Skip to content

Agent Zero AI framework

9.9 KB
Raw Blame
"""Conservative native projection of already-prepared local Responses history."""

from copy import deepcopy
import hashlib
import json
from typing import Any, Callable

from helpers.llm_result import result_from_metadata


PREFIX_HASH = "history_prefix_hash"


def start_prompt(loop_data: Any) -> None:
    loop_data.params_temporary.pop("responses_prompt_replacements", None)
    loop_data.params_temporary.pop("responses_history", None)


def remember_prompt(loop_data: Any, text: str, system_message: Any, protocol: list[dict]) -> None:
    from helpers import history

    loop_data.params_temporary["responses_history"] = {
        "text": text,
        "prefix": [system_message, *history.output_langchain(protocol + loop_data.history_output)],
    }


def prepare_call(agent: Any, call_data: dict) -> dict:
    from langchain_core.prompts import ChatPromptTemplate
    from helpers import history
    from helpers.litellm_transport import ResponsesTransport, TransportMode
    from helpers.secrets import get_secrets_manager

    model, messages = call_data["model"], call_data["messages"]
    params = agent.loop_data.params_temporary
    kwargs = {"responses_prompt_replacements": call_data.get(
        "responses_prompt_replacements", params.get("responses_prompt_replacements", {}),
    )}
    prepared = params.get("responses_history", {})
    if (
        TransportMode.from_value(getattr(model, "kwargs", {}).get("a0_api_mode")) is not TransportMode.RESPONSES
        or not prepared or not hasattr(model, "_convert_messages")
        or any(message.additional_kwargs or getattr(message, "tool_calls", None) for message in messages)
        or ChatPromptTemplate.from_messages(messages).format() != prepared["text"]
    ):
        return kwargs

    def render(record):
        rendered = ResponsesTransport.input_from_model_messages(
            model, history.output_langchain([{**record, "ai": False}]),
        )
        return rendered[0].get("content") if len(rendered) == 1 else None

    mask = get_secrets_manager(agent.context).mask_values
    prompt = ResponsesTransport.input_from_model_messages(model, messages)
    prefix = ResponsesTransport.input_from_model_messages(model, prepared["prefix"])
    kwargs["responses_history_context"] = {
        "prompt": prompt, "prefix": prefix,
        "extras": mask(prompt_extras(prompt, prefix)),
        "groups": prepare_groups(agent.loop_data.history_output, render, mask),
    }
    return kwargs


def _json_bytes(value: Any) -> bytes:
    return json.dumps(value, ensure_ascii=False, sort_keys=True, separators=(",", ":")).encode()


def prefix_hashes(items: list[dict], scope: dict) -> list[str]:
    digest = hashlib.sha256(_json_bytes(scope))
    hashes = [digest.hexdigest()]
    for item in items:
        digest.update(b"\n" + _json_bytes(item))
        hashes.append(digest.hexdigest())
    return hashes


def prompt_extras(items: list[dict], prefix: list[dict]) -> str:
    """Extract only the prepared extras suffix, never a copy of prior input."""
    if len(items) == len(prefix) + 1 and items[:-1] == prefix:
        content = items[-1].get("content")
        return content if items[-1].get("role") == "user" and isinstance(content, str) else ""
    if not prefix or len(items) != len(prefix) or items[:-1] != prefix[:-1]:
        return ""
    before, after = prefix[-1], items[-1]
    if before.get("role") != "user" or {**after, "content": before.get("content")} != before:
        return ""
    old, new = before.get("content"), after.get("content")
    if isinstance(old, str) and isinstance(new, str) and new.startswith(old + "\n"):
        return new[len(old) + 1:]
    if isinstance(old, list) and isinstance(new, list) and len(new) == len(old) + 1 and new[:-1] == old:
        block = new[-1]
        if isinstance(block, dict) and block.get("type") == "input_text" and isinstance(block.get("text"), str):
            return block["text"]
    return ""


def append_extras(items: list[dict], extras: str) -> None:
    if items and items[-1].get("role") == "user":
        content = items[-1].get("content")
        if isinstance(content, str):
            items[-1]["content"] = content + "\n" + extras
            return
        if isinstance(content, list):
            content.append({"type": "input_text", "text": extras})
            return
    items.append({"role": "user", "content": extras})


def _json_object(text: Any) -> dict | None:
    if not isinstance(text, str):
        return None
    try:
        value = json.loads(text)
    except (TypeError, ValueError):
        return None
    return value if isinstance(value, dict) else None


def _contains_secret(value: Any, mask: Callable[[str], str]) -> bool:
    if isinstance(value, str):
        return mask(value) != value
    if isinstance(value, dict):
        return any(_contains_secret(item, mask) for item in value.values())
    if isinstance(value, list):
        return any(_contains_secret(item, mask) for item in value)
    return False


def _replayable_output(item: dict) -> bool:
    if item.get("type") == "function_call":
        return _json_object(item.get("arguments")) is not None
    if item.get("type") == "reasoning":
        return isinstance(item.get("encrypted_content"), str) and bool(item["encrypted_content"])
    content = item.get("content")
    return (
        item.get("type") == "message" and item.get("role") == "assistant"
        and item.get("phase") == "commentary" and item.get("status") in (None, "completed")
        and isinstance(content, list) and bool(content)
        and all(isinstance(block, dict) and block.get("type") == "output_text"
                and isinstance(block.get("text"), str) for block in content)
    )


def prepare_groups(
    records: list[dict], render: Callable[[dict], Any], mask: Callable[[str], str]
) -> list[dict]:
    groups = []
    for index, record in enumerate(records):
        if not record.get("ai"):
            continue
        result = result_from_metadata(record.get("metadata"))
        if not result or result.mode != "responses" or result.state != "local":
            continue
        prefix_hash = result.capability.get(PREFIX_HASH)
        calls = result.function_calls
        if not prefix_hash or not calls:
            continue
        output = [item.to_dict() for item in result.output_items]
        if not all(_replayable_output(item) for item in output):
            continue
        try:
            if (
                _contains_secret(output, mask) or _contains_secret([call.arguments for call in calls], mask)
                or _contains_secret(result.history_extras, mask)
            ):
                continue
        except Exception:
            continue  # Keep prepared text when secret masking cannot be checked.
        call_ids = [call.call_id for call in calls]
        if not all(call_ids) or len(set(call_ids)) != len(call_ids):
            continue
        assistant = render(record)
        if _json_object(assistant) != _json_object(result.function_calls_text()):
            continue
        results = []
        for call_id, following in zip(call_ids, records[index + 1:index + 1 + len(calls)]):
            metadata = result_from_metadata(following.get("metadata"))
            items = metadata.input_items if metadata else []
            text = render(following)
            if (
                following.get("ai") or not isinstance(text, str)
                or len(items) != 1 or not isinstance(items[0], dict)
                or items[0].get("type") != "function_call_output"
                or items[0].get("call_id") != call_id
            ):
                break
            results.append({"type": "function_call_output", "call_id": call_id, "output": text})
        paired = len(results) == len(calls)
        if paired or result.history_extras:
            groups.append({
                PREFIX_HASH: prefix_hash, "assistant": assistant,
                "extras": result.history_extras,
                "output": deepcopy(output) if paired else [], "results": results if paired else [],
            })
    return groups


def project_history(items: list[dict], groups: list[dict], scope: dict) -> list[dict]:
    indices = {value: index for index, value in enumerate(prefix_hashes(items, scope))}
    replacements = {}
    extras = {}
    used_calls, used_items = set(), set()
    for group in groups:
        index = indices.get(group.get(PREFIX_HASH))
        if index is None or index >= len(items):
            continue
        assistant = items[index]
        if assistant.get("role") != "assistant" or assistant.get("content") != group["assistant"]:
            continue
        if group.get("extras"):
            extras[index] = group["extras"]
        if not group["results"] or index + 1 >= len(items):
            continue
        following = items[index + 1]
        content = following.get("content")
        expected = "\n".join(result["output"] for result in group["results"])
        if following.get("role") != "user" or not isinstance(content, str):
            continue
        if content != expected and not content.startswith(expected + "\n"):
            continue
        call_ids = {item["call_id"] for item in group["results"]}
        item_ids = [item["id"] for item in group["output"] if item.get("id")]
        if call_ids & used_calls or set(item_ids) & used_items or len(set(item_ids)) != len(item_ids):
            continue
        used_calls.update(call_ids)
        used_items.update(item_ids)
        suffix = content[len(expected):]
        replacement = [*group["output"], *group["results"]]
        if suffix:
            replacement.append({**following, "content": suffix[1:]})
        replacements[index] = replacement
    result = []
    index = 0
    while index < len(items):
        if index in extras:
            append_extras(result, extras[index])
        if index in replacements:
            result.extend(deepcopy(replacements[index]))
            index += 2
        else:
            result.append(deepcopy(items[index]))
            index += 1
    return result