String deduplication (#1269)

* eliminate a lot of arc clones * replace String by ArcStr (wrapped Arc<str>) for cheap clone and to make it possible to support string deduplication * test string deduplication * implement string deduplication for property values * clean up warnings * expose meta data in core ops and minor cleanup
Pometry · Sep 20, 2023 · 349da6e · 349da6e
1 parent 29d4c06
commit 349da6e
Show file tree

Hide file tree

Showing 64 changed files with 868 additions and 715 deletions.
diff --git a/examples/rust/src/bin/lotr/main.rs b/examples/rust/src/bin/lotr/main.rs
@@ -59,25 +59,22 @@ fn main() {
                 g.add_vertex(
                     lotr.time,
                     lotr.src_id.clone(),
-                    [("type".to_string(), Prop::Str("Character".to_string()))],
+                    [("type", Prop::str("Character"))],
                 )
                 .expect("Failed to add vertex");
 
                 g.add_vertex(
                     lotr.time,
                     lotr.dst_id.clone(),
-                    [("type".to_string(), Prop::Str("Character".to_string()))],
+                    [("type", Prop::str("Character"))],
                 )
                 .expect("Failed to add vertex");
 
                 g.add_edge(
                     lotr.time,
                     lotr.src_id.clone(),
                     lotr.dst_id.clone(),
-                    [(
-                        "type".to_string(),
-                        Prop::Str("Character Co-occurrence".to_string()),
-                    )],
+                    [("type", Prop::str("Character Co-occurrence"))],
                     None,
                 )
                 .expect("Failed to add edge");

diff --git a/js-raphtory/src/graph/edge.rs b/js-raphtory/src/graph/edge.rs
@@ -4,7 +4,7 @@ use raphtory::db::{
     api::view::*,
     graph::{edge::EdgeView, graph::Graph as TGraph},
 };
-use std::{ops::Deref, sync::Arc};
+use std::sync::Arc;
 use wasm_bindgen::prelude::*;
 
 #[wasm_bindgen]
@@ -39,7 +39,7 @@ impl Edge {
         let t_props = self.0.properties();
         let obj = js_sys::Map::new();
         for (k, v) in t_props.iter() {
-            obj.set(&k.deref().into(), &JsProp(v).into());
+            obj.set(&k.to_string().into(), &JsProp(v).into());
         }
         obj
     }

diff --git a/js-raphtory/src/graph/misc.rs b/js-raphtory/src/graph/misc.rs
@@ -21,7 +21,7 @@ impl From<JsProp> for JsValue {
         match value.0 {
             raphtory::core::Prop::U8(v) => v.into(),
             raphtory::core::Prop::U16(v) => v.into(),
-            raphtory::core::Prop::Str(v) => v.into(),
+            raphtory::core::Prop::Str(v) => v.to_string().into(),
             raphtory::core::Prop::I32(v) => v.into(),
             raphtory::core::Prop::I64(v) => v.into(),
             raphtory::core::Prop::U32(v) => v.into(),
@@ -68,6 +68,6 @@ impl From<JsObjectEntry> for Option<(String, Prop)> {
 
         let key = arr.at(0).as_string().unwrap();
         let value = arr.at(1).as_string().unwrap();
-        Some((key, Prop::Str(value)))
+        Some((key, Prop::str(value)))
     }
 }
diff --git a/js-raphtory/src/graph/mod.rs b/js-raphtory/src/graph/mod.rs
@@ -99,7 +99,7 @@ impl Graph {
     #[wasm_bindgen(js_name = addVertex)]
     pub fn add_vertex_js(&self, t: i64, id: JsValue, js_props: Object) -> Result<Vertex, JSError> {
         let rust_props = if js_props.is_string() {
-            vec![("name".to_string(), Prop::Str(js_props.as_string().unwrap()))]
+            vec![("name".to_string(), Prop::str(js_props.as_string().unwrap()))]
         } else if js_props.is_object() {
             Object::entries(&js_props)
                 .iter()

diff --git a/js-raphtory/src/graph/vertex.rs b/js-raphtory/src/graph/vertex.rs
@@ -100,7 +100,7 @@ impl Vertex {
     pub fn properties(&self) -> js_sys::Map {
         let obj = js_sys::Map::new();
         for (k, v) in self.0.properties() {
-            obj.set(&k.into(), &JsProp(v).into());
+            obj.set(&k.to_string().into(), &JsProp(v).into());
         }
         obj
     }

diff --git a/raphtory-graphql/src/data.rs b/raphtory-graphql/src/data.rs
@@ -88,7 +88,7 @@ impl Data {
                 println!("loading graph from {path}");
                 let graph = Graph::load_from_file(&path).expect("Unable to load from graph");
                 graph
-                    .add_constant_properties([("path".to_string(), Prop::Str(path.clone()))])
+                    .add_constant_properties([("path".to_string(), Prop::str(path.clone()))])
                     .expect("Failed to add static property");
                 let maybe_graph_name = graph.properties().get("name");
 

diff --git a/raphtory-graphql/src/lib.rs b/raphtory-graphql/src/lib.rs
@@ -203,18 +203,10 @@ mod graphql_test {
         if let Err(err) = graph.add_vertex(0, "gandalf", NO_PROPS) {
             panic!("Could not add vertex! {:?}", err);
         }
-        if let Err(err) = graph.add_vertex(
-            0,
-            "bilbo",
-            [("food".to_string(), Prop::Str("lots".to_string()))],
-        ) {
+        if let Err(err) = graph.add_vertex(0, "bilbo", [("food".to_string(), Prop::str("lots"))]) {
             panic!("Could not add vertex! {:?}", err);
         }
-        if let Err(err) = graph.add_vertex(
-            0,
-            "frodo",
-            [("food".to_string(), Prop::Str("some".to_string()))],
-        ) {
+        if let Err(err) = graph.add_vertex(0, "frodo", [("food".to_string(), Prop::str("some"))]) {
             panic!("Could not add vertex! {:?}", err);
         }
 

diff --git a/raphtory-graphql/src/model/filters/edge_filter.rs b/raphtory-graphql/src/model/filters/edge_filter.rs
@@ -43,8 +43,7 @@ impl EdgeFilter {
             return edge
                 .ee
                 .layer_names()
-                .iter()
-                .any(|name| name_filter.contains(name));
+                .any(|name| name_filter.contains(&name));
         }
 
         if let Some(property_has_filter) = &self.property_has {

diff --git a/raphtory-graphql/src/model/filters/property_filter.rs b/raphtory-graphql/src/model/filters/property_filter.rs
@@ -62,7 +62,7 @@ fn valid_prop(prop: Prop, value_str: &Option<String>, num_filter: &Option<Number
 
 fn value_neq_str_prop(value: &str, prop: &Prop) -> bool {
     if let Prop::Str(prop_str) = prop {
-        return value != prop_str;
+        return prop_str != value;
     }
 
     false

diff --git a/raphtory-graphql/src/model/graph/edge.rs b/raphtory-graphql/src/model/graph/edge.rs
@@ -1,5 +1,6 @@
 use crate::model::graph::node::Node;
 use dynamic_graphql::{ResolvedObject, ResolvedObjectFields};
+use itertools::Itertools;
 use raphtory::db::{
     api::view::{
         internal::{DynamicGraph, IntoDynamic},
@@ -47,7 +48,7 @@ impl Edge {
     }
 
     async fn layers(&self) -> Vec<String> {
-        self.ee.layer_names()
+        self.ee.layer_names().map_into().collect()
     }
 
     async fn history(&self) -> Vec<i64> {

diff --git a/raphtory-graphql/src/model/graph/graph.rs b/raphtory-graphql/src/model/graph/graph.rs
@@ -46,7 +46,7 @@ impl GraphMeta {
             .properties()
             .constant()
             .into_iter()
-            .map(|(k, v)| Property::new(k, v))
+            .map(|(k, v)| Property::new(k.into(), v))
             .collect()
     }
 
@@ -87,15 +87,15 @@ impl GqlGraph {
     }
 
     async fn layer_names(&self) -> Vec<String> {
-        self.graph.unique_layers()
+        self.graph.unique_layers().map_into().collect()
     }
 
     async fn static_properties(&self) -> Vec<Property> {
         self.graph
             .properties()
             .constant()
             .into_iter()
-            .map(|(k, v)| Property::new(k, v))
+            .map(|(k, v)| Property::new(k.into(), v))
             .collect()
     }
 

diff --git a/raphtory-graphql/src/model/graph/mod.rs b/raphtory-graphql/src/model/graph/mod.rs
@@ -1,5 +1,6 @@
 use itertools::Itertools;
 use raphtory::{
+    core::ArcStr,
     db::{
         api::view::internal::DynamicGraph,
         graph::{edge::EdgeView, vertex::VertexView},
@@ -26,7 +27,7 @@ fn get_expanded_edges(
 
     let mut filtered_fetched_edges = match maybe_layers {
         Some(layers) => {
-            let layer_set: HashSet<String> = layers.into_iter().collect();
+            let layer_set: HashSet<ArcStr> = layers.into_iter().map_into().collect();
             fetched_edges
                 .filter(|e| {
                     e.layer_names()

diff --git a/raphtory-graphql/src/model/graph/node.rs b/raphtory-graphql/src/model/graph/node.rs
@@ -60,7 +60,7 @@ impl Node {
             self.vv
                 .properties()
                 .iter()
-                .map(|(k, v)| Property::new(k.clone(), v))
+                .map(|(k, v)| Property::new(k.into(), v))
                 .collect(),
         )
     }

diff --git a/raphtory-graphql/src/model/mod.rs b/raphtory-graphql/src/model/mod.rs
@@ -155,7 +155,7 @@ impl Mut {
 
         let subgraph = data.get(&graph_name).ok_or("Graph not found")?;
         let mut path = subgraph
-            .static_prop(&"path".to_string())
+            .constant_prop(&"path".to_string())
             .expect("Path is missing")
             .to_string();
 
@@ -197,7 +197,7 @@ impl Mut {
             .add_constant_properties([("name".to_string(), Prop::Str(new_graph_name.clone()))])
             .expect("Failed to add static property");
         new_subgraph
-            .add_constant_properties([("uiProps".to_string(), Prop::Str(props))])
+            .add_constant_properties([("uiProps".to_string(), Prop::str(props))])
             .expect("Failed to add static property");
 
         new_subgraph

diff --git a/raphtory-graphql/src/model/schema/edge_schema.rs b/raphtory-graphql/src/model/schema/edge_schema.rs
@@ -58,10 +58,8 @@ impl<G: GraphViewOps> EdgeSchema<G> {
 }
 
 fn collect_edge_schema<G: GraphViewOps>(edge: EdgeView<G>) -> SchemaAggregate {
-    let pairs = edge
-        .properties()
+    edge.properties()
         .iter()
-        .map(|(key, value)| (key.to_owned(), HashSet::from([value.to_string()])))
-        .collect_vec();
-    HashMap::from_iter(pairs)
+        .map(|(key, value)| (key.to_string(), HashSet::from([value.to_string()])))
+        .collect()
 }
diff --git a/raphtory-graphql/src/model/schema/graph_schema.rs b/raphtory-graphql/src/model/schema/graph_schema.rs
@@ -24,7 +24,6 @@ impl GraphSchema {
 
         let layers = graph
             .unique_layers()
-            .iter()
             .map(|layer_name| graph.layer(layer_name).unwrap().into())
             .collect_vec();
 

diff --git a/raphtory-graphql/src/model/schema/layer_schema.rs b/raphtory-graphql/src/model/schema/layer_schema.rs
@@ -21,10 +21,13 @@ impl<G: GraphViewOps> From<LayeredGraph<G>> for LayerSchema<G> {
 impl<G: GraphViewOps> LayerSchema<G> {
     /// Returns the name of the layer with this schema
     async fn name(&self) -> String {
-        match &self.graph.unique_layers()[..] {
-            [layer] => layer.clone(),
-            _ => panic!("Layered graph outputted more than one layer name"),
-        }
+        let mut layers = self.graph.unique_layers();
+        let layer = layers.next().expect("Layered graph has a layer");
+        debug_assert!(
+            layers.next().is_none(),
+            "Layered graph outputted more than one layer name"
+        );
+        layer.into()
     }
     /// Returns the list of edge schemas for this edge layer
     async fn edges(&self) -> Vec<EdgeSchema<LayeredGraph<G>>> {

diff --git a/raphtory-graphql/src/model/schema/node_schema.rs b/raphtory-graphql/src/model/schema/node_schema.rs
@@ -48,10 +48,9 @@ impl NodeSchema {
 }
 
 fn collect_vertex_schema(vertex: VertexView<DynamicGraph>) -> SchemaAggregate {
-    let pairs = vertex
+    vertex
         .properties()
         .iter()
-        .map(|(key, value)| (key.to_owned(), HashSet::from([value.to_string()])))
-        .collect_vec();
-    HashMap::from_iter(pairs)
+        .map(|(key, value)| (key.to_string(), HashSet::from([value.to_string()])))
+        .collect()
 }