| 1 |
<?php |
| 2 |
/** |
| 3 |
* Records when podcast directory crawlers fetch the feed. |
| 4 |
* |
| 5 |
* @package automattic/jetpack-podcast |
| 6 |
*/ |
| 7 |
|
| 8 |
declare( strict_types = 1 ); |
| 9 |
|
| 10 |
namespace Automattic\Jetpack\Podcast\Feed; |
| 11 |
|
| 12 |
use Automattic\Jetpack\Podcast\Settings; |
| 13 |
|
| 14 |
/** |
| 15 |
* Promotes an existing pending podcatcher `podcasting_show_states` entry to |
| 16 |
* `'active'` the first time we see its UA fetch the feed. Idempotent |
| 17 |
* thereafter. |
| 18 |
*/ |
| 19 |
class Feed_Detection { |
| 20 |
|
| 21 |
/** |
| 22 |
* UA needle → podcatcher slug. First match wins, so order matters: more |
| 23 |
* specific brand needles (e.g. `GooglePodcasts`, `YouTubeMusic`) come |
| 24 |
* before generic ones (e.g. Apple's `Podcasts/`) that they share a |
| 25 |
* substring with. |
| 26 |
* |
| 27 |
* Cross-referenced against opawg/user-agents-v2 (the OPAWG community |
| 28 |
* dataset used by Podtrac, Megaphone, Art19, Chartable). When directories |
| 29 |
* change UAs or new ones emerge, sync the relevant entries here: |
| 30 |
* https://github.com/opawg/user-agents-v2/blob/master/src/{apps,bots}.json |
| 31 |
* |
| 32 |
* @var array<string, string> |
| 33 |
*/ |
| 34 |
private const NEEDLES = array( |
| 35 |
// YouTube / Google Podcasts — listed first so `GooglePodcasts/` |
| 36 |
// doesn't get caught by Apple's broader `Podcasts/` needle below. |
| 37 |
'Google-Podcast' => 'youtube', |
| 38 |
'YouTube-Podcast' => 'youtube', |
| 39 |
'GooglePodcasts' => 'youtube', |
| 40 |
'GoogleChirp' => 'youtube', |
| 41 |
'YouTubeMusic' => 'youtube', |
| 42 |
|
| 43 |
// Apple Podcasts (iOS app, Mac app, automated checks, HomePod, Apple TV). |
| 44 |
'iTMS' => 'apple', |
| 45 |
'AppleCoreMedia' => 'apple', |
| 46 |
'Podcasts/' => 'apple', |
| 47 |
'iTunes' => 'apple', |
| 48 |
'AirPodcasts/' => 'apple', |
| 49 |
|
| 50 |
// Spotify — substring catches `Spotify/…`, `spotify-rss-…`, all variants. |
| 51 |
'Spotify' => 'spotify', |
| 52 |
|
| 53 |
// Pocket Casts. |
| 54 |
'Pocket Casts' => 'pocketcasts', |
| 55 |
'PocketCasts' => 'pocketcasts', |
| 56 |
|
| 57 |
// Amazon — `AmazonMusic` is the listening app; `Amazon Music Podcast` |
| 58 |
// (with spaces) is the actual feed crawler. Keep both. |
| 59 |
'AmazonMusic' => 'amazon', |
| 60 |
'Amazon Music Podcast' => 'amazon', |
| 61 |
|
| 62 |
// Podcast Index — substring `PodcastIndex` catches `Podcastindex.org/`, |
| 63 |
// `PodcastIndexer/`, `PodcastIndexManager/`, `PodcastIndex Classifier/`. |
| 64 |
'PodcastIndex' => 'podcastindex', |
| 65 |
); |
| 66 |
|
| 67 |
/** |
| 68 |
* Inspect the current request's User-Agent and, if it's a directory |
| 69 |
* crawler with an existing pending state, mark its state `'active'`. |
| 70 |
* No-op if the UA is missing, not in the directory allowlist, or the |
| 71 |
* matched directory has no pending state. |
| 72 |
*/ |
| 73 |
public static function detect_and_record(): void { |
| 74 |
$ua = isset( $_SERVER['HTTP_USER_AGENT'] ) |
| 75 |
? sanitize_text_field( wp_unslash( $_SERVER['HTTP_USER_AGENT'] ) ) |
| 76 |
: ''; |
| 77 |
if ( '' === $ua ) { |
| 78 |
return; |
| 79 |
} |
| 80 |
|
| 81 |
$slug = null; |
| 82 |
foreach ( self::NEEDLES as $needle => $candidate ) { |
| 83 |
if ( false !== stripos( $ua, $needle ) ) { |
| 84 |
$slug = $candidate; |
| 85 |
break; |
| 86 |
} |
| 87 |
} |
| 88 |
|
| 89 |
if ( null === $slug || ! isset( Settings::SHOW_URL_HOSTS[ $slug ] ) ) { |
| 90 |
return; |
| 91 |
} |
| 92 |
|
| 93 |
$states = get_option( 'podcasting_show_states', array() ); |
| 94 |
if ( ! is_array( $states ) ) { |
| 95 |
$states = array(); |
| 96 |
} |
| 97 |
|
| 98 |
if ( ! isset( $states[ $slug ] ) || 'pending' !== $states[ $slug ] ) { |
| 99 |
return; |
| 100 |
} |
| 101 |
|
| 102 |
// Concurrent first-fetches from different apps can race the read-modify-write. |
| 103 |
// Benign: the loser's next poll (multi-hour cadence) re-writes its key. Worst |
| 104 |
// case is one app sitting in `pending` for one extra poll cycle. |
| 105 |
$states[ $slug ] = 'active'; |
| 106 |
update_option( 'podcasting_show_states', $states ); |
| 107 |
} |
| 108 |
} |
| 109 |
|
