v0.11.0: public torrent-site catalog with one-click install

.gitignore

@@ -55,3 +55,6 @@ session.secret
 # Logs
 *.log
 /logs/
+
+# Worktrees
+.worktrees/

TODO.md

@@ -0,0 +1,9 @@
+# TODO
+
+## Catalog fixtures
+
+- [ ] Capture `<slug>-search.html` fixtures for the 11 remaining catalog slugs.
+  Pattern: `backend/tests/fixtures/catalog/<slug>-search.html`.
+  Extraction test lives in `backend/tests/test_catalog_fixtures.py` and
+  is parametrized over whichever fixtures are present. When captured,
+  the test runs `_extract_release` against each row and asserts ≥1 result.

backend/migrations/versions/0016_indexer_catalog_slug.py

@@ -0,0 +1,38 @@
+"""indexer.catalog_slug column
+
+Revision ID: 0016
+Revises: 0015
+Create Date: 2026-04-20
+
+"""
+
+from __future__ import annotations
+
+import sqlalchemy as sa
+import sqlmodel
+from alembic import op
+
+revision: str = "0016"
+down_revision: str | None = "0015"
+branch_labels: str | None = None
+depends_on: str | None = None
+
+
+def upgrade() -> None:
+    with op.batch_alter_table("indexer") as batch:
+        batch.add_column(
+            sa.Column(
+                "catalog_slug",
+                sqlmodel.sql.sqltypes.AutoString(length=64),
+                nullable=True,
+            )
+        )
+    op.create_index(
+        "ix_indexer_catalog_slug", "indexer", ["catalog_slug"], unique=False
+    )
+
+
+def downgrade() -> None:
+    op.drop_index("ix_indexer_catalog_slug", table_name="indexer")
+    with op.batch_alter_table("indexer") as batch:
+        batch.drop_column("catalog_slug")

backend/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "trove"
-version = "0.10.4"
+version = "0.11.0"
 description = "A modern replacement for FlexGet with multi-indexer search, Usenet support, and optional local AI."
 readme = "../README.md"
 requires-python = ">=3.12"

backend/src/trove/api/catalog.py

@@ -0,0 +1,119 @@
+from __future__ import annotations
+
+from fastapi import APIRouter, Depends, HTTPException, status
+from pydantic import BaseModel, Field
+from sqlmodel import Session, select
+
+from trove.api.deps import current_user, db_session
+from trove.api.indexers import IndexerOut, _to_out
+from trove.clients.base import Protocol
+from trove.indexers.base import Category
+from trove.indexers.cardigann import load_definition_yaml
+from trove.models.indexer import IndexerRow
+from trove.models.user import User
+from trove.services import catalog, indexer_registry
+
+router = APIRouter()
+
+
+class CatalogEntryOut(BaseModel):
+    slug: str
+    display_name: str
+    description: str
+    categories: list[Category]
+    mirrors: list[str]
+    default_mirror: str
+    protocol: Protocol
+    logo: str | None = None
+    already_installed: bool
+
+
+class CatalogInstallRequest(BaseModel):
+    base_url: str = Field(min_length=1, max_length=512)
+    name: str | None = Field(default=None, max_length=64)
+
+
+@router.get("", response_model=list[CatalogEntryOut])
+async def list_catalog(
+    session: Session = Depends(db_session),
+    _user: User = Depends(current_user),
+) -> list[CatalogEntryOut]:
+    installed_slugs = set(
+        session.exec(
+            select(IndexerRow.catalog_slug).where(
+                IndexerRow.catalog_slug.is_not(None)  # type: ignore[union-attr]
+            )
+        ).all()
+    )
+    out: list[CatalogEntryOut] = []
+    for entry in catalog.list_entries():
+        out.append(
+            CatalogEntryOut(
+                slug=entry.slug,
+                display_name=entry.display_name,
+                description=entry.description,
+                categories=entry.categories,
+                mirrors=entry.mirrors,
+                default_mirror=entry.default_mirror,
+                protocol=entry.protocol,
+                logo=entry.logo,
+                already_installed=entry.slug in installed_slugs,
+            )
+        )
+    return out
+
+
+def _dedup_name(session: Session, base: str) -> str:
+    candidate = base
+    suffix = 2
+    while session.exec(select(IndexerRow).where(IndexerRow.name == candidate)).first() is not None:
+        candidate = f"{base}-{suffix}"
+        suffix += 1
+    return candidate
+
+
+@router.post("/{slug}", response_model=IndexerOut, status_code=status.HTTP_201_CREATED)
+async def install_catalog_entry(
+    slug: str,
+    payload: CatalogInstallRequest,
+    session: Session = Depends(db_session),
+    _user: User = Depends(current_user),
+) -> IndexerOut:
+    try:
+        entry = catalog.get_entry(slug)
+    except KeyError:
+        raise HTTPException(status_code=status.HTTP_404_NOT_FOUND, detail="unknown_slug") from None
+
+    if payload.base_url not in entry.mirrors:
+        raise HTTPException(
+            status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
+            detail="base_url_not_in_catalog_mirrors",
+        )
+
+    try:
+        yaml_text = catalog.read_yaml(slug)
+        load_definition_yaml(yaml_text)
+    except Exception as e:
+        raise HTTPException(
+            status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+            detail=f"catalog_yaml_broken: {e}",
+        ) from e
+
+    base_name = payload.name or entry.display_name
+    name = _dedup_name(session, base_name)
+
+    row = IndexerRow(
+        name=name,
+        type="cardigann",
+        protocol=entry.protocol.value,
+        base_url=payload.base_url,
+        credentials_cipher=indexer_registry.encrypt_credentials({}),
+        definition_yaml=yaml_text,
+        enabled=True,
+        priority=50,
+        catalog_slug=slug,
+    )
+    session.add(row)
+    session.commit()
+    session.refresh(row)
+    return _to_out(row)

backend/src/trove/docs/03-indexers.md

@@ -89,17 +89,36 @@ For trackers that don't have a native API, Cardigann uses YAML definition files
 5. Save and test
 
 **What works**:
-- Static `search.paths[0].path`
-- `search.rows.selector` (CSS selectors)
+- `search.paths[0].path` with Go-template expansion (`{{ .Keywords }}`, `{{ if .Keywords }}…{{ else }}…{{ end }}`, `{{ .Config.X }}`)
+- `search.rows.selector` (CSS selectors, including template-expanded ones)
 - Field extraction: `title`, `download`, `size`, `infohash`, `category`
-- Basic filters: `replace`, `regexp`, `append`, `prepend`
+- Basic filters: `replace`, `regexp`, `append`, `prepend`, `trim`, `tolower`, `re_replace`
+- `settings:` block defaults populated automatically as `{{ .Config.X }}` values
 
 **What doesn't work yet**:
 - Login flows (cookie, form, POST)
 - Multi-page pagination
 - Custom headers beyond what httpx sends by default
 - JavaScript-heavy sites that need Playwright/Puppeteer
 
+## Catalog (public sites, one click)
+
+For public, no-account torrent sites, Trove ships with a curated catalog of pre-configured definitions including 1337x, TorrentGalaxy, LimeTorrents, Nyaa, EZTV, and others. Five slugs map to substitute definitions (ExtraTorrent, KickAssTorrents, TorrentDownload, TorrentProject2, Tokyo Toshokan) because the original sites have no upstream Cardigann YAML — each entry honestly declares its actual source in the description.
+
+The Pirate Bay and YTS are not in the current catalog because both return JSON responses that require a native driver rather than the HTML-scraping Cardigann parser. They are candidates for a follow-up JSON-indexer batch.
+
+**How to install:**
+
+1. On `/indexers`, click **Browse catalog** (next to **Add indexer**).
+2. Pick a mirror from the dropdown on the site's tile — catalog entries list multiple known mirrors for sites that have them.
+3. Click **Add**. The tile flips to **Installed** and the site appears on `/indexers` as a normal Cardigann indexer.
+
+The onboarding wizard also surfaces these sites as an optional step — pick any you want in one go.
+
+**Behind the scenes**: a catalog-installed entry is an ordinary `type=cardigann` indexer with a vendored YAML definition. The Test, Edit, and Delete buttons work the same way they do for hand-added indexers. If you ever need to override the URL or rename the entry, use **Edit** — nothing is special about catalog rows.
+
+**Updating definitions**: the shipped YAMLs track `Prowlarr/Indexers`. When a site changes its HTML, searches will start returning 0 results. Upstream usually has a fix within days. A maintainer can run `scripts/update-catalog.py diff` to see what's changed upstream, then `sync` to pull — this is manual today, typically done alongside a release.
+
 ## Priority and ordering
 
 Each indexer has a **priority** field (default 50). Lower numbers run first in the fan-out, but since Trove queries all enabled indexers in parallel, priority mainly matters for tie-breaking when the same release appears from multiple sources — the higher-priority indexer's copy is kept.