owen/lms
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Simplified queries + added some hints for the query planner for bad quality indexes

Browse Source
This commit is contained in:
emeric
2024-04-17 19:01:28 +02:00
parent ba1fb4a6d6
commit 87daa5c767
1 changed files with 21 additions and 28 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/libs/database/impl/Release.cpp
+21 -28
View File
@@ -148,8 +148,7 @@ namespace lms::db
else if (params.clusters.size() > 1)
{
std::ostringstream oss;
oss << "r.id IN (SELECT DISTINCT r.id FROM release r"
" INNER JOIN track t ON t.release_id = r.id"
oss << "r.id IN (SELECT DISTINCT t.release_id FROM track t"
" INNER JOIN track_cluster t_c ON t_c.track_id = t.id";
WhereClause clusterClause;
@@ -334,17 +333,14 @@ namespace lms::db
{
session.checkReadTransaction();
return utils::fetchQuerySingleResult(createQuery<int>(session, "COUNT(DISTINCT r.id)", params));
return utils::fetchQuerySingleResult(createQuery<int>(session, "COUNT(r.id)", params));
}
std::size_t Release::getDiscCount() const
{
assert(session());
int res{ utils::fetchQuerySingleResult(session()->query<int>("SELECT COUNT(DISTINCT disc_number) FROM track t")
.join("release r ON r.id = t.release_id")
.where("r.id = ?")
.bind(getId())) };
.where("t.release_id = ?").bind(getId()))};
return res;
}
@@ -352,13 +348,10 @@ namespace lms::db
{
assert(session());
using ResultType = std::tuple<int, std::string>;
const auto query{ session()->query<ResultType>("SELECT DISTINCT disc_number, disc_subtitle FROM track t")
.join("release r ON r.id = t.release_id")
.where("r.id = ?")
.orderBy("disc_number")
.bind(getId()) };
.where("t.release_id = ?").bind(getId())
.orderBy("disc_number")};
std::vector<DiscInfo> discs;
utils::forEachQueryResult(query, [&](ResultType&& res)
@@ -384,8 +377,8 @@ namespace lms::db
assert(session());
const char* field{ original ? "original_date" : "date" };
auto query{ (session()->query<Wt::WDate>(std::string {"SELECT "} + "t." + field + " FROM track t INNER JOIN release r ON r.id = t.release_id")
.where("r.id = ?")
auto query{ (session()->query<Wt::WDate>(std::string {"SELECT "} + "t." + field + " FROM track t")
.where("t.release_id = ?")
.groupBy(field)
.bind(getId())) };
@@ -413,8 +406,8 @@ namespace lms::db
assert(session());
const char* field{ original ? "original_year" : "year" };
auto query{ session()->query<std::optional<int>>(std::string {"SELECT "} + "t." + field + " FROM track t INNER JOIN release r ON r.id = t.release_id")
.where("r.id = ?").bind(getId())
auto query{ session()->query<std::optional<int>>(std::string {"SELECT "} + "t." + field + " FROM track t")
.where("t.release_id = ?").bind(getId())
.groupBy(field) };
const auto years{ utils::fetchQueryResults(query) };
@@ -477,12 +470,12 @@ namespace lms::db
assert(session());
const auto query{ session()->query<Wt::Dbo::ptr<Artist>>(
"SELECT DISTINCT a FROM artist a"
"SELECT a FROM artist a"
" INNER JOIN track_artist_link t_a_l ON t_a_l.artist_id = a.id"
" INNER JOIN track t ON t.id = t_a_l.track_id"
" INNER JOIN release r ON r.id = t.release_id")
.where("r.id = ?").bind(getId())
.where("t_a_l.type = ?").bind(linkType) };
" INNER JOIN track t ON t.id = t_a_l.track_id")
.where("t.release_id = ?").bind(getId())
.where("+t_a_l.type = ?").bind(linkType) // adding + since the query planner does not a good job when analyze is not performed
.groupBy("a.id") };
return utils::fetchQueryResults<Artist::pointer>(query);
}
@@ -560,16 +553,16 @@ namespace lms::db
using milli = std::chrono::duration<int, std::milli>;
return utils::fetchQuerySingleResult(session()->query<milli>("SELECT COALESCE(SUM(duration), 0) FROM track t INNER JOIN release r ON t.release_id = r.id")
.where("r.id = ?").bind(getId()));
return utils::fetchQuerySingleResult(session()->query<milli>("SELECT COALESCE(SUM(duration), 0) FROM track t")
.where("t.release_id = ?").bind(getId()));
}
Wt::WDateTime Release::getLastWritten() const
{
assert(session());
return utils::fetchQuerySingleResult(session()->query<Wt::WDateTime>("SELECT COALESCE(MAX(file_last_write), '1970-01-01T00:00:00') FROM track t INNER JOIN release r ON t.release_id = r.id")
.where("r.id = ?").bind(getId()));
return utils::fetchQuerySingleResult(session()->query<Wt::WDateTime>("SELECT COALESCE(MAX(file_last_write), '1970-01-01T00:00:00') FROM track t")
.where("t.release_id = ?").bind(getId()));
}
std::vector<std::vector<Cluster::pointer>> Release::getClusterGroups(const std::vector<ClusterTypeId>& clusterTypeIds, std::size_t size) const
@@ -580,13 +573,13 @@ namespace lms::db
std::ostringstream oss;
oss << "SELECT c from cluster c INNER JOIN track t ON c.id = t_c.cluster_id INNER JOIN track_cluster t_c ON t_c.track_id = t.id INNER JOIN cluster_type c_type ON c.cluster_type_id = c_type.id INNER JOIN release r ON t.release_id = r.id ";
oss << "SELECT c from cluster c INNER JOIN track t ON c.id = t_c.cluster_id INNER JOIN track_cluster t_c ON t_c.track_id = t.id INNER JOIN cluster_type c_type ON c.cluster_type_id = c_type.id ";
where.And(WhereClause("r.id = ?")).bind(getId().toString());
where.And(WhereClause("t.release_id = ?")).bind(getId().toString());
{
WhereClause clusterClause;
for (const ClusterTypeId clusterTypeId : clusterTypeIds)
clusterClause.Or(WhereClause("c_type.id = ?")).bind(clusterTypeId.toString());
clusterClause.Or(WhereClause("+c_type.id = ?")).bind(clusterTypeId.toString()); // Exclude this since the query planner does not do a good job when db is not analyzed
where.And(clusterClause);
}
oss << " " << where.get();