From 0b50baad9854c70175c6e80f75c51c240e55c402 Mon Sep 17 00:00:00 2001
From: Dustin Carlino <dabreegster@gmail.com>
Date: Mon, 18 Sep 2023 12:31:44 +0100
Subject: [PATCH] Optionally use centroids for origin zones without any points.
 #27

---
 aggregate_routes/src/config.rs |  2 ++
 aggregate_routes/src/od.rs     | 25 ++++++++++++++++++++-----
 aggregate_routes/src/timer.rs  |  5 ++++-
 examples/liverpool/config.json |  3 ++-
 examples/york/config.json      |  3 ++-
 5 files changed, 30 insertions(+), 8 deletions(-)

diff --git a/aggregate_routes/src/config.rs b/aggregate_routes/src/config.rs
index 7de36bd..e21ece2 100644
--- a/aggregate_routes/src/config.rs
+++ b/aggregate_routes/src/config.rs
@@ -54,6 +54,8 @@ pub enum ODPattern {
         csv_path: String,
         /// Path to a GeoJSON file containing Points with a "name" property
         destinations_path: String,
+        /// If a zone doesn't have any matching origin points, use the zone's centroid instead.
+        origin_zone_centroid_fallback: bool,
     },
 }
 
diff --git a/aggregate_routes/src/od.rs b/aggregate_routes/src/od.rs
index a15ba9f..10da155 100644
--- a/aggregate_routes/src/od.rs
+++ b/aggregate_routes/src/od.rs
@@ -3,7 +3,7 @@ use std::io::BufReader;
 
 use anyhow::Result;
 use fs_err::File;
-use geo::{BoundingRect, Contains, MultiPolygon};
+use geo::{BoundingRect, Centroid, Contains, MultiPolygon};
 use geojson::{FeatureReader, Value};
 use indicatif::HumanCount;
 use nanorand::{Rng, WyRand};
@@ -81,8 +81,9 @@ pub fn generate(
             let zones = load_zones(&zones_path)?;
             timer.stop();
             timer.start("Matching points to zones");
-            let origins_per_zone = points_per_polygon("origin", origins, &zones)?;
-            let destinations_per_zone = points_per_polygon("destination", destinations, &zones)?;
+            let origins_per_zone = points_per_polygon("origin", origins, &zones, false)?;
+            let destinations_per_zone =
+                points_per_polygon("destination", destinations, &zones, false)?;
             timer.stop();
 
             timer.start(format!("Generating requests from {csv_path}"));
@@ -118,6 +119,7 @@ pub fn generate(
             zones_path,
             csv_path,
             destinations_path,
+            origin_zone_centroid_fallback,
         } => {
             let zones_path = format!("{input_directory}/{zones_path}");
             let csv_path = format!("{input_directory}/{csv_path}");
@@ -130,7 +132,8 @@ pub fn generate(
             let destinations = load_named_points(&destinations_path)?;
             timer.stop();
             timer.start("Matching points to zones");
-            let origins_per_zone = points_per_polygon("origin", origins, &zones)?;
+            let origins_per_zone =
+                points_per_polygon("origin", origins, &zones, origin_zone_centroid_fallback)?;
             timer.stop();
 
             timer.start(format!("Generating requests from {csv_path}"));
@@ -209,6 +212,7 @@ fn points_per_polygon(
     name: &str,
     points: Vec<(f64, f64)>,
     polygons: &HashMap<String, MultiPolygon<f64>>,
+    use_centroids_for_empty_zones: bool,
 ) -> Result<HashMap<String, Vec<(f64, f64)>>> {
     let tree = RTree::bulk_load(points);
 
@@ -231,9 +235,20 @@ fn points_per_polygon(
         output.insert(key.clone(), pts_inside);
     }
 
-    if !empty.is_empty() {
+    if !empty.is_empty() && !use_centroids_for_empty_zones {
         bail!("Some zones have no matching {name} points: {:?}", empty);
     }
+    println!(
+        "{} zones have no matching {name} points. Using centroid instead.",
+        HumanCount(empty.len() as u64)
+    );
+    for key in empty {
+        if let Some(centroid) = polygons[key].centroid() {
+            output.insert(key.clone(), vec![centroid.into()]);
+        } else {
+            bail!("{key} had no matching {name} points, and couldn't calculate its centroid");
+        }
+    }
 
     Ok(output)
 }
diff --git a/aggregate_routes/src/timer.rs b/aggregate_routes/src/timer.rs
index 108fb5a..2bb033c 100644
--- a/aggregate_routes/src/timer.rs
+++ b/aggregate_routes/src/timer.rs
@@ -74,7 +74,10 @@ impl Timer {
 impl Drop for Timer {
     fn drop(&mut self) {
         if let Some(current) = self.stack.last() {
-            println!("WARNING: Dropping timer during block {}. Probably crashing.", current.name);
+            println!(
+                "WARNING: Dropping timer during block {}. Probably crashing.",
+                current.name
+            );
             return;
         }
 
diff --git a/examples/liverpool/config.json b/examples/liverpool/config.json
index c3a05a0..419b727 100644
--- a/examples/liverpool/config.json
+++ b/examples/liverpool/config.json
@@ -5,7 +5,8 @@
 	"ZoneToPoint": {
 	  "zones_path": "zones.geojson",
 	  "destinations_path": "destinations.geojson",
-	  "csv_path": "od.csv"
+	  "csv_path": "od.csv",
+	  "origin_zone_centroid_fallback": true
         }
       }
     }
diff --git a/examples/york/config.json b/examples/york/config.json
index c3a05a0..b22cad8 100644
--- a/examples/york/config.json
+++ b/examples/york/config.json
@@ -5,7 +5,8 @@
 	"ZoneToPoint": {
 	  "zones_path": "zones.geojson",
 	  "destinations_path": "destinations.geojson",
-	  "csv_path": "od.csv"
+	  "csv_path": "od.csv",
+	  "origin_zone_centroid_fallback": false
         }
       }
     }