<?php
declare(strict_types=1);

/*
 * Merge PDFs through the public bearer API. Requires PHP 8.1+ with ext-curl.
 * Usage: php merge.php MAX_CREDITS OUTPUT_DIR FIRST.pdf SECOND.pdf [MORE.pdf ...]
 * Set SOFTWAREFOOL_API_KEY and, for local development only, optionally set
 * SOFTWAREFOOL_API_BASE_URL=http://localhost:8001/api/v1.
 */

function apiBase(): array
{
    $raw = rtrim(getenv('SOFTWAREFOOL_API_BASE_URL') ?: 'https://softwarefool.com/api/v1', '/');
    $parts = parse_url($raw);
    if ($parts === false || !isset($parts['scheme'], $parts['host'])
        || ($parts['path'] ?? '') !== '/api/v1'
        || isset($parts['user']) || isset($parts['pass'])
        || isset($parts['query']) || isset($parts['fragment'])) {
        throw new RuntimeException('Use a base URL ending in /api/v1.');
    }
    $scheme = strtolower($parts['scheme']);
    $host = strtolower($parts['host']);
    if ($scheme !== 'https' && !($scheme === 'http' && in_array($host, ['localhost', '127.0.0.1', '[::1]'], true))) {
        throw new RuntimeException('Use HTTPS, or HTTP on a loopback development server.');
    }
    $origin = $scheme . '://' . $parts['host'] . (isset($parts['port']) ? ':' . $parts['port'] : '');
    return [$origin . '/api/v1', $origin];
}

function apiLink(string $origin, mixed $path): string
{
    if (!is_string($path) || !preg_match('~\A/api/v1/[A-Za-z0-9/_-]+/\z~D', $path)) {
        throw new RuntimeException('The API returned an unexpected link.');
    }
    return $origin . $path;
}

function readJsonResponse($handle, mixed $body): array
{
    if ($body === false) {
        throw new RuntimeException('HTTP request failed: ' . curl_error($handle));
    }
    $status = curl_getinfo($handle, CURLINFO_RESPONSE_CODE);
    $decoded = json_decode($body, true);
    if ($status < 200 || $status >= 300) {
        $code = is_array($decoded) ? ($decoded['error']['code'] ?? 'request_failed') : 'request_failed';
        throw new RuntimeException("API returned HTTP $status ($code).");
    }
    if (!is_array($decoded) || !array_key_exists('data', $decoded)) {
        throw new RuntimeException('The API returned an unexpected JSON response.');
    }
    return $decoded;
}

function jsonRequest(string $url, string $apiKey, ?array $payload = null, ?string $idempotencyKey = null): array
{
    $handle = curl_init($url);
    if ($handle === false) {
        throw new RuntimeException('Could not start an HTTP request.');
    }
    $headers = ['Authorization: Bearer ' . $apiKey, 'Accept: application/json'];
    if ($idempotencyKey !== null) {
        $headers[] = 'Idempotency-Key: ' . $idempotencyKey;
    }
    $options = [
        CURLOPT_HTTPHEADER => $headers,
        CURLOPT_RETURNTRANSFER => true,
        CURLOPT_FOLLOWLOCATION => false,
        CURLOPT_CONNECTTIMEOUT => 15,
        CURLOPT_TIMEOUT => 30,
    ];
    if ($payload !== null) {
        $headers[] = 'Content-Type: application/json';
        $options[CURLOPT_HTTPHEADER] = $headers;
        $options[CURLOPT_POST] = true;
        $options[CURLOPT_POSTFIELDS] = json_encode($payload, JSON_THROW_ON_ERROR);
    }
    try {
        curl_setopt_array($handle, $options);
        return readJsonResponse($handle, curl_exec($handle));
    } finally {
        curl_close($handle);
    }
}

function writeAll($stream, string $text): void
{
    $offset = 0;
    while ($offset < strlen($text)) {
        $written = fwrite($stream, substr($text, $offset));
        if ($written === false || $written === 0) {
            throw new RuntimeException('Could not prepare the multipart upload.');
        }
        $offset += $written;
    }
}

function confirmQuote(string $url, string $apiKey, string $key, string $token, array $files): array
{
    // A PHP CURLOPT_POSTFIELDS map cannot express repeated fields named "files".
    // Build a disk-backed multipart stream to preserve order without loading PDFs into RAM.
    $boundary = 'softwarefool-' . bin2hex(random_bytes(16));
    $body = tmpfile();
    if ($body === false) {
        throw new RuntimeException('Could not create a temporary upload stream.');
    }
    try {
        writeAll($body, "--$boundary\r\nContent-Disposition: form-data; name=\"quote_token\"\r\n\r\n$token\r\n");
        foreach ($files as $index => $file) {
            $number = $index + 1;
            writeAll($body, "--$boundary\r\nContent-Disposition: form-data; name=\"files\"; filename=\"input-$number.pdf\"\r\nContent-Type: application/pdf\r\n\r\n");
            $source = fopen($file['path'], 'rb');
            if ($source === false) {
                throw new RuntimeException('An input PDF is no longer readable.');
            }
            try {
                $copied = stream_copy_to_stream($source, $body);
            } finally {
                fclose($source);
            }
            if ($copied !== $file['bytes']) {
                throw new RuntimeException('An input PDF changed after quoting. Start again with a new quote.');
            }
            writeAll($body, "\r\n");
        }
        writeAll($body, "--$boundary--\r\n");
        $length = ftell($body);
        if ($length === false || !rewind($body)) {
            throw new RuntimeException('Could not read the prepared upload.');
        }
        for ($attempt = 1; $attempt <= 3; $attempt++) {
            rewind($body);
            $handle = curl_init($url);
            if ($handle === false) {
                throw new RuntimeException('Could not start an HTTP request.');
            }
            try {
                curl_setopt_array($handle, [
                    CURLOPT_HTTPHEADER => [
                        'Authorization: Bearer ' . $apiKey,
                        'Idempotency-Key: ' . $key,
                        'Accept: application/json',
                        'Content-Type: multipart/form-data; boundary=' . $boundary,
                    ],
                    CURLOPT_UPLOAD => true,
                    CURLOPT_CUSTOMREQUEST => 'POST',
                    CURLOPT_INFILESIZE => $length,
                    CURLOPT_READFUNCTION => static function ($handle, $stream, int $count) use ($body): string {
                        $chunk = fread($body, $count);
                        return $chunk === false ? '' : $chunk;
                    },
                    CURLOPT_RETURNTRANSFER => true,
                    CURLOPT_FOLLOWLOCATION => false,
                    CURLOPT_CONNECTTIMEOUT => 15,
                    CURLOPT_TIMEOUT => 300,
                ]);
                $response = curl_exec($handle);
                $status = curl_getinfo($handle, CURLINFO_RESPONSE_CODE);
                // A lost response may hide an accepted job. Retry the same body,
                // quote token, and idempotency key, never a new submission.
                if ($attempt < 3 && ($response === false || in_array($status, [429, 500, 502, 503, 504], true))) {
                    sleep($attempt);
                    continue;
                }
                return readJsonResponse($handle, $response);
            } finally {
                curl_close($handle);
            }
        }
    } finally {
        fclose($body);
    }
}

function downloadResult(string $url, string $apiKey, string $outputDir, array $result): string
{
    $name = $result['filename'] ?? null;
    $expected = $result['bytes'] ?? null;
    if (!is_string($name) || !preg_match('/\A[A-Za-z0-9][A-Za-z0-9._ -]{0,239}\z/D', $name)
        || str_contains($name, '..') || !is_int($expected) || $expected <= 0) {
        throw new RuntimeException('The API returned invalid result metadata.');
    }
    if (!is_dir($outputDir) && !mkdir($outputDir, 0770, true) && !is_dir($outputDir)) {
        throw new RuntimeException('Could not create the output directory.');
    }
    $destination = rtrim($outputDir, '/\\') . DIRECTORY_SEPARATOR . $name;
    $output = fopen($destination, 'x+b'); // Never overwrite an earlier result.
    if ($output === false) {
        throw new RuntimeException('The result destination already exists or cannot be created.');
    }
    $handle = curl_init($url);
    if ($handle === false) {
        fclose($output);
        unlink($destination);
        throw new RuntimeException('Could not start a download.');
    }
    $received = 0;
    try {
        curl_setopt_array($handle, [
            CURLOPT_HTTPHEADER => ['Authorization: Bearer ' . $apiKey],
            CURLOPT_FOLLOWLOCATION => false,
            CURLOPT_CONNECTTIMEOUT => 15,
            CURLOPT_TIMEOUT => 300,
            CURLOPT_WRITEFUNCTION => static function ($handle, string $chunk) use ($output, $expected, &$received): int {
                $size = strlen($chunk);
                if ($received + $size > $expected) {
                    return 0;
                }
                $written = fwrite($output, $chunk);
                if ($written === false) {
                    return 0;
                }
                $received += $written;
                return $written;
            },
        ]);
        $response = curl_exec($handle);
        $status = curl_getinfo($handle, CURLINFO_RESPONSE_CODE);
        if ($response === false || $status !== 200 || $received !== $expected) {
            throw new RuntimeException('The result download failed or was incomplete.');
        }
    } catch (Throwable $error) {
        fclose($output);
        unlink($destination);
        throw $error;
    } finally {
        curl_close($handle);
    }
    fclose($output);
    return $destination;
}

function main(array $argv): void
{
    if (count($argv) < 5) {
        throw new RuntimeException('Usage: php merge.php MAX_CREDITS OUTPUT_DIR FIRST.pdf SECOND.pdf [MORE.pdf ...]');
    }
    if (!extension_loaded('curl')) {
        throw new RuntimeException('Enable the PHP cURL extension.');
    }
    $apiKey = getenv('SOFTWAREFOOL_API_KEY');
    if (!is_string($apiKey) || $apiKey === '' || strpbrk($apiKey, "\r\n") !== false) {
        throw new RuntimeException('Set SOFTWAREFOOL_API_KEY to a valid bearer key.');
    }
    $budget = filter_var($argv[1], FILTER_VALIDATE_INT, ['options' => ['min_range' => 1]]);
    if ($budget === false) {
        throw new RuntimeException('MAX_CREDITS must be a positive whole number.');
    }
    [$base, $origin] = apiBase();
    $files = [];
    foreach (array_slice($argv, 3) as $path) {
        if (!is_file($path) || !is_readable($path)) {
            throw new RuntimeException("Cannot read input: $path");
        }
        $bytes = filesize($path);
        $hash = hash_file('sha256', $path);
        if (!is_int($bytes) || $bytes <= 0 || !is_string($hash)) {
            throw new RuntimeException("Cannot inspect input: $path");
        }
        $files[] = ['path' => $path, 'bytes' => $bytes, 'sha256' => $hash];
    }
    $key = bin2hex(random_bytes(16));
    $manifest = [
        'files' => array_map(static fn(array $file): array => [
            'bytes' => $file['bytes'], 'sha256' => $file['sha256'],
        ], $files),
        'parameters' => (object) [],
    ];
    $quote = jsonRequest($base . '/tools/merge-pdf/quotes/', $apiKey, $manifest, $key)['data'];
    $cost = $quote['cost_credits'] ?? null;
    $token = $quote['token'] ?? null;
    if (($quote['tool'] ?? null) !== 'merge-pdf' || !is_int($cost) || $cost < 1
        || !is_string($token) || $token === '' || strpbrk($token, "\r\n") !== false
        || !is_bool($quote['can_afford'] ?? null)
        || ($quote['file_count'] ?? null) !== count($files)
        || ($quote['total_bytes'] ?? null) !== array_sum(array_column($files, 'bytes'))) {
        throw new RuntimeException('The quote does not match the selected files.');
    }
    printf("Quote: %d credits; available: %d credits; budget: %d credits.\n",
        $cost, $quote['wallet']['available_credits'] ?? 0, $budget);
    if (!$quote['can_afford'] || $cost > $budget) {
        throw new RuntimeException('The quote is unaffordable or exceeds your budget.');
    }
    echo "Confirm this quoted job? Type yes: ";
    if (strtolower(trim((string) fgets(STDIN))) !== 'yes') {
        echo "Cancelled before upload; no credits reserved.\n";
        return;
    }
    foreach ($files as $file) {
        if (filesize($file['path']) !== $file['bytes']
            || hash_file('sha256', $file['path']) !== $file['sha256']) {
            throw new RuntimeException('An input PDF changed after quoting. Start again with a new quote.');
        }
    }
    echo "Idempotency key: $key\n";
    $confirmed = confirmQuote($base . '/tools/merge-pdf/quotes/confirm/', $apiKey, $key, $token, $files);
    $job = $confirmed['data'];
    $jobId = $job['id'] ?? null;
    if (!is_string($jobId) || !preg_match('/\A[0-9a-f-]{36}\z/D', $jobId)
        || ($job['accepted_credits'] ?? null) !== $cost) {
        throw new RuntimeException('The API returned an unexpected job.');
    }
    echo "Accepted job: $jobId\n";
    $jobUrl = apiLink($origin, $job['links']['self'] ?? null);
    $deadline = time() + 300;
    $delay = 2;
    while (true) {
        $job = jsonRequest($jobUrl, $apiKey)['data'];
        if (($job['id'] ?? null) !== $jobId) {
            throw new RuntimeException('The API returned a different job while polling.');
        }
        $status = $job['status'] ?? null;
        if ($status === 'succeeded') {
            break;
        }
        if ($status === 'failed' || $status === 'cancelled') {
            throw new RuntimeException("Job $jobId $status; reserved credits were released.");
        }
        if (!in_array($status, ['queued', 'running'], true)) {
            throw new RuntimeException('The API returned an unknown job status.');
        }
        if (time() >= $deadline) {
            throw new RuntimeException("Job $jobId is still $status. Poll its job URL later.");
        }
        sleep(min($delay, max(1, $deadline - time())));
        $delay = min($delay * 2, 8);
    }
    $results = jsonRequest(apiLink($origin, $job['links']['results'] ?? null), $apiKey)['data'];
    if (!is_array($results) || $results === []) {
        throw new RuntimeException('The completed job has no results.');
    }
    $saved = 0;
    foreach ($results as $result) {
        if (($result['status'] ?? null) !== 'available') {
            continue;
        }
        $path = downloadResult(apiLink($origin, $result['links']['download'] ?? null), $apiKey, $argv[2], $result);
        echo "Saved: $path\n";
        $saved++;
    }
    if ($saved === 0) {
        throw new RuntimeException('No available result files remain to download.');
    }
}

try {
    main($argv);
} catch (Throwable $error) {
    fwrite(STDERR, $error->getMessage() . PHP_EOL);
    exit(1);
}