From fdebc857539e8a73a391ba03d588b0bdcd2d676e Mon Sep 17 00:00:00 2001
From: dull b <dull.bananas0@gmail.com>
Date: Sat, 16 Dec 2023 05:15:08 +0000
Subject: [PATCH] stuff including batch_upsert function

---
 crates/db_schema/src/impls/comment.rs | 26 +++++++++++++++++++++-----
 crates/db_schema/src/utils.rs         | 15 ++++++++++++++-
 scripts/start_dev_db.sh               | 17 +++++++++++++++--
 3 files changed, 50 insertions(+), 8 deletions(-)

diff --git a/crates/db_schema/src/impls/comment.rs b/crates/db_schema/src/impls/comment.rs
index aef399c59..f59afb6aa 100644
--- a/crates/db_schema/src/impls/comment.rs
+++ b/crates/db_schema/src/impls/comment.rs
@@ -57,21 +57,37 @@ impl Comment {
     comment_form: &CommentInsertForm,
     parent_path: Option<&Ltree>,
   ) -> Result<Comment, Error> {
+    Comment::create_batch(pool, &[(comment_form, parent_path)]).await?.into_iter().next().ok_or(Error::NotFound)
+  }
+  
+  pub async fn create_batch(
+    pool: &mut DbPool<'_>,
+    items: &[(&CommentInsertForm, Option<&Ltree>)],
+  ) -> Result<Vec<Comment>, Error> {
     let conn = &mut get_conn(pool).await?;
 
     conn
       .build_transaction()
       .run(|conn| {
         Box::pin(async move {
-          // Insert, to get the id
-          let inserted_comment = insert_into(comment)
-            .values(comment_form)
-            .on_conflict(ap_id)
+          let forms = items
+            .iter()
+            .map(|&(form, _)| form)
+            .collect::<Vec<_>>();
+
+          // Insert, to get the ids
+          let inserted_comments = insert_into(comment)
+            .values(&forms)
+            /*.on_conflict(ap_id)
             .do_update()
-            .set(comment_form)
+            .set()*/
             .get_result::<Self>(conn)
             .await?;
 
+          // `ap_id` unique constraint violation is handled individually for each row
+          // because batch upsert requires having the same `set` argument for all rows
+          
+
           let comment_id = inserted_comment.id;
 
           // You need to update the ltree column
diff --git a/crates/db_schema/src/utils.rs b/crates/db_schema/src/utils.rs
index b181edd7f..7adf009a4 100644
--- a/crates/db_schema/src/utils.rs
+++ b/crates/db_schema/src/utils.rs
@@ -19,7 +19,7 @@ use diesel::{
   sql_types::{Text, Timestamptz},
   IntoSql,
   PgConnection,
-  RunQueryDsl,
+  RunQueryDsl, Insertable, Table, Column, AsChangeset, Expression, SelectableExpression, expression::NonAggregate, query_builder::QueryFragment,
 };
 use diesel_async::{
   pg::AsyncPgConnection,
@@ -153,6 +153,19 @@ macro_rules! try_join_with_pool {
   }};
 }
 
+pub async fn batch_upsert<T, U, Target, R>(conn: &mut AsyncPgConnection, target: T, records: U, conflict_target: Target) -> Result<Vec<R>, DieselError>
+where
+  T: Table,
+  T::AllColumns: Expression + SelectableExpression<T> + NonAggregate + QueryFragment<Pg>,
+  U: IntoIterator + Clone,
+  Vec<U::Item>: Insertable<T>,
+  U::Item: Insertable<T> + AsChangeset<Target = T>,
+  Target: Column<Table = T>,
+{
+  let result = diesel::insert_into(target).values(records.clone().into_iter().collect::<Vec<_>>()).load::<R>(conn).await;
+  
+}
+
 pub fn fuzzy_search(q: &str) -> String {
   let replaced = q.replace('%', "\\%").replace('_', "\\_").replace(' ', "%");
   format!("%{replaced}%")
diff --git a/scripts/start_dev_db.sh b/scripts/start_dev_db.sh
index c16969ed7..d25b15e75 100644
--- a/scripts/start_dev_db.sh
+++ b/scripts/start_dev_db.sh
@@ -18,10 +18,23 @@ then
 fi
 
 # Create cluster
-initdb --username=postgres --auth=trust --no-instructions
+initdb --username=postgres --auth=trust --no-instructions \
+  # Only listen to socket in current directory
+  -c listen_addresses= -c unix_socket_directories=$PWD \
+  # Write logs to a file in $PGDATA/log
+  -c logging_collector=on \
+  # Log all query plans by default
+  -c session_preload_libraries=auto_explain -c auto_explain.log_min_duration=0
+  # Include actual row amounts and run times for query plan nodes
+  -c auto_explain.log_analyze=on
+  # Avoid sequential scans so query plans show what index scans can be done
+  # (index scan is normally avoided in some cases, such as the table being small enough)
+  -c enable_seqscan=off
+  # Don't log parameter values
+  -c auto_explain.log_parameter_max_length=0
 
 # Start server that only listens to socket in current directory
-pg_ctl start --options="-c listen_addresses= -c unix_socket_directories=$PWD -c logging_collector=on -c session_preload_libraries=auto_explain -c auto_explain.log_min_duration=0 -c auto_explain.log_parameter_max_length=0 -c auto_explain.log_analyze=on -c enable_seqscan=off" > /dev/null
+pg_ctl start
 
 # Setup database
 psql -c "CREATE USER lemmy WITH PASSWORD 'password' SUPERUSER;" -U postgres