Add dumb: a privacy focused genius frontend alternative

api/_bootstrap.php

@@ -101,3 +101,109 @@ function request_track() {
     $title  = $_POST['title']  ?? $_GET['title']  ?? '';
     return [trim((string) $artist), trim((string) $title)];
 }
+
+/**
+ * Return the inner HTML of a balanced <div> whose opening tag starts at $lt.
+ * Correctly handles nested <div> elements. Shared by the Genius and dumb
+ * lyrics scrapers (both extract a lyrics container out of a raw HTML page).
+ */
+function inner_balanced_div($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 $text;
+}
+
+/**
+ * dumb (https://github.com/rramiachraf/dumb): a self-hosted/public Genius
+ * frontend that server-renders search + song pages. When USE_DUMB is on,
+ * these replace the official Genius API as the lyrics/artwork source (no
+ * GENIUS_TOKEN needed) — same idea as the Genius HTML scrape elsewhere in
+ * this codebase, just against dumb's own pages instead of genius.com.
+ * Fragile by nature (scrapes another project's HTML): if dumb changes its
+ * templates, these regexes may need updating.
+ */
+
+/** Search dumb for a track; returns the first song result's page path (e.g. "/Artist-title-lyrics"), or '' if none. */
+function dumb_search($artist, $title, $timeout) {
+    $base = cfg('dumb.url');
+    if ($base === '') return '';
+    $query = rawurlencode(trim($artist . ' - ' . $title, ' -'));
+    $html = http_get($base . '/search?q=' . $query, $timeout);
+    if ($html === null) return '';
+    if (!preg_match('/<h2>\s*Songs\s*<\/h2>.*?<a[^>]+id="search-item"[^>]+href="([^"]+)"/is', $html, $m)) return '';
+    return html_entity_decode($m[1], ENT_QUOTES | ENT_HTML5, 'UTF-8');
+}
+
+/**
+ * Fetch + parse a dumb song page ($path from dumb_search()). Returns
+ * ['lyrics' => plain text, 'album' => name, 'art' => absolute album art url,
+ * 'song_art' => absolute song thumbnail url] (any field may be ''), or null
+ * if the page couldn't be fetched.
+ */
+function dumb_song($path, $timeout) {
+    $base = cfg('dumb.url');
+    if ($base === '' || $path === '') return null;
+    $html = http_get($base . $path, $timeout);
+    if ($html === null) return null;
+
+    $out = ['lyrics' => '', 'album' => '', 'art' => '', 'song_art' => ''];
+
+    // <div id="lyrics"> wraps the raw Genius lyrics HTML that dumb already
+    // extracted server-side (no data-lyrics-container markers left in it).
+    $idPos = strpos($html, 'id="lyrics"');
+    if ($idPos !== false) {
+        $lt = strrpos(substr($html, 0, $idPos), '<div');
+        if ($lt !== false) $out['lyrics'] = trim(container_to_text(inner_balanced_div($html, $lt)));
+    }
+
+    // Two <img id="album-artwork"> tags can appear: the song thumbnail
+    // (#metadata, always present) and the album cover (#lyrics-album-
+    // container, only when the song belongs to an album — marked with a
+    // title="Album: …" attribute so the two are distinguishable).
+    if (preg_match_all('/<img\b[^>]*>/i', $html, $imgs)) {
+        foreach ($imgs[0] as $img) {
+            if (strpos($img, 'id="album-artwork"') === false) continue;
+            if (!preg_match('/\bsrc="([^"]+)"/i', $img, $src)) continue;
+            $url = $base . html_entity_decode($src[1], ENT_QUOTES | ENT_HTML5, 'UTF-8');
+            if (preg_match('/\btitle="Album:\s*([^"]*)"/i', $img, $t)) {
+                $out['album'] = html_entity_decode($t[1], ENT_QUOTES | ENT_HTML5, 'UTF-8');
+                $out['art'] = $url;
+            } elseif ($out['song_art'] === '') {
+                $out['song_art'] = $url;
+            }
+        }
+    }
+
+    return $out;
+}