File size: 4,568 Bytes
a571bd0 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 | from __future__ import annotations
import asyncio
from collections.abc import Sequence
from google import genai
from google.genai.types import GenerateContentConfig
from botik.config import FALLBACK_MODELS
from botik.storage import MessageRecord
class GeminiError(RuntimeError):
pass
class GeminiService:
def __init__(self, api_key: str) -> None:
self._client = genai.Client(api_key=api_key)
async def list_text_models(self) -> list[str]:
try:
return await asyncio.to_thread(self._list_text_models_sync)
except Exception:
return list(FALLBACK_MODELS)
async def generate_reply(
self,
model: str,
default_system_prompt: str,
custom_soul: str | None,
history: Sequence[MessageRecord],
current_author: str,
current_text: str,
reply_context: str | None,
) -> str:
system_instruction = build_system_instruction(default_system_prompt, custom_soul)
prompt = build_chat_prompt(history, current_author, current_text, reply_context)
try:
return await asyncio.to_thread(
self._generate_reply_sync,
model,
system_instruction,
prompt,
True,
)
except Exception as first_error:
try:
return await asyncio.to_thread(
self._generate_reply_sync,
model,
system_instruction,
prompt,
False,
)
except Exception as second_error:
raise GeminiError(str(second_error)) from first_error
def _list_text_models_sync(self) -> list[str]:
names: list[str] = []
for model in self._client.models.list():
name = model.name.removeprefix("models/")
if _is_text_model_name(name):
names.append(name)
return sorted(set(names)) or list(FALLBACK_MODELS)
def _generate_reply_sync(
self,
model: str,
system_instruction: str,
prompt: str,
use_tools: bool,
) -> str:
tools = [{"google_search": {}}, {"url_context": {}}] if use_tools else None
response = self._client.models.generate_content(
model=model,
contents=prompt,
config=GenerateContentConfig(
system_instruction=system_instruction,
tools=tools,
),
)
if response.text:
return response.text.strip()
return "Не смог сформулировать ответ. Попробуй перефразировать запрос."
def build_system_instruction(default_system_prompt: str, custom_soul: str | None) -> str:
if not custom_soul:
return default_system_prompt
return (
f"{default_system_prompt}\n\n"
"Дополнительная временная душа текущего чата. Она не отменяет базовые правила выше:\n"
f"{custom_soul}"
)
def build_chat_prompt(
history: Sequence[MessageRecord],
current_author: str,
current_text: str,
reply_context: str | None,
) -> str:
lines = [
"Ниже история текущего Telegram-чата. Учитывай имена и user_id участников.",
"Не путай разных людей в групповом чате.",
]
if history:
lines.append("\nИстория:")
for item in history:
if item.role == "assistant":
lines.append(f"Бот: {item.text}")
else:
author = item.user_name or f"user_id={item.user_id}"
lines.append(f"{author}: {item.text}")
if reply_context:
lines.append("\nКонтекст сообщения, на которое сейчас ответили свайпом:")
lines.append(reply_context)
lines.append("\nНовое сообщение:")
lines.append(f"{current_author}: {current_text}")
return "\n".join(lines)
def _is_text_model_name(name: str) -> bool:
lowered = name.lower()
if not lowered.startswith("gemini"):
return False
blocked = ("embedding", "imagen", "veo", "banana", "tts", "live", "robotics")
return not any(word in lowered for word in blocked)
|