Improve the performance of reverse dependencies using the default_versions

Author: eth3lbert

src/models/krate.rs

@@ -17,7 +17,6 @@ use crate::models::{
 };
 use crate::util::errors::{version_not_found, AppResult};
 
-use crate::models::helpers::with_count::*;
 use crate::schema::*;
 use crate::sql::canon_crate_name;
 
@@ -433,18 +432,32 @@ impl Crate {
         conn: &mut PgConnection,
         options: PaginationOptions,
     ) -> QueryResult<(Vec<ReverseDependency>, i64)> {
+        use diesel::dsl::{exists, not, sql};
         use diesel::sql_query;
         use diesel::sql_types::{BigInt, Integer};
 
         let offset = options.offset().unwrap_or_default();
-        let rows: Vec<WithCount<ReverseDependency>> =
+        let rows: Vec<ReverseDependency> =
             sql_query(include_str!("krate_reverse_dependencies.sql"))
                 .bind::<Integer, _>(self.id)
                 .bind::<BigInt, _>(offset)
                 .bind::<BigInt, _>(options.per_page)
                 .load(conn)?;
 
-        Ok(rows.records_and_total())
+        let dep_versions_subq = dependencies::table
+            .filter(dependencies::crate_id.eq(self.id))
+            .select(dependencies::version_id);
+        let yanked_subq = versions::table
+            .filter(versions::id.eq(default_versions::version_id))
+            .filter(versions::yanked)
+            .select(sql::<Integer>("1"));
+        let total = default_versions::table
+            .filter(default_versions::version_id.eq_any(dep_versions_subq))
+            .filter(not(exists(yanked_subq)))
+            .count()
+            .get_result(conn)?;
+
+        Ok((rows, total))
     }
 
     /// Gather all the necessary data to write an index metadata file

src/models/krate_reverse_dependencies.sql

@@ -1,34 +1,20 @@
 SELECT
-    dependencies.*, crate_downloads, crate_name, total
+    dependencies.*, crate_downloads, crate_name
 FROM (
-    -- Apply pagination to the crates
-    SELECT *, COUNT(*) OVER () as total FROM (
-        SELECT
-            crate_downloads.downloads AS crate_downloads,
-            crates.name AS crate_name,
-            versions.id AS version_id
-        FROM
-        -- We only want the crates whose *max* version is dependent, so we join on a
-        -- subselect that includes the versions with their ordinal position
-        (
-            SELECT DISTINCT ON (crate_id)
-               crate_id, semver_no_prerelease, id
+    SELECT
+        crate_downloads.downloads AS crate_downloads,
+        crates.name AS crate_name,
+        version_id
+    FROM default_versions
+    INNER JOIN crates
+      ON crates.id = default_versions.crate_id
+    INNER JOIN crate_downloads using (crate_id)
+    WHERE version_id IN (SELECT version_id FROM dependencies WHERE crate_id = $1)
+        AND NOT EXISTS (
+            SELECT 1
             FROM versions
-            WHERE NOT yanked
-            ORDER BY
-                crate_id,
-                semver_no_prerelease DESC NULLS LAST,
-                id DESC
-        ) versions
-        INNER JOIN crates
-          ON crates.id = versions.crate_id
-        INNER JOIN crate_downloads
-          ON crate_downloads.crate_id = crates.id
-        WHERE versions.id IN (SELECT version_id FROM dependencies WHERE crate_id = $1)
-    ) c
-    ORDER BY
-        crate_downloads DESC,
-        crate_name ASC
+            WHERE id = version_id and yanked
+    )
 ) crates
 -- Multiple dependencies can exist, we only want first one
 CROSS JOIN LATERAL (
@@ -38,5 +24,8 @@ CROSS JOIN LATERAL (
     ORDER BY id ASC
     LIMIT 1
 ) dependencies
+ORDER BY
+    crate_downloads DESC,
+    crate_name ASC
 OFFSET $2
 LIMIT $3