Method _dfs

miniwob++/html_tools/html_parser.py:191–262

(node: html.HtmlElement, keep: list[str]=[], obs: list[str]=[], 
                 parent_chain: bool=False, get_new_label: bool=False, par_keep: bool=False)

            return True
        
        def _dfs(node: html.HtmlElement, keep: list[str]=[], obs: list[str]=[], 
                 parent_chain: bool=False, get_new_label: bool=False, par_keep: bool=False) -> (str, dict[str]):
            # basic information
            bid = node.attrib.get(self.id_attr, '')
            tag = node.tag
            label = node.attrib.get(self.label_attr, '')
            
            # element which is keeped equivalent to visible
            visible = is_visible(bid)
            in_keep_list = bid in keep
            in_obs_list = (bid in obs or len(label) > 0) and visible
            keep_element = in_keep_list or in_obs_list or visible or par_keep
            
            # mark label
            bids2label, labeled_elems = {}, []
            have_label = False
            if in_keep_list or in_obs_list:
                if label is None or len(label) == 0 or get_new_label:
                    label = self.identifier.generate()
                    node.attrib[self.label_attr] = label
                bids2label[bid] = label
                bids2label[label] = bid
                have_label = True
            
            # get text or alt_text of current element
            text = get_text(node.text)
            
            classes = {}
            # keep attributes if needed
            keep_all_attrs = len(self.keep_attrs) == 0
            keep_attrs = node.attrib.keys() if keep_all_attrs else self.keep_attrs
            
            # traverse attributes
            for attr in keep_attrs:
                if attr not in node.attrib or not check_attr(attr, node):
                    continue
                if attr in [self.id_attr, self.label_attr]:
                    continue
                val = get_text(node.attrib[attr])
                if len(val) > 0 or keep_all_attrs:
                    classes[attr] = val

            have_text = len(text) > 0 or len(classes) > 0
            
            parts = []
            clickable_count = 0
            children = node.getchildren()
            for child in children:
                cres, cmsg = _dfs(child, keep, obs, parent_chain, get_new_label)
                clickable_count += 1 if cmsg.get('have_clickable', False) else 0
                bids2label.update(cmsg.get('bids2label', {}))
                labeled_elems.extend(cmsg.get('label_element', []))
                if len(cres) != 0:
                    parts.append(cres)

            dom = self.prompt.subtree_constructor(parts)
            
            keep_element = keep_element and (clickable_count > 1 or have_text or have_label)