Meilleure détection de word et pdf

app.py

@@ -2698,36 +2698,54 @@ else:
         st.markdown("---")
         st.markdown("**Ajouter un document (max 50 Mo) depuis un lien public**")
 
-        from urllib.parse import urlparse, parse_qs
-
-        def make_direct_link(file_url: str) -> str:
+        def make_direct_link(file_url: str, prefer_format: str = "docx") -> str:
             """
-            Convertit les liens Google Drive / Dropbox / OneDrive en liens directs téléchargeables.
+            Convertit les liens Google Docs / Google Drive / Dropbox / OneDrive en liens directs téléchargeables.
+            - prefer_format: "docx" ou "pdf" pour Google Docs.
             Retourne le lien original si aucun cas particulier n'est détecté.
             """
-            # --- Google Drive ---
-            if "drive.google.com" in file_url:
-                if "/d/" in file_url:
-                    file_id = file_url.split("/d/")[1].split("/")[0]
-                elif "id=" in file_url:
-                    file_id = parse_qs(urlparse(file_url).query).get("id", [None])[0]
+            u = file_url.strip()
+
+            # --- Google Docs (document, pas un fichier Drive) ---
+            if "docs.google.com/document" in u:
+                m = re.search(r"/document/d/([^/]+)/", u)
+                doc_id = m.group(1) if m else None
+                if doc_id:
+                    fmt = "docx" if prefer_format == "docx" else "pdf"
+                    return f"https://docs.google.com/document/d/{doc_id}/export?format={fmt}"
+                return u
+
+            # --- Google Drive (fichiers) ---
+            if "drive.google.com" in u:
+                file_id = None
+                if "/d/" in u:
+                    file_id = u.split("/d/")[1].split("/")[0]
                 else:
-                    file_id = None
+                    qs_id = parse_qs(urlparse(u).query).get("id", [None])[0]
+                    file_id = qs_id or file_id
                 if file_id:
                     return f"https://drive.google.com/uc?export=download&id={file_id}"
+                return u
 
             # --- Dropbox ---
-            if "dropbox.com" in file_url:
-                return file_url.replace("?dl=0", "?dl=1")
+            if "dropbox.com" in u:
+                parsed = urlparse(u)
+                # Force le download binaire
+                if "dl=" in parsed.query:
+                    return re.sub(r"dl=\d", "dl=1", u)
+                # sinon ajouter dl=1
+                sep = "&" if parsed.query else "?"
+                return u + f"{sep}dl=1"
 
             # --- OneDrive ---
-            if "1drv.ms" in file_url or "onedrive.live.com" in file_url:
-                if "redir?" in file_url:
-                    return file_url.replace("redir?", "download?")
-                elif "download" not in file_url:
-                    return file_url + "?download=1"
-
-            return file_url
+            if "1drv.ms" in u or "onedrive.live.com" in u:
+                if "download=" not in u:
+                    parsed = urlparse(u)
+                    sep = "&" if parsed.query else "?"
+                    return u + f"{sep}download=1"
+                return u
+
+            return u
 
         file_url = st.text_input(
             "🔗 Collez ici le lien public vers votre document (Google Drive, Dropbox, OneDrive, etc.) :",