movmaker-webui/lib/generate_data_sync.php

<?php
// lib/generate_data_sync.php
// Run generate_data.sh synchronously (no-append) and return the generated JSON/description.
require __DIR__ . '/../auth.php';
mvlog_require_login();

$MVLOG_ROOT = dirname(__DIR__);
const DESCRIBE_DEFAULT_MAX_FRAMES = 32;
header('Content-Type: application/json; charset=utf-8');

$id = strtolower(trim((string)($_POST['id'] ?? '')));
$job = trim((string)($_POST['job'] ?? ''));
$max_frames = isset($_POST['max_frames']) ? intval($_POST['max_frames']) : DESCRIBE_DEFAULT_MAX_FRAMES;
$additional_prompt = trim((string)($_POST['additional_prompt'] ?? ''));
$personality = strtolower(trim((string)($_POST['personality'] ?? 'auto')));
if (!preg_match('/^[a-z0-9_-]+$/', $personality)) $personality = 'auto';

function mvlog_article_id_is_valid(string $id): bool {
    return (bool)preg_match('/^[0-9]{14}[a-f0-9]{16}$/', $id);
}

function mvlog_legacy_job_is_valid(string $job): bool {
    return (bool)preg_match('/^[0-9]{8}[-_][A-Za-z0-9._-]+$/', $job);
}

function mvlog_read_article_id(string $jobdir): string {
    $path = $jobdir . '/.mvlog-id';
    if (!is_file($path)) return '';
    $value = strtolower(trim((string)@file_get_contents($path)));
    return mvlog_article_id_is_valid($value) ? $value : '';
}

function mvlog_find_jobdir_by_id(string $mvlogRoot, string $articleId): ?string {
    if (!mvlog_article_id_is_valid($articleId)) return null;

    $cacheFile = $mvlogRoot . '/cache/articles-index.json';
    if (is_file($cacheFile)) {
        $cache = json_decode((string)@file_get_contents($cacheFile), true);
        if (is_array($cache) && is_array($cache['by_id'] ?? null)) {
            $name = (string)($cache['by_id'][$articleId] ?? '');
            if ($name !== '') {
                $candidate = $mvlogRoot . '/in-dir/' . basename($name);
                if (is_dir($candidate)) return $candidate;
            }
        }
    }

    foreach (glob($mvlogRoot . '/in-dir/*', GLOB_ONLYDIR) ?: [] as $dir) {
        if (mvlog_read_article_id($dir) === $articleId) return $dir;
    }

    return null;
}

$jobdir = null;
if ($id !== '' && mvlog_article_id_is_valid($id)) {
    $jobdir = mvlog_find_jobdir_by_id($MVLOG_ROOT, $id);
}
if ($jobdir === null && $job !== '' && mvlog_legacy_job_is_valid($job)) {
    $candidate = $MVLOG_ROOT . '/in-dir/' . basename($job);
    if (is_dir($candidate)) $jobdir = $candidate;
}

if ($jobdir === null || !is_dir($jobdir)) {
    http_response_code(404);
    echo json_encode(['status'=>'error','message'=>'job not found']);
    exit;
}

$describe_prompt_file = $jobdir . '/.mvlog-describe-prompt.txt';
if ($additional_prompt !== '') {
    @file_put_contents($describe_prompt_file, $additional_prompt . PHP_EOL);
    @chmod($describe_prompt_file, 0664);
    @chown($describe_prompt_file, 'www-data');
    @chgrp($describe_prompt_file, 'www-data');
}

// Read existing data.txt for Title and Location (case-insensitive aliases)
$data_file = $jobdir . '/data.txt';
$title = '';
$location = '';
if (is_file($data_file)) {
    $lines = file($data_file, FILE_IGNORE_NEW_LINES);
    $i = 0;
    while ($i < count($lines)) {
        $raw = $lines[$i];
        $line = trim($raw);
        if ($line === '' || str_starts_with($line, '#')) { $i++; continue; }
        if (!str_contains($line, ':')) { $i++; continue; }
        [$key, $value] = array_map('trim', explode(':', $line, 2));
        $low = strtolower($key);
        if ($value === '|' && in_array($low, ['description','desc','synopsis'], true)) {
            // Skip indented block
            $i++;
            while ($i < count($lines)) {
                $next = $lines[$i];
                if (trim($next) !== '' && $next[0] !== ' ' && $next[0] !== "\t") break;
                $i++;
            }
            continue;
        }
        if (in_array($low, ['title','name'], true) && $title === '') $title = $value;
        if (in_array($low, ['place','location','where'], true) && $location === '') $location = $value;
        $i++;
    }
}

// Run generator synchronously but in no-append mode so it doesn't modify data.txt directly.
$script = $MVLOG_ROOT . '/bin/generate_data.sh';
if (!is_file($script)) {
    http_response_code(500);
    echo json_encode(['status' => 'error', 'message' => 'generate_data.sh not found']);
    exit;
}

// Build command and execute. Pass optional additional prompt and chosen personality.
$prompt_arg = $additional_prompt !== '' ? $additional_prompt : 'no-append';
$cmd = escapeshellcmd($script) . ' ' . escapeshellarg($jobdir) . ' ' . escapeshellarg($title) . ' ' . escapeshellarg($location) . ' ' . escapeshellarg((string)$max_frames) . ' ' . escapeshellarg($prompt_arg) . ' ' . escapeshellarg($personality) . ' 2>&1';

// Allow long-running
@set_time_limit(0);

$gfile = dirname($jobdir) . '/' . basename($jobdir) . '.json';
$gfileExistedBefore = is_file($gfile);
$gfileMtimeBefore = $gfileExistedBefore ? @filemtime($gfile) : null;

exec($cmd, $out_lines, $rc);
$log = implode("\n", $out_lines);
clearstatcache(true, $gfile);

if ($rc !== 0) {
    http_response_code(500);
    echo json_encode([
        'status' => 'error',
        'message' => 'generator failed',
        'log' => $log,
        'rc' => $rc,
        'output_file' => basename($gfile),
        'stale_output_present' => is_file($gfile),
    ], JSON_UNESCAPED_UNICODE);
    exit;
}

if (!is_file($gfile)) {
    http_response_code(500);
    echo json_encode([
        'status' => 'error',
        'message' => 'no gemini output',
        'log' => $log,
        'rc' => $rc,
    ], JSON_UNESCAPED_UNICODE);
    exit;
}

// If gemini_generated.json exists, parse and return
if (is_file($gfile)) {
    $raw = @file_get_contents($gfile);
    $json_data = @json_decode($raw, true);

    // Handle case where Gemini returns an array of candidates or a single object.
    // We will always take the first valid description object we can find.
    $json = null;
    if (is_array($json_data) && isset($json_data[0]) && is_array($json_data[0])) {
        $json = $json_data[0]; // It's an array of descriptions, take the first.
    } elseif (is_array($json_data)) {
        $json = $json_data; // It's already a single description object.
    }

    $description = '';
    $teaser = '';
    $quote_da = '';
    $personality = '';
    $tags = [];
    if (is_array($json)) {
        $description = isset($json['description']) ? (string)$json['description'] : '';
        $teaser = isset($json['teaser']) ? (string)$json['teaser'] : '';
        $quote_da = isset($json['quote_da']) ? (string)$json['quote_da'] : '';
        $quote_da = trim($quote_da, "\"“”");
        $personality = isset($json['personality']) ? (string)$json['personality'] : '';
        $tags = isset($json['tags']) && is_array($json['tags']) ? $json['tags'] : [];
    }

    echo json_encode([
        'status' => 'ok',
        'description' => $description,
        'teaser' => $teaser,
        'quote_da' => $quote_da,
        'personality' => $personality,
        'tags' => $tags,
        'raw_json' => $json_data, // Send original raw data back for reference
        'log' => $log,
        'rc' => $rc,
    ], JSON_UNESCAPED_UNICODE);
    exit;
}

// No gemini file — return error and include log
http_response_code(500);
echo json_encode(['status'=>'error','message'=>'no gemini output','log'=>$log,'rc'=>$rc], JSON_UNESCAPED_UNICODE);
exit;