File size: 2,616 Bytes
ff069bf
5f26e9c
 
 
a2104ab
ff069bf
 
 
5f26e9c
ff069bf
5f26e9c
 
 
 
 
 
 
 
 
 
 
ea1580e
5f26e9c
 
ff069bf
5f26e9c
a2104ab
8a6c511
 
a2104ab
 
5f26e9c
 
 
 
 
 
 
 
 
 
 
 
 
dba77ac
5f26e9c
 
 
dba77ac
5f26e9c
 
 
 
c187bdf
5f26e9c
 
 
dba77ac
 
 
c187bdf
5f26e9c
 
 
c187bdf
dba77ac
 
 
 
5f26e9c
 
 
 
 
 
 
ff069bf
5f26e9c
ff069bf
5f26e9c
 
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
import os
import time
import random
import logging
from openai import OpenAI
from dotenv import load_dotenv
from utils import read_config

# --- Load environment & config ---
load_dotenv()
_config = read_config()["llm"]

# --- Logging setup ---
LOG_LEVEL = os.getenv("LOG_LEVEL", "INFO").upper()
logger = logging.getLogger("polLLM")
logger.setLevel(LOG_LEVEL)
handler = logging.StreamHandler()
handler.setFormatter(logging.Formatter("%(asctime)s [%(levelname)s] %(message)s"))
logger.addHandler(handler)

# --- LLM settings from config.yaml ---
_DEFAULT_MODEL     = "openai-large" #_config.get("model", "openai-large")
_SYSTEM_TEMPLATE   = _config.get("system_prompt", "")
_CHAR              = _config.get("char", "Eve")

# --- OpenAI client init ---
client = OpenAI(
    base_url    = "https://text.pollinations.ai/openai",
    api_key     = "OPENAI_API_KEY"
)

def _build_system_prompt() -> str:
    """
    Substitute {char} into the system prompt template.
    """
    return _SYSTEM_TEMPLATE.replace("{char}", _CHAR)

def generate_llm(
    prompt: str,
) -> str:
    """
    Send a chat-completion request to the LLM, with retries and backoff.
    Reads defaults from config.yaml, but can be overridden per-call.
    """
    model = _DEFAULT_MODEL
    system_prompt = _build_system_prompt()
    messages = [
        {"role": "system", "content": system_prompt},
        {"role": "user", "content": prompt},
    ]

    backoff = 1
    for attempt in range(1, 6):
        try:
            seed = random.randint(0, 2**31 - 1)
            logger.debug(f"LLM call attempt={attempt}, model={model}, seed={seed}")
            resp = client.chat.completions.create(
                model=model,
                messages=messages,
                seed=seed,
            )
            text = resp.choices[0].message.content.strip()
            logger.debug("LLM response received")
            return text
        except Exception as e:
            # Check if the error is a 400 Bad Request
            if hasattr(e, "status_code") and e.status_code == 400:
                logger.error("LLM error 400 (Bad Request): Not retrying.")
                return "Error: Bad Request (400)"
            logger.error(f"LLM error on attempt {attempt}: {e}")
            if attempt < 5:
                time.sleep(backoff)
                backoff *= 2
            else:
                logger.critical("LLM failed after 5 attempts, raising")
                raise

# Example local test
if __name__ == "__main__":
    logger.info("Testing generate_llm() with a sample prompt")
    print(generate_llm("Say hello in a poetic style."))