Build a graph of all artists played on a BBC Radio show and their last.fm tags

fetch-broadcasts.php

<?php

define('LASTFM_API_KEY', 'EDIT_THIS'); // get an API key from http://www.last.fm/api

$brand = 'EDIT_THIS'; // PID of the "brand" e.g. b00c72y1 for Marc Riley

$url = sprintf('http://open.live.bbc.co.uk/aps/programmes/%s/episodes/upcoming.json', $brand);
$data = get_json($url);

$count = count($data->broadcasts);
$programme = $data->broadcasts[$count - 1]->programme; // most recent broadcast

$output = fopen($brand . '-edges.csv', 'w');
fputcsv($output, array('Source', 'Target', 'Weight'));

$pids = [];
$tags = [];
$plays = [];
$artists = [];
$tagged = [];

do {
    $details = get_programme($programme->pid);
    $canonicalPID = canonical_version($details);

    if (!array_key_exists($canonicalPID, $pids)) {
        $pids[$canonicalPID] = true;

        $canonical = get_version($canonicalPID);

        foreach ($canonical->segment_events as $event) {
            $segment = $event->segment;

            if ($segment->type !== 'music') {
                continue;
            }

            if (!$contributor = $segment->primary_contributor) {
                continue;
            }

            if (!$mbid = $contributor->musicbrainz_gid) {
                continue;
            }

            if (!$name = $contributor->name) {
                continue;
            }

            if (!array_key_exists($mbid, $artists)) {
                $artists[$mbid] = $name;
            }

            if (!array_key_exists($mbid, $plays)) {
                $plays[$mbid] = 0;
            }

            $plays[$mbid]++;

            if (!array_key_exists($mbid, $tagged)) {
                $tagged[$mbid] = true;

                foreach (get_tags($mbid) as $tag) {
                    if ($tag->count < 10) {
                        break;
                    }

                    if (!array_key_exists($tag->name, $tags)) {
                        $tags[$tag->name] = 0;
                    }

                    $tags[$tag->name]++;

                    fputcsv($output, array($mbid, $tag->name, $tag->count));
                }
            }
        }
    }

    $programme = $details->peers->previous;
} while ($programme);

fclose($output);

$output = fopen($brand . '-nodes.csv', 'w');
fputcsv($output, array('ID', 'Label', 'Type', 'Plays'));

foreach ($artists as $mbid => $name) {
    fputcsv($output, array($mbid, $name, 'artist', $plays[$mbid]));
}

foreach ($tags as $tag => $count) {
    fputcsv($output, array($tag, $tag, 'tag'));
}

fclose($output);

function get_json($url) {
    print $url . "\n";
    $json = file_get_contents($url);

    return json_decode($json);
}

function get_tags($mbid) {
    $url = 'http://ws.audioscrobbler.com/2.0/?' . http_build_query(array(
        'method' => 'artist.gettoptags',
        'mbid' => $mbid,
        'api_key' => LASTFM_API_KEY,
        'format' => 'json',
    ));

    $data = get_json($url);

    if (!$data->toptags) {
        return [];
    }

    return $data->toptags->tag;
}

function get_programme($pid) {
    $url = sprintf('http://open.live.bbc.co.uk/aps/programmes/%s.json', $pid);
    $data = get_json($url);

    return $data->programme;
}

function get_version($pid) {
    $url = sprintf('http://open.live.bbc.co.uk/aps/programmes/%s.json', $pid);
    $data = get_json($url);

    return $data->version;
}

function canonical_version($details) {
    foreach ($details->versions as $version) {
        if ($version->canonical) {
            return $version->pid;
        }
    }
}