Add browser_tool example to fetch webpages

requirements.txt

@@ -45,3 +45,4 @@ typing_extensions==4.8.0
 urllib3==1.26.18
 wikipedia==1.4.0
 yarl==1.9.2
+html2text==2024.2.26

tool_use_package/browser_example.py

@@ -0,0 +1,35 @@
+# This contains the code for browser tool that opens a single webpage at a url
+import html2text
+import requests
+
+from .tools.base_tool import BaseTool
+from .tool_user import ToolUser
+
+class BrowserTool(BaseTool):
+  """Retrieves a webpage and formats to text for Claude"""
+
+  def use_tool(self, url):
+    try:
+      response = requests.get(url)
+      html = response.text
+      h = html2text.HTML2Text()
+      text = h.handle(html)
+      return text
+    except requests.exceptions.RequestException:
+      return f"There was an error fetching {url}"
+
+tool_name = "get_webpage"
+tool_description = """The get_webpage tool will return the text of a webpage."""
+tool_parameters = [
+    {"name": "url", "type": "str", "description": "The URL of the webpage to get."}
+]
+
+browser_tool = BrowserTool(tool_name, tool_description, tool_parameters)
+
+# Pass the tool instance into the ToolUser
+tool_user = ToolUser([browser_tool])
+
+# Call the tool_user with a prompt to get a version of Claude that can use your tools!
+if __name__ == '__main__':
+    messages = [{"role":"user", "content":f"Summarize http://docs.python.org"}]
+    print(tool_user.use_tools(messages, execution_mode="automatic"))