add neighbor components

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "neighborhood_analysis"
-version = "0.2.2"
+version = "0.2.3"
 authors = ["Mr-Milk <zym.zym1220@gmail.com>"]
 edition = "2018"
 license = "MIT"

neighborhood_analysis/__init__.py

@@ -1,4 +1,5 @@
-from .neighborhood_analysis import (get_bbox,
+from .neighborhood_analysis import (neighbor_components,
+                                    get_bbox,
                                     get_bbox_neighbors,
                                     get_point_neighbors,
                                     comb_bootstrap,

neighborhood_analysis/__init__.pyi

@@ -1,5 +1,6 @@
 from typing import List, Tuple, Dict, Optional
 
+def neighbor_components(neighbors: Dict[int, List[int]], types: Dict[int, str]) -> (List[int], List[str], List[List[int]]): ...
 
 def get_bbox(points_collections: List[List[Tuple[float, float]]]) -> List[Tuple[float, float, float, float]]: ...
 

src/lib.rs

@@ -7,6 +7,7 @@ use rand::thread_rng;
 use std::collections::HashMap;
 
 use kdbush::KDBush;
+use counter::Counter;
 use rayon::prelude::*;
 use rstar::{RTree, RTreeObject, AABB};
 use spade::BoundingRect;
@@ -23,9 +24,42 @@ fn neighborhood_analysis(_py: Python, m: &PyModule) -> PyResult<()> {
     m.add_wrapped(wrap_pyfunction!(get_bbox_neighbors))?;
     m.add_class::<CellCombs>()?;
     m.add_wrapped(wrap_pyfunction!(comb_bootstrap))?;
+    m.add_wrapped(wrap_pyfunction!(neighbor_components))?;
     Ok(())
 }
 
+
+
+#[pyfunction]
+pub fn neighbor_components(neighbors: HashMap<usize, Vec<usize>>, types: HashMap<usize, &str>)
+    -> (Vec<usize>, Vec<&str>, Vec<Vec<usize>>) {
+
+    let mut uni_types: HashMap<&str, i64> = HashMap::new();
+    for (_, t) in &types {
+        uni_types.entry(*t).or_insert(0);
+    }
+    let uni_types: Vec<&str> = uni_types.keys().map(|k| *k).collect_vec();
+    let mut cent_order: Vec<usize> = vec![];
+    let result: Vec<Vec<usize>> = neighbors.iter().map(|(cent, neigh)| {
+        let count: HashMap<&&str, usize> = neigh.iter().map(|i| &types[i]).collect::<Counter<_>>().into_map();
+        let mut result_v: Vec<usize> = vec![];
+        for t in &uni_types {
+            let v = match count.get(t) {
+                Some(v) => *v,
+                None => {0}
+            };
+            result_v.push(v);
+        }
+        cent_order.push(*cent);
+        result_v
+        }).collect();
+
+    (cent_order, uni_types, result)
+
+}
+
+
+
 /// get_bbox(points_collections)
 /// --
 ///

tests.py

@@ -1,5 +1,10 @@
 import numpy as np
-from neighborhood_analysis import CellCombs, get_bbox, get_point_neighbors, get_bbox_neighbors, comb_bootstrap
+from neighborhood_analysis import (CellCombs,
+                                   get_bbox,
+                                   get_point_neighbors,
+                                   get_bbox_neighbors,
+                                   comb_bootstrap,
+                                   neighbor_components)
 from time import time
 
 types = [str(i) for i in range(30)]
@@ -17,30 +22,36 @@
 start = time()
 bbox = get_bbox(polygons)
 end = time()
-print(f"Get bbox used {(end-start):.5f}s")
+print(f"Get bbox used {(end - start):.5f}s")
 
 start = time()
 neighbors = get_bbox_neighbors(bbox, 2, labels=[i for i in range(0, len(bbox))])
 end = time()
-print(f"search bbox neighbors used {(end-start):.5f}s")
-
+print(f"search bbox neighbors used {(end - start):.5f}s")
 
 start = time()
 neighbors = get_point_neighbors(points, 10.0)
 end = time()
-print(f"search point neighbors used {(end-start):.5f}s")
+print(f"search point neighbors used {(end - start):.5f}s")
 
 start = time()
 
 cc = CellCombs(types, False)
 results = cc.bootstrap(corr_types, neighbors, times=1000, ignore_self=True)
 
 end = time()
-print(f"neighborhood used {(end-start):.5f}s")
+print(f"neighborhood used {(end - start):.5f}s")
 
 s1 = time()
 X = [bool(i) for i in np.random.choice([True, False], 10000)]
 Y = [bool(i) for i in np.random.choice([True, False], 10000)]
 v = comb_bootstrap(X, Y, neighbors, times=1000, ignore_self=True)
 s2 = time()
-print(f"marker co-exp used {(s2-s1):.5f}s")
+print(f"marker co-exp used {(s2 - s1):.5f}s")
+
+cents = [i for i in range(len(points))]
+s1 = time()
+ix, col, data = neighbor_components(dict(zip(cents, neighbors)), dict(zip(cents, corr_types)))
+s2 = time()
+print(ix, col, data)
+print(f"neighbor components used {(s2 - s1):.5f}s")