takeone-youtube-clone/app/Services/LlmLyricsService.php

<?php

namespace App\Services;

use App\Models\Setting;
use Illuminate\Support\Facades\Cache;
use Illuminate\Support\Facades\Http;
use Illuminate\Support\Facades\Log;

/**
 * Optional LLM helper for the lyrics pipeline.
 *
 * Supports multiple providers configured in Admin → Settings → AI / LLM
 * (local Ollama, hosted Anthropic Claude, or any OpenAI-compatible endpoint).
 * Picks the provider flagged "Active" and dispatches the request through the
 * matching adapter. Results are cached so a regenerate doesn't re-bill / re-hit
 * the local model.
 */
class LlmLyricsService
{
    public function isEnabled(): bool
    {
        return Setting::get('llm_enabled', 'false') === 'true'
            && $this->activeProvider() !== null;
    }

    public function cleanLyricsEnabled(): bool
    {
        return $this->isEnabled() && Setting::get('llm_clean_lyrics', 'true') === 'true';
    }

    public function decorateEnabled(): bool
    {
        return $this->isEnabled() && Setting::get('llm_decorate_lyrics', 'false') === 'true';
    }

    /** The currently selected provider config, or null if none. */
    public function activeProvider(): ?array
    {
        $providers = json_decode((string) Setting::get('llm_providers', '[]'), true) ?: [];
        if (! $providers) return null;
        $activeId = (string) Setting::get('llm_active_id', '');
        foreach ($providers as $p) {
            if (($p['id'] ?? null) === $activeId) {
                $kind = $p['kind'] ?? 'ollama';
                // An Ollama provider doesn't need a key; the others do.
                if ($kind !== 'ollama' && trim((string) ($p['api_key'] ?? '')) === '') return null;
                if (trim((string) ($p['model'] ?? '')) === '') return null;
                return $p;
            }
        }
        return null;
    }

    /** Returns clean lyric lines extracted by the LLM, or [] on any failure. */
    public function cleanDescription(?string $description): array
    {
        if (! $description || ! $this->cleanLyricsEnabled()) return [];

        $provider = $this->activeProvider();
        // v2 cache key: the prompt was rewritten to stop dropping English/Thai
        // lyric lines that happened to carry leading/trailing emoji decoration.
        $cacheKey = 'llm_lyrics_clean_v2:' . ($provider['id'] ?? '') . ':' . sha1($description);
        return Cache::remember($cacheKey, now()->addDays(30), function () use ($description) {
            $prompt = "Extract the SUNG lyric lines from this song description, preserving every\n"
                . "language exactly as written. Songs are often MULTILINGUAL (e.g. mixed English\n"
                . "and Thai, English and Italian, English and Arabic) — KEEP EVERY LANGUAGE.\n\n"
                . "KEEP a line when it contains real lyric words, even if it's wrapped in or\n"
                . "  punctuated by emojis. Example:  '🛡️💻 Met behind the firewalls 🌌'  →  KEEP.\n"
                . "  Strip ONLY the emojis themselves; the lyric words stay untouched.\n"
                . "DROP a line ONLY when it is one of:\n"
                . "  • the song title or artist credit\n"
                . "  • a pure section header (Verse / Chorus / Bridge / Verso / Ritornello /\n"
                . "    Pre-Chorus / Outro / Intro / 副歌 / 후렴 / كورس / ท่อน / etc.) — typically\n"
                . "    one or two words, possibly numbered\n"
                . "  • an instrument or production note inside 【…】 or 〔…〕 brackets\n"
                . "  • a row that is ONLY emojis / separators / decorative symbols with no words\n"
                . "  • commentary or social-media call-to-action (subscribe, follow, link in bio)\n\n"
                . "Hard rules:\n"
                . "  - DO NOT translate. DO NOT re-script (no romanising Thai/Arabic, no converting\n"
                . "    English to Thai). The output of each kept line must be in the SAME language\n"
                . "    and script as the original line.\n"
                . "  - DO NOT merge or split lines. One source lyric line → one output entry.\n"
                . "  - Preserve original punctuation (drop only the emojis).\n"
                . "  - Maintain the original order.\n\n"
                . "Respond with ONLY a JSON array of strings. No prose, no markdown, no code fence.\n\n"
                . "DESCRIPTION:\n" . $description;

            $raw = $this->call($prompt, 8192);
            if ($raw === '') return [];

            $raw = trim(preg_replace('/^```(?:json)?\s*|\s*```$/m', '', $raw));
            $arr = json_decode($raw, true);
            if (! is_array($arr)) return [];
            $out = [];
            foreach ($arr as $line) {
                $line = trim((string) $line);
                if ($line === '') continue;
                $out[] = $line;
            }
            return $out;
        });
    }

    /**
     * Rewrite each lyric line with heavy, expressive emoji styling. Emojis go
     * inside the line AND at the end; multiple per line where it fits. The
     * original words are NEVER changed — emojis are layered on top.
     *
     * Returns [index => decoratedLineText]. The caller swaps line.text and
     * re-distributes the word timings across the new tokens.
     */
    public function decorateLines(array $lines): array
    {
        if (! $lines || ! $this->decorateEnabled()) return [];

        $provider = $this->activeProvider();
        $cacheKey = 'llm_lyrics_deco_v3:' . ($provider['id'] ?? '') . ':' . sha1(json_encode($lines));
        return Cache::remember($cacheKey, now()->addDays(30), function () use ($lines) {
            $numbered = [];
            foreach ($lines as $i => $l) $numbered[] = ($i + 1) . '. ' . $l;

            $prompt = "Decorate the following song lyrics with heavy, expressive emoji styling.\n\n"
                . "Strict instructions:\n"
                . "- Add emojis to almost every line (rich and visually striking, not minimal).\n"
                . "- Place emojis both WITHIN lines and AT THE END where they enhance meaning.\n"
                . "- Use 2–4 emojis per line on average, more on emotional peaks.\n"
                . "- Match emojis to the line's specific emotion, action, image, or vibe.\n"
                . "- VARIETY IS CRITICAL. Across the WHOLE song you must use a wide palette:\n"
                . "    • Aim for 30+ distinct emojis across the song.\n"
                . "    • Never reuse the same emoji on two adjacent lines.\n"
                . "    • Do NOT lean on the same 5–6 staples (🔥💔✨🎵❤️). Reach for less obvious\n"
                . "      ones that fit: ⚡🌊🌙🕯️🪞🥀🦋🌪️🗡️👁️🩸🦅🌀💎🪽🌑🪐🩹🌹🫧🌧️🔮🧨🪞🛡️\n"
                . "      ⚔️🏹🪄💫🥷🧿🪙🥀🎭🩰🪦⛓️🌌🚪🧊🌠💢🪶🩷🫀🪐🕊️ and many others.\n"
                . "- Keep the original lyrics 100% UNCHANGED — no rewriting, no translation, no\n"
                . "  re-spelling, no script conversion. Preserve every original word verbatim.\n"
                . "- Style should feel bold, dramatic, pop-star, Gen Z, visually addictive — like a\n"
                . "  designed lyric post or viral TikTok caption.\n"
                . "- Do NOT add section headers, titles, intros, or any new lines. Every input line\n"
                . "  must map to exactly one output line, in the same order, with the same words.\n\n"
                . "Output format: ONLY a JSON object mapping the 1-based line number to the fully\n"
                . "decorated line text. No prose, no markdown, no code fence.\n\n"
                . "LINES:\n" . implode("\n", $numbered);

            $raw = $this->call($prompt, 8192);
            if ($raw === '') return [];
            $raw = trim(preg_replace('/^```(?:json)?\s*|\s*```$/m', '', $raw));
            $obj = json_decode($raw, true);
            if (! is_array($obj)) return [];

            $out = [];
            foreach ($obj as $k => $v) {
                if (! is_string($v)) continue;
                $v = trim($v);
                if ($v === '') continue;
                $idx = ((int) $k) - 1;
                if ($idx < 0 || ! isset($lines[$idx])) continue;
                // Cheap safety check: the original words must survive verbatim
                // (the LLM should only LAYER emojis on top). Drop the
                // decoration if too many original characters are missing.
                if (! self::preservesOriginal($lines[$idx], $v)) continue;
                $out[$idx] = $v;
            }
            return $out;
        });
    }

    /**
     * Verify the decorated line still contains every alphanumeric character of
     * the original (in the same order). Stops the LLM from quietly rewording
     * a line — we keep only decorations that strictly add emojis on top.
     */
    private static function preservesOriginal(string $orig, string $decorated): bool
    {
        $strip = fn (string $s) => mb_strtolower(preg_replace('/[^\p{L}\p{N}]+/u', '', $s) ?? '');
        $a = $strip($orig);
        $b = $strip($decorated);
        if ($a === '') return true;
        // Sequential subsequence check: every char of $a must appear in $b in order.
        $aLen = mb_strlen($a); $bLen = mb_strlen($b);
        $j = 0;
        for ($i = 0; $i < $aLen; $i++) {
            $needle = mb_substr($a, $i, 1);
            $found = false;
            for (; $j < $bLen; $j++) {
                if (mb_substr($b, $j, 1) === $needle) { $found = true; $j++; break; }
            }
            if (! $found) return false;
        }
        return true;
    }


    /** Dispatch to the active provider's adapter. */
    private function call(string $prompt, int $maxTokens): string
    {
        $p = $this->activeProvider();
        if (! $p) return '';
        try {
            return match ($p['kind']) {
                'anthropic' => $this->callAnthropic($p, $prompt, $maxTokens),
                'openai'    => $this->callOpenAI($p, $prompt, $maxTokens),
                default     => $this->callOllama($p, $prompt, $maxTokens),
            };
        } catch (\Throwable $e) {
            Log::error('LLM call failed: ' . $e->getMessage(), ['provider' => $p['name'] ?? '?']);
            return '';
        }
    }

    private function callOllama(array $p, string $prompt, int $maxTokens): string
    {
        $endpoint = rtrim((string) ($p['endpoint'] ?? 'http://localhost:11434'), '/');
        $resp = Http::timeout(180)->acceptJson()->post($endpoint . '/api/chat', [
            'model'    => $p['model'],
            'messages' => [['role' => 'user', 'content' => $prompt]],
            'stream'   => false,
            'options'  => ['num_predict' => $maxTokens, 'temperature' => 0.2],
        ]);
        if (! $resp->successful()) {
            Log::warning('Ollama API error', ['status' => $resp->status(), 'body' => substr($resp->body(), 0, 500)]);
            return '';
        }
        $j = $resp->json();
        return (string) ($j['message']['content'] ?? '');
    }

    private function callAnthropic(array $p, string $prompt, int $maxTokens): string
    {
        $endpoint = rtrim((string) ($p['endpoint'] ?? 'https://api.anthropic.com'), '/');
        $resp = Http::timeout(120)->withHeaders([
            'x-api-key'         => (string) $p['api_key'],
            'anthropic-version' => '2023-06-01',
            'content-type'      => 'application/json',
        ])->post($endpoint . '/v1/messages', [
            'model'      => $p['model'],
            'max_tokens' => $maxTokens,
            'messages'   => [['role' => 'user', 'content' => $prompt]],
        ]);
        if (! $resp->successful()) {
            Log::warning('Anthropic API error', ['status' => $resp->status(), 'body' => substr($resp->body(), 0, 500)]);
            return '';
        }
        $j = $resp->json();
        return (string) ($j['content'][0]['text'] ?? '');
    }

    private function callOpenAI(array $p, string $prompt, int $maxTokens): string
    {
        $endpoint = rtrim((string) ($p['endpoint'] ?? 'https://api.openai.com'), '/');
        $resp = Http::timeout(120)->withToken((string) $p['api_key'])
            ->acceptJson()
            ->post($endpoint . '/v1/chat/completions', [
                'model'       => $p['model'],
                'messages'    => [['role' => 'user', 'content' => $prompt]],
                'max_tokens'  => $maxTokens,
                'temperature' => 0.2,
            ]);
        if (! $resp->successful()) {
            Log::warning('OpenAI API error', ['status' => $resp->status(), 'body' => substr($resp->body(), 0, 500)]);
            return '';
        }
        $j = $resp->json();
        return (string) ($j['choices'][0]['message']['content'] ?? '');
    }
}