fix bug

Author: vvincent1234

src/agent/custom_agent.py

@@ -270,7 +270,8 @@ async def step(self, step_info: Optional[CustomAgentStepInfo] = None) -> None:
             self._last_result = result
             self._last_actions = actions
             if len(result) > 0 and result[-1].is_done:
-                self.extracted_content += step_info.memory
+                if not self.extracted_content:
+                    self.extracted_content = step_info.memory
                 result[-1].extracted_content = self.extracted_content
                 logger.info(f"📄 Result: {result[-1].extracted_content}")
 
@@ -346,7 +347,10 @@ async def run(self, max_steps: int = 100) -> AgentHistoryList:
                     break
             else:
                 logger.info("❌ Failed to complete task in maximum steps")
-                self.history.history[-1].result[-1].extracted_content = self.extracted_content
+                if not self.extracted_content:
+                    self.history.history[-1].result[-1].extracted_content = step_info.memory
+                else:
+                    self.history.history[-1].result[-1].extracted_content = self.extracted_content
 
             return self.history
 

src/controller/custom_controller.py

@@ -1,3 +1,5 @@
+import pdb
+
 import pyperclip
 from typing import Optional, Type
 from pydantic import BaseModel
@@ -21,10 +23,11 @@
 
 logger = logging.getLogger(__name__)
 
+
 class CustomController(Controller):
     def __init__(self, exclude_actions: list[str] = [],
-                output_model: Optional[Type[BaseModel]] = None
-                ):
+                 output_model: Optional[Type[BaseModel]] = None
+                 ):
         super().__init__(exclude_actions=exclude_actions, output_model=output_model)
         self._register_custom_actions()
 
@@ -44,20 +47,25 @@ async def paste_from_clipboard(browser: BrowserContext):
             await page.keyboard.type(text)
 
             return ActionResult(extracted_content=text)
-        
+
         @self.registry.action(
             'Extract page content to get the pure text or markdown with links if include_links is set to true',
             param_model=ExtractPageContentAction,
             requires_browser=True,
         )
         async def extract_content(params: ExtractPageContentAction, browser: BrowserContext):
             page = await browser.get_current_page()
+            # use jina reader
+            url = page.url
+            jina_url = f"https://r.jina.ai/{url}"
+            await page.goto(jina_url)
             output_format = 'markdown' if params.include_links else 'text'
             content = MainContentExtractor.extract(  # type: ignore
                 html=await page.content(),
                 output_format=output_format,
             )
-            title = await page.title()
-            msg = f'📄  Page url: {page.url}, Page title: {title}, Extracted page content as {output_format}\n: {content}\n'
+            # go back to org url
+            await page.go_back()
+            msg = f'📄  Extracted page content as {output_format}\n: {content}\n'
             logger.info(msg)
             return ActionResult(extracted_content=msg)

src/utils/utils.py

@@ -3,6 +3,7 @@
 import time
 from pathlib import Path
 from typing import Dict, Optional
+import requests
 
 from langchain_anthropic import ChatAnthropic
 from langchain_mistralai import ChatMistralAI