HNSW Implementation

.github/workflows/test.yml

@@ -18,6 +18,8 @@ env:
   RUSTC_WRAPPER: "sccache"
 
 jobs:
+
+
   run-rust-tests:
     runs-on: ubuntu-latest
 
@@ -69,15 +71,32 @@ jobs:
           exit 1
         fi
 
+    - name: Detect changed rust files in workspace
+      uses: dorny/paths-filter@v3
+      id: changes
+      with:
+        filters: |
+          ai:
+            - 'ahnlich/ai/**'
+          types:
+            - 'ahnlich/types/**'
+
     - name: Build and Format Rust Example
       working-directory: ./examples/rust/image-search
       run: |
         cargo b && cargo fmt
 
-    - name: Run Test
+
+    - name: Run Full Workspace Tests
+      if: steps.changes.outputs.ai == 'true' || steps.changes.outputs.types == 'true'
       working-directory: ./ahnlich
-      run: |
-        make test
+      run: make test
+
+    - name: Run Workspace Tests (excluding AI)
+      if: steps.changes.outputs.ai != 'true' && steps.changes.outputs.types != 'true'
+      working-directory: ./ahnlich
+      run: cargo nextest run --workspace --exclude ai --no-capture
+
 
     - name: Upload Test Results
       uses: actions/upload-artifact@v4

ahnlich/Cargo.toml

@@ -46,9 +46,12 @@ fallible_collections = "0.4.9"
 dirs = "5.0.1"
 strum = { version = "0.26", features = ["derive"] }
 papaya = { version = "0.2.0", features = ["serde"] }
+parking_lot = "0.12"
 tonic = "0.12.3"
 http = "1.2.0"
 prost = "0.13"
+pulp = "0.21.4"
+smallvec = "1.13"
 
 [profile.release]
 lto = true

ahnlich/client/src/db.rs

@@ -836,7 +836,7 @@ mod test {
                 stores: vec![StoreInfo {
                     name: "Main".to_string(),
                     len: 2,
-                    size_in_bytes: 1356,
+                    size_in_bytes: 1340,
                 }]
             }
         );
@@ -870,7 +870,7 @@ mod test {
                 stores: vec![StoreInfo {
                     name: "Main".to_string(),
                     len: 1,
-                    size_in_bytes: 1244,
+                    size_in_bytes: 1236,
                 }]
             }
         );

ahnlich/db/Cargo.toml

@@ -39,7 +39,7 @@ async-trait.workspace = true
 rayon.workspace = true
 log.workspace = true
 fallible_collections.workspace = true
-pulp = "0.21.4"
+pulp.workspace = true
 papaya.workspace = true
 ahnlich_types = { path = "../types", version = "*"  }
 tonic.workspace = true

ahnlich/db/src/algorithm/heap.rs

@@ -1,8 +1,7 @@
-#![allow(dead_code)]
 use super::LinearAlgorithm;
-use std::cmp::Reverse;
-use std::collections::BinaryHeap;
-use std::num::NonZeroUsize;
+
+// Re-export bounded heaps from similarity crate
+pub use ahnlich_similarity::heap::{BoundedMaxHeap, BoundedMinHeap};
 
 pub enum HeapOrder {
     Min,
@@ -19,79 +18,3 @@ impl From<&LinearAlgorithm> for HeapOrder {
         }
     }
 }
-
-/// A bounded max heap that maintains at most `capacity` elements.
-/// When the heap is full, incoming elements replace the smallest element if they are larger.
-pub struct BoundedMaxHeap<T: Ord> {
-    heap: BinaryHeap<Reverse<T>>,
-    capacity: usize,
-}
-
-impl<T: Ord> BoundedMaxHeap<T> {
-    pub fn new(capacity: NonZeroUsize) -> Self {
-        Self {
-            heap: BinaryHeap::with_capacity(capacity.get()),
-            capacity: capacity.get(),
-        }
-    }
-
-    /// Insert an element into the bounded heap.
-    /// If the heap is at capacity and the new element is larger than the smallest,
-    /// the smallest element is removed and the new element is added.
-    pub fn push(&mut self, item: T) {
-        if self.heap.len() < self.capacity {
-            self.heap.push(Reverse(item));
-        } else if let Some(mut smallest) = self.heap.peek_mut() {
-            // Only insert if the new item is better (larger) than the worst (smallest) item
-            if item > smallest.0 {
-                *smallest = Reverse(item);
-            }
-        }
-    }
-
-    /// Convert the bounded heap into a Vec, consuming the heap.
-    /// Elements are returned in descending order (largest first).
-    pub fn into_sorted_vec(self) -> Vec<T> {
-        let mut vec: Vec<_> = self.heap.into_iter().map(|r| r.0).collect();
-        vec.sort_by(|a, b| b.cmp(a)); // Sort descending
-        vec
-    }
-}
-
-/// A bounded min heap that maintains at most `capacity` elements.
-/// When the heap is full, incoming elements replace the largest element if they are smaller.
-pub struct BoundedMinHeap<T: Ord> {
-    heap: BinaryHeap<T>,
-    capacity: usize,
-}
-
-impl<T: Ord> BoundedMinHeap<T> {
-    pub fn new(capacity: NonZeroUsize) -> Self {
-        Self {
-            heap: BinaryHeap::with_capacity(capacity.get()),
-            capacity: capacity.get(),
-        }
-    }
-
-    /// Insert an element into the bounded heap.
-    /// If the heap is at capacity and the new element is smaller than the largest,
-    /// the largest element is removed and the new element is added.
-    pub fn push(&mut self, item: T) {
-        if self.heap.len() < self.capacity {
-            self.heap.push(item);
-        } else if let Some(mut largest) = self.heap.peek_mut() {
-            // Only insert if the new item is better (smaller) than the worst (largest) item
-            if item < *largest {
-                *largest = item;
-            }
-        }
-    }
-
-    /// Convert the bounded heap into a Vec, consuming the heap.
-    /// Elements are returned in ascending order (smallest first).
-    pub fn into_sorted_vec(self) -> Vec<T> {
-        let mut vec: Vec<_> = self.heap.into_iter().collect();
-        vec.sort(); // Sort ascending
-        vec
-    }
-}