perf(nextcloud): batch oc:fileid resolution to kill PROPFIND N+1

Resolving the stable numeric oc:fileid for every child in a NextCloud
listing issued one `INSERT ... ON CONFLICT DO UPDATE` per entry — a write
(row rewrite + WAL + dead tuple) even when the mapping already existed.
A Depth:1 PROPFIND of a folder with N children meant N sequential write
round-trips on a read-only operation that sync clients repeat constantly.

- Repository: replace the single `get_or_create` (DO UPDATE) with
  `get_or_create_many` — one idempotent bulk `INSERT ... SELECT unnest(...)
  ON CONFLICT DO NOTHING` (existing rows untouched) plus a single
  `SELECT ... WHERE object_id = ANY(...)`. Two statements instead of N.
- Service: add an Arc-backed moka cache (uuid -> i64; the mapping is
  immutable, so warm entries never go stale) and batch APIs
  `get_or_create_file_ids` / `get_or_create_folder_ids` that only query
  the misses. Warm listings cost zero queries.
- Handlers (PROPFIND, REPORT favorites/search, trashbin, OCS unified
  search): pre-resolve all ids in two batched queries — file and folder
  run concurrently via `tokio::join!` — and turn the XML/JSON emission
  into a synchronous map lookup.

https://claude.ai/code/session_01Dp3oWon5GBMVn4j3QXZdgx
This commit is contained in:
Claude
2026-06-10 08:39:06 +00:00
parent da6dcb3771
commit 616e48b338
6 changed files with 304 additions and 113 deletions
+41 -22
View File
@@ -9,7 +9,7 @@ use quick_xml::{
Writer,
events::{BytesEnd, BytesStart, BytesText, Event},
};
use std::collections::HashSet;
use std::collections::{HashMap, HashSet};
use std::sync::Arc;
use crate::application::adapters::webdav_adapter::{PropFindRequest, WebDavAdapter};
@@ -1001,6 +1001,26 @@ async fn write_nc_multistatus<W: std::io::Write>(
file_id_svc: Option<&Arc<NextcloudFileIdService>>,
favorite_ids: &HashSet<String>,
) -> Result<(), String> {
// When folder is None, files are the target resource itself (single-file
// PROPFIND) and must always be emitted. When folder is Some, files/subfolders
// are children and should only be listed when depth > 0.
let emit_children = folder.is_none() || depth != "0";
// Pre-resolve every oc:fileid up front in two batch queries (one per
// object type) instead of one INSERT round-trip per child. The XML writing
// below is then a pure synchronous map lookup.
let mut file_uuids: Vec<String> = Vec::new();
let mut folder_uuids: Vec<String> = Vec::new();
if let Some(f) = folder {
folder_uuids.push(f.id.clone());
}
if emit_children {
file_uuids.extend(files.iter().map(|f| f.id.clone()));
folder_uuids.extend(subfolders.iter().map(|sf| sf.id.clone()));
}
let (file_id_map, folder_id_map) =
batch_resolve_ids(file_id_svc, &file_uuids, &folder_uuids).await;
let mut xml = Writer::new(writer);
// Root element with all required namespaces.
@@ -1015,7 +1035,7 @@ async fn write_nc_multistatus<W: std::io::Write>(
// §5.2 + strict NC-client enforcement — see `nc_collection_href`).
if let Some(f) = folder {
let href = nc_collection_href(username, subpath);
let file_id = resolve_folder_id(file_id_svc, &f.id).await;
let file_id = folder_id_map.get(&f.id).copied();
let oc_id = file_id.map(|id| format_oc_id(id, file_id_svc));
write_folder_response(
&mut xml,
@@ -1028,11 +1048,6 @@ async fn write_nc_multistatus<W: std::io::Write>(
)?;
}
// When folder is None, files are the target resource itself (single-file
// PROPFIND) and must always be emitted. When folder is Some, files/subfolders
// are children and should only be listed when depth > 0.
let emit_children = folder.is_none() || depth != "0";
if emit_children {
// Files.
for file in files {
@@ -1045,7 +1060,7 @@ async fn write_nc_multistatus<W: std::io::Write>(
format!("{}/{}", subpath.trim_end_matches('/'), file.name)
};
let href = nc_href(username, &child_sub);
let file_id = resolve_file_id(file_id_svc, &file.id).await;
let file_id = file_id_map.get(&file.id).copied();
let oc_id = file_id.map(|id| format_oc_id(id, file_id_svc));
write_file_response(
&mut xml,
@@ -1066,7 +1081,7 @@ async fn write_nc_multistatus<W: std::io::Write>(
format!("{}/{}", subpath.trim_end_matches('/'), sf.name)
};
let href = nc_collection_href(username, &child_sub);
let file_id = resolve_folder_id(file_id_svc, &sf.id).await;
let file_id = folder_id_map.get(&sf.id).copied();
let oc_id = file_id.map(|id| format_oc_id(id, file_id_svc));
write_folder_response(
&mut xml,
@@ -1273,20 +1288,24 @@ pub fn write_text_element<W: std::io::Write>(
Ok(())
}
pub async fn resolve_file_id(
/// Resolve every `oc:fileid` for a listing in two batch queries (one per
/// object type) instead of one INSERT round-trip per child. Returns
/// `(file_map, folder_map)` keyed by object UUID; entries are absent when the
/// service is disabled or an id can't be resolved, mirroring the previous
/// per-call `Option` behaviour. The two batches run concurrently.
pub async fn batch_resolve_ids(
svc: Option<&Arc<NextcloudFileIdService>>,
file_uuid: &str,
) -> Option<i64> {
let svc = svc?;
svc.get_or_create_file_id(file_uuid).await.ok()
}
pub async fn resolve_folder_id(
svc: Option<&Arc<NextcloudFileIdService>>,
folder_uuid: &str,
) -> Option<i64> {
let svc = svc?;
svc.get_or_create_folder_id(folder_uuid).await.ok()
file_uuids: &[String],
folder_uuids: &[String],
) -> (HashMap<String, i64>, HashMap<String, i64>) {
let Some(svc) = svc else {
return (HashMap::new(), HashMap::new());
};
let (files, folders) = tokio::join!(
svc.get_or_create_file_ids(file_uuids),
svc.get_or_create_folder_ids(folder_uuids),
);
(files.unwrap_or_default(), folders.unwrap_or_default())
}
pub fn format_oc_id(id: i64, svc: Option<&Arc<NextcloudFileIdService>>) -> String {