(self)
| 280 | return dom, cmsg |
| 281 | |
| 282 | def parse_tree(self) -> dict[str]: |
| 283 | # start from here |
| 284 | stt = time.time() |
| 285 | root = self.get_root(self.dom_tree) |
| 286 | dom, cmsg = self.parse(root, self.keep, self.obs, self.parent_chain, self.get_new_label) |
| 287 | self.bids2label = cmsg.get('bids2label', {}) |
| 288 | self.keep = list(set(self.keep + cmsg.get('label_element', []))) |
| 289 | |
| 290 | obj = { |
| 291 | 'html': dom, |
| 292 | 'parse_time': time.time() - stt |
| 293 | } |
| 294 | |
| 295 | return obj |
| 296 | |
| 297 | # From mind2web, https://github.com/OSU-NLP-Group/Mind2Web/blob/main/src/data_utils/dom_utils.py |
| 298 | def get_keep_elements(self, tree: html.HtmlElement, keep: list[str], max_depth: int, max_children: int, |
no test coverage detected