Fast location queries using KDBush.

package.json

@@ -15,6 +15,7 @@
     "@testing-library/react": "^9.3.2",
     "@testing-library/user-event": "^7.1.2",
     "@turf/points-within-polygon": "^5.1.5",
+    "@turf/turf": "^5.1.6",
     "alasql": "^0.5.4",
     "console": "^0.7.2",
     "date-fns": "^2.0.0",

src/database.js

@@ -1,9 +1,12 @@
 import React, { useEffect, useState, useContext } from 'react';
 import NodeEnvironment from 'jest-environment-node';
 
-import { add_geo_area_to_store, get_geo_area_from_store } from './geo_store.js';
-import * as turf from '@turf/turf'
+import {
+  add_geo_area_to_store, get_geo_area_from_store,
+  initialize_image_index, image_in_area
+} from './geo_store.js';
 
+import * as turf from '@turf/turf'
 import KDBush from 'kdbush';
 
 export async function sqljs_async_queries(sqljs_object, queries) {
@@ -52,11 +55,8 @@ export function regexp_match(string, regex) {
   return string.match(regex) != null;
 }
 
-export function is_in_geo_polygon_from_store(lat, long, polygon_hash) {
-  const area = get_geo_area_from_store(polygon_hash);
-  var gju = require('geojson-utils');
-  return gju.pointInPolygon({ "type": "Point", "coordinates": [long, lat] },
-    area.geojson);
+export function is_in_geo_polygon_from_store(image_id, lat, long, polygon_hash) {
+  return image_in_area(image_id, polygon_hash);
 }
 
 // Digikam stores its tree of tags as individual tags,
@@ -109,15 +109,18 @@ export async function add_full_tag_info(db) {
 }
 
 function polygons_benchmark(database) {
-  var img_query = "SELECT Images.id, ImagePositions.latitudeNumber, ImagePositions.longitudeNumber FROM Images "
-    + "LEFT JOIN ImagePositions ON ImagePositions.imageid=Images.id WHERE ImagePositions.latitudeNumber NOT NULL GROUP BY Images.id;";
+  var img_query = "SELECT GROUP_CONCAT(Images.id), ImagePositions.latitudeNumber, ImagePositions.longitudeNumber FROM Images "
+    + "LEFT JOIN ImagePositions ON ImagePositions.imageid=Images.id WHERE ImagePositions.latitudeNumber NOT NULL GROUP BY ImagePositions.latitudeNumber, ImagePositions.longitudeNumber;";
 
   sqljs_async_queries(database, [img_query]).then(res => {
+    console.log("Images query result:", res);
+
     fetch("https://nominatim.openstreetmap.org/search?polygon_geojson=1&polygon_threshold=0.001&format=json&limit=1&q=Australia")
       .then(res => res.json())
       .then(jsonres => {
         var polies;
-        var points = [];
+        var points = []; // Contains all unique geometric points as [lat long]
+        var ids_per_point = []; // Contains a list of ids for each unique geometric point
 
         console.log("Nominatim geo answer:", jsonres);
 
@@ -130,10 +133,12 @@ function polygons_benchmark(database) {
           var data = res[0].values;
           data.forEach(row => {
             points.push([parseFloat(row[cols.indexOf("longitudeNumber")]), parseFloat(row[cols.indexOf("latitudeNumber")])]);
+            ids_per_point.push(row[cols.indexOf("GROUP_CONCAT(Images.id)")].split(',').map(e => parseInt(e)));
           });
         }
 
         console.log("Points: ", points);
+        console.log("IDs per point: ", ids_per_point);
         console.log("Nominatim GEOJSON: ", polies);
 
         {
@@ -151,6 +156,8 @@ function polygons_benchmark(database) {
           const index = new KDBush(points, p => p[0], p => p[1], 16, Float64Array);
           console.timeEnd("Build KDBush index");
 
+          console.time("Total KDBush search time");
+
           // Get bounding boxes for all subpolygons
           var boxes = [];
           console.time("Build KDBush poly boxes");
@@ -184,21 +191,26 @@ function polygons_benchmark(database) {
           // Test hit points exactly
           console.time("Get exact hits after KDBush");
           var realhits = [];
-          var hitcache = {};
           hits.forEach(hit => {
             const point = points[hit];
-            if(point in hitcache && hitcache[point]) {
-              realhits.push(hit);
-              return;
-            }
             const is_real_hit = turf.booleanPointInPolygon(turf.point(points[hit]), polies);
-            hitcache[point] = is_real_hit;
             if (is_real_hit) {
               realhits.push(hit);
             }
           });
           console.log("Real: ", realhits);
           console.timeEnd("Get exact hits after KDBush");
+
+          // Expand to all image IDs
+          console.time("Expand exact hits to IDs");
+          var hit_ids = [];
+          realhits.forEach(idx => {
+            hit_ids = hit_ids.concat(ids_per_point[idx]);
+          });
+          console.log("Real hit IDs: ", hit_ids);
+          console.timeEnd("Expand exact hits to IDs");
+
+          console.timeEnd("Total KDBush search time");
         }
       });
   })
@@ -213,14 +225,17 @@ export function ProvideDB(props) {
   useEffect(() => {
     fetch_sqljs_db_from_sqlite(db_url)
       .then(db => {
-        add_full_tag_info(db).then((newdb) => {
+        add_full_tag_info(db)
+          .then((newdb) => {
+            initialize_image_index(newdb).then(() => {
               db.create_function("REGEXP", regexp_match);
               db.create_function("IS_IN_GEO", is_in_geo_polygon_from_store);
-          polygons_benchmark(db);
+              //polygons_benchmark(db);
               setError(false);
               setDb(newdb);
             })
           })
+      })
       .catch(error => { setError(error); });
   }, [])
 

src/geo_store.js

@@ -1,4 +1,13 @@
-var g_GeoStore = {};
+import KDBush from 'kdbush';
+
+import { sqljs_async_queries } from './database.js';
+import * as turf from '@turf/turf';
+
+var g_GeoStore = {
+    areas: {},
+    area_query_results: {},
+    image_index: {},
+};
 
 export function hash_geo_area(geo_area) {
     var hash = require('object-hash');
@@ -8,7 +17,7 @@ export function hash_geo_area(geo_area) {
 export function add_geo_area_to_store(area) {
     var h = hash_geo_area(area);
     if (!(h in g_GeoStore)) {
-        g_GeoStore[h] = area;
+        g_GeoStore["areas"][h] = area;
     }
 
     return h;
@@ -16,7 +25,119 @@ export function add_geo_area_to_store(area) {
 
 export function get_geo_area_from_store(hash) {
     if (hash in g_GeoStore) {
-        return g_GeoStore[hash];
+        return g_GeoStore["areas"][hash];
     }
     throw new Error("Requested non-existent geo area from store.");
 }
+
+export function initialize_image_index(database) {
+    return new Promise((resolve, reject) => {
+        var img_query = "SELECT GROUP_CONCAT(Images.id), ImagePositions.latitudeNumber, ImagePositions.longitudeNumber FROM Images "
+            + "LEFT JOIN ImagePositions ON ImagePositions.imageid=Images.id WHERE ImagePositions.latitudeNumber NOT NULL GROUP BY ImagePositions.latitudeNumber, ImagePositions.longitudeNumber;";
+
+        sqljs_async_queries(database, [img_query])
+            .then(res => {
+                var points = []; // will contain a list of unique [long, lat] points
+                var ids_per_point = []; // will contain a list of image ids per point in "points", same indexing
+                if (res && Array.isArray(res) && res.length > 0) {
+                    var cols = res[0].columns;
+                    var data = res[0].values;
+                    data.forEach(row => {
+                        points.push([parseFloat(row[cols.indexOf("longitudeNumber")]), parseFloat(row[cols.indexOf("latitudeNumber")])]);
+                        ids_per_point.push(row[cols.indexOf("GROUP_CONCAT(Images.id)")].split(',').map(e => parseInt(e)));
+                    });
+                }
+
+                // Store all info we will need later into the image index.
+                g_GeoStore["image_index"] = {
+                    points: points,
+                    ids_per_point: ids_per_point,
+                    kdbush: new KDBush(points, p => p[0], p => p[1], 16, Float64Array)
+                };
+
+                resolve();
+            })
+            .catch(e => { reject(e); });
+    });
+}
+
+// Do a query whether an image is in an area. If not done before, this will
+// calculate all images in said area and cache the result for later calls.
+export function image_in_area(image_id, area_hash) {
+    if (area_hash in g_GeoStore["area_query_results"]) {
+        return g_GeoStore["area_query_results"][area_hash].includes(parseInt(image_id));
+    }
+    if (!(area_hash in g_GeoStore["areas"])) {
+        throw new Error("Queried for images in an unknown area.");
+    }
+
+    //console.time("Total KDBush search time");
+
+    // Get bounding boxes for all (sub-)polygons of the area
+    const full_area = g_GeoStore["areas"][area_hash];
+    //console.log("Search area:", full_area);
+    const area = full_area.geojson;
+    var boxes = [];
+    const add_box = poly => {
+        var minx = Number.POSITIVE_INFINITY;
+        var miny = Number.POSITIVE_INFINITY;
+        var maxx = Number.NEGATIVE_INFINITY;
+        var maxy = Number.NEGATIVE_INFINITY;
+        for (let j = 0; j < poly.length; j++) {
+            minx = Math.min(minx, poly[j][0]);
+            miny = Math.min(miny, poly[j][1]);
+            maxx = Math.max(maxx, poly[j][0]);
+            maxy = Math.max(maxy, poly[j][1]);
+        }
+        boxes.push([minx, miny, maxx, maxy]);
+    }
+    if (area.type == "Polygon") {
+        add_box(area.coordinates[0]);
+    } else if (area.type == "MultiPolygon") {
+        for (let i = 0; i < area.coordinates.length; i++) {
+            add_box(area.coordinates[i][0]);
+        }
+    }
+
+    // Test points in KD tree against each subpolygon bounding box
+    //console.time("Find box points in KDBush");
+    let hits = new Set();
+    const index = g_GeoStore["image_index"].kdbush;
+    for (let i = 0; i < boxes.length; i++) {
+        const idxs = index.range(boxes[i][0], boxes[i][1], boxes[i][2], boxes[i][3]);
+        idxs.forEach(e => hits.add(e));
+    }
+    //console.timeEnd("Find box points in KDBush");
+    //console.log("Hits: ", hits);
+
+    // Test hit points exactly
+    // console.time("Get exact hits after KDBush");
+    var realhits = [];
+    const points = g_GeoStore["image_index"].points;
+    hits.forEach(hit => {
+        const point = points[hit];
+        const is_real_hit = turf.booleanPointInPolygon(turf.point(point), area);
+        if (is_real_hit) {
+            realhits.push(hit);
+        }
+    });
+    //console.log("Real hits: ", realhits);
+    // console.timeEnd("Get exact hits after KDBush");
+
+    // Expand to all image IDs
+    // console.time("Expand exact hits to IDs");
+    var hit_ids = [];
+    const ids_per_point = g_GeoStore["image_index"].ids_per_point;
+    realhits.forEach(idx => {
+        hit_ids = hit_ids.concat(ids_per_point[idx]);
+    });
+    //console.log("Real hit ImageIds: ", hit_ids);
+    // console.timeEnd("Expand exact hits to IDs");
+
+    //console.timeEnd("Total KDBush search time");
+
+    g_GeoStore["area_query_results"][area_hash] = hit_ids;
+
+    //console.log("Is ", parseInt(image_id), " inside ", g_GeoStore["area_query_results"][area_hash], "?");
+    return g_GeoStore["area_query_results"][area_hash].includes(parseInt(image_id));
+}

src/queries.js

@@ -13,7 +13,7 @@ export function do_image_query(query, database, collection_path, collection_thum
         var queries = [];
         queries.push(query);
         sqljs_async_queries(database, queries).then(res => {
-            //console.log("Query result: ", res);
+            console.log("Query result: ", res);
             var photos = [];
             if (res && Array.isArray(res) && res.length > 0) {
                 var cols = res[0].columns;
@@ -320,7 +320,7 @@ export class LocationFilter extends ResultFilter {
         // it will have access to said polygon.
         const hash = add_geo_area_to_store(this.geo_area);
         return '(ImagePositions.latitudeNumber NOT NULL AND ImagePositions.longitudeNumber NOT NULL AND '
-            + 'IS_IN_GEO(ImagePositions.latitudeNumber, ImagePositions.longitudeNumber, "' + hash + '"))';
+            + 'IS_IN_GEO(Images.id, ImagePositions.latitudeNumber, ImagePositions.longitudeNumber, "' + hash + '"))';
     }
 
     simplify() { return this; }