Merge pull request 'fix(connect): queue new files with the audiobookshelf watcher instead of a full scan' (#10) from bugfix/abs-targeted-scan into main

Reviewed-on: #10

static/app.js

@@ -81,6 +81,11 @@ function libOptions(mediaType) {
     .join("");
 }
 
+// the server keeps a copy; the url in the query makes a changed cover a new address
+function coverSrc(r) {
+  return `/api/requests/${r.id}/cover?u=${encodeURIComponent(r.cover_url)}`;
+}
+
 // ---- missing badge ----
 async function refreshMissingBadge() {
   try {
@@ -93,7 +98,7 @@ async function refreshMissingBadge() {
 
 // ---- navigation ----
 // the view lives in the hash so reload, back button and links keep it
-const VIEWS = ["search", "missing", "imported", "import", "activity", "settings", "system"];
+const VIEWS = ["search", "missing", "imported", "import", "series", "activity", "settings", "system"];
 
 function showView() {
   const hash = location.hash.slice(1);
@@ -103,6 +108,7 @@ function showView() {
   if (name === "missing" || name === "imported") loadRequests(name);
   if (name === "activity") loadHistory();
   if (name === "system") loadSystem();
+  if (name === "series") loadWatches();
   if (name === "settings") loadLibraries().catch((err) => toast(err.message, true));
 }
 
@@ -320,6 +326,8 @@ function renderSearchResults() {
 // ---- request a whole series ----
 let seriesEpisodes = [];
 let seriesType = "audiobook";
+let seriesResult = null;
+let seriesLanguage = "";
 
 function renderSeriesList() {
   const from = parseInt($("#series-from").value);
@@ -366,6 +374,8 @@ function updateSeriesCount() {
 async function openSeriesDialog(result, type = "audiobook", language = "") {
   const dlg = $("#series-dialog");
   seriesType = type;
+  seriesResult = result;
+  seriesLanguage = language;
   const unit = t(type === "ebook" ? "series.unitVolumes" : "series.unitEpisodes");
   $("#series-title").textContent = result.series || t("series.fallbackTitle");
   $("#series-info").innerHTML =
@@ -411,6 +421,31 @@ $("#series-none").addEventListener("click", () => {
 });
 $("#series-cancel").addEventListener("click", () => $("#series-dialog").close());
 
+$("#series-watch").addEventListener("click", async () => {
+  const libId = $("#series-library").value;
+  if (!libId) {
+    toast(t("series.needLibrary", { type: seriesType === "ebook" ? "Ebook" : "Audiobook" }), true);
+    return;
+  }
+  const btn = $("#series-watch");
+  btn.disabled = true;
+  try {
+    await api("/api/series", {
+      method: "POST",
+      body: JSON.stringify({
+        media_type: seriesType, library_id: parseInt(libId),
+        series_id: seriesResult.series_id, source: seriesResult.source || "",
+        title: seriesResult.series || seriesResult.title, language: seriesLanguage,
+      }),
+    });
+    toast(t("watch.added"));
+  } catch (err) {
+    toast(err.status === 409 ? t("watch.already") : err.message, true);
+  } finally {
+    btn.disabled = false;
+  }
+});
+
 $("#series-submit").addEventListener("click", async () => {
   const libId = $("#series-library").value;
   if (!libId) {
@@ -579,7 +614,7 @@ function renderRequests(status) {
   box.innerHTML = reqs
     .map(
       (r) => `<div class="card clickable" data-detail="${r.id}" tabindex="0">
-        ${r.cover_url ? `<img src="${esc(r.cover_url)}" alt="" loading="lazy">` : '<div class="nocover">?</div>'}
+        ${r.cover_url ? `<img src="${esc(coverSrc(r))}" alt="" loading="lazy">` : '<div class="nocover">?</div>'}
         <div class="card-body">
           <strong>${esc(r.title)}${r.volume != null ? ` · ${t("common.volume")} ${r.volume}` : ""}</strong>
           <span>${esc(r.authors)}</span>
@@ -701,6 +736,84 @@ function renderHistory() {
     .join("");
 }
 
+// ---- watched series ----
+let watches = [];
+
+async function loadWatches(path = "/api/series", opts = {}) {
+  try {
+    watches = await api(path, opts);
+  } catch (err) {
+    toast(err.message, true);
+    return;
+  }
+  renderWatches();
+}
+
+function renderWatches() {
+  const n = watches.reduce((sum, w) => sum + w.new.length, 0);
+  $("#watch-badge").textContent = n;
+  $("#watch-badge").hidden = n === 0;
+  const box = $("#watch-list");
+  if (!watches.length) {
+    box.innerHTML = `<p class="muted">${t("watch.empty")}</p>`;
+    return;
+  }
+  const locale = lang === "de" ? "de-DE" : "en-GB";
+  box.innerHTML = watches
+    .map((w) => {
+      const checked = w.checked_at ? new Date(w.checked_at).toLocaleString(locale) : "—";
+      const parts = w.new
+        .map((p) => `<tr>
+            <td>${p.volume ?? ""}</td>
+            <td>${esc(p.title)}${p.year ? ` <span class="muted">(${p.year})</span>` : ""}</td>
+            <td class="muted">${esc(p.authors)}</td>
+            <td class="row">
+              <button data-part="${p.id}" data-action="request">${t("common.request")}</button>
+              <button class="secondary" data-part="${p.id}" data-action="ignore">${t("watch.ignore")}</button>
+            </td>
+          </tr>`)
+        .join("");
+      return `<div class="watch">
+        <div class="row">
+          <strong>${esc(w.title)}</strong>
+          <span class="muted">${esc(w.library_name)} · ${esc(w.media_type)} · ${t("watch.checked", { at: checked })}</span>
+          <button class="danger" data-unwatch="${w.id}">${t("watch.remove")}</button>
+        </div>
+        ${w.error ? `<p class="muted error-text">${esc(t("watch.error", { error: w.error }))}</p>` : ""}
+        ${parts ? `<table><tbody>${parts}</tbody></table>` : `<p class="muted">${t("watch.nothingNew")}</p>`}
+      </div>`;
+    })
+    .join("");
+}
+
+$("#watch-list").addEventListener("click", async (e) => {
+  const btn = e.target.closest("button");
+  if (!btn) return;
+  try {
+    if (btn.dataset.part) {
+      btn.disabled = true;
+      await api("/api/series/parts", {
+        method: "POST",
+        body: JSON.stringify({ ids: [parseInt(btn.dataset.part)], action: btn.dataset.action }),
+      });
+      if (btn.dataset.action === "request") refreshMissingBadge();
+    } else if (btn.dataset.unwatch) {
+      if (!confirm(t("watch.confirmRemove"))) return;
+      await api(`/api/series/${btn.dataset.unwatch}`, { method: "DELETE" });
+    } else return;
+  } catch (err) {
+    toast(err.message, true);
+  }
+  loadWatches();
+});
+
+$("#watch-check").addEventListener("click", async () => {
+  const btn = $("#watch-check");
+  btn.disabled = true;
+  await loadWatches("/api/series/check", { method: "POST" });
+  btn.disabled = false;
+});
+
 // ---- system ----
 let healthIssues = [];
 let logRows = [];
@@ -780,7 +893,7 @@ function openDetail(id, status) {
   const img = $("#detail-cover");
   img.hidden = !r.cover_url;
   $("#detail-nocover").hidden = !!r.cover_url;
-  if (r.cover_url) img.src = r.cover_url;
+  if (r.cover_url) img.src = coverSrc(r);
   $("#detail-retag").hidden = !(r.status === "imported" && r.media_type === "audiobook");
   $("#detail-relocate").hidden = r.status !== "imported";
   $("#detail-dialog").showModal();
@@ -1962,6 +2075,7 @@ function onLangChange() {
   renderHistory();
   renderHealth();
   renderLogs();
+  renderWatches();
   if (libraries.length) loadLibraries();
   const editId = $("#lib-form").dataset.editId;
   editId ? startLibraryEdit(parseInt(editId)) : endLibraryEdit();
@@ -1987,5 +2101,6 @@ updateLangButton();
 endLibraryEdit();
 loadLibraries();
 refreshMissingBadge();
+loadWatches();
 showVersion();
 showView();

static/i18n.js

@@ -47,6 +47,19 @@ const STRINGS = {
     "nav.import": "Import",
     "nav.settings": "Libraries",
     "nav.activity": "Aktivität",
+    "nav.series": "Serien",
+    "watch.button": "Beobachten",
+    "watch.buttonTitle": "Täglich nach neuen Teilen dieser Serie schauen",
+    "watch.added": "Serie wird beobachtet",
+    "watch.already": "Diese Serie wird schon beobachtet",
+    "watch.check": "Jetzt prüfen",
+    "watch.empty": "Noch keine Serie beobachtet. In der Suche unter „Ganze Serie…“ auf „Beobachten“ klicken.",
+    "watch.checked": "geprüft {at}",
+    "watch.error": "Letzte Prüfung fehlgeschlagen: {error}",
+    "watch.nothingNew": "Nichts Neues.",
+    "watch.ignore": "Ignorieren",
+    "watch.remove": "Nicht mehr beobachten",
+    "watch.confirmRemove": "Diese Serie nicht mehr beobachten?",
     "nav.system": "System",
     "system.health": "Zustand",
     "system.healthy": "Alles in Ordnung.",
@@ -309,6 +322,19 @@ const STRINGS = {
     "nav.import": "Import",
     "nav.settings": "Libraries",
     "nav.activity": "Activity",
+    "nav.series": "Series",
+    "watch.button": "Watch",
+    "watch.buttonTitle": "Look for new parts of this series every day",
+    "watch.added": "Series is being watched",
+    "watch.already": "This series is already watched",
+    "watch.check": "Check now",
+    "watch.empty": "No series watched yet. In the search, open “Whole series…” and click “Watch”.",
+    "watch.checked": "checked {at}",
+    "watch.error": "Last check failed: {error}",
+    "watch.nothingNew": "Nothing new.",
+    "watch.ignore": "Ignore",
+    "watch.remove": "Stop watching",
+    "watch.confirmRemove": "Stop watching this series?",
     "nav.system": "System",
     "system.health": "Health",
     "system.healthy": "All good.",

static/index.html

@@ -20,6 +20,10 @@
         </button>
         <button data-view="imported" data-i18n="nav.imported"></button>
         <button data-view="import" data-i18n="nav.import"></button>
+        <button data-view="series">
+          <span data-i18n="nav.series"></span>
+          <span id="watch-badge" class="badge" hidden></span>
+        </button>
         <button data-view="activity" data-i18n="nav.activity"></button>
         <button data-view="settings" data-i18n="nav.settings"></button>
         <button data-view="system">
@@ -97,6 +101,7 @@
           </div>
           <div class="row">
             <button type="button" id="series-cancel" class="secondary" data-i18n="common.cancel"></button>
+            <button type="button" id="series-watch" class="secondary" data-i18n="watch.button" data-i18n-title="watch.buttonTitle"></button>
             <button type="button" id="series-submit" data-i18n="common.request"></button>
           </div>
         </dialog>
@@ -287,6 +292,13 @@
         </dialog>
       </section>
 
+      <section id="view-series" hidden>
+        <div class="bar">
+          <button type="button" id="watch-check" data-i18n="watch.check"></button>
+        </div>
+        <div id="watch-list"></div>
+      </section>
+
       <section id="view-activity" hidden>
         <div class="table-scroll">
           <table id="history-table">

static/style.css

@@ -77,6 +77,10 @@ td.state.running { color: var(--muted); }
 td.state.warn { color: var(--warn); }
 .badge.warn { background: var(--warn); color: #000; }
 .badge.error { background: var(--danger); }
+.watch { background: var(--panel); border: 1px solid var(--border); border-radius: 6px; padding: 0.7rem 0.9rem; margin-bottom: 0.8rem; }
+.watch .row:first-child { margin-top: 0; }
+.watch table { margin-top: 0.5rem; }
+.watch .error-text { color: #e08585; }
 #health-list { list-style: none; margin: 0.5rem 0 1rem; }
 #health-list li { background: var(--panel); border-left: 3px solid var(--warn); border-radius: 4px; padding: 0.5rem 0.8rem; margin: 0.3rem 0; }
 #health-list li.error { border-left-color: var(--danger); }

tests/test_import_flow.py

@@ -1,4 +1,5 @@
 import importlib
+import json
 import os
 import shutil
 from pathlib import Path
@@ -1228,12 +1229,13 @@ def test_connections_hear_about_committed_rows(client, monkeypatch):
     from wordarr import connect
 
     calls = []
+    watcher = {"disableWatcher": False}
 
     def handler(request):
-        calls.append((request.method, str(request.url)))
+        calls.append((request.method, str(request.url), request.content))
         if request.url.path == "/api/libraries":
             return httpx.Response(200, json={"libraries": [
-                {"id": "lib-audio", "name": "Audio",
+                {"id": "lib-audio", "name": "Audio", "settings": watcher,
                  "folders": [{"fullPath": str(client.tmp_path / "library")}]},
                 {"id": "lib-pod", "name": "Pod", "folders": [{"fullPath": "/podcasts"}]},
             ]})
@@ -1261,16 +1263,29 @@ def test_connections_hear_about_committed_rows(client, monkeypatch):
         "path": str(src), "is_dir": False, "files": [str(src)], "request_id": req_id,
     }]}).json()["results"][0]["ok"]
 
-    assert calls == [
+    # no full scan while the watcher runs: the two raced and added the book twice
+    dest = str(root / "Momo" / "Momo.epub")
+    assert [c[:2] for c in calls] == [
+        ("POST", "http://hook/in"),
+        ("GET", "http://abs/api/libraries"),
+        ("POST", "http://abs/api/watcher/update"),
+    ]
+    assert json.loads(calls[2][2]) == {
+        "libraryId": "lib-audio", "path": dest, "type": "add"}
+
+    calls.clear()
+    client.delete(f"/api/requests/{req_id}")
+    assert [c[:2] for c in calls] == [("POST", "http://hook/in")]  # nothing new on disk
+
+    calls.clear()
+    watcher["disableWatcher"] = True
+    connect.notify([{"event": "imported", "dest": dest}])
+    assert [c[:2] for c in calls] == [
         ("POST", "http://hook/in"),
         ("GET", "http://abs/api/libraries"),
         ("POST", "http://abs/api/libraries/lib-audio/scan"),
     ]
 
-    calls.clear()
-    client.delete(f"/api/requests/{req_id}")
-    assert calls == [("POST", "http://hook/in")]  # nothing new on disk, no scan
-
 
 def test_bulk_move_delete_and_count(client):
     def lib(name, media_type):
@@ -1359,3 +1374,103 @@ def test_recent_logs_newest_first(client):
     rows = client.get("/api/system/logs").json()
     assert [r["message"] for r in rows[:2]] == ["second", "first"]
     assert rows[0]["level"] == "WARNING"
+
+
+def test_watched_series_suggests_only_new_parts(client, monkeypatch):
+    import asyncio
+    import datetime
+
+    from wordarr import db
+    from wordarr.api import series
+    from wordarr.metadata import SERIES_PROVIDERS
+    from wordarr.metadata.base import MetadataResult
+
+    def part(volume, external_id=""):
+        return MetadataResult(media_type="ebook", title=f"Band {volume}", volume=volume,
+                              series="Saga", external_id=external_id, source="hardcover")
+
+    hits = [part(1, "a1"), part(2, "a2")]
+    calls = []
+
+    async def lookup(series_id, title, language, source):
+        calls.append(series_id)
+        if hits is None:
+            raise RuntimeError("offline")
+        return hits
+
+    monkeypatch.setitem(SERIES_PROVIDERS, "ebook", lookup)
+    lib = client.post("/api/libraries", json={
+        "name": "Saga", "media_type": "ebook", "root_path": str(client.tmp_path)}).json()
+    client.post("/api/requests", json={"library_id": lib["id"], "title": "Band 3", "volume": 3})
+    watch = {"media_type": "ebook", "library_id": lib["id"], "series_id": "s1",
+             "source": "hardcover", "title": "Saga"}
+
+    resp = client.post("/api/series", json=watch)
+    assert resp.status_code == 200
+    assert resp.json()["new"] == []  # what exists today is not news
+    assert client.post("/api/series", json=watch).status_code == 409
+
+    hits += [part(3), part(4, "a4"), part(5, "a5")]  # 3 is requested already
+    [w] = client.post("/api/series/check").json()
+    new = {p["title"]: p["id"] for p in w["new"]}
+    assert set(new) == {"Band 4", "Band 5"}
+
+    client.post("/api/series/parts", json={"ids": [new["Band 4"]], "action": "request"})
+    client.post("/api/series/parts", json={"ids": [new["Band 5"]], "action": "ignore"})
+    missing = client.get("/api/requests?status=missing").json()
+    assert {r["title"] for r in missing} == {"Band 3", "Band 4"}
+    assert client.post("/api/series/check").json()[0]["new"] == []
+
+    hits = None
+    [w] = client.post("/api/series/check").json()
+    assert w["error"] == "offline"
+    assert client.post("/api/series", json={**watch, "series_id": "s2"}).status_code == 502
+
+    # the daily run only picks up what was checked more than a day ago
+    hits, calls[:] = [], []
+    asyncio.run(series.check_due())
+    assert calls == []
+    with db.SessionLocal() as session:
+        session.get(db.WatchedSeries, w["id"]).checked_at = (
+            datetime.datetime.now(datetime.UTC) - datetime.timedelta(days=2))
+        session.commit()
+    asyncio.run(series.check_due())
+    assert calls == ["s1"]
+
+    assert client.delete(f"/api/series/{w['id']}").status_code == 200
+    assert client.get("/api/series").json() == []
+
+
+def test_covers_are_fetched_once_and_kept(client, monkeypatch):
+    import httpx
+    calls = []
+
+    async def fake_get(self, url, **kwargs):
+        calls.append(url)
+        if "dead" in url:
+            raise httpx.ConnectError("gone")
+        ctype = "text/html" if "page" in url else "image/png"
+        return httpx.Response(200, headers={"content-type": ctype}, content=b"PNG",
+                              request=httpx.Request("GET", url))
+
+    monkeypatch.setattr(httpx.AsyncClient, "get", fake_get)
+    from wordarr import config
+    monkeypatch.setattr(config, "CONFIG_DIR", client.tmp_path / "config")
+    lib = client.post("/api/libraries", json={
+        "name": "Covers", "media_type": "ebook", "root_path": str(client.tmp_path)}).json()
+
+    def request(cover_url):
+        return client.post("/api/requests?allow_duplicate=true", json={
+            "library_id": lib["id"], "title": "Buch", "cover_url": cover_url}).json()["id"]
+
+    good = request("https://covers.example/a.png")
+    for _ in range(2):
+        resp = client.get(f"/api/requests/{good}/cover")
+        assert resp.status_code == 200
+        assert resp.content == b"PNG"
+        assert resp.headers["content-type"] == "image/png"
+    assert calls == ["https://covers.example/a.png"]  # the second came from disk
+    assert [p.suffix for p in (client.tmp_path / "config" / "covers").iterdir()] == [".png"]
+
+    for url in ("https://covers.example/page", "https://covers.example/dead", "", "file:///etc/passwd"):
+        assert client.get(f"/api/requests/{request(url)}/cover").status_code == 404

wordarr/api/requests.py

@@ -1,11 +1,12 @@
 from pathlib import Path
 
 from fastapi import APIRouter, BackgroundTasks, Depends, HTTPException
+from fastapi.responses import FileResponse
 from pydantic import BaseModel, ConfigDict
 from sqlalchemy import func, select
 from sqlalchemy.orm import Session
 
-from .. import config, connect
+from .. import config, connect, covers
 from ..db import BookRequest, Library, get_session, record
 from ..importer import mover, tagger
 
@@ -237,6 +238,15 @@ def update_request(request_id: int, data: RequestIn, session: Session = Depends(
     return _out(req)
 
 
+@router.get("/{request_id}/cover")
+async def request_cover(request_id: int, session: Session = Depends(get_session)):
+    req = session.get(BookRequest, request_id)
+    if not req or not (path := await covers.cached(req.cover_url)):
+        raise HTTPException(404, "no cover")
+    # the UI puts the cover url into the query, so a changed cover is a new url
+    return FileResponse(path, headers={"Cache-Control": "max-age=31536000, immutable"})
+
+
 @router.post("/{request_id}/retag")
 def retag_request(request_id: int, background: BackgroundTasks,
                   session: Session = Depends(get_session)):

wordarr/api/series.py

@@ -0,0 +1,164 @@
+import asyncio
+import datetime
+import logging
+
+from fastapi import APIRouter, Depends, HTTPException
+from pydantic import BaseModel
+from sqlalchemy import or_, select
+from sqlalchemy.orm import Session
+
+from .. import db
+from ..db import BookRequest, Library, SeriesPart, WatchedSeries, get_session
+from ..metadata import SERIES_PROVIDERS
+
+router = APIRouter(prefix="/api/series", tags=["series"])
+log = logging.getLogger(__name__)
+
+PART_FIELDS = ("title", "authors", "narrator", "external_id", "year", "series",
+               "volume", "cover_url")
+
+
+def _now() -> datetime.datetime:
+    return datetime.datetime.now(datetime.UTC)
+
+
+def _key(r) -> str:
+    return r.external_id or f"{r.title.strip().lower()}|{r.volume}"
+
+
+async def check_watch(session: Session, w: WatchedSeries) -> None:
+    """Look the series up again. The first lookup only records what exists, so
+    just later parts become suggestions; so do parts nobody has requested yet."""
+    try:
+        hits = await SERIES_PROVIDERS[w.media_type](w.series_id, w.title, w.language, w.source)
+    except Exception as exc:
+        log.warning("series %r lookup failed: %s", w.title, exc)
+        w.error, w.checked_at = str(exc), _now()
+        return
+    initial = not w.parts
+    reqs = session.scalars(select(BookRequest).where(BookRequest.media_type == w.media_type)).all()
+    requested = {r.external_id for r in reqs if r.external_id}
+    requested |= {(r.title.strip().lower(), r.volume) for r in reqs}
+    seen = {p.key for p in w.parts}
+    for hit in hits:
+        if (key := _key(hit)) in seen:
+            continue
+        seen.add(key)
+        known = initial or hit.external_id in requested or (
+            hit.title.strip().lower(), hit.volume) in requested
+        w.parts.append(SeriesPart(key=key, state="known" if known else "new",
+                                  **{f: getattr(hit, f) for f in PART_FIELDS}))
+    w.error, w.checked_at = "", _now()
+
+
+async def check_due() -> None:
+    cutoff = (_now() - datetime.timedelta(days=1)).replace(tzinfo=None)  # SQLite stores naive
+    with db.SessionLocal() as session:
+        due = session.scalars(select(WatchedSeries).where(
+            or_(WatchedSeries.checked_at.is_(None), WatchedSeries.checked_at < cutoff))).all()
+        for w in due:
+            await check_watch(session, w)
+            session.commit()
+        if due:
+            log.info("checked %d watched series", len(due))
+
+
+async def run_daily() -> None:
+    """Wakes hourly and checks what is a day old, so restarts do not reset the clock."""
+    while True:
+        try:
+            await check_due()
+        except Exception:
+            log.exception("series check failed")
+        await asyncio.sleep(3600)
+
+
+def _out(w: WatchedSeries) -> dict:
+    checked = w.checked_at and w.checked_at.replace(tzinfo=datetime.UTC).isoformat()
+    return {
+        "id": w.id, "media_type": w.media_type, "library_id": w.library_id,
+        "library_name": w.library.name, "series_id": w.series_id, "source": w.source,
+        "title": w.title, "checked_at": checked, "error": w.error,
+        "new": [{"id": p.id, **{f: getattr(p, f) for f in PART_FIELDS}}
+                for p in w.parts if p.state == "new"],
+    }
+
+
+def _all(session: Session) -> list[dict]:
+    return [_out(w) for w in session.scalars(select(WatchedSeries).order_by(WatchedSeries.title))]
+
+
+@router.get("")
+def list_watches(session: Session = Depends(get_session)):
+    return _all(session)
+
+
+class WatchIn(BaseModel):
+    media_type: str
+    library_id: int
+    series_id: str
+    source: str = ""
+    title: str
+    language: str = ""
+
+
+@router.post("")
+async def create_watch(data: WatchIn, session: Session = Depends(get_session)):
+    if data.media_type not in SERIES_PROVIDERS:
+        raise HTTPException(400, f"no series lookup for media_type: {data.media_type}")
+    lib = session.get(Library, data.library_id)
+    if not lib:
+        raise HTTPException(404, "library not found")
+    if lib.media_type != data.media_type:
+        raise HTTPException(400, "library has a different media type")
+    if session.scalar(select(WatchedSeries).where(
+            WatchedSeries.media_type == data.media_type,
+            WatchedSeries.source == data.source,
+            WatchedSeries.series_id == data.series_id)):
+        raise HTTPException(409, "series is already watched")
+    w = WatchedSeries(**data.model_dump())
+    session.add(w)
+    await check_watch(session, w)
+    if w.error:  # without a first lookup every part would turn up as new later
+        session.rollback()
+        raise HTTPException(502, f"metadata provider error: {w.error}")
+    session.commit()
+    return _out(w)
+
+
+@router.delete("/{watch_id}")
+def delete_watch(watch_id: int, session: Session = Depends(get_session)):
+    w = session.get(WatchedSeries, watch_id)
+    if not w:
+        raise HTTPException(404, "series not found")
+    session.delete(w)
+    session.commit()
+    return {"ok": True}
+
+
+@router.post("/check")
+async def check_all(session: Session = Depends(get_session)):
+    for w in session.scalars(select(WatchedSeries)).all():
+        await check_watch(session, w)
+        session.commit()
+    return _all(session)
+
+
+class PartsIn(BaseModel):
+    ids: list[int]
+    action: str  # request | ignore
+
+
+@router.post("/parts")
+def handle_parts(data: PartsIn, session: Session = Depends(get_session)):
+    if data.action not in ("request", "ignore"):
+        raise HTTPException(400, f"invalid action: {data.action}")
+    parts = session.scalars(select(SeriesPart).where(
+        SeriesPart.id.in_(data.ids), SeriesPart.state == "new")).all()
+    for p in parts:
+        if data.action == "request":
+            session.add(BookRequest(media_type=p.watch.media_type, library_id=p.watch.library_id,
+                                    **{f: getattr(p, f) for f in PART_FIELDS}))
+        p.state = "requested" if data.action == "request" else "ignored"
+    session.commit()
+    return {"done": len(parts)}

wordarr/connect.py

@@ -1,9 +1,10 @@
 """Tell other services about history rows once they are committed: a generic
-webhook gets every row, Audiobookshelf a scan of the libraries holding new
+webhook gets every row, Audiobookshelf the new files of the libraries holding
-files. Runs as a background task, so a slow or dead service never fails the
+them. Runs as a background task, so a slow or dead service never fails the
 request that caused it; errors only reach the log."""
 import logging
 import os
+from pathlib import Path
 
 import httpx
 from fastapi import BackgroundTasks
@@ -46,15 +47,39 @@ def notify(events: list[dict]) -> None:
 
 
 def _scan_audiobookshelf(client: httpx.Client, dests: list[str]) -> None:
-    """Scan only the libraries whose folders contain a destination. That needs
+    """Tell the libraries whose folders contain a destination about the new
-    both containers to mount the media under the same path."""
+    files. That needs both containers to mount the media under the same path."""
     headers = {"Authorization": f"Bearer {ABS_TOKEN}"}
     resp = client.get(f"{ABS_URL}/api/libraries", headers=headers)
     resp.raise_for_status()
     for lib in resp.json().get("libraries") or []:
         folders = [str(f.get("fullPath") or "").rstrip("/")
                    for f in lib.get("folders") or []]
-        if any(f and (d == f or d.startswith(f + "/")) for f in folders for d in dests):
+        mine = [d for d in dests
+                if any(f and (d == f or d.startswith(f + "/")) for f in folders)]
+        if not mine:
+            continue
+        if (lib.get("settings") or {}).get("disableWatcher"):
+            # the watcher endpoint drops paths of unwatched libraries
             client.post(f"{ABS_URL}/api/libraries/{lib.get('id')}/scan",
                         headers=headers).raise_for_status()
             log.info("audiobookshelf scans library %s", lib.get("name"))
+            continue
+        # a full scan racing the watcher adds the same folder twice, and it
+        # walks the whole library; this queues the files with the watcher
+        for path in _files(mine):
+            client.post(f"{ABS_URL}/api/watcher/update", headers=headers,
+                        json={"libraryId": lib.get("id"), "path": path,
+                              "type": "add"}).raise_for_status()
+        log.info("audiobookshelf watches new files in library %s", lib.get("name"))
+
+
+def _files(dests: list[str]) -> list[str]:
+    """Imports are flat: a destination is a file or a folder of files."""
+    out = []
+    for d in map(Path, dests):
+        if d.is_file():
+            out.append(str(d))
+        elif d.is_dir():
+            out += sorted(str(f) for f in d.iterdir() if f.is_file())
+    return out

wordarr/covers.py

@@ -0,0 +1,44 @@
+"""Covers are fetched once and kept under the config dir, so a dead link or a
+slow host does not take the picture with it."""
+import hashlib
+import logging
+import mimetypes
+import tempfile
+from pathlib import Path
+
+import httpx
+
+from . import config
+
+MAX_BYTES = 5 * 1024 * 1024
+log = logging.getLogger(__name__)
+
+
+# shortcut: covers of deleted requests stay on disk, add a sweep if the dir grows
+async def cached(url: str) -> Path | None:
+    """The file for this url, downloaded on first use; None if it cannot be had."""
+    if not url.startswith(("http://", "https://")):
+        return None
+    folder = config.CONFIG_DIR / "covers"
+    name = hashlib.sha256(url.encode()).hexdigest()
+    if hit := next(folder.glob(name + ".*"), None):
+        return hit
+    try:
+        async with httpx.AsyncClient(timeout=15, follow_redirects=True) as client:
+            resp = await client.get(url)
+            resp.raise_for_status()
+    except httpx.HTTPError as exc:
+        log.warning("cover %s failed: %s", url, exc)
+        return None
+    ctype = resp.headers.get("content-type", "").split(";")[0].strip()
+    if not ctype.startswith("image/") or len(resp.content) > MAX_BYTES:
+        log.warning("cover %s is no usable image (%s, %d bytes)", url, ctype, len(resp.content))
+        return None
+    folder.mkdir(parents=True, exist_ok=True)
+    # written aside and renamed, so a parallel request never serves half a file
+    with tempfile.NamedTemporaryFile(dir=folder, suffix=".tmp", delete=False) as tmp:
+        tmp.write(resp.content)
+    path = folder / (name + (mimetypes.guess_extension(ctype) or ".img"))
+    Path(tmp.name).chmod(0o644)  # NamedTemporaryFile creates 0600
+    Path(tmp.name).replace(path)
+    return path

wordarr/db.py

@@ -23,6 +23,8 @@ class Library(Base):
     language: Mapped[str] = mapped_column(String, default="")
 
     requests: Mapped[list["BookRequest"]] = relationship(back_populates="library")
+    watches: Mapped[list["WatchedSeries"]] = relationship(
+        back_populates="library", cascade="all, delete-orphan")
 
 
 class BookRequest(Base):
@@ -68,6 +70,47 @@ class History(Base):
     message: Mapped[str] = mapped_column(String, default="")
 
 
+class WatchedSeries(Base):
+    """A series looked up again now and then; new parts become suggestions."""
+
+    __tablename__ = "watched_series"
+
+    id: Mapped[int] = mapped_column(primary_key=True)
+    media_type: Mapped[str] = mapped_column(String)
+    library_id: Mapped[int] = mapped_column(ForeignKey("libraries.id"))
+    series_id: Mapped[str] = mapped_column(String)
+    source: Mapped[str] = mapped_column(String, default="")
+    title: Mapped[str] = mapped_column(String)
+    language: Mapped[str] = mapped_column(String, default="")
+    checked_at: Mapped[datetime.datetime | None] = mapped_column(default=None)
+    error: Mapped[str] = mapped_column(String, default="")
+
+    library: Mapped[Library] = relationship(back_populates="watches")
+    parts: Mapped[list["SeriesPart"]] = relationship(
+        back_populates="watch", cascade="all, delete-orphan")
+
+
+class SeriesPart(Base):
+    """Every part a watch has seen, so a later lookup can tell what is new."""
+
+    __tablename__ = "series_parts"
+
+    id: Mapped[int] = mapped_column(primary_key=True)
+    watch_id: Mapped[int] = mapped_column(ForeignKey("watched_series.id"))
+    key: Mapped[str] = mapped_column(String)
+    state: Mapped[str] = mapped_column(String)  # known | new | requested | ignored
+    title: Mapped[str] = mapped_column(String)
+    authors: Mapped[str] = mapped_column(String, default="")
+    narrator: Mapped[str] = mapped_column(String, default="")
+    external_id: Mapped[str] = mapped_column(String, default="")
+    year: Mapped[int | None] = mapped_column(default=None)
+    series: Mapped[str] = mapped_column(String, default="")
+    volume: Mapped[int | None] = mapped_column(default=None)
+    cover_url: Mapped[str] = mapped_column(String, default="")
+
+    watch: Mapped[WatchedSeries] = relationship(back_populates="parts")
+
+
 def record(session, event: str, req: BookRequest, **fields: str) -> History:
     """Add a history row; it is saved with the caller's next commit."""
     row = History(event=event, request_id=req.id, title=req.title,

wordarr/main.py

@@ -1,3 +1,4 @@
+import asyncio
 from contextlib import asynccontextmanager
 from pathlib import Path
 
@@ -5,7 +6,7 @@ from fastapi import FastAPI
 from fastapi.staticfiles import StaticFiles
 
 from . import db, logs, version
-from .api import fs, history, imports, libraries, requests, search, system
+from .api import fs, history, imports, libraries, requests, search, series, system
 
 logs.install()
 
@@ -13,7 +14,9 @@ logs.install()
 @asynccontextmanager
 async def lifespan(app: FastAPI):
     db.init_db()
+    watcher = asyncio.create_task(series.run_daily())
     yield
+    watcher.cancel()
 
 
 app = FastAPI(title="wordarr", lifespan=lifespan)
@@ -25,6 +28,7 @@ app.include_router(imports.router)
 app.include_router(fs.router)
 app.include_router(history.router)
 app.include_router(system.router)
+app.include_router(series.router)
 
 
 @app.get("/api/version")