feat(radiobrowser): add search, paging and favourite lookup

beefy_player/providers/radiobrowser/__init__.py

@@ -1,8 +1,9 @@
 """Radio Browser: a directory of the community's stations rather than a station.
 
-Three hundred of the most-clicked working relays, keyless and public-domain.
-There is no now-playing API, so this catalogue offers breadth, not history: pick
-a station and the stream speaks for itself.
+The most-clicked working relays are loaded a page at a time, and the rest of
+the 60,000-station directory is searched on the server as you type.  There is no
+now-playing API, so this catalogue offers breadth, not history: pick a station
+and the stream speaks for itself.
 """
 
 from __future__ import annotations
@@ -11,7 +12,7 @@ import httpx
 
 from ... import config
 from ...models import Song, Station
-from ..base import RadioProvider
+from ..base import RadioProvider, StationPage
 from . import api, fallback
 from . import hero as hero_art
 
@@ -23,8 +24,8 @@ class RadioBrowserProvider(RadioProvider):
     name = "Radio Browser"
     homepage = "radio-browser.info"
     brand = "RADIO BROWSER"
-    tagline = "community directory · 50,000 stations"
-    station_source = "radio-browser.info (top 300)"
+    tagline = "community directory · live search"
+    station_source = "radio-browser.info (paged + search)"
     history_source = ""
 
     hero = hero_art.HERO
@@ -33,6 +34,14 @@ class RadioBrowserProvider(RadioProvider):
     # A directory has no show schedule and no Bandcamp worth scraping.
     bandcamp = False
 
+    # The full directory is much larger than the list fetched at startup; the
+    # app asks the API when the filter changes.
+    server_search = True
+
+    # The directory is loaded a page at a time, in click-count order.
+    paginated = True
+    preserve_order = True
+
     @property
     def user_agent(self) -> str:
         return f"{config.USER_AGENT}; +https://www.radio-browser.info"
@@ -40,6 +49,21 @@ class RadioBrowserProvider(RadioProvider):
     async def fetch_stations(self, client: httpx.AsyncClient) -> list[Station]:
         return await api.fetch_stations(client)
 
+    async def fetch_station_page(
+        self, client: httpx.AsyncClient, offset: int = 0
+    ) -> StationPage:
+        return await api.fetch_station_page(client, offset)
+
+    async def fetch_stations_by_ids(
+        self, client: httpx.AsyncClient, station_ids: list[str]
+    ) -> list[Station]:
+        return await api.fetch_stations_by_ids(client, station_ids)
+
+    async def search_stations(
+        self, client: httpx.AsyncClient, query: str
+    ) -> list[Station]:
+        return await api.search_stations(client, query)
+
     def load_cached_stations(self) -> list[Station]:
         return api.load_cached_stations()
 

beefy_player/providers/radiobrowser/api.py

@@ -2,13 +2,15 @@
 
 ``https://de1.api.radio-browser.info/json/stations/search`` is keyless, CORS-open
 and public-domain data.  It knows nothing about what a station is playing —
-there is no per-station history — so this catalogue is a *directory*: three
-hundred of the most-clicked working stations, ready to tune into.  Whatever the
-relay puts in its stream is what you hear; the app has no metadata to show.
+there is no per-station history — so this catalogue is a *directory*: the most
+clicked working stations, a page at a time.  Whatever the relay puts in its
+stream is what you hear; the app has no metadata to show.
 """
 
 from __future__ import annotations
 
+import asyncio
+import re
 from typing import Any
 from urllib.parse import urlencode
 
@@ -16,18 +18,34 @@ import httpx
 
 from ... import config
 from ...models import Station, to_int
-from ..base import fetch_json
+from ..base import StationPage, fetch_json
 
 API_URL = "https://de1.api.radio-browser.info/json/stations/search"
+BY_UUID_URL = "https://de1.api.radio-browser.info/json/stations/byuuid"
 
-#: Three hundred is a comfortable directory: enough to browse, quick to fetch.
-STATION_LIMIT = 300
+#: Stations per directory page.  The app loads the first page at startup and
+#: the next ones as the station list nears its end.
+PAGE_SIZE = 100
+
+#: Results per field for a directory search.  The four fields are merged.
+SEARCH_LIMIT = 100
+
+#: The search endpoint has no full-text parameter, so a free-text query is
+#: sent once per field and the answers are merged.
+_SEARCH_FIELDS = ("name", "tag", "country", "language")
+
+#: Radio Browser ids are UUIDs; anything else is another catalogue's favourite.
+_UUID_PATTERN = re.compile(
+    r"^[0-9a-fA-F]{8}(?:-[0-9a-fA-F]{4}){3}-[0-9a-fA-F]{12}$"
+)
+
+#: How many UUIDs to ask for in one ``byuuid`` request.
+LOOKUP_BATCH = 50
 
 _PARAMS = {
     "hidebroken": "true",
     "order": "clickcount",
     "reverse": "true",
-    "limit": str(STATION_LIMIT),
 }
 
 _CODECS = {"mp3": "mp3", "aac": "aac", "aac+": "aacp", "ogg": "ogg", "opus": "opus"}
@@ -81,7 +99,7 @@ def station_from_entry(entry: dict[str, Any]) -> Station | None:
     )
 
 
-def parse_stations(payload: Any) -> list[Station]:
+def parse_stations(payload: Any, *, sort: bool = True) -> list[Station]:
     if not isinstance(payload, list):
         return []
     stations: list[Station] = []
@@ -92,22 +110,129 @@ def parse_stations(payload: Any) -> list[Station]:
             continue
         seen.add(station.id)
         stations.append(station)
-    stations.sort(key=lambda station: station.title.casefold())
+    if sort:
+        stations.sort(key=lambda station: station.title.casefold())
     return stations
 
 
+def _page_url(offset: int) -> str:
+    params = {**_PARAMS, "limit": str(PAGE_SIZE), "offset": str(offset)}
+    return f"{API_URL}?{urlencode(params)}"
+
+
+async def fetch_station_page(
+    client: httpx.AsyncClient, offset: int = 0
+) -> StationPage:
+    """One page of the directory, most-clicked first."""
+    try:
+        payload = await fetch_json(client, _page_url(offset))
+    except RuntimeError as exc:
+        raise RuntimeError(f"could not fetch the Radio Browser directory: {exc}") from None
+
+    stations = parse_stations(payload, sort=False)
+    rows = len(payload) if isinstance(payload, list) else 0
+    next_offset = offset + PAGE_SIZE if rows >= PAGE_SIZE else None
+    if offset == 0 and stations and isinstance(payload, list):
+        config.write_json(config.channels_cache_path("radiobrowser"), payload)
+    return StationPage(stations=stations, next_offset=next_offset)
+
+
 async def fetch_stations(client: httpx.AsyncClient) -> list[Station]:
-    url = f"{API_URL}?{urlencode(_PARAMS)}"
-    try:
-        payload = await fetch_json(client, url)
-    except RuntimeError as exc:
-        raise RuntimeError(f"could not fetch the Radio Browser directory: {exc}") from None
-    stations = parse_stations(payload)
-    if not stations:
+    page = await fetch_station_page(client, 0)
+    if not page.stations:
         raise RuntimeError("Radio Browser returned no stations")
-    config.write_json(config.channels_cache_path("radiobrowser"), payload)
-    return stations
+    return page.stations
 
 
 def load_cached_stations() -> list[Station]:
-    return parse_stations(config.read_json(config.channels_cache_path("radiobrowser")))
+    return parse_stations(
+        config.read_json(config.channels_cache_path("radiobrowser")), sort=False
+    )
+
+
+async def _search_field(
+    client: httpx.AsyncClient, query: str, field: str
+) -> list[Station]:
+    params = {**_PARAMS, "limit": str(SEARCH_LIMIT), field: query}
+    payload = await fetch_json(client, f"{API_URL}?{urlencode(params)}")
+    return parse_stations(payload, sort=False)
+
+
+async def search_stations(client: httpx.AsyncClient, query: str) -> list[Station]:
+    """Search the directory by name, tag, country and language.
+
+    The API has no single full-text search, so the query is sent once per field
+    and the results are merged.  A station that matches a genre rarely has that
+    genre in its name, and vice versa.  Fields that fail are ignored as long as
+    one of them still answers; only a complete failure is raised.
+    """
+    text = query.strip()
+    if not text:
+        return []
+
+    batches = await asyncio.gather(
+        *(_search_field(client, text, field) for field in _SEARCH_FIELDS),
+        return_exceptions=True,
+    )
+
+    stations: list[Station] = []
+    seen: set[str] = set()
+    failures = 0
+    for batch in batches:
+        if isinstance(batch, BaseException):
+            failures += 1
+            continue
+        for station in batch:
+            if station.id in seen:
+                continue
+            seen.add(station.id)
+            stations.append(station)
+
+    if failures == len(batches):
+        raise RuntimeError("could not search the Radio Browser directory")
+
+    return stations
+
+
+async def _fetch_by_uuids(
+    client: httpx.AsyncClient, station_ids: list[str]
+) -> list[Station]:
+    params = urlencode({"uuids": ",".join(station_ids)}, safe=",")
+    payload = await fetch_json(client, f"{BY_UUID_URL}?{params}")
+    return parse_stations(payload, sort=False)
+
+
+async def fetch_stations_by_ids(
+    client: httpx.AsyncClient, station_ids: list[str]
+) -> list[Station]:
+    """Look up specific stations, for favourites outside the first page."""
+    ids = list(
+        dict.fromkeys(
+            station_id.strip()
+            for station_id in station_ids
+            if _UUID_PATTERN.match(station_id.strip())
+        )
+    )
+    if not ids:
+        return []
+
+    batches = [
+        ids[index : index + LOOKUP_BATCH]
+        for index in range(0, len(ids), LOOKUP_BATCH)
+    ]
+    results = await asyncio.gather(
+        *(_fetch_by_uuids(client, batch) for batch in batches),
+        return_exceptions=True,
+    )
+
+    stations: list[Station] = []
+    seen: set[str] = set()
+    for batch in results:
+        if isinstance(batch, BaseException):
+            continue
+        for station in batch:
+            if station.id in seen:
+                continue
+            seen.add(station.id)
+            stations.append(station)
+    return stations