Diego

examples/serialize_bench.rs

@@ -10,19 +10,20 @@ static ALLOCATOR: jemallocator::Jemalloc = jemallocator::Jemalloc;
 
 fn main() {
     let args: Vec<String> = env::args().collect();
-    if args.len() <= 1 {
-        panic!("Usage: cargo run --example serialize_bench <number_of_universities>");
+    if args.len() <= 3 {
+        panic!("Usage: cargo run --example serialize_bench <rdf_path> <zarr_path> <shard_size>");
     }
-    let number_of_universities: &String = &args[1];
-    let zarr_path = format!("{}-lubm", number_of_universities);
-
+    let rdf_path: &String = &args[1];
+    let zarr_path: &String = &args[2];
+    let shard_size: &String = &args[3];
+
     let before = Instant::now();
 
     LocalStorage::new(MatrixLayout)
         .serialize(
-            format!("{}.zarr", zarr_path).as_str(),
-            format!("../lubm-uba-improved/out/{}.ttl", zarr_path).as_str(),
-            ChunkingStrategy::Sharding(10240),
+                    &zarr_path.as_str(),
+            &rdf_path.as_str(),
+            ChunkingStrategy::Sharding(shard_size.parse::<u64>().unwrap()),
         )
         .unwrap();
 

src/engine/array.rs

@@ -1,21 +1,26 @@
-use sprs::CsVec;
+use sprs::{CsMat, TriMat};
 
 use crate::storage::ZarrArray;
 
 use super::{EngineResult, EngineStrategy};
 
-impl EngineStrategy<CsVec<u8>> for ZarrArray {
-    fn get_subject(&self, index: usize) -> EngineResult<CsVec<u8>> {
-        let selection = CsVec::new(self.rows(), vec![index], vec![1]);
-        Ok(&self.transpose_view() * &selection)
+impl EngineStrategy<CsMat<u8>> for ZarrArray {
+    fn get_subject(&self, index: usize) -> EngineResult<CsMat<u8>> {
+        let mut matrix = TriMat::new((self.rows(), self.rows()));
+        matrix.add_triplet(index, index, 1);
+        let matrix = matrix.to_csc();
+        Ok(&matrix * self)
     }
 
-    fn get_predicate(&self, index: usize) -> EngineResult<CsVec<u8>> {
+    fn get_predicate(&self, value: u8) ->  EngineResult<CsMat<u8>>  {
+
         unimplemented!()
     }
 
-    fn get_object(&self, index: usize) -> EngineResult<CsVec<u8>> {
-        let selection = CsVec::new(self.cols(), vec![index], vec![1]);
-        Ok(self * &selection)
+    fn get_object(&self, index: usize) -> EngineResult<CsMat<u8>> {     
+        let mut matrix = TriMat::new((self.cols(), self.cols()));
+        matrix.add_triplet(index, index, 1);
+        let matrix = matrix.to_csc();
+        Ok(self * &matrix)
     }
 }

src/engine/chunk.rs

@@ -23,7 +23,7 @@
         }
     }
 
-    fn get_predicate(&self, index: usize) -> EngineResult<Vec<u8>> {
+    fn get_predicate(&self, index: u8) -> EngineResult<Vec<u8>> {
         unimplemented!()
     }
 

src/engine/mod.rs

@@ -7,6 +7,6 @@ pub type EngineResult<T> = Result<T, EngineError>;
 
 pub trait EngineStrategy<T> {
     fn get_subject(&self, index: usize) -> EngineResult<T>;
-    fn get_predicate(&self, index: usize) -> EngineResult<T>;
+    fn get_predicate(&self, index: u8) -> EngineResult<T>;
     fn get_object(&self, index: usize) -> EngineResult<T>;
 }

tests/common/mod.rs

@@ -68,6 +68,7 @@ impl Predicate {
     pub fn get_idx(self, dictionary: &Dictionary) -> u8 {
         dictionary.get_predicate_idx_unchecked(self.into()) as u8
     }
+
 }
 
 impl From<Predicate> for &str {
@@ -180,7 +181,9 @@ impl Graph {
             Object::GCHQ.get_idx(dictionary),
             Predicate::Manufacturer.get_idx(dictionary),
         );
-
         ans.to_csc()
+
+
+
     }
 }

tests/get_object_test.rs

@@ -2,8 +2,7 @@ use remote_hdt::{
     engine::EngineStrategy,
     storage::{matrix::MatrixLayout, tabular::TabularLayout, ChunkingStrategy, LocalStorage},
 };
-use sprs::CsVec;
-
+use sprs:: TriMat;
 mod common;
 
 #[test]
@@ -49,5 +48,10 @@ fn get_object_tabular_test() {
         .get_object(common::Object::Alan.get_idx(&storage.get_dictionary()))
         .unwrap();
 
-    assert_eq!(actual, CsVec::new(4, vec![1], vec![3]))
+    let mut expected = TriMat::new((4,9));
+    expected.add_triplet(1,3,3);
+    let expected = expected.to_csc();
+    assert_eq!(actual, expected )
 }
+
+

tests/get_subject_test.rs

@@ -2,8 +2,7 @@ use remote_hdt::{
     engine::EngineStrategy,
     storage::{matrix::MatrixLayout, tabular::TabularLayout, ChunkingStrategy, LocalStorage},
 };
-use sprs::CsVec;
-
+use sprs:: TriMat;
 mod common;
 
 #[test]
@@ -49,8 +48,13 @@ fn get_subject_tabular_test() {
         .get_subject(common::Subject::Alan.get_idx(&storage.get_dictionary()))
         .unwrap();
 
-    assert_eq!(
-        actual,
-        CsVec::new(9, vec![0, 1, 2, 7, 8], vec![2, 4, 5, 7, 8])
-    )
+        let mut result = TriMat::new((4,9));
+        result.add_triplet(0,0,2);
+        result.add_triplet(0,1,4);
+        result.add_triplet(0,2,5);
+        result.add_triplet(0,7,7);
+        result.add_triplet(0,8,8);
+        let result = result.to_csc();
+        assert_eq!(actual, result)
 }
+