impplement pageing between native host and browser extension

browser_cli/native_host.py

@@ -22,7 +22,26 @@ from browser_cli.platform import DEFAULT_ALIAS, endpoint_for_alias, is_windows,
 SOCKET_PATH: str = ""          # set after hello handshake
 PENDING: dict[str, queue.Queue] = {}
 PENDING_LOCK = threading.Lock()
+WRITE_LOCK = threading.Lock()
 REGISTRY_PATH = registry_path()
+PAGE_SIZE = int(os.environ.get("BROWSER_CLI_PAGE_SIZE", "100"))
+PAGEABLE_COMMANDS = {
+    "tabs.list",
+    "tabs.filter",
+    "tabs.query",
+    "group.list",
+    "group.tabs",
+    "group.query",
+    "windows.list",
+    "dom.query",
+    "dom.text",
+    "dom.attr",
+    "extract.links",
+    "extract.images",
+    "extract.json",
+    "cookies.list",
+    "session.list",
+}
 
 # --- Native Messaging protocol (4-byte LE length prefix + UTF-8 JSON) ---
 
@@ -143,21 +162,7 @@ def handle_cli_connection(conn, listener=None) -> None:
         if "id" not in cmd:
             cmd["id"] = str(uuid.uuid4())
 
-        msg_id = cmd["id"]
-        response_queue: queue.Queue = queue.Queue()
-
-        with PENDING_LOCK:
-            PENDING[msg_id] = response_queue
-
-        write_native_message(sys.stdout.buffer, cmd)
-
-        try:
-            result = response_queue.get(timeout=30)
-        except queue.Empty:
-            result = {"id": msg_id, "success": False, "error": "timeout waiting for browser response"}
-
-        with PENDING_LOCK:
-            PENDING.pop(msg_id, None)
+        result = _handle_browser_command(cmd)
 
         response = json.dumps(result).encode("utf-8")
         if is_windows():
@@ -179,6 +184,69 @@ def handle_cli_connection(conn, listener=None) -> None:
             listener.close()
 
 
+def _handle_browser_command(cmd: dict) -> dict:
+    command = cmd.get("command")
+    if command in PAGEABLE_COMMANDS:
+        return _collect_paged_browser_command(cmd)
+    return _send_browser_command(cmd)
+
+
+def _send_browser_command(cmd: dict, timeout: int = 30) -> dict:
+    msg_id = cmd.get("id") or str(uuid.uuid4())
+    cmd["id"] = msg_id
+    response_queue: queue.Queue = queue.Queue()
+
+    with PENDING_LOCK:
+        PENDING[msg_id] = response_queue
+
+    try:
+        with WRITE_LOCK:
+            write_native_message(sys.stdout.buffer, cmd)
+
+        try:
+            return response_queue.get(timeout=timeout)
+        except queue.Empty:
+            return {"id": msg_id, "success": False, "error": "timeout waiting for browser response"}
+    finally:
+        with PENDING_LOCK:
+            PENDING.pop(msg_id, None)
+
+
+def _collect_paged_browser_command(cmd: dict) -> dict:
+    original_id = cmd.get("id") or str(uuid.uuid4())
+    offset = 0
+    items = []
+    total = None
+
+    while True:
+        page_cmd = dict(cmd)
+        page_cmd["id"] = str(uuid.uuid4())
+        page_args = dict(cmd.get("args") or {})
+        page_args["__page"] = {"offset": offset, "limit": PAGE_SIZE}
+        page_cmd["args"] = page_args
+
+        result = _send_browser_command(page_cmd)
+        result["id"] = original_id
+        if not result.get("success", True):
+            return result
+
+        data = result.get("data")
+        if not isinstance(data, dict) or data.get("__browserCliPage") is not True:
+            return result
+
+        page_items = data.get("items") or []
+        if not isinstance(page_items, list):
+            return {"id": original_id, "success": False, "error": "invalid paged response from browser"}
+        items.extend(page_items)
+        total = data.get("total", total)
+        next_offset = data.get("nextOffset")
+        if next_offset is None:
+            break
+        offset = int(next_offset)
+
+    return {"id": original_id, "success": True, "data": items, "pageSize": PAGE_SIZE, "total": total}
+
+
 # --- Socket helpers (length-prefixed framing) ---
 
 def _send_all(conn: socket.socket, data: bytes) -> None: