Skip to content

Sources and Channels

aivrar edited this page Aug 1, 2026 · 1 revision

Sources and channels

World Media currently registers eleven adapters. Each adapter converts a public provider response into the same item shape (stable ID, title, type, stream or download capability, artwork, metadata, tags, and rights note). The source is queried directly by the adapter through the local server's approved catalog boundary; the app never pretends that a provider's directory is its own content.

The source list is dynamic. A source can add, remove, rename, rate-limit, or temporarily lose a station/channel without a World Media release. Use the provider's Source link for authoritative rights, schedule, and availability.

Source reference

Source What appears in World Media Live/on-demand behavior Download/record behavior Important caveat
Radio Browser Community internet-radio stations with country, language, tags, click rank, and optional favicon Live audio; mirror is resolved and re-resolved if a mirror fails Record audio; no finite download is claimed for a station stream Community entries can be offline, malformed, or missing artwork
iptv-org Publicly listed TV channels joined from streams, channels, and logos data Live video; up to eight validated stream candidates are retained per channel Record video; HLS candidates are probed and fail over Directory presence does not prove the broadcaster is reachable
Internet Archive Archive films and audio from curated collections and search On-demand; per-item metadata resolves the playable derivative lazily Download finite original/derivative when the item exposes one Rights vary by item; Archive search and metadata can be transiently unavailable
NASA Image and Video Library NASA mission video and audio records with descriptions, dates, and assets On-demand Download when a playable finite asset resolves NASA work is generally public domain in the U.S.; item guidance still applies
Wikimedia Commons Searchable free-licensed/public-domain audio and video files On-demand Download the file URL when available Every file carries its own license and attribution requirements
LibriVox Public-domain audiobook records and chapter/track media On-demand audio Download chapter audio or the available full-audiobook archive RSS, cover, or older mirror endpoints may fail independently
media.ccc.de / C3VOC Technical talks, event recordings, and current conference live rooms VOD recordings plus independently refreshed live snapshots Download VOD; record live C3VOC audio/video An empty live snapshot between events is valid and does not end VOD collection
Library of Congress U.S. cultural-heritage audio and film/video records and resources On-demand when a usable resource is found Conditional; only exposed after rights/resource validation Shared one-request/6-second pacing; CAPTCHA or 429 enters a long cooldown
gPodder Podcasts Podcast directory shows resolved through publisher RSS/Atom feeds On-demand episodes plus feeds that declare a current live enclosure Download finite episodes; record only explicitly live items Dead/slow/malformed publisher feeds are isolated; publisher terms apply
PeerTube Federated video indexed by public SepiaSearch and resolved at exact origins On-demand and live HLS video Conditional download; record live video Independent instances vary in rate limits, licenses, and uptime
Owncast Verified independent self-hosted live video directory entries Live video snapshots refreshed about every two minutes Record live video Entries without an exact rating join fail closed; servers can disappear

Provider pacing and caching

The shared catalog scheduler allows independent source progress while respecting adapter policies:

Adapter Same-source policy in code Cache/refresh highlights
Radio Browser Default one lane Mirror and results coalesced for the session
iptv-org Default one lane Joined streams/channels/logos snapshot once per session
Internet Archive Default one lane plus a 5-request/sec metadata gate Curated browse cursors rotate collections; item metadata is session-cached
NASA Default one lane Search pages are finite; assets resolve lazily
Wikimedia Commons Default one lane Audio/video searches are paged independently and may run in parallel
LibriVox Default one lane Audiobook pages finite; RSS and cover resolution cached in session
media.ccc.de Two lanes, at least 500 ms between starts Recent/search ~5 minutes, details ~1 hour, live ~10 seconds; live refresh about every minute
Library of Congress One lane, at least 6 seconds between starts Catalog ~30 minutes, details ~6 hours; CAPTCHA/429 keeps the cursor and cools down
gPodder Four lanes Directory/feed data ~30 minutes; searches ~5 minutes; at most four feeds active globally and one per host
PeerTube Two lanes, at least 500 ms between starts Index ~5 minutes, origin details ~10 minutes; Retry-After is honored
Owncast One lane Verified snapshot refresh about every 2 minutes; last-known-good data can be stale

These are provider-protection policies, not an item cap. The Library can keep collecting while a source has more authoritative pages. A short or empty page is not treated as exhaustion unless the adapter returns an explicit end marker.

Live snapshots versus finite catalogs

C3VOC and Owncast are snapshot sources. They can legitimately return an empty live set and later return new rooms/servers. The snapshot manager tracks fresh, refreshing, stale, retrying, and live states independently of the finite browse cursor. A Done label applies only to a finite catalog lane; it does not mean a live snapshot will never change.

Content ratings

PeerTube, gPodder, iptv-org, and Owncast normalize provider ratings. The global setting is the only authority that reveals marked content. Filtering happens in Library, Grid, Tuner, Discovery, and favorites views. Enabling the setting never permits unsafe URLs or unverified dynamic origins.

Rights and channel responsibility

World Media passes through public metadata and streams. It does not grant a license, bypass DRM, or verify that a broadcaster has permission to transmit a program. Before downloading or recording, open the source page and follow its license, attribution, broadcaster terms, and local law.

For exact endpoint allowlists, response limits, cache TTLs, retry semantics, and the implementation evidence behind the five newer sources, see docs/PROVIDERS.md and the linked research/execution logs.