File size: 15,072 Bytes
a4b70d9 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 |
from __future__ import annotations
import os
import re
import json
import asyncio
import time
import datetime
from pathlib import Path
from typing import Optional, AsyncIterator, Iterator, Dict, Any, Tuple, List, Union
from ..typing import Messages
from ..providers.helper import filter_none
from ..providers.asyncio import to_async_iterator
from ..providers.response import Reasoning, FinishReason, Sources, Usage, ProviderInfo
from ..providers.types import ProviderType
from ..cookies import get_cookies_dir
from .web_search import do_search, get_search_message
from .files import read_bucket, get_bucket_dir
from .. import debug
# Constants
BUCKET_INSTRUCTIONS = """
Instruction: Make sure to add the sources of cites using [[domain]](Url) notation after the reference. Example: [[a-z0-9.]](http://example.com)
"""
TOOL_NAMES = {
"SEARCH": "search_tool",
"CONTINUE": "continue_tool",
"BUCKET": "bucket_tool"
}
class ToolHandler:
"""Handles processing of different tool types"""
@staticmethod
def validate_arguments(data: dict) -> dict:
"""Validate and parse tool arguments"""
if "arguments" in data:
if isinstance(data["arguments"], str):
data["arguments"] = json.loads(data["arguments"])
if not isinstance(data["arguments"], dict):
raise ValueError("Tool function arguments must be a dictionary or a json string")
else:
return filter_none(**data["arguments"])
else:
return {}
@staticmethod
async def process_search_tool(messages: Messages, tool: dict) -> Messages:
"""Process search tool requests"""
messages = messages.copy()
args = ToolHandler.validate_arguments(tool["function"])
messages[-1]["content"], sources = await do_search(
messages[-1]["content"],
**args
)
return messages, sources
@staticmethod
def process_continue_tool(messages: Messages, tool: dict, provider: Any) -> Tuple[Messages, Dict[str, Any]]:
"""Process continue tool requests"""
kwargs = {}
if provider not in ("OpenaiAccount", "HuggingFaceAPI"):
messages = messages.copy()
last_line = messages[-1]["content"].strip().splitlines()[-1]
content = f"Carry on from this point:\n{last_line}"
messages.append({"role": "user", "content": content})
else:
# Enable provider native continue
kwargs["action"] = "continue"
return messages, kwargs
@staticmethod
def process_bucket_tool(messages: Messages, tool: dict) -> Messages:
"""Process bucket tool requests"""
messages = messages.copy()
def on_bucket(match):
return "".join(read_bucket(get_bucket_dir(match.group(1))))
has_bucket = False
for message in messages:
if "content" in message and isinstance(message["content"], str):
new_message_content = re.sub(r'{"bucket_id":\s*"([^"]*)"}', on_bucket, message["content"])
if new_message_content != message["content"]:
has_bucket = True
message["content"] = new_message_content
last_message_content = messages[-1]["content"]
if has_bucket and isinstance(last_message_content, str):
if "\nSource: " in last_message_content:
messages[-1]["content"] = last_message_content + BUCKET_INSTRUCTIONS
return messages
@staticmethod
async def process_tools(messages: Messages, tool_calls: List[dict], provider: Any) -> Tuple[Messages, Dict[str, Any]]:
"""Process all tool calls and return updated messages and kwargs"""
if not tool_calls:
return messages, {}
extra_kwargs = {}
messages = messages.copy()
sources = None
for tool in tool_calls:
if tool.get("type") != "function":
continue
function_name = tool.get("function", {}).get("name")
debug.log(f"Processing tool call: {function_name}")
if function_name == TOOL_NAMES["SEARCH"]:
messages, sources = await ToolHandler.process_search_tool(messages, tool)
elif function_name == TOOL_NAMES["CONTINUE"]:
messages, kwargs = ToolHandler.process_continue_tool(messages, tool, provider)
extra_kwargs.update(kwargs)
elif function_name == TOOL_NAMES["BUCKET"]:
messages = ToolHandler.process_bucket_tool(messages, tool)
return messages, sources, extra_kwargs
class AuthManager:
"""Handles API key management"""
aliases = {
"GeminiPro": "Gemini",
"PollinationsAI": "Pollinations",
"OpenaiAPI": "Openai",
"PuterJS": "Puter",
}
@classmethod
def load_api_key(cls, provider: ProviderType) -> Optional[str]:
"""Load API key from config file"""
if not provider.needs_auth and not hasattr(provider, "login_url"):
return None
provider_name = provider.get_parent()
env_var = f"{provider_name.upper()}_API_KEY"
api_key = os.environ.get(env_var)
if not api_key and provider_name in cls.aliases:
env_var = f"{cls.aliases[provider_name].upper()}_API_KEY"
api_key = os.environ.get(env_var)
if api_key:
debug.log(f"Loading API key for {provider_name} from environment variable {env_var}")
return api_key
return None
class ThinkingProcessor:
"""Processes thinking chunks"""
@staticmethod
def process_thinking_chunk(chunk: str, start_time: float = 0) -> Tuple[float, List[Union[str, Reasoning]]]:
"""Process a thinking chunk and return timing and results."""
results = []
# Handle non-thinking chunk
if not start_time and "<think>" not in chunk and "</think>" not in chunk:
return 0, [chunk]
# Handle thinking start
if "<think>" in chunk and "`<think>`" not in chunk:
before_think, *after = chunk.split("<think>", 1)
if before_think:
results.append(before_think)
results.append(Reasoning(status="🤔 Is thinking...", is_thinking="<think>"))
if after:
if "</think>" in after[0]:
after, *after_end = after[0].split("</think>", 1)
results.append(Reasoning(after))
results.append(Reasoning(status="", is_thinking="</think>"))
if after_end:
results.append(after_end[0])
return 0, results
else:
results.append(Reasoning(after[0]))
return time.time(), results
# Handle thinking end
if "</think>" in chunk:
before_end, *after = chunk.split("</think>", 1)
if before_end:
results.append(Reasoning(before_end))
thinking_duration = time.time() - start_time if start_time > 0 else 0
status = f"Thought for {thinking_duration:.2f}s" if thinking_duration > 1 else ""
results.append(Reasoning(status=status, is_thinking="</think>"))
# Make sure to handle text after the closing tag
if after and after[0].strip():
results.append(after[0])
return 0, results
# Handle ongoing thinking
if start_time:
return start_time, [Reasoning(chunk)]
return start_time, [chunk]
async def perform_web_search(messages: Messages, web_search_param: Any) -> Tuple[Messages, Optional[Sources]]:
"""Perform web search and return updated messages and sources"""
messages = messages.copy()
sources = None
if not web_search_param:
return messages, sources
try:
search_query = web_search_param if isinstance(web_search_param, str) and web_search_param != "true" else None
messages[-1]["content"], sources = await do_search(messages[-1]["content"], search_query)
except Exception as e:
debug.error(f"Couldn't do web search:", e)
return messages, sources
async def async_iter_run_tools(
provider: ProviderType,
model: str,
messages: Messages,
tool_calls: Optional[List[dict]] = None,
**kwargs
) -> AsyncIterator:
"""Asynchronously run tools and yield results"""
# Process web search
sources = None
web_search = kwargs.get('web_search')
if web_search:
debug.log(f"Performing web search with value: {web_search}")
messages, sources = await perform_web_search(messages, web_search)
# Get API key
if not kwargs.get("api_key"):
api_key = AuthManager.load_api_key(provider)
if api_key:
kwargs["api_key"] = api_key
# Process tool calls
if tool_calls:
messages, sources, extra_kwargs = await ToolHandler.process_tools(messages, tool_calls, provider)
kwargs.update(extra_kwargs)
# Generate response
response = to_async_iterator(provider.async_create_function(model=model, messages=messages, **kwargs))
try:
model_info = model
async for chunk in response:
if isinstance(chunk, ProviderInfo):
model_info = getattr(chunk, 'model', model_info)
elif isinstance(chunk, Usage):
usage = {"user": kwargs.get("user"), "model": model_info, "provider": provider.get_parent(), **chunk.get_dict()}
usage_dir = Path(get_cookies_dir()) / ".usage"
usage_file = usage_dir / f"{datetime.date.today()}.jsonl"
usage_dir.mkdir(parents=True, exist_ok=True)
with usage_file.open("a" if usage_file.exists() else "w") as f:
f.write(f"{json.dumps(usage)}\n")
yield chunk
provider.live += 1
except:
provider.live -= 1
raise
# Yield sources if available
if sources:
yield sources
def iter_run_tools(
provider: ProviderType,
model: str,
messages: Messages,
tool_calls: Optional[List[dict]] = None,
**kwargs
) -> Iterator:
"""Run tools synchronously and yield results"""
# Process web search
web_search = kwargs.get('web_search')
sources = None
if web_search:
debug.log(f"Performing web search with value: {web_search}")
try:
messages = messages.copy()
search_query = web_search if isinstance(web_search, str) and web_search != "true" else None
# Note: Using asyncio.run inside sync function is not ideal, but maintaining original pattern
messages[-1]["content"], sources = asyncio.run(do_search(messages[-1]["content"], search_query))
except Exception as e:
debug.error(f"Couldn't do web search:", e)
# Get API key if needed
if not kwargs.get("api_key"):
api_key = AuthManager.load_api_key(provider)
if api_key:
kwargs["api_key"] = api_key
# Process tool calls
if tool_calls:
for tool in tool_calls:
if tool.get("type") == "function":
function_name = tool.get("function", {}).get("name")
debug.log(f"Processing tool call: {function_name}")
if function_name == TOOL_NAMES["SEARCH"]:
tool["function"]["arguments"] = ToolHandler.validate_arguments(tool["function"])
messages[-1]["content"] = get_search_message(
messages[-1]["content"],
raise_search_exceptions=True,
**tool["function"]["arguments"]
)
elif function_name == TOOL_NAMES["CONTINUE"]:
if provider.__name__ not in ("OpenaiAccount", "HuggingFace"):
last_line = messages[-1]["content"].strip().splitlines()[-1]
content = f"Carry on from this point:\n{last_line}"
messages.append({"role": "user", "content": content})
else:
# Enable provider native continue
kwargs["action"] = "continue"
elif function_name == TOOL_NAMES["BUCKET"]:
def on_bucket(match):
return "".join(read_bucket(get_bucket_dir(match.group(1))))
has_bucket = False
for message in messages:
if "content" in message and isinstance(message["content"], str):
new_message_content = re.sub(r'{"bucket_id":"([^"]*)"}', on_bucket, message["content"])
if new_message_content != message["content"]:
has_bucket = True
message["content"] = new_message_content
last_message = messages[-1]["content"]
if has_bucket and isinstance(last_message, str):
if "\nSource: " in last_message:
messages[-1]["content"] = last_message + BUCKET_INSTRUCTIONS
# Process response chunks
thinking_start_time = 0
processor = ThinkingProcessor()
model_info = model
try:
for chunk in provider.create_function(model=model, messages=messages, provider=provider, **kwargs):
if isinstance(chunk, FinishReason):
if sources is not None:
yield sources
sources = None
yield chunk
continue
elif isinstance(chunk, Sources):
sources = None
elif isinstance(chunk, ProviderInfo):
model_info = getattr(chunk, 'model', model_info)
elif isinstance(chunk, Usage):
usage = {"user": kwargs.get("user"), "model": model_info, "provider": provider.get_parent(), **chunk.get_dict()}
usage_dir = Path(get_cookies_dir()) / ".usage"
usage_file = usage_dir / f"{datetime.date.today()}.jsonl"
usage_dir.mkdir(parents=True, exist_ok=True)
with usage_file.open("a" if usage_file.exists() else "w") as f:
f.write(f"{json.dumps(usage)}\n")
if not isinstance(chunk, str):
yield chunk
continue
thinking_start_time, results = processor.process_thinking_chunk(chunk, thinking_start_time)
for result in results:
yield result
provider.live += 1
except:
provider.live -= 1
raise
if sources is not None:
yield sources
|