_deserialize_from_provided_data() {
+fn test_deserialize_without_runs_from_provided_data() {
     assert_eq!(RoaringBitmap::deserialize_from(BITMAP_WITHOUT_RUNS).unwrap(), test_data_bitmap());
 }
 
+#[test]
+fn test_deserialize_with_runs_from_provided_data() {
+    assert_eq!(
+        RoaringBitmap::deserialize_from(&mut &BITMAP_WITH_RUNS[..]).unwrap(),
+        test_data_bitmap()
+    );
+}
+
 #[test]
 fn test_serialize_into_provided_data() {
     let bitmap = test_data_bitmap();