Gtnotebook (Pometry#882)

* impl generic taint v2

* fix merge compile/test issues

* fix test

* intro v2

* impl hits with new apis

* fix example

* fix lotr

* update ingestion

* impl triplet count using new api

* rid old triangle count impl

* impl reciprocity using new api

* impl cluster using new api

* fix test

examples/py/enron/enron.ipynb

@@ -461,9 +461,7 @@
    "cell_type": "code",
    "execution_count": 10,
    "id": "5e8af6e0",
-   "metadata": {
-    "scrolled": false
-   },
+   "metadata": {},
    "outputs": [
     {
      "data": {
@@ -823,9 +821,9 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "raphtory",
+   "display_name": "Python [conda env:raphtory] *",
    "language": "python",
-   "name": "raphtory"
+   "name": "conda-env-raphtory-py"
   },
   "language_info": {
    "codemirror_mode": {

examples/rust/src/bin/crypto/main.rs

@@ -1,7 +1,11 @@
+use chrono::NaiveDateTime;
 use itertools::Itertools;
 use raphtory::algorithms::generic_taint::generic_taint;
 use raphtory::algorithms::pagerank::unweighted_page_rank;
+use raphtory::core::time::TryIntoTime;
+use raphtory::db::view_api::internal::GraphViewInternalOps;
 use raphtory::db::view_api::layer::LayerOps;
+use raphtory::db::view_api::time::WindowSet;
 use raphtory::db::view_api::*;
 use raphtory_io::graph_loader::example::stable_coins::stable_coin_graph;
 use serde::Deserialize;
@@ -31,18 +35,11 @@ fn main() {
     let g = stable_coin_graph(data_dir, 1);
 
     assert_eq!(g.num_vertices(), 1523333);
-    assert_eq!(g.num_edges(), 2871269);
+    assert_eq!(g.num_edges(), 2814155);
 
     assert_eq!(
         g.get_unique_layers().into_iter().sorted().collect_vec(),
-        vec![
-            "0x6b175474e89094c44da98b954eedeac495271d0f",
-            "0x8e870d67f660d95d5be530380d0ec0bd388289e1",
-            "0xa0b86991c6218b36c1d19d4a2e9eb0ce3606eb48",
-            "0xa47c8bf37f92abed4a126bda807a7b7498661acd",
-            "0xd2877702675e6ceb975b4a1dff9fb7baf4c91ea9",
-            "0xdac17f958d2ee523a2206206994597c13d831ec7"
-        ]
+        vec!["Dai", "LUNC", "USD", "USDP", "USDT", "USTC"]
     );
 
     println!("Pagerank");
@@ -52,7 +49,16 @@ fn main() {
 
     let now = Instant::now();
     let _ = unweighted_page_rank(
-        &g.layer("0xdac17f958d2ee523a2206206994597c13d831ec7")
+        &g,
+        20,
+        None,
+        None,
+    );
+    println!("Time taken: {} secs", now.elapsed().as_secs());
+
+    let now = Instant::now();
+    let _ = unweighted_page_rank(
+        &g.layer("USDT")
             .unwrap(),
         20,
         None,
@@ -63,12 +69,13 @@ fn main() {
     println!("Generic taint");
     let now = Instant::now();
     let _ = generic_taint(
-        &g.layer("0xdac17f958d2ee523a2206206994597c13d831ec7")
+        &g.layer("USDT")
             .unwrap(),
+        None,
         20,
         1651105815,
         vec!["0xd30b438df65f4f788563b2b3611bd6059bff4ad9"],
         vec![],
     );
-    println!("Time taken: {}", now.elapsed().as_secs());
+    println!("Time taken: {} secs", now.elapsed().as_secs());
 }

examples/rust/src/bin/hulongbay/main.rs

@@ -6,7 +6,7 @@ use std::path::Path;
 
 use itertools::Itertools;
 use raphtory::algorithms::connected_components::weakly_connected_components;
-use raphtory::algorithms::triangle_count::triangle_counting_fast;
+use raphtory::algorithms::triangle_count::triangle_count;
 use raphtory::core::{Direction, Prop};
 use raphtory::db::graph::Graph;
 use raphtory::db::view_api::*;
@@ -112,7 +112,7 @@ fn try_main() -> Result<(), Box<dyn Error>> {
     let max_time = graph.end().ok_or(GraphEmptyError)?;
     let mid_time = (min_time + max_time) / 2;
     let now = Instant::now();
-    let actual_tri_count = triangle_counting_fast(&graph, None);
+    let actual_tri_count = triangle_count(&graph, None);
 
     println!("Actual triangle count: {:?}", actual_tri_count);
 

examples/rust/src/bin/lotr/main.rs

@@ -106,7 +106,7 @@ fn main() {
     assert!(graph.has_vertex(gandalf));
     assert_eq!(graph.vertex(gandalf).unwrap().name(), "Gandalf");
 
-    let r = generic_taint(&graph, 20, 31930, vec!["Gandalf"], vec![]);
+    let r = generic_taint(&graph, None, 20, 31930, vec!["Gandalf"], vec![]);
     assert_eq!(
         r.keys().sorted().collect_vec(),
         vec!["Gandalf", "Saruman", "Wormtongue"]

python/src/algorithms.rs

@@ -52,7 +52,7 @@ pub(crate) fn local_triangle_count(g: &PyGraphView, v: &PyAny) -> PyResult<Optio
 pub(crate) fn weakly_connected_components(
     g: &PyGraphView,
     iter_count: usize,
-) -> PyResult<FxHashMap<String, u64>> {
+) -> PyResult<HashMap<String, u64>> {
     Ok(connected_components::weakly_connected_components(
         &g.graph, iter_count, None,
     ))
@@ -63,7 +63,7 @@ pub(crate) fn pagerank(
     g: &PyGraphView,
     iter_count: usize,
     max_diff: Option<f32>,
-) -> PyResult<FxHashMap<String, f32>> {
+) -> PyResult<HashMap<String, f32>> {
     Ok(unweighted_page_rank(&g.graph, iter_count, None, max_diff))
 }
 
@@ -86,6 +86,7 @@ pub(crate) fn generic_taint(
 
     Ok(generic_taint_rs(
         &g.graph,
+        None,
         iter_count,
         start_time,
         infected_nodes?,
@@ -181,7 +182,7 @@ pub(crate) fn min_in_degree(g: &PyGraphView) -> usize {
 /// it could imply a less reciprocal or more one-sided relationship.
 #[pyfunction]
 pub(crate) fn global_reciprocity(g: &PyGraphView) -> f64 {
-    global_reciprocity_rs(&g.graph)
+    global_reciprocity_rs(&g.graph, None)
 }
 
 /// Reciprocity - measure of the symmetry of relationships in a graph.
@@ -202,8 +203,8 @@ pub(crate) fn global_reciprocity(g: &PyGraphView) -> f64 {
 /// it could imply a less reciprocal or more one-sided relationship.
 ///
 #[pyfunction]
-pub(crate) fn all_local_reciprocity(g: &PyGraphView) -> HashMap<u64, f64> {
-    all_local_reciprocity_rs(&g.graph)
+pub(crate) fn all_local_reciprocity(g: &PyGraphView) -> HashMap<String, f64> {
+    all_local_reciprocity_rs(&g.graph, None)
 }
 
 /// Computes the number of both open and closed triplets within a graph
@@ -212,7 +213,7 @@ pub(crate) fn all_local_reciprocity(g: &PyGraphView) -> HashMap<u64, f64> {
 /// A closed triplet is one where a node has two neighbors, and an edge between them.
 #[pyfunction]
 pub(crate) fn triplet_count(g: &PyGraphView) -> usize {
-    raphtory::algorithms::triplet_count::triplet_count(&g.graph)
+    raphtory::algorithms::triplet_count::triplet_count(&g.graph, None)
 }
 
 /// Computes the global clustering coefficient of a graph. The global clustering coefficient is

python/src/edge.rs

@@ -8,7 +8,7 @@ use crate::dynamic::{DynamicGraph, IntoDynamic};
 use crate::types::repr::{iterator_repr, Repr};
 use crate::utils::*;
 use crate::vertex::{PyVertex, PyVertexIterable};
-use crate::wrappers::iterators::{OptionPropIterable, I64Iterable};
+use crate::wrappers::iterators::{I64Iterable, OptionPropIterable};
 use crate::wrappers::prop::Prop;
 use chrono::NaiveDateTime;
 use itertools::Itertools;
@@ -198,7 +198,7 @@ impl PyEdge {
     ///
     /// Returns:
     ///     the start datetime of the Edge.
-    pub fn start_date_time(&self) ->Option<NaiveDateTime> {
+    pub fn start_date_time(&self) -> Option<NaiveDateTime> {
         let start_time = self.edge.start()?;
         Some(NaiveDateTime::from_timestamp_millis(start_time).unwrap())
     }
@@ -215,7 +215,7 @@ impl PyEdge {
     ///
     /// Returns:
     ///    The end datetime of the Edge
-    pub fn end_date_time(&self) ->Option<NaiveDateTime> {
+    pub fn end_date_time(&self) -> Option<NaiveDateTime> {
         let end_time = self.edge.end()?;
         Some(NaiveDateTime::from_timestamp_millis(end_time).unwrap())
     }
@@ -335,7 +335,7 @@ impl PyEdge {
     pub fn layer_name(&self) -> String {
         self.edge.layer_name()
     }
-    
+
     /// Gets the datetime of an exploded edge.
     ///
     /// Returns:
@@ -457,18 +457,24 @@ impl PyEdges {
 
     /// Returns the earliest time of the edges.
     fn earliest_time(&self) -> I64Iterable {
-        let edges: Arc<dyn Fn() -> Box<dyn Iterator<Item = EdgeView<DynamicGraph>> + Send> + Send + Sync> = self.builder.clone();
+        let edges: Arc<
+            dyn Fn() -> Box<dyn Iterator<Item = EdgeView<DynamicGraph>> + Send> + Send + Sync,
+        > = self.builder.clone();
         (move || edges().earliest_time()).into()
     }
 
     /// Returns the latest time of the edges.
     fn latest_time(&self) -> I64Iterable {
-        let edges: Arc<dyn Fn() -> Box<dyn Iterator<Item = EdgeView<DynamicGraph>> + Send> + Send + Sync> = self.builder.clone();
+        let edges: Arc<
+            dyn Fn() -> Box<dyn Iterator<Item = EdgeView<DynamicGraph>> + Send> + Send + Sync,
+        > = self.builder.clone();
         (move || edges().latest_time()).into()
     }
 
     fn property(&self, name: String, include_static: Option<bool>) -> OptionPropIterable {
-        let edges: Arc<dyn Fn() -> Box<dyn Iterator<Item = EdgeView<DynamicGraph>> + Send> + Send + Sync> = self.builder.clone();
+        let edges: Arc<
+            dyn Fn() -> Box<dyn Iterator<Item = EdgeView<DynamicGraph>> + Send> + Send + Sync,
+        > = self.builder.clone();
         (move || edges().property(name.clone(), include_static.unwrap_or(true))).into()
     }
 

python/src/graph_view.rs

@@ -6,11 +6,11 @@ use crate::utils::{
     PyWindowSet,
 };
 use crate::vertex::{PyVertex, PyVertices};
+use chrono::prelude::*;
 use pyo3::prelude::*;
 use raphtory::db::view_api::layer::LayerOps;
 use raphtory::db::view_api::*;
 use raphtory::*;
-use chrono::prelude::*;
 
 /// Graph view is a read-only version of a graph at a certain point in time.
 #[pyclass(name = "GraphView", frozen, subclass)]
@@ -55,7 +55,7 @@ impl PyGraphView {
     ///
     /// Returns:
     ///     the datetime of the earliest activity in the graph
-    pub fn earliest_date_time(&self) ->Option<NaiveDateTime> {
+    pub fn earliest_date_time(&self) -> Option<NaiveDateTime> {
         let earliest_time = self.graph.earliest_time()?;
         Some(NaiveDateTime::from_timestamp_millis(earliest_time).unwrap())
     }
@@ -72,7 +72,7 @@ impl PyGraphView {
     ///
     /// Returns:
     ///     the datetime of the latest activity in the graph
-    pub fn latest_date_time(&self) ->Option<NaiveDateTime> {
+    pub fn latest_date_time(&self) -> Option<NaiveDateTime> {
         let latest_time = self.graph.latest_time()?;
         Some(NaiveDateTime::from_timestamp_millis(latest_time).unwrap())
     }
@@ -183,7 +183,7 @@ impl PyGraphView {
     ///
     /// Returns:
     ///     the default start datetime for perspectives over the view
-    pub fn start_date_time(&self) ->Option<NaiveDateTime> {
+    pub fn start_date_time(&self) -> Option<NaiveDateTime> {
         let start_time = self.graph.start()?;
         Some(NaiveDateTime::from_timestamp_millis(start_time).unwrap())
     }
@@ -200,12 +200,12 @@ impl PyGraphView {
     pub fn window_size(&self) -> Option<u64> {
         self.graph.window_size()
     }
-    
+
     /// Returns the default end datetime for perspectives over the view
     ///
     /// Returns:
     ///    the default end datetime for perspectives over the view
-    pub fn end_date_time(&self) ->Option<NaiveDateTime> {
+    pub fn end_date_time(&self) -> Option<NaiveDateTime> {
         let end_time = self.graph.end()?;
         Some(NaiveDateTime::from_timestamp_millis(end_time).unwrap())
     }

python/src/lib.rs

@@ -47,7 +47,10 @@ fn raphtory(py: Python<'_>, m: &PyModule) -> PyResult<()> {
     algorithm_module.add_function(wrap_pyfunction!(min_out_degree, algorithm_module)?)?;
     algorithm_module.add_function(wrap_pyfunction!(min_in_degree, algorithm_module)?)?;
     algorithm_module.add_function(wrap_pyfunction!(pagerank, algorithm_module)?)?;
-    algorithm_module.add_function(wrap_pyfunction!(weakly_connected_components, algorithm_module)?)?;
+    algorithm_module.add_function(wrap_pyfunction!(
+        weakly_connected_components,
+        algorithm_module
+    )?)?;
     m.add_submodule(algorithm_module)?;
 
     let graph_loader_module = PyModule::new(py, "graph_loader")?;

python/src/vertex.rs

@@ -97,7 +97,7 @@ impl PyVertex {
     ///
     /// Returns:
     ///     The earliest datetime that the vertex exists as an integer.
-    pub fn earliest_date_time(&self) ->Option<NaiveDateTime> {
+    pub fn earliest_date_time(&self) -> Option<NaiveDateTime> {
         let earliest_time = self.vertex.earliest_time()?;
         Some(NaiveDateTime::from_timestamp_millis(earliest_time).unwrap())
     }
@@ -117,12 +117,11 @@ impl PyVertex {
     ///
     /// Returns:
     ///     The latest datetime that the vertex exists as an integer.
-    pub fn latest_date_time(&self) ->Option<NaiveDateTime> {
+    pub fn latest_date_time(&self) -> Option<NaiveDateTime> {
         let latest_time = self.vertex.latest_time()?;
         Some(NaiveDateTime::from_timestamp_millis(latest_time).unwrap())
     }
 
-
     /// Gets the property value of this vertex given the name of the property.
     ///
     /// Arguments:
@@ -321,7 +320,7 @@ impl PyVertex {
     ///
     /// Returns:
     ///     The earliest datetime that this vertex is valid or None if the vertex is valid for all times.
-    pub fn start_date_time(&self) ->Option<NaiveDateTime> {
+    pub fn start_date_time(&self) -> Option<NaiveDateTime> {
         let start_time = self.vertex.start()?;
         Some(NaiveDateTime::from_timestamp_millis(start_time).unwrap())
     }
@@ -338,7 +337,7 @@ impl PyVertex {
     ///
     /// Returns:
     ///     The latest datetime that this vertex is valid or None if the vertex is valid for all times.
-    pub fn end_date_time(&self) ->Option<NaiveDateTime> {
+    pub fn end_date_time(&self) -> Option<NaiveDateTime> {
         let end_time = self.vertex.end()?;
         Some(NaiveDateTime::from_timestamp_millis(end_time).unwrap())
     }