def current_viewport_html()

in evals/elsuite/multistep_web_tasks/webarena/browser_env/processors.py [0:0]

• 52 lines of code
• 14 McCabe index (conditional complexity)

    def current_viewport_html(self, info: BrowserState) -> str:
        # adopted from [natbot](https://github.com/nat/natbot)
        tree = info["DOMTree"]
        strings = tree["strings"]
        document = tree["documents"][0]
        nodes = document["nodes"]
        attributes = nodes["attributes"]
        node_value = nodes["nodeValue"]
        parent = nodes["parentIndex"]
        node_names = nodes["nodeName"]

        layout = document["layout"]
        layout_node_cursor = layout["nodeIndex"]
        union_bounds = layout["unionBounds"]

        graph = defaultdict(lambda: [])
        for node_idx in range(len(node_names)):
            parent_idx = parent[node_idx]
            if parent_idx != -1:
                graph[parent_idx].append(node_idx)

        def dfs(idx: int) -> str:
            node_name = strings[node_names[idx]].lower().strip()
            can_skip = "#" in node_name or "::" in node_name

            inner_text = ""
            node_value_idx = node_value[idx]
            if node_value_idx >= 0 and node_value_idx < len(strings):
                inner_text = " ".join(strings[node_value_idx].split())
            node_attributes = [strings[i] for i in attributes[idx]]
            node_attributes_str = ""
            for i in range(0, len(node_attributes), 2):
                a = node_attributes[i]
                b = node_attributes[i + 1]
                b = " ".join(b.split())
                node_attributes_str += f'{a}="{b}" '
            node_attributes_str = node_attributes_str.strip()

            html = ""
            if not can_skip:
                html += f"<{node_name}"
                if {node_attributes_str}:
                    html += f" {node_attributes_str}"
                html += f">{inner_text}"
            else:
                html += f"{inner_text}"

            for child_idx in graph[idx]:
                if child_idx in layout_node_cursor:
                    cursor = layout_node_cursor.index(child_idx)
                    union_bound = union_bounds[cursor]
                    if not self.partially_in_viewport(union_bound, info["config"]):
                        continue
                    html += dfs(child_idx)

            if not can_skip:
                html += f"</{node_name}>"

            return html

        html = dfs(0)
        return html