configuration et al: implemented new Config format

This commit is contained in:
juk0de 2023-09-06 22:52:03 +02:00
parent 74a26b8c2f
commit 8dbe8b5f25
4 changed files with 105 additions and 31 deletions

View File

@ -33,18 +33,23 @@ class AI(Protocol):
The base class for AI clients.
"""
ID: str
name: str
config: AIConfig
def request(self,
question: Message,
context: Chat,
chat: Chat,
num_answers: int = 1,
otags: Optional[set[Tag]] = None) -> AIResponse:
"""
Make an AI request, asking the given question with the given
context (i. e. chat history). The nr. of requested answers
corresponds to the nr. of messages in the 'AIResponse'.
Make an AI request. Parameters:
* question: the question to ask
* chat: the chat history to be added as context
* num_answers: nr. of requested answers (corresponds
to the nr. of messages in the 'AIResponse')
* otags: the output tags, i. e. the tags that all
returned messages should contain
"""
raise NotImplementedError

View File

@ -3,7 +3,8 @@ Creates different AI instances, based on the given configuration.
"""
import argparse
from .configuration import Config
from typing import cast
from .configuration import Config, OpenAIConfig, default_ai_ID
from .ai import AI, AIError
from .ais.openai import OpenAI
@ -12,9 +13,14 @@ def create_ai(args: argparse.Namespace, config: Config) -> AI:
"""
Creates an AI subclass instance from the given args and configuration.
"""
if args.ai == 'openai':
# FIXME: create actual 'OpenAIConfig' and set values from 'args'
# FIXME: use actual name from config
return OpenAI("openai", config.openai)
if args.ai:
ai_conf = config.ais[args.ai]
elif default_ai_ID in config.ais:
ai_conf = config.ais[default_ai_ID]
else:
raise AIError("No AI name given and no default exists")
if ai_conf.name == 'openai':
return OpenAI(cast(OpenAIConfig, ai_conf))
else:
raise AIError(f"AI '{args.ai}' is not supported")

View File

@ -17,9 +17,11 @@ class OpenAI(AI):
The OpenAI AI client.
"""
def __init__(self, name: str, config: OpenAIConfig) -> None:
self.name = name
def __init__(self, config: OpenAIConfig) -> None:
self.ID = config.ID
self.name = config.name
self.config = config
openai.api_key = config.api_key
def request(self,
question: Message,
@ -31,8 +33,7 @@ class OpenAI(AI):
chat history. The nr. of requested answers corresponds to the
nr. of messages in the 'AIResponse'.
"""
# FIXME: use real 'system' message (store in OpenAIConfig)
oai_chat = self.openai_chat(chat, "system", question)
oai_chat = self.openai_chat(chat, self.config.system, question)
response = openai.ChatCompletion.create(
model=self.config.model,
messages=oai_chat,

View File

@ -1,16 +1,28 @@
import yaml
from typing import Type, TypeVar, Any
from dataclasses import dataclass, asdict
from pathlib import Path
from typing import Type, TypeVar, Any, Optional
from dataclasses import dataclass, asdict, field
ConfigInst = TypeVar('ConfigInst', bound='Config')
AIConfigInst = TypeVar('AIConfigInst', bound='AIConfig')
OpenAIConfigInst = TypeVar('OpenAIConfigInst', bound='OpenAIConfig')
supported_ais: list[str] = ['openai']
default_ai_ID: str = 'default'
default_config_path = '.config.yaml'
class ConfigError(Exception):
pass
@dataclass
class AIConfig:
"""
The base class of all AI configurations.
"""
ID: str
name: str
@ -19,13 +31,18 @@ class OpenAIConfig(AIConfig):
"""
The OpenAI section of the configuration file.
"""
api_key: str
model: str
temperature: float
max_tokens: int
top_p: float
frequency_penalty: float
presence_penalty: float
# all members have default values, so we can easily create
# a default configuration
ID: str = 'default'
name: str = 'openai'
api_key: str = '0123456789'
system: str = 'You are an assistant'
model: str = 'gpt-3.5-turbo-16k'
temperature: float = 1.0
max_tokens: int = 4000
top_p: float = 1.0
frequency_penalty: float = 0.0
presence_penalty: float = 0.0
@classmethod
def from_dict(cls: Type[OpenAIConfigInst], source: dict[str, Any]) -> OpenAIConfigInst:
@ -33,7 +50,9 @@ class OpenAIConfig(AIConfig):
Create OpenAIConfig from a dict.
"""
return cls(
name='OpenAI',
ID='openai',
name='openai',
system=str(source['system']),
api_key=str(source['api_key']),
model=str(source['model']),
max_tokens=int(source['max_tokens']),
@ -43,36 +62,79 @@ class OpenAIConfig(AIConfig):
presence_penalty=float(source['presence_penalty'])
)
def as_dict(self) -> dict[str, Any]:
return asdict(self)
def ai_config_instance(name: str, conf_dict: Optional[dict[str, Any]] = None) -> AIConfig:
"""
Creates an AIConfig instance of the given name.
"""
if name.lower() == 'openai':
if conf_dict is None:
return OpenAIConfig()
else:
return OpenAIConfig.from_dict(conf_dict)
else:
raise ConfigError(f"AI '{name}' is not supported")
def create_default_ai_configs() -> dict[str, AIConfig]:
"""
Create a dict containing default configurations for all supported AIs.
"""
return {ai_config_instance(name).ID: ai_config_instance(name) for name in supported_ais}
@dataclass
class Config:
"""
The configuration file structure.
"""
system: str
db: str
openai: OpenAIConfig
# all members have default values, so we can easily create
# a default configuration
db: str = './db/'
ais: dict[str, AIConfig] = field(default_factory=create_default_ai_configs)
@classmethod
def from_dict(cls: Type[ConfigInst], source: dict[str, Any]) -> ConfigInst:
"""
Create Config from a dict.
"""
# create the correct AI type instances
ais: dict[str, AIConfig] = {}
for ID, conf in source['ais'].items():
ai_conf = ai_config_instance(conf['name'], conf)
ais[ID] = ai_conf
return cls(
system=str(source['system']),
db=str(source['db']),
openai=OpenAIConfig.from_dict(source['openai'])
ais=ais
)
@classmethod
def create_default(self, file_path: Path) -> None:
"""
Creates a default Config in the given file.
"""
conf = Config()
conf.to_file(file_path)
@classmethod
def from_file(cls: Type[ConfigInst], path: str) -> ConfigInst:
with open(path, 'r') as f:
source = yaml.load(f, Loader=yaml.FullLoader)
# add the AI ID to the config (for easy internal access)
for ID, conf in source['ais'].items():
conf['ID'] = ID
return cls.from_dict(source)
def to_file(self, path: str) -> None:
with open(path, 'w') as f:
yaml.dump(asdict(self), f, sort_keys=False)
def to_file(self, file_path: Path) -> None:
# remove the AI name from the config (for a cleaner format)
data = self.as_dict()
for conf in data['ais'].values():
del (conf['ID'])
with open(file_path, 'w') as f:
yaml.dump(data, f, sort_keys=False)
def as_dict(self) -> dict[str, Any]:
return asdict(self)