msft-trns/process.php

<?php
// Enable error reporting for debugging
ini_set('display_errors', 1);
ini_set('display_startup_errors', 1);
error_reporting(E_ALL);

include 'config.php';
include 'logger.php';
include 'azure_storage.php';


// Create log file if it doesn't exist
$logFile = 'app.log';
if (!file_exists($logFile)) {
    file_put_contents($logFile, "Log started at " . date('Y-m-d H:i:s') . PHP_EOL);
    chmod($logFile, 0666); // Make writable by web server
}

header('Content-Type: application/json');

logMessage("Received request to process.php");

if ($_SERVER['REQUEST_METHOD'] === 'POST') {
    try {
        logMessage("Received POST request");
        
        if (!isset($_FILES['file'])) {
            logMessage("No file received", 'ERROR');
            echo json_encode(['error' => 'No file received']);
            exit;
        }
        
        $file = $_FILES['file'];
        logMessage("File received: " . $file['name']);
        
        // Check for upload errors
        if ($file['error'] !== UPLOAD_ERR_OK) {
            $uploadErrors = [
                1 => 'The uploaded file exceeds the upload_max_filesize directive in php.ini',
                2 => 'The uploaded file exceeds the MAX_FILE_SIZE directive that was specified in the HTML form',
                3 => 'The uploaded file was only partially uploaded',
                4 => 'No file was uploaded',
                6 => 'Missing a temporary folder',
                7 => 'Failed to write file to disk',
                8 => 'A PHP extension stopped the file upload'
            ];
            $errorMessage = isset($uploadErrors[$file['error']]) ? $uploadErrors[$file['error']] : 'Unknown upload error';
            logMessage("File upload error: " . $errorMessage, 'ERROR');
            echo json_encode(['error' => 'File upload error: ' . $errorMessage]);
            exit;
        }
        
        // Check if the file exists and is readable
        if (!file_exists($file['tmp_name']) || !is_readable($file['tmp_name'])) {
            logMessage("Temporary file does not exist or is not readable: " . $file['tmp_name'], 'ERROR');
            echo json_encode(['error' => 'Temporary file does not exist or is not readable']);
            exit;
        }
        
        $source_lang = $_POST['source_lang'] ?? '';
        $target_lang = $_POST['target_lang'] ?? '';
        $formality = $_POST['formality'] ?? 'default';
        
        logMessage("Source Language: $source_lang");
        logMessage("Target Language: $target_lang");
        logMessage("Formality: $formality");
        
        // Initialize Azure Storage Helper
        logMessage("Initializing Azure Storage Helper");
        $azureStorage = new AzureStorageHelper();
        
        // Ensure containers exist
        logMessage("Ensuring Azure Storage containers exist");
        if (!$azureStorage->ensureContainersExist()) {
            logMessage("Failed to verify Azure Storage containers", 'ERROR');
            echo json_encode(['error' => 'Failed to verify Azure Storage containers']);
            exit;
        }
        
        // Upload file to Azure Blob Storage
        logMessage("Uploading file to Azure Blob Storage: " . $file['name']);
        $sourceBlob = $azureStorage->uploadSourceFile($file['tmp_name'], $file['name']);
        if (!$sourceBlob) {
            logMessage("Failed to upload file to Azure Storage", 'ERROR');
            echo json_encode(['error' => 'Failed to upload file to Azure Storage']);
            exit;
        }
        
        // Set up target blob for the translated file
        logMessage("Setting up target blob for translated file");
        $targetBlob = $azureStorage->getTargetBlob($file['name'], $target_lang);
        if (!$targetBlob) {
            logMessage("Failed to set up target blob", 'ERROR');
            echo json_encode(['error' => 'Failed to set up target blob']);
            exit;
        }
        
        // Prepare the request to Microsoft Document Translation API
        $batchTranslationUrl = MS_API_ENDPOINT . '/translator/document/batches?api-version=' . MS_API_VERSION;
        logMessage("Prepared translation API URL: $batchTranslationUrl");
        
        // Format the source and target language codes correctly
        $formattedSourceLang = $source_lang !== 'AUTO' ? strtolower($source_lang) : '';
        $formattedTargetLang = strtolower($target_lang);
        
        // Handle special cases for Microsoft's language codes
        if ($formattedTargetLang === 'en-gb') {
            $formattedTargetLang = 'en-GB';
        } else if ($formattedTargetLang === 'en-us') {
            $formattedTargetLang = 'en-US';
        } else if ($formattedTargetLang === 'pt-pt') {
            $formattedTargetLang = 'pt-PT';
        } else if ($formattedTargetLang === 'pt-br') {
            $formattedTargetLang = 'pt-BR';
        }
        
        // For Microsoft Document Translation, we need container-level URLs
        // The sourceUrl should be the container URL with SAS token
        // Extract account name from connection string
        preg_match('/AccountName=([^;]+)/', AZURE_STORAGE_CONNECTION_STRING, $nameMatches);
        $accountName = !empty($nameMatches[1]) ? $nameMatches[1] : 'opticaltranslations';
        
        // Use the container URLs directly with SAS tokens
        $sasToken = defined('AZURE_STORAGE_SAS_TOKEN') ? AZURE_STORAGE_SAS_TOKEN : $sourceBlob['sas_token'];
        $sourceUrl = $sourceBlob['blob_url'] . '?' . $sasToken;
        $targetUrl = $targetBlob['blob_url'] . '?' . $sasToken;
        
        logMessage("Source Container URL with SAS (partial): " . substr($sourceUrl, 0, 100) . "...");
        logMessage("Target Container URL with SAS (partial): " . substr($targetUrl, 0, 100) . "...");
        
        // Let's try a different approach with direct blob URLs
        // Extract account name from connection string
        preg_match('/AccountName=([^;]+)/', AZURE_STORAGE_CONNECTION_STRING, $nameMatches);
        $accountName = !empty($nameMatches[1]) ? $nameMatches[1] : 'opticaltranslations';
        $sasToken = defined('AZURE_STORAGE_SAS_TOKEN') ? AZURE_STORAGE_SAS_TOKEN : '';
        
        // Format the blob name to be URL-safe (replace spaces with %20)
        $safeBlobName = str_replace(' ', '%20', $sourceBlob['blob_name']);
        
        // Direct blob URLs with SAS token - make sure the URL is properly formed
        $sourceDirectUrl = "https://$accountName.blob.core.windows.net/".AZURE_STORAGE_CONTAINER_SOURCE."/$safeBlobName?$sasToken";
        
        // Target folder URL - make sure to include the container name
        $targetDirectUrl = "https://$accountName.blob.core.windows.net/".AZURE_STORAGE_CONTAINER_TARGET."?$sasToken";
        
        // Use container-level URLs with filter parameter, similar to the test script
        // Instead of pointing directly to the blob, point to the container and specify the blob with filter
        $sourceContainerUrl = "https://$accountName.blob.core.windows.net/".AZURE_STORAGE_CONTAINER_SOURCE."?$sasToken";
        $targetContainerUrl = "https://$accountName.blob.core.windows.net/".AZURE_STORAGE_CONTAINER_TARGET."?$sasToken";
        
        // Modified request body for Document Translation API
        $requestBody = [
            'inputs' => [
                [
                    'source' => [
                        'sourceUrl' => $sourceContainerUrl,
                        'storageSource' => 'AzureBlob',
                        'language' => $formattedSourceLang ?: 'en', // Default to English if auto-detect
                        'filter' => [
                            'prefix' => $sourceBlob['blob_name'],
                            'includeSubfolders' => false
                        ]
                    ],
                    'targets' => [
                        [
                            'targetUrl' => $targetContainerUrl,
                            'storageSource' => 'AzureBlob',
                            'language' => $formattedTargetLang,
                            'category' => 'general'
                        ]
                    ]
                ]
            ]
        ];
        
        // Log the full request body for debugging
        logMessage("Full request body: " . json_encode($requestBody, JSON_PRETTY_PRINT));
        
        // Add source language if specified (not AUTO)
        if (!empty($formattedSourceLang)) {
            $requestBody['inputs'][0]['source']['language'] = $formattedSourceLang;
        }
        
        // Set formality if supported by Microsoft Translator
        if ($formality === 'more' || $formality === 'less') {
            $requestBody['inputs'][0]['targets'][0]['formality'] = ($formality === 'more') ? 'Formal' : 'Informal';
        }
        
        logMessage("Prepared request body: " . json_encode($requestBody));
        
        $ch = curl_init();
        curl_setopt($ch, CURLOPT_URL, $batchTranslationUrl);
        curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
        curl_setopt($ch, CURLOPT_POST, 1);
        curl_setopt($ch, CURLOPT_HTTPHEADER, [
            'Ocp-Apim-Subscription-Key: ' . MS_API_KEY,
            'Ocp-Apim-Subscription-Region: ' . MS_API_REGION,
            'Content-Type: application/json'
        ]);
        
        $jsonBody = json_encode($requestBody);
        curl_setopt($ch, CURLOPT_POSTFIELDS, $jsonBody);
        
        logMessage("Sending request to Microsoft Translator API");
        $response = curl_exec($ch);
        $httpCode = curl_getinfo($ch, CURLINFO_HTTP_CODE);
        $curlError = curl_error($ch);
        
        logMessage("CURL Info: " . json_encode(curl_getinfo($ch)));
        
        logApiCall($batchTranslationUrl, 'POST', [
            'Ocp-Apim-Subscription-Key: ' . MS_API_KEY,
            'Ocp-Apim-Subscription-Region: ' . MS_API_REGION,
            'Content-Type: application/json'
        ], $requestBody, $response, $httpCode);
        
        if ($response === false) {
            $error = curl_error($ch);
            logMessage("cURL Error: " . $error, 'ERROR');
            echo json_encode(['error' => 'cURL error: ' . $error]);
        } else {
            logMessage("API Response Code: $httpCode");
            logMessage("API Response: $response");
            
            if ($httpCode >= 400) {
                logMessage("HTTP Error Response: $httpCode - $response", 'ERROR');
                echo json_encode(['error' => "HTTP Error: $httpCode", 'details' => $response]);
            } else {
                $result = json_decode($response, true);
                
                if (json_last_error() === JSON_ERROR_NONE) {
                    // Store the blob information for later use
                    if (isset($result['id'])) {
                        // Create a custom response that matches what our frontend expects
                        $translationResponse = [
                            'document_id' => $result['id'], // Microsoft's batch ID
                            'document_key' => json_encode([
                                'source_blob' => $sourceBlob['blob_name'],
                                'target_blob' => $targetBlob['blob_name'],
                                'ms_response' => $result
                            ])
                        ];
                        echo json_encode($translationResponse);
                        logMessage("Translation job submitted successfully with ID: " . $result['id']);
                    } else {
                        logMessage("Microsoft Translator API error: " . json_encode($result), 'ERROR');
                        echo json_encode(['error' => 'Microsoft Translator API response missing ID', 'details' => $result]);
                    }
                } else {
                    logMessage("Invalid JSON response: " . json_last_error_msg(), 'ERROR');
                    echo json_encode(['error' => 'Invalid JSON response: ' . json_last_error_msg(), 'raw_response' => $response]);
                }
            }
        }
        
        curl_close($ch);
    
    } catch (Exception $e) {
        logMessage("Exception: " . $e->getMessage() . "\nStack trace: " . $e->getTraceAsString(), 'ERROR');
        echo json_encode(['error' => 'Server error: ' . $e->getMessage()]);
    }
} else {
    logMessage("Received non-POST request", 'ERROR');
    echo json_encode(['error' => 'Invalid request method']);
}