add docx support

file_cleaning_ui.py

@@ -53,6 +53,18 @@ def extract_and_clean_txt(txt_path: str) -> str:
     ]
     return "\n".join(cleaned)
 
+def extract_and_clean_docx(docx_path: str) -> str:
+    """Lit un fichier type Word et le nettoie."""
+    print(f"[+] Lecture du fichier docx : {docx_path}")
+    with open(docx_path, "r", encoding="utf-8") as f:
+        lines = f.readlines()
+    cleaned = [
+        clean_text_for_rag(line.strip())
+        for line in lines
+        if line.strip()
+    ]
+    return "\n".join(cleaned)
+
 def process_file(input_file: gr.File, output_name: str) -> str:
     """
     - Detecte le type (PDF ou texte)