dobetternorge-tools/scripts/translate-pages.php

<?php
/**
 * translate-pages.php
 *
 * CLI utility: translates the 'en' entries in a translations/*.php file
 * into Norwegian (no), Ukrainian (uk), and Polish (pl) using Azure OpenAI gpt-4o.
 *
 * Usage:
 *   php scripts/translate-pages.php translations/transcribe-about.php
 *
 * The file is updated in-place — existing no/uk/pl keys are replaced.
 *
 * Config (env vars or constants below):
 *   AZURE_OPENAI_KEY      — Azure OpenAI API key
 *   AZURE_OPENAI_ENDPOINT — e.g. https://exos-openai-test.openai.azure.com
 *   AZURE_OPENAI_DEPLOY   — deployment name, default gpt-4o
 *   AZURE_OPENAI_VERSION  — API version, default 2024-02-01
 */

declare(strict_types=1);

// ── Config ──────────────────────────────────────────────────────────────────
$apiKey    = getenv('AZURE_OPENAI_KEY')      ?: '7cb23c8ce75741c798a7dfe24827c95e';
$endpoint  = rtrim(getenv('AZURE_OPENAI_ENDPOINT') ?: 'https://exos-openai-test.openai.azure.com', '/');
$deploy    = getenv('AZURE_OPENAI_DEPLOY')   ?: 'gpt-4o';
$apiVer    = getenv('AZURE_OPENAI_VERSION')  ?: '2024-02-01';
$batchSize = 20; // keys per API call

$targetLangs = [
    'no' => 'Norwegian Bokmål (nn is Nynorsk; this is bokmål). Legal and formal register. Preserve Norwegian legal terminology.',
    'uk' => 'Ukrainian. Formal register suitable for legal documents. Use standard Ukrainian orthography.',
    'pl' => 'Polish. Formal register suitable for legal documents.',
];

// ── Entry point ──────────────────────────────────────────────────────────────
if (PHP_SAPI !== 'cli') {
    http_response_code(403);
    exit('CLI only');
}

if ($argc < 2) {
    fwrite(STDERR, "Usage: php scripts/translate-pages.php translations/<file>.php\n");
    exit(1);
}

$filePath = realpath($argv[1]);
if (!$filePath || !is_file($filePath)) {
    fwrite(STDERR, "File not found: {$argv[1]}\n");
    exit(1);
}

echo "Loading: $filePath\n";
$allLangs = require $filePath;

if (!is_array($allLangs) || !isset($allLangs['en'])) {
    fwrite(STDERR, "File must return an array with an 'en' key.\n");
    exit(1);
}

$englishStrings = $allLangs['en'];
echo 'English keys: ' . count($englishStrings) . "\n";

// ── Translate each target language ──────────────────────────────────────────
foreach ($targetLangs as $langCode => $langDesc) {
    echo "\nTranslating → $langCode ($langDesc)...\n";
    $translated = translateAll($englishStrings, $langCode, $langDesc, $endpoint, $deploy, $apiVer, $apiKey, $batchSize);
    $allLangs[$langCode] = $translated;
    echo "  Done. Keys translated: " . count($translated) . "\n";
}

// ── Write back ───────────────────────────────────────────────────────────────
$phpCode = "<?php\nreturn " . varExportShort($allLangs) . ";\n";
file_put_contents($filePath, $phpCode);
echo "\nWritten: $filePath\n";

// ── Functions ────────────────────────────────────────────────────────────────

function translateAll(
    array $strings,
    string $langCode,
    string $langDesc,
    string $endpoint,
    string $deploy,
    string $apiVer,
    string $apiKey,
    int $batchSize
): array {
    $keys    = array_keys($strings);
    $batches = array_chunk($keys, $batchSize);
    $result  = [];

    foreach ($batches as $i => $batchKeys) {
        $batchNum = $i + 1;
        $total    = count($batches);
        echo "  Batch $batchNum/$total (" . count($batchKeys) . " keys)...\n";

        $batchStrings = array_intersect_key($strings, array_flip($batchKeys));
        $translated   = callAzureOpenAI($batchStrings, $langCode, $langDesc, $endpoint, $deploy, $apiVer, $apiKey);
        $result       = array_merge($result, $translated);
    }

    return $result;
}

function callAzureOpenAI(
    array $strings,
    string $langCode,
    string $langDesc,
    string $endpoint,
    string $deploy,
    string $apiVer,
    string $apiKey
): array {
    $systemPrompt = <<<PROMPT
You are a professional legal translator specialising in Scandinavian and Eastern European languages.

Translate the following PHP string values from English to {$langDesc}.

Rules:
- Preserve all HTML tags exactly (<strong>, <em>, <code>, <br> etc.)
- Preserve all bracketed tokens exactly: [FATHER], [PERSON], [SOCIAL WORKER], [ORG] etc.
- Preserve all arrow characters (→, ←, →) and typographic symbols
- Preserve placeholder values like "Ola Nordmann", "00:01:24" unchanged
- Keep URLs, file extensions (.docx, .php), and technical terms (PDF, DOCX, GPT-4o, Whisper, pyannote) unchanged
- Translate UI labels, descriptions, and body text naturally into {$langDesc}
- Return ONLY a valid JSON object with the same keys as input — no commentary, no markdown fences
PROMPT;

    $userContent = json_encode($strings, JSON_UNESCAPED_UNICODE | JSON_PRETTY_PRINT);

    $payload = json_encode([
        'messages' => [
            ['role' => 'system', 'content' => $systemPrompt],
            ['role' => 'user',   'content' => $userContent],
        ],
        'temperature'     => 0.2,
        'response_format' => ['type' => 'json_object'],
    ]);

    $url = "{$endpoint}/openai/deployments/{$deploy}/chat/completions?api-version={$apiVer}";

    $ch = curl_init($url);
    curl_setopt_array($ch, [
        CURLOPT_RETURNTRANSFER => true,
        CURLOPT_POST           => true,
        CURLOPT_POSTFIELDS     => $payload,
        CURLOPT_HTTPHEADER     => [
            'Content-Type: application/json',
            "api-key: {$apiKey}",
        ],
        CURLOPT_TIMEOUT        => 120,
    ]);

    $response = curl_exec($ch);
    $httpCode = curl_getinfo($ch, CURLINFO_HTTP_CODE);
    curl_close($ch);

    if ($response === false || $httpCode !== 200) {
        fwrite(STDERR, "API error HTTP $httpCode: $response\n");
        exit(1);
    }

    $data = json_decode($response, true);
    $content = $data['choices'][0]['message']['content'] ?? '';
    $translated = json_decode($content, true);

    if (!is_array($translated)) {
        fwrite(STDERR, "Failed to parse JSON from model response:\n$content\n");
        exit(1);
    }

    return $translated;
}

function varExportShort(mixed $data, int $indent = 0): string
{
    $pad  = str_repeat('  ', $indent);
    $pad1 = str_repeat('  ', $indent + 1);

    if (!is_array($data)) {
        return var_export($data, true);
    }

    $isIndexed = array_keys($data) === range(0, count($data) - 1);
    $lines     = [];

    foreach ($data as $key => $value) {
        $keyStr = $isIndexed ? '' : var_export($key, true) . ' => ';
        $lines[] = $pad1 . $keyStr . varExportShort($value, $indent + 1);
    }

    return "array (\n" . implode(",\n", $lines) . ",\n{$pad})";
}