v3

lang/ar.php

@@ -22,7 +22,7 @@ $lang = [
 
     // Manual Email
     "option_1" => "الخيار 1: رفع ملف بريد إلكتروني",
-    "choose_email_file" => "اختر ملف البريد الإلكتروني (.eml أو .txt)",
+    "choose_email_file" => "اختر ملفًا (.eml, .txt, .pdf, .doc, .docx)",
     "analyze_and_populate" => "تحليل الملف وتعبئة الحقول",
     "option_2" => "الخيار 2: إدخال يدوي",
     "from" => "من",
@@ -43,6 +43,10 @@ $lang = [
     "language" => "اللغة",
     "arabic" => "العربية",
     "english" => "English",
-    "french" => "Français"
+    "french" => "Français",
+
+    // أخطاء تحميل الملفات
+    "unsupported_file_type" => "نوع الملف غير مدعوم. يرجى رفع ملف من نوع .eml, .txt, .pdf, .doc, or .docx.",
+    "error_parsing_file" => "تعذر استخراج النص من الملف. قد يكون فارغًا أو تالفًا أو محميًا بكلمة مرور."
 ];
 ?>

lang/en.php

@@ -22,7 +22,7 @@ $lang = [
 
     // Manual Email
     "option_1" => "Option 1: Upload Email File",
-    "choose_email_file" => "Choose an email file (.eml or .txt)",
+    "choose_email_file" => "Choose a file (.eml, .txt, .pdf, .doc, .docx)",
     "analyze_and_populate" => "Analyze and Populate Fields",
     "option_2" => "Option 2: Manual Entry",
     "from" => "From",
@@ -43,6 +43,10 @@ $lang = [
     "language" => "Language",
     "arabic" => "العربية",
     "english" => "English",
-    "french" => "Français"
+    "french" => "Français",
+
+    // File Upload Errors
+    "unsupported_file_type" => "Unsupported file type. Please upload a .eml, .txt, .pdf, .doc, or .docx file.",
+    "error_parsing_file" => "Could not extract text from the file. It might be empty, corrupted, or password-protected."
 ];
 ?>

lang/fr.php

@@ -22,7 +22,7 @@ $lang = [
 
     // Manual Email
     "option_1" => "Option 1 : Télécharger un fichier e-mail",
-    "choose_email_file" => "Choisissez un fichier e-mail (.eml ou .txt)",
+    "choose_email_file" => "Choisissez un fichier (.eml, .txt, .pdf, .doc, .docx)",
     "analyze_and_populate" => "Analyser et remplir les champs",
     "option_2" => "Option 2 : Saisie manuelle",
     "from" => "De",
@@ -43,6 +43,10 @@ $lang = [
     "language" => "Langue",
     "arabic" => "العربية",
     "english" => "English",
-    "french" => "Français"
+    "french" => "Français",
+
+    // Erreurs de téléchargement de fichiers
+    "unsupported_file_type" => "Type de fichier non pris en charge. Veuillez télécharger un fichier .eml, .txt, .pdf, .doc ou .docx.",
+    "error_parsing_file" => "Impossible d'extraire le texte du fichier. Il est peut-être vide, corrompu ou protégé par un mot de passe."
 ];
 ?>

manual_email.php

@@ -11,32 +11,62 @@ $upload_error = '';
 // Handle File Upload and Parsing
 if ($_SERVER['REQUEST_METHOD'] === 'POST' && isset($_FILES['email_file']) && $_FILES['email_file']['error'] == UPLOAD_ERR_OK) {
     $file_tmp_path = $_FILES['email_file']['tmp_name'];
-    $file_content = file_get_contents($file_tmp_path);
+    $file_name = $_FILES['email_file']['name'];
+    $file_ext = strtolower(pathinfo($file_name, PATHINFO_EXTENSION));
 
-    if ($file_content) {
+    $message = ''; // Initialize message
-        // Basic EML parsing
+    $subject = pathinfo($file_name, PATHINFO_FILENAME); // Use filename as subject by default
-        $headers = [];
+    $from = ''; // From is not available in these files
-        $body_started = false;
-        $body_lines = [];
-        $lines = explode("\n", $file_content);
 
-        foreach ($lines as $line) {
+    // IMPORTANT: Ensure the temp file path is properly escaped to prevent command injection
-            if (!$body_started) {
+    $escaped_path = escapeshellarg($file_tmp_path);
-                if (preg_match('/^From:\s*(.*)/i', $line, $matches)) {
+
-                    $from = trim($matches[1]);
+    switch ($file_ext) {
-                } elseif (preg_match('/^Subject:\s*(.*)/i', $line, $matches)) {
+        case 'pdf':
-                    $subject = trim($matches[1]);
+            $message = shell_exec("pdftotext " . $escaped_path . " -");
-                } elseif (trim($line) === '') {
+            break;
-                    $body_started = true; // Headers are done, body begins
+        case 'docx':
+            $message = shell_exec("docx2txt < " . $escaped_path);
+            break;
+        case 'doc':
+            $message = shell_exec("antiword " . $escaped_path);
+            break;
+        case 'eml':
+        case 'txt':
+            $file_content = file_get_contents($file_tmp_path);
+            if ($file_content) {
+                // Basic EML parsing
+                $body_started = false;
+                $body_lines = [];
+                $lines = explode("\n", $file_content);
+
+                foreach ($lines as $line) {
+                    if (!$body_started) {
+                        if (preg_match('/^From:\s*(.*)/i', $line, $matches)) {
+                            $from = trim($matches[1]);
+                        } elseif (preg_match('/^Subject:\s*(.*)/i', $line, $matches)) {
+                            $subject = trim($matches[1]);
+                        } elseif (trim($line) === '') {
+                            $body_started = true; // Headers are done, body begins
+                        }
+                    } else {
+                        $body_lines[] = $line;
+                    }
                 }
+                $message = implode("\n", $body_lines);
             } else {
-                $body_lines[] = $line;
+                $upload_error = 'Failed to read uploaded file.';
             }
-        }
+            break;
-        $message = implode("\n", $body_lines);
+        default:
-    } else {
+            $upload_error = 'Unsupported file type. Please upload a .eml, .txt, .pdf, .doc, or .docx file.';
-        $upload_error = 'Failed to read uploaded file.';
+            break;
     }
+
+    if (empty($message) && empty($upload_error)) {
+        $upload_error = 'Could not extract text from the file. It might be empty, corrupted, or password-protected.';
+    }
+
 } elseif ($_SERVER['REQUEST_METHOD'] === 'POST') {
     // Handle Manual Form Submission
     $from = isset($_POST['from']) ? $_POST['from'] : '';
@@ -129,7 +159,7 @@ try {
                 <form action="manual_email.php" method="POST" enctype="multipart/form-data">
                     <div class="mb-3">
                         <label for="email_file" class="form-label"><?php echo t('choose_email_file'); ?></label>
-                        <input type="file" class="form-control" id="email_file" name="email_file" accept=".eml,.txt">
+                        <input type="file" class="form-control" id="email_file" name="email_file" accept=".eml,.txt,.pdf,.doc,.docx">
                     </div>
                     <button type="submit" class="btn btn-secondary"><?php echo t('analyze_and_populate'); ?></button>
                 </form>