Parse the html tree into a string text
(dom_tree: DOMTree)
| 439 | |
| 440 | @staticmethod |
| 441 | def parse_html(dom_tree: DOMTree) -> tuple[str, dict[str, Any]]: |
| 442 | """Parse the html tree into a string text""" |
| 443 | |
| 444 | obs_nodes_info = {} |
| 445 | nodeid_to_cursor = { |
| 446 | node["nodeId"]: idx for idx, node in enumerate(dom_tree) |
| 447 | } |
| 448 | |
| 449 | def dfs(node_cursor: int, depth: int) -> str: |
| 450 | tree_str = "" |
| 451 | node = dom_tree[node_cursor] |
| 452 | indent = "\t" * depth |
| 453 | valid_node = True |
| 454 | try: |
| 455 | node_str = f"[{node_cursor}] <{node['nodeName']}" |
| 456 | if node["attributes"]: |
| 457 | node_str += f" {node['attributes']}" |
| 458 | node_str += f"> {node['nodeValue']}" |
| 459 | valid_node = bool(node["attributes"] or node["nodeValue"]) |
| 460 | |
| 461 | if valid_node: |
| 462 | obs_nodes_info[str(node_cursor)] = { |
| 463 | "backend_id": node["backendNodeId"], |
| 464 | "union_bound": node["union_bound"], |
| 465 | "text": node_str, |
| 466 | } |
| 467 | tree_str += f"{indent}{node_str}\n" |
| 468 | |
| 469 | except Exception as e: |
| 470 | valid_node = False |
| 471 | |
| 472 | for child_ids in node["childIds"]: |
| 473 | child_cursor = nodeid_to_cursor[child_ids] |
| 474 | child_depth = depth + 1 if valid_node else depth |
| 475 | child_str = dfs(child_cursor, child_depth) |
| 476 | tree_str += child_str |
| 477 | |
| 478 | return tree_str |
| 479 | |
| 480 | html = dfs(0, 0) |
| 481 | return html, obs_nodes_info |
| 482 | |
| 483 | def fetch_page_accessibility_tree( |
| 484 | self, |
nothing calls this directly
no outgoing calls
no test coverage detected