Use fnv instead of twox.

src/lib.rs

@@ -1,48 +1,169 @@
+extern crate fnv;
+
+use std::ops;
+use std::iter::FromIterator;
+
+use fnv::FnvHashMap;
+
+const BITS: usize = 64;
+
+type Block = u64;
+type Storage = FnvHashMap<usize, Block>;
+
+#[inline]
+fn block_bit(x: usize, d: usize) -> (usize, usize) {
+    (x / d, x % d)
+}
+
+#[derive(Debug)]
 pub struct BitSet {
-    data: Box<[u64]>,
+    blocks: Storage,
+    nbits: usize,
 }
 
 impl BitSet {
-    pub fn with_capacity(max: usize) -> BitSet {
+    pub fn new() -> BitSet {
         BitSet {
-            data: vec![0; max / 64 + if max % 64 == 0 { 0 } else { 1 }].into_boxed_slice()
+            blocks: Storage::default(),
+            nbits: 0,
         }
     }
 
-    pub fn capacity(&self) -> u64 {
-        self.data.len() as u64 * 64
+    pub fn with_capacity(nbits: usize) -> BitSet {
+        let (mut block, bit) = block_bit(nbits, BITS);
+        block += (bit > 0) as usize;
+
+        BitSet {
+            blocks: Storage::with_capacity_and_hasher(block, Default::default()),
+            nbits: 0,
+        }
     }
 
-    pub fn insert(&mut self, value: usize) {
-        let block = value / 64 as usize;
-        let bit = 2u64.pow((value - (block * 64)) as u32);
-
-        self.data[block] |= bit;
+    pub fn capacity(&self) -> usize {
+        self.blocks.len() * BITS
     }
 
-    pub fn remove(&mut self, value: usize) {
-        let block = value / 64 as usize;
-        let bit = 2u64.pow((value - (block * 64)) as u32);
+    pub fn len(&self) -> usize {
+        self.nbits
+    }
 
-        self.data[block] &= !bit;
+    pub fn is_empty(&self) -> bool {
+        self.nbits == 0
+    }
+
+    pub fn insert(&mut self, value: usize) -> bool {
+        let (block, bit) = block_bit(value, BITS);
+        let block = self.blocks.entry(block).or_insert(0);
+
+        if (*block & (1 << bit)) == 0 {
+            *block |= 1 << bit;
+            self.nbits += 1;
+
+            true
+        } else {
+            false
+        }
+    }
+
+    pub fn remove(&mut self, value: usize) -> bool {
+        let (block, bit) = block_bit(value, BITS);
+        let block = self.blocks.entry(block).or_insert(0);
+
+        if (*block & (1 << bit)) != 0 {
+            *block &= !(1 << bit);
+            self.nbits -= 1;
+
+            true
+        } else {
+            false
+        }
     }
 
     pub fn contains(&self, value: usize) -> bool {
-        let block = value / 64 as usize;
-        let bit = 2u64.pow((value - (block * 64)) as u32);
+        let (block, bit) = block_bit(value, BITS);
 
-        self.data[block] & bit == bit
+        match self.blocks.get(&block) {
+            None => false,
+            Some(block) => (block & (1 << bit)) != 0,
+        }
     }
 }
 
+impl Default for BitSet {
+    #[inline]
+    fn default() -> BitSet {
+        BitSet::new()
+    }
+}
+
+impl FromIterator<usize> for BitSet {
+    fn from_iter<I: IntoIterator<Item = usize>>(iter: I) -> BitSet {
+        let mut set = BitSet::new();
+        set.extend(iter);
+        set
+    }
+}
+
+impl Extend<usize> for BitSet {
+    #[inline]
+    fn extend<I: IntoIterator<Item = usize>>(&mut self, iter: I) {
+        for i in iter {
+            self.insert(i);
+        }
+    }
+}
+
+impl ops::BitOr for BitSet {
+    type Output = Self;
+
+    fn bitor(self, rhs: Self) -> Self {
+        let mut blocks = self.blocks.clone();
+
+        for (key, value) in &rhs.blocks {
+            *blocks.entry(*key).or_insert(0) |= value;
+        }
+
+        let nbits = blocks
+            .values()
+            .map(|block| block.count_ones() as usize)
+            .sum();
+
+        BitSet {
+            blocks: blocks,
+            nbits: nbits,
+        }
+    }
+}
+
+impl<'a> ops::BitOr<&'a Self> for BitSet {
+    type Output = Self;
+
+    fn bitor(self, rhs: &'a Self) -> Self {
+        let mut blocks = self.blocks.clone();
+
+        for (key, value) in &rhs.blocks {
+            *blocks.entry(*key).or_insert(0) |= value;
+        }
+
+        let nbits = blocks
+            .values()
+            .map(|block| block.count_ones() as usize)
+            .sum();
+
+        BitSet {
+            blocks: blocks,
+            nbits: nbits,
+        }
+    }
+}
 
 #[cfg(test)]
 mod tests {
     use super::*;
 
     #[test]
-    fn test() {
-        let mut set = BitSet::with_capacity(100);
+    fn insert() {
+        let mut set = BitSet::with_capacity(10);
 
         assert_eq!(set.contains(0), false);
         assert_eq!(set.contains(10), false);
@@ -52,5 +173,40 @@ mod tests {
 
         assert_eq!(set.contains(0), true);
         assert_eq!(set.contains(10), true);
+
+        assert_eq!(set.contains(100), false);
+
+        set.insert(100);
+
+        assert_eq!(set.contains(100), true);
+    }
+
+    #[test]
+    fn from_iter() {
+        let set = [1, 2, 3, 10, 100].iter().cloned().collect::<BitSet>();
+
+        assert_eq!(set.len(), 5);
+
+        assert_eq!(set.contains(1), true);
+        assert_eq!(set.contains(2), true);
+        assert_eq!(set.contains(3), true);
+        assert_eq!(set.contains(10), true);
+        assert_eq!(set.contains(100), true);
+    }
+
+    #[test]
+    fn bitor() {
+        let set_a = [1, 2, 3].iter().cloned().collect::<BitSet>();
+        let set_b = [3, 4, 5].iter().cloned().collect::<BitSet>();
+
+        let set = set_a | set_b;
+
+        assert_eq!(set.len(), 5);
+
+        assert_eq!(set.contains(1), true);
+        assert_eq!(set.contains(2), true);
+        assert_eq!(set.contains(3), true);
+        assert_eq!(set.contains(4), true);
+        assert_eq!(set.contains(5), true);
     }
 }