Optimize top frecent query (#6969)

* Optimize top frecent query

* add index to get perf speedup

Author: skhamis

components/places/sql/create_shared_schema.sql

@@ -49,6 +49,12 @@ CREATE INDEX IF NOT EXISTS idx_places_outgoing_by_frecency
 ON moz_places(frecency DESC)
 WHERE hidden = 0 AND (sync_change_counter > 0 OR sync_status != 2); -- 2 is SyncStatus::Normal
 
+-- partial index for get_top_frecent_site_infos
+CREATE INDEX IF NOT EXISTS top_frecent_cover_idx
+ON moz_places(frecency DESC, id DESC)
+WHERE hidden = 0
+  AND (last_visit_date_local + last_visit_date_remote) != 0
+  AND (url GLOB 'http:*' OR url GLOB 'https:*');
 
 CREATE TABLE IF NOT EXISTS moz_places_tombstones (
     guid TEXT PRIMARY KEY
@@ -84,6 +90,9 @@ CREATE INDEX IF NOT EXISTS fromindex ON moz_historyvisits(from_visit);
 CREATE INDEX IF NOT EXISTS dateindex ON moz_historyvisits(visit_date);
 CREATE INDEX IF NOT EXISTS islocalindex ON moz_historyvisits(is_local);
 
+-- Speeds up queries frecency queries, specifically get_top_frecent_site_infos
+CREATE INDEX IF NOT EXISTS idx_visits_place_type ON moz_historyvisits(place_id, visit_type);
+
 -- Greatly helps the multi-join query in frecency.
 CREATE INDEX IF NOT EXISTS visits_from_type_idx ON moz_historyvisits(from_visit, visit_type);
 

components/places/src/db/schema.rs

@@ -14,7 +14,7 @@ use sql_support::ConnExt;
 
 use super::db::{Pragma, PragmaGuard};
 
-pub const VERSION: u32 = 19;
+pub const VERSION: u32 = 20;
 
 // Shared schema and temp tables for the read-write and Sync connections.
 const CREATE_SHARED_SCHEMA_SQL: &str = include_str!("../../sql/create_shared_schema.sql");
@@ -333,6 +333,14 @@ pub fn upgrade_from(db: &Connection, from: u32) -> rusqlite::Result<()> {
             // Manually call analyze so the planner can start using the indexes immediately
             db.execute("ANALYZE moz_places", [])?;
         }
+        19 => {
+            // Create the new indexes by just calling the shared schema file
+            // top_frecent_cover_idx, idx_visits_place_type
+            db.execute_batch(CREATE_SHARED_SCHEMA_SQL)?;
+            // Manually call analyze so the planner can start using the indexes immediately
+            db.execute("ANALYZE moz_places", [])?;
+            db.execute("ANALYZE moz_historyvisits", [])?;
+        }
         // Add more migrations here...
 
         // Any other from value indicates that something very wrong happened

components/places/src/storage/history.rs

@@ -1357,17 +1357,18 @@ pub fn get_top_frecent_site_infos(
     let infos = db.query_rows_and_then_cached(
         "SELECT h.frecency, h.title, h.url
         FROM moz_places h
-        WHERE EXISTS (
-            SELECT v.visit_type
+        WHERE h.hidden = 0
+        AND (h.last_visit_date_local + h.last_visit_date_remote) != 0
+        AND (h.url LIKE 'http:%' OR h.url LIKE 'https:%')
+        AND h.frecency >= :frecency_threshold
+        AND EXISTS (
+            SELECT 1
             FROM moz_historyvisits v
-            WHERE h.id = v.place_id
-              AND (SUBSTR(h.url, 1, 6) == 'https:' OR SUBSTR(h.url, 1, 5) == 'http:')
-              AND (h.last_visit_date_local + h.last_visit_date_remote) != 0
-              AND ((1 << v.visit_type) & :allowed_types) != 0
-              AND h.frecency >= :frecency_threshold AND
-              NOT h.hidden
+            WHERE v.place_id = h.id
+            AND ((1 << v.visit_type) & :allowed_types) != 0
+            LIMIT 1
         )
-        ORDER BY h.frecency DESC
+        ORDER BY h.frecency DESC, h.id DESC
         LIMIT :limit",
         rusqlite::named_params! {
             ":limit": num_items,

testing/separated/places-bench/src/bench.rs

@@ -8,14 +8,17 @@ mod database;
 mod matching;
 
 use criterion::{criterion_group, criterion_main};
-use database::{bench_match_url, bench_outgoing_candidates, bench_search_frecent};
+use database::{
+    bench_match_url, bench_outgoing_candidates, bench_search_frecent, bench_top_frecent,
+};
 use matching::bench_match_anywhere;
 
 criterion_group!(
     bench_db,
     bench_search_frecent,
     bench_match_url,
     bench_outgoing_candidates,
+    bench_top_frecent
 );
 criterion_group!(bench_mem, bench_match_anywhere);
 criterion_main!(bench_db, bench_mem);