radio/api/lyrics.php

<?php
/**
 * api/lyrics.php
 * --------------
 * Optional lyrics lookup. Finds the song on Genius, then scrapes the lyric
 * containers from the public song page. Returns JSON:
 *
 *   { "found": true|false, "lyrics": "plain text\nwith newlines", "url": "…" }
 *
 * Lyrics are returned as plain text (tags stripped) so the front-end can render
 * them with textContent — no HTML injection from a third-party page.
 *
 * Note: this scrapes Genius' HTML, which can change at any time. If Genius
 * restructures their page the extraction may need updating.
 */

require_once __DIR__ . '/_bootstrap.php';

$empty = ['found' => false, 'lyrics' => '', 'url' => ''];

if (!cfg('lyrics.enabled') || cfg('genius.token') === '') {
    json_out($empty);
}

[$artist, $title] = request_track();
if ($artist === '' && $title === '') {
    json_out($empty);
}

$token   = cfg('genius.token');
$timeout = 5;
$headers = ['Authorization: Bearer ' . $token];

/**
 * Return the inner HTML of a balanced <div> whose opening tag starts at $lt.
 * Correctly handles nested <div> elements.
 */
function inner_balanced_div($html, $lt) {
    $open = strpos($html, '>', $lt);
    if ($open === false) return '';
    $i = $open + 1;
    $start = $i;
    $len = strlen($html);
    $depth = 1;
    while ($i < $len && $depth > 0) {
        $next = strpos($html, '<', $i);
        if ($next === false) break;
        if (strncasecmp(substr($html, $next, 5), '</div', 5) === 0) {
            $depth--;
            if ($depth === 0) return substr($html, $start, $next - $start);
            $i = $next + 5;
        } elseif (strncasecmp(substr($html, $next, 4), '<div', 4) === 0) {
            $depth++;
            $i = $next + 4;
        } else {
            $i = $next + 1;
        }
    }
    return substr($html, $start, max(0, $i - $start));
}

/** Turn one container's HTML into plain text with newlines. */
function container_to_text($fragment) {
    // Line breaks -> newlines before stripping tags.
    $fragment = preg_replace('#<br\s*/?>#i', "\n", $fragment);
    // Drop any stray script/style.
    $fragment = preg_replace('#<(script|style)\b[^>]*>.*?</\1>#is', '', $fragment);
    // Block-level tags imply a line break.
    $fragment = preg_replace('#</(p|div)>#i', "\n", $fragment);
    $text = strip_tags($fragment);
    $text = html_entity_decode($text, ENT_QUOTES | ENT_HTML5, 'UTF-8');
    return $text;
}

/* -- Find the song on Genius ----------------------------------------------- */
$query  = rawurlencode(trim($artist . ' - ' . $title, ' -'));
$search = http_get('https://api.genius.com/search?q=' . $query, $timeout, $headers);
if ($search === null) json_out($empty);

$json = json_decode($search, true);
$hits = dot_get($json, 'response.hits', []);
if (!is_array($hits) || count($hits) === 0) json_out($empty);

$pageUrl = (string) dot_get($hits, '0.result.url', '');

// Optionally prefer the first English-language match (a few extra API calls).
if (cfg('lyrics.prefer_english')) {
    $checked = 0;
    foreach ($hits as $h) {
        if ($checked++ >= 5) break;
        $id = dot_get($h, 'result.id', null);
        if (!$id) continue;
        $songRaw = http_get('https://api.genius.com/songs/' . rawurlencode((string) $id), $timeout, $headers);
        if ($songRaw === null) continue;
        if (dot_get(json_decode($songRaw, true), 'response.song.language', '') === 'en') {
            $pageUrl = (string) dot_get($h, 'result.url', $pageUrl);
            break;
        }
    }
}
if ($pageUrl === '') json_out($empty);

/* -- Scrape the lyric containers ------------------------------------------- */
$page = http_get($pageUrl, $timeout);
if ($page === null) {
    json_out(['found' => false, 'lyrics' => '', 'url' => $pageUrl]);
}

$marker = 'data-lyrics-container="true"';
$pieces = [];
$offset = 0;
while (($mpos = strpos($page, $marker, $offset)) !== false) {
    // Backtrack to the start of this opening <div ...> tag.
    $lt = strrpos(substr($page, 0, $mpos), '<div');
    if ($lt === false) { $offset = $mpos + strlen($marker); continue; }
    $inner = inner_balanced_div($page, $lt);
    $text  = trim(container_to_text($inner));
    if ($text !== '') $pieces[] = $text;
    $offset = $mpos + strlen($marker);
}

$lyrics = trim(implode("\n\n", $pieces));
// Trim stray indentation left on each line, then collapse blank-line runs.
$lyrics = implode("\n", array_map('trim', explode("\n", $lyrics)));
$lyrics = preg_replace("/\n{3,}/", "\n\n", $lyrics);

if ($lyrics === '') {
    json_out(['found' => false, 'lyrics' => '', 'url' => $pageUrl]);
}

json_out(['found' => true, 'lyrics' => $lyrics, 'url' => $pageUrl]);