File size: 4,740 Bytes
82bf89e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 |
from typing import Any, Callable, Optional
from agents import Agent, Runner, RunResult
from agents.run_context import TContext
class ResearchAgent(Agent[TContext]):
"""
This is a custom implementation of the OpenAI Agent class that supports output parsing
for models that don't support structured output types. The user can specify an output_parser
function that will be called with the raw output from the agent. This can run custom logic
such as cleaning up the output and converting it to a structured JSON object.
Needs to be run with the ResearchRunner to work.
"""
def __init__(
self, *args, output_parser: Optional[Callable[[str], Any]] = None, **kwargs
):
# The output_parser is a function that only takes effect if output_type is not specified
self.output_parser = output_parser
# If both are specified, we raise an error - they can't be used together
if self.output_parser and kwargs.get("output_type"):
raise ValueError("Cannot specify both output_parser and output_type")
super().__init__(*args, **kwargs)
async def parse_output(self, run_result: RunResult) -> RunResult:
"""
Process the RunResult by applying the output_parser to its final_output if specified.
This preserves the RunResult structure while modifying its content.
"""
if self.output_parser:
raw_output = run_result.final_output
parsed_output = self.output_parser(raw_output)
run_result.final_output = parsed_output
return run_result
class ResearchRunner(Runner):
"""
Custom implementation of the OpenAI Runner class that supports output parsing
for models that don't support structured output types with tools.
Needs to be run with the ResearchAgent class.
"""
@classmethod
async def run(cls, *args, **kwargs) -> RunResult:
"""
Run the agent and process its output with the custom parser if applicable.
"""
# Call the original run method
result = await Runner.run(*args, **kwargs)
# Get the starting agent
starting_agent = kwargs.get("starting_agent") or args[0]
# If the starting agent is of type ResearchAgent, parse the output
if isinstance(starting_agent, ResearchAgent):
return await starting_agent.parse_output(result)
return result
class GeneralAgent(Agent[TContext]):
"""
This is a custom implementation of the OpenAI Agent class that supports output parsing
for models that don't support structured output types. The user can specify an output_parser
function that will be called with the raw output from the agent. This can run custom logic
such as cleaning up the output and converting it to a structured JSON object.
Needs to be run with the GeneralRunner to work.
"""
def __init__(
self, *args, output_parser: Optional[Callable[[str], Any]] = None, **kwargs
):
# The output_parser is a function that only takes effect if output_type is not specified
self.output_parser = output_parser
# If both are specified, we raise an error - they can't be used together
if self.output_parser and kwargs.get("output_type"):
raise ValueError("Cannot specify both output_parser and output_type")
super().__init__(*args, **kwargs)
async def parse_output(self, run_result: RunResult) -> RunResult:
"""
Process the RunResult by applying the output_parser to its final_output if specified.
This preserves the RunResult structure while modifying its content.
"""
if self.output_parser:
raw_output = run_result.final_output
parsed_output = self.output_parser(raw_output)
run_result.final_output = parsed_output
return run_result
class GeneralRunner(Runner):
"""
Custom implementation of the OpenAI Runner class that supports output parsing
for models that don't support structured output types with tools.
Needs to be run with the ResearchAgent class.
"""
@classmethod
async def run(cls, *args, **kwargs) -> RunResult:
"""
Run the agent and process its output with the custom parser if applicable.
"""
# Call the original run method
result = await Runner.run(*args, **kwargs)
# Get the starting agent
starting_agent = kwargs.get("starting_agent") or args[0]
# If the starting agent is of type ResearchAgent, parse the output
if isinstance(starting_agent, GeneralAgent):
return await starting_agent.parse_output(result)
return result
|