feat: lazy retention sweep on /web/cases

New module retention::sweep_user_retention is invoked at the top of
handle_my_cases, using the current user's retention policy. Per case:
step 1 auto-closes when the newest .m4a mtime is older than
auto_close_days; step 2 auto-purges when closed_at is older than
auto_delete_days. Order matters - a case auto-closed in the same sweep
has closed_at ~= now and naturally survives the purge check,
preserving the full grace period after a vacation.

Race guard: purge re-reads the close marker immediately before
remove_dir_all so a concurrent upload that reopens the case cancels
the purge. Every automatic action emits an info! line with reason and
threshold; failures warn but never abort the sweep.

Seven integration tests cover both threshold paths, the disable-via-0
escape hatch, the newest-recording rule, the vacation scenario, and
the open-case skip. Tests age the comparison objects (mtime via
filetime, closed_at as handwritten RFC3339) instead of faking "now" -
no clock abstraction needed.
This commit is contained in:
2026-04-21 10:51:27 +02:00
parent e5a62d8f10
commit f358da215d
4 changed files with 469 additions and 0 deletions
+1
View File
@@ -8,6 +8,7 @@ pub mod gazetteer;
pub mod magic_link;
pub mod models;
pub mod paths;
pub mod retention;
pub mod routes;
pub mod transcribe;
pub mod web_session;
+194
View File
@@ -0,0 +1,194 @@
//! Lazy retention sweep: runs at the top of `scan_user_cases` for the
//! signed-in user. Auto-closes stale cases and auto-purges old closed
//! cases — no background scheduler, no cron. The natural "someone
//! opened the case list" trigger is sufficient for doctate's single-
//! digit-user scale.
//!
//! Logging is deliberate per `feedback_log_autonomous_server_actions`:
//! every automatic action — and every race-guarded skip — emits an
//! `info!` line with the reason and the applied thresholds.
//!
//! Sweep order per case: auto-close first, auto-purge second. A case
//! that was just auto-closed has `closed_at ≈ now` and naturally
//! survives the purge check — giving the user the full `auto_delete_days`
//! grace period even after a long absence.
use std::path::Path;
use time::OffsetDateTime;
use time::format_description::well_known::Rfc3339;
use tokio::fs;
use tracing::{info, warn};
use doctate_common::timestamp::now_rfc3339;
use crate::config::RetentionUserSettings;
use crate::events::{self, CaseEventKind, EventSender};
use crate::paths::{CloseMarker, is_closed, read_close_marker, write_close_marker};
/// Scan `user_root` once, applying auto-close and auto-purge under the
/// user's retention policy. Idempotent and best-effort: every I/O
/// failure is logged but never blocks the remaining cases or the
/// enclosing list view.
pub async fn sweep_user_retention(
user_root: &Path,
slug: &str,
settings: &RetentionUserSettings,
events_tx: &EventSender,
) {
if settings.auto_close_days == 0 && settings.auto_delete_days == 0 {
return;
}
let now = OffsetDateTime::now_utc();
let mut entries = match fs::read_dir(user_root).await {
Ok(r) => r,
// User dir may not exist yet (new user, no uploads). Nothing to sweep.
Err(_) => return,
};
while let Ok(Some(entry)) = entries.next_entry().await {
let case_path = entry.path();
if !case_path.is_dir() {
continue;
}
if settings.auto_close_days > 0 && !is_closed(&case_path).await {
try_auto_close(&case_path, slug, settings.auto_close_days, now, events_tx).await;
}
if settings.auto_delete_days > 0 && is_closed(&case_path).await {
try_auto_purge(&case_path, slug, settings.auto_delete_days, now, events_tx).await;
}
}
}
async fn try_auto_close(
case_path: &Path,
slug: &str,
threshold_days: u32,
now: OffsetDateTime,
events_tx: &EventSender,
) {
let Some(latest_m4a_mtime) = newest_m4a_mtime(case_path).await else {
// No .m4a yet → not a candidate.
return;
};
let age_days = (now - latest_m4a_mtime).whole_days();
if age_days <= threshold_days as i64 {
return;
}
let marker = CloseMarker {
closed_at: now_rfc3339(),
};
let case_id = events::case_id_of(case_path);
match write_close_marker(case_path, &marker).await {
Ok(_) => {
info!(
slug = %slug,
case_id = %case_id,
last_recording_age_days = age_days,
threshold_days = threshold_days,
"auto-closed case: last recording older than threshold"
);
events::emit(events_tx, slug, case_id, CaseEventKind::CaseClosed);
}
Err(e) => {
warn!(
slug = %slug,
case_id = %case_id,
error = %e,
"auto-close failed to write marker"
);
}
}
}
async fn try_auto_purge(
case_path: &Path,
slug: &str,
threshold_days: u32,
now: OffsetDateTime,
events_tx: &EventSender,
) {
let Some(marker) = read_close_marker(case_path).await else {
// Malformed marker → do not purge. Admin can manually fix.
return;
};
let Ok(closed_at) = OffsetDateTime::parse(&marker.closed_at, &Rfc3339) else {
warn!(
slug = %slug,
case_dir = ?case_path,
closed_at = %marker.closed_at,
"auto-purge: cannot parse closed_at; skipping"
);
return;
};
let age_days = (now - closed_at).whole_days();
if age_days <= threshold_days as i64 {
return;
}
// Race-guard: re-read the marker immediately before `remove_dir_all`.
// If a concurrent upload has cleared the marker, the case has been
// reopened and must not be purged.
if read_close_marker(case_path).await.is_none() {
info!(
slug = %slug,
case_dir = ?case_path,
reason = "reopened during purge",
"auto-purge skipped"
);
return;
}
let case_id = events::case_id_of(case_path);
match fs::remove_dir_all(case_path).await {
Ok(_) => {
info!(
slug = %slug,
case_id = %case_id,
closed_days_ago = age_days,
threshold_days = threshold_days,
"auto-purged case: closed marker older than threshold"
);
events::emit(events_tx, slug, case_id, CaseEventKind::CasePurged);
}
Err(e) => {
warn!(
slug = %slug,
case_id = %case_id,
error = %e,
"auto-purge failed to remove dir"
);
}
}
}
/// Newest `.m4a` mtime in `case_dir`, as a UTC `OffsetDateTime`.
/// `None` when no `.m4a` is present. `.m4a.failed` counts too — a
/// recording that failed ffmpeg is still user activity.
async fn newest_m4a_mtime(case_dir: &Path) -> Option<OffsetDateTime> {
let mut entries = fs::read_dir(case_dir).await.ok()?;
let mut newest: Option<OffsetDateTime> = None;
while let Ok(Some(entry)) = entries.next_entry().await {
let name = entry.file_name();
let Some(s) = name.to_str() else { continue };
if !s.ends_with(".m4a") && !s.ends_with(".m4a.failed") {
continue;
}
let Ok(meta) = entry.metadata().await else {
continue;
};
let Ok(mtime) = meta.modified() else {
continue;
};
let ts = OffsetDateTime::from(mtime);
newest = Some(match newest {
Some(cur) if cur > ts => cur,
_ => ts,
});
}
newest
}
+9
View File
@@ -312,6 +312,15 @@ pub async fn handle_my_cases(
auto_trigger::try_enqueue_all_for_user(&user_root, &pipeline.analyze_tx, &config, &events_tx)
.await;
// Lazy retention sweep: at most one disk scan per /web/cases visit,
// reusing the user's retention policy from users.toml. Runs before
// the listing scan so any auto-close/auto-purge actions are
// reflected in the same response.
if let Some(u) = config.users.iter().find(|u| u.slug == user.slug) {
crate::retention::sweep_user_retention(&user_root, &user.slug, &u.retention, &events_tx)
.await;
}
let busy = pipeline.analyze_busy.0.load(Ordering::Acquire);
let cases = scan_user_cases(&config, &user.slug, busy).await;
let total = cases.len();