Method find_curl_from_content

integuru/agent.py:292–403

Find the cURL command that contains the dynamic parts

(self, state: AgentState)

        return simplest_curl

    def find_curl_from_content(self, state: AgentState) -> AgentState:
        """
        Find the cURL command that contains the dynamic parts
        """
        search_string_list = state[self.IN_PROCESS_NODE_DYNAMIC_PARTS_KEY]
        search_string_list_leftovers = search_string_list.copy()

        in_process_node_id = state[self.IN_PROCESS_NODE_KEY]
        new_to_be_processed_nodes = []

        # Handle cookies
        for search_string in search_string_list_leftovers[:]:
            cookie_key = self.find_key_by_string_in_value(
                self.cookie_dict, search_string
            )
            if cookie_key:
                search_string_list_leftovers.remove(search_string)
                if cookie_key in self.cookie_to_id_dict:
                    cookie_node_id = self.cookie_to_id_dict[cookie_key]
                else:
                    cookie_node_id = self.dag_manager.add_node(
                        node_type="cookie",  # Specify node type
                        content={
                            "key": cookie_key,
                            "value": search_string
                        }, 
                        extracted_parts=[search_string]
                    )
                    self.cookie_to_id_dict[cookie_key] = cookie_node_id
                    #dont need to add node to to_be_processed_nodes because cookies dont need further processing
                self.dag_manager.add_edge(in_process_node_id, cookie_node_id)

        # Handle curls
        if search_string_list_leftovers:
            for search_string in search_string_list_leftovers[:]:
                requests_with_search_string = []

                for request, response in self.req_to_res_map.items():
                    curl = str(request)
                    if (
                        (
                            isinstance(curl, str)
                            and search_string.lower() in response["text"].lower()
                        )
                        and (search_string.lower() not in curl.lower())
                    ) or (
                        urllib.parse.unquote(search_string) in curl
                        and (urllib.parse.unquote(search_string) not in curl)
                    ):
                        requests_with_search_string.append(request)
                simplest_request = ""

                # Get simplest curl to reduce number of dependencies
                if len(requests_with_search_string) > 1:
                    simplest_request = self.get_simplest_request(requests_with_search_string)
                elif len(requests_with_search_string) == 1:
                    simplest_request = requests_with_search_string[0]
                else: