core.section_processor

class celi_framework.core.section_processor.SectionProcessor(current_section, system_message, initial_user_message, tool_descriptions, tool_implementations, primary_model_name, llm_cache, monitor_instructions, max_tokens, force_tool_every_n, seed=777, callback=None, model_url=None, token_counter=None)

Bases: object

This is the core processor that uses iterative calls to LLMs to manage the drafting process. Each SectionProcessor runs independently, working through a task list described in the JobDescription and using the tools it provides.

async add_human_input(input)
async builtin_review(edit_system_message=True)
callback: Optional[CELIUpdateCallback] = None
static check_for_duplicates(ongoing_chat)
current_section: str
force_tool_every_n: int
static format_chat_messages(msgs)
static format_message_content(m)
initial_user_message: str
llm_cache: bool
async make_tool_calls(response)
max_tokens: int
model_url: Optional[str] = None
monitor_instructions: str
primary_model_name: str
async process_iteration()

Run one iteration of the main execution loop for drafting documents.

Within each iteration of the loop, the method performs the following operations:

  • Prepares a prompt for the language model based on the current conversation history.

  • Processes the language model’s response, including handling any specified function calls, updating

    the draft content, and managing the ongoing conversation context.

  • Checks if the iteration is in a loop.

  • If the section is complete, call builtin_review to check the resul and decide whether a redo is required.

Returns:

True if the drafting is still in progress and another iteration should be run. False if the process should stop.

static responses_since_last_tool_call(ongoing_chat)
async run()
seed: int = 777
system_message: str
token_counter: Optional[TokenCounter] = None
tool_descriptions: List[ToolDescription]
tool_implementations: ToolImplementations
async celi_framework.core.section_processor.ask_split(user_prompt, system_message, model_name, max_tokens=0, seed=777, verbose=False, max_retries=7, wait_between_retries=2, temperature=0.0, timeout=120, tool_descriptions=None, model_url=None, json_mode=False, response_format=None, token_counter=None, force_tool_use=False)

Sends a prompt to the OpenAI API and returns the response, with retries on error.

user_prompt can be either a string or a list of messages.