adjustments

crates/schema/database/mysql_queries/src/queries/media_files/job/list_video_media_files_without_thumbnails_for_job.rs

@@ -1,14 +1,18 @@
-use chrono::{TimeDelta, Utc};
 use sqlx::MySqlPool;
 use enums::by_table::media_files::media_file_class::MediaFileClass;
 use tokens::tokens::media_files::MediaFileToken;
 
 /// The serialized value for `MediaFileClass::Video` as stored in the database.
 const MEDIA_CLASS_VIDEO: &str = MediaFileClass::Video.to_str();
-const PAGE_SIZE: i64 = 100;
+const DEFAULT_PAGE_SIZE: i64 = 100;
+const DEFAULT_MAX_LOOKBACK_HOURS: i32 = 4;
 
 pub struct ListVideoMediaFilesWithoutThumbnailsArgs<'a> {
-  pub time_delta: TimeDelta,
+  /// Override the maximum number of hours in our lookback window.
+  pub custom_max_lookback_hours: Option<i32>,
+  /// Override how many results to fetch (given a full page of results)
+  pub custom_page_size: Option<i64>,
+  /// Cursor to continue paginating with
   pub maybe_id_cursor: Option<i64>,
   pub pool: &'a MySqlPool,
 }
@@ -30,9 +34,13 @@ pub struct VideoMediaFilesWithoutThumbnails {
 pub async fn list_video_media_files_without_thumbnails_for_job(
   args: ListVideoMediaFilesWithoutThumbnailsArgs<'_>,
 ) -> Result<VideoMediaFilesWithoutThumbnails, sqlx::Error> {
-  let cutoff = Utc::now() - args.time_delta;
   let cursor = args.maybe_id_cursor.unwrap_or(i64::MAX);
 
+  let page_size = args.custom_page_size.unwrap_or(DEFAULT_PAGE_SIZE);
+  let max_lookback_hours = args.custom_max_lookback_hours.unwrap_or(DEFAULT_MAX_LOOKBACK_HOURS);
+
+  // NB: `COALESCE ...` helps us force the index for a more performant query plan (otherwise it explodes into a table scan).
+  // NB: `NOW() - INTERVAL ? HOUR` uses the database clock to avoid client/server clock skew.
   let media_files = sqlx::query_as!(
     VideoMediaFileWithoutThumbnail,
     r#"
@@ -45,7 +53,7 @@ SELECT
     maybe_public_bucket_extension
 FROM media_files
 WHERE
-    id >= (SELECT COALESCE(MIN(id), 0) FROM media_files WHERE created_at >= ?)
+    id >= (SELECT COALESCE(MIN(id), 0) FROM media_files WHERE created_at >= NOW() - INTERVAL ? HOUR)
     AND id < ?
     AND media_class = ?
     AND maybe_thumbnail_version IS NULL
@@ -54,15 +62,15 @@ WHERE
 ORDER BY id DESC
 LIMIT ?
     "#,
-    cutoff,
+    max_lookback_hours,
     cursor,
     MEDIA_CLASS_VIDEO,
-    PAGE_SIZE,
+    page_size,
   )
     .fetch_all(args.pool)
     .await?;
 
-  let next_cursor = if media_files.len() as i64 == PAGE_SIZE {
+  let next_cursor = if media_files.len() as i64 == page_size {
     media_files.last().map(|f| f.id)
   } else {
     None