Files
letta-server/letta/llm_api/helpers.py
2025-12-15 12:02:19 -08:00

397 lines
16 KiB
Python

import copy
import json
import logging
from collections import OrderedDict
from typing import Any, List, Optional, Union
import requests
from letta.constants import OPENAI_CONTEXT_WINDOW_ERROR_SUBSTRING
from letta.helpers.json_helpers import json_dumps
from letta.log import get_logger
from letta.schemas.message import Message
from letta.schemas.openai.chat_completion_response import ChatCompletionResponse, Choice
from letta.schemas.response_format import (
JsonObjectResponseFormat,
JsonSchemaResponseFormat,
ResponseFormatType,
ResponseFormatUnion,
TextResponseFormat,
)
from letta.settings import summarizer_settings
from letta.utils import printd
logger = get_logger(__name__)
def _convert_to_structured_output_helper(property: dict) -> dict:
"""Convert a single JSON schema property to structured output format (recursive)"""
# Handle anyOf structures
if "anyOf" in property and "type" not in property:
# Check if this is a simple anyOf that can be flattened to type array
types = []
has_complex = False
for option in property["anyOf"]:
if "type" in option:
opt_type = option["type"]
if opt_type in ["object", "array"]:
has_complex = True
break
types.append(opt_type)
elif "$ref" in option:
# Has unresolved $ref, treat as complex
has_complex = True
break
# If it's simple primitives only (string, null, integer, boolean, etc), flatten to type array
if not has_complex and types:
param_description = property.get("description")
property_dict = {"type": types}
if param_description is not None:
property_dict["description"] = param_description
if "default" in property:
property_dict["default"] = property["default"]
# Preserve other fields like enum, format, etc
for key in ["enum", "format", "pattern", "minLength", "maxLength", "minimum", "maximum"]:
if key in property:
property_dict[key] = property[key]
return property_dict
# Otherwise, preserve anyOf and recursively process each option
property_dict = {"anyOf": [_convert_to_structured_output_helper(opt) for opt in property["anyOf"]]}
if "description" in property:
property_dict["description"] = property["description"]
if "default" in property:
property_dict["default"] = property["default"]
if "title" in property:
property_dict["title"] = property["title"]
return property_dict
if "type" not in property:
raise ValueError(f"Property {property} is missing a type and doesn't have anyOf")
param_type = property["type"]
param_description = property.get("description")
# Handle type arrays (e.g., ["string", "null"])
if isinstance(param_type, list):
property_dict = {"type": param_type}
if param_description is not None:
property_dict["description"] = param_description
if "default" in property:
property_dict["default"] = property["default"]
# Preserve other fields
for key in ["enum", "format", "pattern", "minLength", "maxLength", "minimum", "maximum", "title"]:
if key in property:
property_dict[key] = property[key]
return property_dict
if param_type == "object":
if "properties" not in property:
raise ValueError(f"Property {property} of type object is missing properties")
properties = property["properties"]
property_dict = {
"type": "object",
"properties": {k: _convert_to_structured_output_helper(v) for k, v in properties.items()},
"additionalProperties": False,
"required": list(properties.keys()),
}
if param_description is not None:
property_dict["description"] = param_description
if "title" in property:
property_dict["title"] = property["title"]
return property_dict
elif param_type == "array":
if "items" not in property:
raise ValueError(f"Property {property} of type array is missing items")
items = property["items"]
property_dict = {
"type": "array",
"items": _convert_to_structured_output_helper(items),
}
if param_description is not None:
property_dict["description"] = param_description
if "title" in property:
property_dict["title"] = property["title"]
return property_dict
else:
property_dict = {
"type": param_type, # simple type
}
if param_description is not None:
property_dict["description"] = param_description
# Preserve other fields
for key in ["enum", "format", "pattern", "minLength", "maxLength", "minimum", "maximum", "default", "title"]:
if key in property:
property_dict[key] = property[key]
return property_dict
def convert_to_structured_output(openai_function: dict, allow_optional: bool = False) -> dict:
"""Convert function call objects to structured output objects.
See: https://platform.openai.com/docs/guides/structured-outputs/supported-schemas
Supports:
- Simple type arrays: type: ["string", "null"]
- anyOf with primitives (flattened to type array)
- anyOf with complex objects (preserved as anyOf)
- Nested structures with recursion
For OpenAI strict mode, optional fields (not in required) must have explicit default values.
"""
description = openai_function.get("description", "")
structured_output = {
"name": openai_function["name"],
"description": description,
"strict": True,
"parameters": {
"type": "object",
"properties": {},
"additionalProperties": False,
"required": [],
},
}
for param, details in openai_function["parameters"]["properties"].items():
# Use the helper for all parameter types - it now handles anyOf, type arrays, objects, arrays, etc.
structured_output["parameters"]["properties"][param] = _convert_to_structured_output_helper(details)
# Determine which fields are required
# For OpenAI strict mode, ALL fields must be in the required array
# This is a requirement for strict: true schemas
if not allow_optional:
# All fields are required for strict mode
structured_output["parameters"]["required"] = list(structured_output["parameters"]["properties"].keys())
else:
# Use the input's required list if provided, otherwise empty
structured_output["parameters"]["required"] = openai_function["parameters"].get("required", [])
return structured_output
def convert_response_format_to_responses_api(
response_format: Optional["ResponseFormatUnion"],
) -> Optional[dict]:
"""
Convert Letta's ResponseFormatUnion to OpenAI Responses API text.format structure.
The Responses API uses a different structure than Chat Completions:
text={
"format": {
"type": "json_schema",
"name": "...",
"strict": True,
"schema": {...}
}
}
Args:
response_format: Letta ResponseFormatUnion object
Returns:
Dict with format structure for Responses API, or None
"""
if response_format is None:
return None
# Text format - return None since it's the default
if isinstance(response_format, TextResponseFormat):
return None
# JSON object format - not directly supported in Responses API
# Users should use json_schema instead
elif isinstance(response_format, JsonObjectResponseFormat):
logger.warning(
"json_object response format is not supported in Responses API. "
"Use json_schema with a proper schema instead. Skipping response_format."
)
return None
# JSON schema format - this is what Responses API supports
elif isinstance(response_format, JsonSchemaResponseFormat):
json_schema_dict = response_format.json_schema
# Ensure required fields are present
if "schema" not in json_schema_dict:
logger.warning("json_schema missing 'schema' field, skipping response_format")
return None
return {
"type": "json_schema",
"name": json_schema_dict.get("name", "response_schema"),
"schema": json_schema_dict["schema"],
"strict": json_schema_dict.get("strict", True), # Default to strict mode
}
def make_post_request(url: str, headers: dict[str, str], data: dict[str, Any]) -> dict[str, Any]:
printd(f"Sending request to {url}")
try:
# Make the POST request
response = requests.post(url, headers=headers, json=data)
printd(f"Response status code: {response.status_code}")
# Raise for 4XX/5XX HTTP errors
response.raise_for_status()
# Check if the response content type indicates JSON and attempt to parse it
content_type = response.headers.get("Content-Type", "")
if "application/json" in content_type.lower():
try:
response_data = response.json() # Attempt to parse the response as JSON
printd(f"Response JSON: {response_data}")
except ValueError as json_err:
# Handle the case where the content type says JSON but the body is invalid
error_message = f"Failed to parse JSON despite Content-Type being {content_type}: {json_err}"
printd(error_message)
raise ValueError(error_message) from json_err
else:
error_message = f"Unexpected content type returned: {response.headers.get('Content-Type')}"
printd(error_message)
raise ValueError(error_message)
# Process the response using the callback function
return response_data
except requests.exceptions.HTTPError as http_err:
# HTTP errors (4XX, 5XX)
error_message = f"HTTP error occurred: {http_err}"
if http_err.response is not None:
error_message += f" | Status code: {http_err.response.status_code}, Message: {http_err.response.text}"
printd(error_message)
raise requests.exceptions.HTTPError(error_message) from http_err
except requests.exceptions.Timeout as timeout_err:
# Handle timeout errors
error_message = f"Request timed out: {timeout_err}"
printd(error_message)
raise requests.exceptions.Timeout(error_message) from timeout_err
except requests.exceptions.RequestException as req_err:
# Non-HTTP errors (e.g., connection, SSL errors)
error_message = f"Request failed: {req_err}"
printd(error_message)
raise requests.exceptions.RequestException(error_message) from req_err
except ValueError as val_err:
# Handle content-type or non-JSON response issues
error_message = f"ValueError: {val_err}"
printd(error_message)
raise ValueError(error_message) from val_err
except Exception as e:
# Catch any other unknown exceptions
error_message = f"An unexpected error occurred: {e}"
printd(error_message)
raise Exception(error_message) from e
# TODO update to use better types
def add_inner_thoughts_to_functions(
functions: List[dict],
inner_thoughts_key: str,
inner_thoughts_description: str,
inner_thoughts_required: bool = True,
put_inner_thoughts_first: bool = True,
) -> List[dict]:
"""Add an inner_thoughts kwarg to every function in the provided list, ensuring it's the first parameter"""
new_functions = []
for function_object in functions:
new_function_object = copy.deepcopy(function_object)
new_properties = OrderedDict()
# For chat completions, we want inner thoughts to come later
if put_inner_thoughts_first:
# Create with inner_thoughts as the first item
new_properties[inner_thoughts_key] = {
"type": "string",
"description": inner_thoughts_description,
}
# Add the rest of the properties
new_properties.update(function_object["parameters"]["properties"])
else:
new_properties.update(function_object["parameters"]["properties"])
new_properties[inner_thoughts_key] = {
"type": "string",
"description": inner_thoughts_description,
}
# Cast OrderedDict back to a regular dict
new_function_object["parameters"]["properties"] = dict(new_properties)
# Update required parameters if necessary
if inner_thoughts_required:
required_params = new_function_object["parameters"].get("required", [])
if inner_thoughts_key not in required_params:
if put_inner_thoughts_first:
required_params.insert(0, inner_thoughts_key)
else:
required_params.append(inner_thoughts_key)
new_function_object["parameters"]["required"] = required_params
new_functions.append(new_function_object)
return new_functions
def unpack_all_inner_thoughts_from_kwargs(
response: ChatCompletionResponse,
inner_thoughts_key: str,
) -> ChatCompletionResponse:
"""Strip the inner thoughts out of the tool call and put it in the message content"""
if len(response.choices) == 0:
raise ValueError("Unpacking inner thoughts from empty response not supported")
new_choices = []
for choice in response.choices:
new_choices.append(unpack_inner_thoughts_from_kwargs(choice, inner_thoughts_key))
# return an updated copy
new_response = response.model_copy(deep=True)
new_response.choices = new_choices
return new_response
def unpack_inner_thoughts_from_kwargs(choice: Choice, inner_thoughts_key: str) -> Choice:
message = choice.message
rewritten_choice = choice # inner thoughts unpacked out of the function
if message.role == "assistant" and message.tool_calls and len(message.tool_calls) >= 1:
if len(message.tool_calls) > 1:
logger.warning(f"Unpacking inner thoughts from more than one tool call ({len(message.tool_calls)}) is not supported")
# TODO support multiple tool calls
tool_call = message.tool_calls[0]
try:
# Sadly we need to parse the JSON since args are in string format
func_args = dict(json.loads(tool_call.function.arguments))
if inner_thoughts_key in func_args:
# extract the inner thoughts
inner_thoughts = func_args.pop(inner_thoughts_key)
# replace the kwargs
new_choice = choice.model_copy(deep=True)
new_choice.message.tool_calls[0].function.arguments = json_dumps(func_args)
# also replace the message content
if new_choice.message.content is not None:
logger.warning(f"Overwriting existing inner monologue ({new_choice.message.content}) with kwarg ({inner_thoughts})")
new_choice.message.content = inner_thoughts
# update the choice object
rewritten_choice = new_choice
else:
logger.warning(f"Did not find inner thoughts in tool call: {str(tool_call)}")
except json.JSONDecodeError as e:
logger.warning(f"Failed to strip inner thoughts from kwargs: {e}")
logger.error(f"Failed to strip inner thoughts from kwargs: {e}, Tool call arguments: {tool_call.function.arguments}")
raise e
else:
logger.warning(f"Did not find tool call in message: {str(message)}")
return rewritten_choice