80 lines
3.5 KiB
Python
80 lines
3.5 KiB
Python
import requests
|
|
import json
|
|
import logging
|
|
from typing import Dict, Optional, List
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
class LLMClient:
|
|
def __init__(self, endpoint: str = 'http://localhost:11434', model: str = 'llama3', use_local: bool = True, lm_studio_host: str = None):
|
|
self.endpoint = endpoint
|
|
self.model = model
|
|
self.use_local = use_local
|
|
|
|
self.lm_studio_endpoints = {
|
|
'plato': {'url': 'http://192.168.1.74:1234', 'model': 'openai/gpt-oss-20b'},
|
|
'postgres': {'url': 'http://192.168.1.159:1234', 'model': 'mistralai/devstral-small-2507'},
|
|
'local': {'url': 'http://localhost:11434', 'model': 'llama3'}
|
|
}
|
|
|
|
self.lm_studio_host = lm_studio_host or 'postgres'
|
|
studio_config = self.lm_studio_endpoints.get(self.lm_studio_host, self.lm_studio_endpoints['postgres'])
|
|
self.lm_studio_endpoint = studio_config['url']
|
|
self.lm_studio_model = studio_config['model']
|
|
|
|
def summarize(self, text: str, max_length: int = 200) -> Dict:
|
|
prompt = f"Summarize this concisely in under {max_length} characters:\n\n{text[:3000]}"
|
|
return self._query(prompt)
|
|
|
|
def extract_topics(self, text: str) -> Dict:
|
|
prompt = f"Extract 5-10 key topics/tags. Return ONLY comma-separated words:\n\n{text[:3000]}"
|
|
result = self._query(prompt)
|
|
if result.get('success'):
|
|
topics = [t.strip() for t in result['text'].split(',')]
|
|
result['topics'] = topics[:10]
|
|
return result
|
|
|
|
def extract_intent(self, text: str) -> Dict:
|
|
prompt = f"What is the main purpose/intent of this code/document? Answer in 1-2 sentences:\n\n{text[:3000]}"
|
|
return self._query(prompt)
|
|
|
|
def detect_project_type(self, text: str, file_list: List[str]) -> Dict:
|
|
files_str = ', '.join(file_list[:20])
|
|
prompt = f"Based on these files: {files_str}\nAnd this content:\n{text[:2000]}\n\nWhat type of project is this? (e.g. web app, ml/ai, transcription, data processing, etc.)"
|
|
return self._query(prompt)
|
|
|
|
def _query(self, prompt: str, timeout: int = 30) -> Dict:
|
|
try:
|
|
if self.use_local:
|
|
response = requests.post(
|
|
f'{self.endpoint}/api/generate',
|
|
json={'model': self.model, 'prompt': prompt, 'stream': False},
|
|
timeout=timeout
|
|
)
|
|
if response.status_code == 200:
|
|
data = response.json()
|
|
return {'success': True, 'text': data.get('response', '').strip()}
|
|
else:
|
|
response = requests.post(
|
|
f'{self.lm_studio_endpoint}/v1/chat/completions',
|
|
json={
|
|
'model': self.lm_studio_model,
|
|
'messages': [{'role': 'user', 'content': prompt}],
|
|
'max_tokens': 500,
|
|
'temperature': 0.7
|
|
},
|
|
timeout=timeout
|
|
)
|
|
if response.status_code == 200:
|
|
data = response.json()
|
|
return {'success': True, 'text': data['choices'][0]['message']['content'].strip()}
|
|
|
|
return {'success': False, 'error': f'HTTP {response.status_code}'}
|
|
|
|
except requests.Timeout:
|
|
logger.warning(f'LLM request timeout after {timeout}s')
|
|
return {'success': False, 'error': 'timeout'}
|
|
except Exception as e:
|
|
logger.error(f'LLM query failed: {e}')
|
|
return {'success': False, 'error': str(e)}
|