Method process

webarena/browser_env/processors.py:702–797

(self, page: Page, client: CDPSession, context: str)

        return "\n".join(clean_lines)

    def process(self, page: Page, client: CDPSession, context: str) -> str:
        # get the tab info
        open_tabs = page.context.pages
        # try:
        #     tab_titles = [tab.title() for tab in open_tabs]
        #     current_tab_idx = open_tabs.index(page)
        #     for idx in range(len(open_tabs)):
        #         if idx == current_tab_idx:
        #             tab_titles[
        #                 idx
        #             ] = f"Tab {idx} (current): {open_tabs[idx].title()}"
        #         else:
        #             tab_titles[idx] = f"Tab {idx}: {open_tabs[idx].title()}"
        #     tab_title_str = " | ".join(tab_titles)
        # except Exception:
        #     tab_title_str = " | ".join(
        #         ["Tab {idx}" for idx in range(len(open_tabs))]
        #     )

        try:
            tab_titles = [tab.title() for tab in open_tabs]
            current_tab_idx = open_tabs.index(page)
            for idx in range(len(open_tabs)):
                if idx == current_tab_idx:
                    tab_titles[
                        idx
                    ] = f"{idx+1}. {open_tabs[idx].title()} <-- current tab"
                else:
                    tab_titles[idx] = f"{idx+1}. {open_tabs[idx].title()}"
            tab_title_str = "\n".join(tab_titles)
        except Exception:
            tab_title_str = "\n".join(
                [f"{idx+1}. Default" for idx in range(len(open_tabs))]
            )

        
        try:
            browser_info = self.fetch_browser_info(page, client)
        except Exception:
            page.wait_for_load_state("load", timeout=500)
            browser_info = self.fetch_browser_info(page, client)
        
        if self.observation_type == "html":
            import time
            stt = time.time()
            dom_tree = self.fetch_page_html(
                browser_info,
                page,
                client,
                current_viewport_only=self.current_viewport_only,
            )
            
            print('[fetch]', time.time() - stt)
            
            stt = time.time()
            raw_html, content, obs_nodes_info, hp = self.parse_my_html(dom_tree)
            print('[parse]', time.time() - stt)