From 495b74f52ed6d22d8def7e91a7c5dc23aef40894 Mon Sep 17 00:00:00 2001
From: missinglink <insomnia@rcpt.at>
Date: Fri, 3 Dec 2021 17:29:01 +0100
Subject: [PATCH] feat(perf): rewrite query for performance

---
 lib/Queries.js                                |  8 ++--
 .../match_subject_object_geom_intersects.sql  | 47 ++++++++++---------
 ...ct_object_geom_intersects_autocomplete.sql | 25 ----------
 3 files changed, 29 insertions(+), 51 deletions(-)
 delete mode 100644 query/match_subject_object_geom_intersects_autocomplete.sql

diff --git a/lib/Queries.js b/lib/Queries.js
index 5be380fa..9e0e4663 100644
--- a/lib/Queries.js
+++ b/lib/Queries.js
@@ -152,9 +152,9 @@ module.exports.matchSubjectObjectGeomIntersects = function( subject, object, cb
     if( '' === object.trim() ){ return cb( null, [] ); }
 
     this._queryAll(
-      this.prepare( query.match_subject_object_geom_intersects_autocomplete ),
+      this.prepare( query.match_subject_object_geom_intersects ),
       {
-        subject: subject,
+        subject: `"${subject}"`,
         object: `"${object}" OR "${object}"*`,
         threshold: RTREE_THRESHOLD,
         limit: MAX_RESULTS
@@ -163,9 +163,9 @@ module.exports.matchSubjectObjectGeomIntersects = function( subject, object, cb
     );
   } else {
     this._queryAll(
-      this.prepare( query.match_subject_object_geom_intersects_autocomplete ),
+      this.prepare( query.match_subject_object_geom_intersects ),
       {
-        subject: subject,
+        subject: `"${subject}"`,
         object: `"${object}"`,
         threshold: RTREE_THRESHOLD,
         limit: MAX_RESULTS
diff --git a/query/match_subject_object_geom_intersects.sql b/query/match_subject_object_geom_intersects.sql
index 95560335..8dd94c03 100644
--- a/query/match_subject_object_geom_intersects.sql
+++ b/query/match_subject_object_geom_intersects.sql
@@ -1,24 +1,27 @@
-SELECT t1.id AS subjectId, t2.id as objectId
-FROM rtree AS r1, rtree AS r2
-  JOIN tokens AS t1 ON t1.id = r1.id
-  JOIN tokens AS t2 ON t2.id = r2.id
-WHERE t1.token = $subject
-AND t2.token = $object
-AND (
-  t1.lang = t2.lang OR
-  t1.lang IN ( 'eng', 'und' ) OR
-  t2.lang IN ( 'eng', 'und' )
-)
--- https://silentmatt.com/rectangle-intersection/
-AND (
-  r1.maxZ < r2.minZ AND
-  r1.minX - $threshold < r2.maxX AND
-  r1.maxX + $threshold > r2.minX AND
-  r1.minY - $threshold < r2.maxY AND
-  r1.maxY + $threshold > r2.minY
-)
--- AND t1.tag NOT IN ( 'colloquial' )
--- AND t2.tag NOT IN ( 'colloquial' )
+SELECT
+  t1.id AS subjectId,
+  t2.id as objectId
+FROM fulltext f1
+  JOIN tokens t1 ON f1.rowid = t1.rowid
+    JOIN rtree AS r1 ON t1.id = r1.id
+      JOIN rtree AS r2 ON (
+        r1.maxZ < r2.minZ AND
+        (r1.minX - $threshold) < r2.maxX AND
+        (r1.maxX + $threshold) > r2.minX AND
+        (r1.minY - $threshold) < r2.maxY AND
+        (r1.maxY + $threshold) > r2.minY
+      )
+        JOIN fulltext AS f2 ON f2.fulltext MATCH $object
+          JOIN tokens t2 ON (
+            f2.rowid = t2.rowid
+            AND r2.id = t2.id
+            AND (
+              t1.lang = t2.lang OR
+              t1.lang IN ('eng', 'und') OR
+              t2.lang IN ('eng', 'und')
+            )
+          )
+WHERE f1.fulltext MATCH $subject
 GROUP BY t1.id, t2.id
 ORDER BY t1.id ASC, t2.id ASC
-LIMIT $limit
\ No newline at end of file
+LIMIT $limit
diff --git a/query/match_subject_object_geom_intersects_autocomplete.sql b/query/match_subject_object_geom_intersects_autocomplete.sql
deleted file mode 100644
index d167c2b3..00000000
--- a/query/match_subject_object_geom_intersects_autocomplete.sql
+++ /dev/null
@@ -1,25 +0,0 @@
-SELECT t1.id AS subjectId, t2.id as objectId
-FROM rtree AS r1, rtree AS r2
-  JOIN tokens AS t1 ON t1.id = r1.id
-  JOIN tokens AS t2 ON t2.id = r2.id
-    JOIN fulltext AS f2 ON f2.rowid = t2.rowid
-WHERE t1.token = $subject
-AND f2.fulltext MATCH $object
-AND (
-  t1.lang = t2.lang OR
-  t1.lang IN ( 'eng', 'und' ) OR
-  t2.lang IN ( 'eng', 'und' )
-)
--- https://silentmatt.com/rectangle-intersection/
-AND (
-  r1.maxZ < r2.minZ AND
-  r1.minX - $threshold < r2.maxX AND
-  r1.maxX + $threshold > r2.minX AND
-  r1.minY - $threshold < r2.maxY AND
-  r1.maxY + $threshold > r2.minY
-)
--- AND t1.tag NOT IN ( 'colloquial' )
--- AND t2.tag NOT IN ( 'colloquial' )
-GROUP BY t1.id, t2.id
-ORDER BY t1.id ASC, t2.id ASC
-LIMIT $limit